fix: share thread signal stacks with macOS

Comparison textures can carry valid HTILE descriptors before any depth-target binding. Keep encoding checks independent of cached image metadata, matching SDK texture classification and shadPS4 depth promotion. Preserve HTILE address and control validation, and replace cache-history test fixtures with captured first-use coverage.

.github/workflows/build.yml

@@ -208,34 +208,66 @@ jobs:
         run: |
           archive="$RUNNER_TEMP/MoltenVK-macos.tar"
           package="$RUNNER_TEMP/MoltenVK"
+          install_dir="_Build/macos/install"
+          app_dir="$install_dir/KytyPS5.app"
+          moltenvk="$install_dir/libMoltenVK.dylib"
           curl --fail --location --retry 3 \
             --output "$archive" \
             "https://github.com/KhronosGroup/MoltenVK/releases/download/$MOLTENVK_VERSION/MoltenVK-macos.tar"
           echo "$MOLTENVK_SHA256  $archive" | shasum -a 256 --check
           tar -xf "$archive" -C "$RUNNER_TEMP"
+          test -d "$app_dir"
           install -m 755 \
             "$package/MoltenVK/dynamic/dylib/macOS/libMoltenVK.dylib" \
-            _Build/macos/install/libMoltenVK.dylib
-          install -m 644 "$package/LICENSE" _Build/macos/install/LICENSE.MoltenVK
-          codesign --force --sign - --timestamp=none _Build/macos/install/libMoltenVK.dylib
+            "$moltenvk"
+          install -m 644 "$package/LICENSE" "$install_dir/LICENSE.MoltenVK"
+          codesign --force --sign - --timestamp=none "$moltenvk"
+          install -d -m 755 "$app_dir/Contents/Frameworks" "$app_dir/Contents/Resources"
+          install -m 755 "$moltenvk" "$app_dir/Contents/Frameworks/libMoltenVK.dylib"
+          install -m 644 "$package/LICENSE" "$app_dir/Contents/Resources/LICENSE.MoltenVK"
+          codesign --force --sign - --timestamp=none "$app_dir"
 
       - name: Verify artifacts
         shell: bash
         run: |
-          file _Build/macos/install/launcher
-          file _Build/macos/install/kyty_emulator
-          file _Build/macos/install/libMoltenVK.dylib
-          lipo _Build/macos/install/launcher -verify_arch x86_64
-          lipo _Build/macos/install/kyty_emulator -verify_arch x86_64
-          lipo _Build/macos/install/libMoltenVK.dylib -verify_arch x86_64
-          codesign --verify --strict _Build/macos/install/kyty_emulator
-          codesign --verify --strict _Build/macos/install/libMoltenVK.dylib
+          install_dir="_Build/macos/install"
+          app_dir="$install_dir/KytyPS5.app"
+          test -d "$app_dir"
+          for binary in \
+            "$install_dir/kyty_emulator" \
+            "$install_dir/libMoltenVK.dylib" \
+            "$app_dir/Contents/MacOS/KytyPS5" \
+            "$app_dir/Contents/MacOS/kyty_emulator" \
+            "$app_dir/Contents/Frameworks/libMoltenVK.dylib"; do
+            file "$binary"
+            lipo "$binary" -verify_arch x86_64
+          done
+          for binary in \
+            "$install_dir/kyty_emulator" \
+            "$install_dir/libMoltenVK.dylib" \
+            "$app_dir/Contents/MacOS/KytyPS5" \
+            "$app_dir/Contents/MacOS/kyty_emulator" \
+            "$app_dir/Contents/Frameworks/libMoltenVK.dylib"; do
+            codesign --verify --strict "$binary"
+          done
+          codesign --verify --deep --strict --verbose=2 "$app_dir"
+
+      - name: Package macOS artifacts
+        shell: bash
+        run: |
+          install_dir="_Build/macos/install"
+          archive="$RUNNER_TEMP/KytyPS5-macOS-x86_64.zip"
+          (
+            cd "$install_dir"
+            zip -qry "$archive" .
+          )
+          unzip -t "$archive"
 
       - name: Upload macOS artifacts
         uses: actions/upload-artifact@v4
         with:
           name: KytyPS5-macOS-x86_64
-          path: _Build/macos/install/**
+          path: ${{ runner.temp }}/KytyPS5-macOS-x86_64.zip
           if-no-files-found: error
 
   linux:
@@ -424,16 +456,9 @@ jobs:
         if: needs.macos.result == 'success'
         shell: bash
         run: |
-          macos_dir="artifacts/KytyPS5-macOS-x86_64"
-          test -d "$macos_dir"
-          chmod a+x \
-            "$macos_dir/launcher" \
-            "$macos_dir/kyty_emulator" \
-            "$macos_dir/libMoltenVK.dylib"
-          (
-            cd "$macos_dir"
-            zip -r "$GITHUB_WORKSPACE/$RELEASE_NAME-macOS-x86_64.zip" .
-          )
+          macos_archive="artifacts/KytyPS5-macOS-x86_64/KytyPS5-macOS-x86_64.zip"
+          test -f "$macos_archive"
+          cp "$macos_archive" "$RELEASE_NAME-macOS-x86_64.zip"
 
       - name: Package Linux build
         if: needs.linux.result == 'success'

.gitignore

@@ -5,3 +5,4 @@ build/
 _Build/vscode-clang/
 _Build/
 cmake-build-debug/
+_PipelineCache/

.vscode/launch.json

@@ -10,7 +10,14 @@
             "stopAtEntry": false,
             "cwd": "${workspaceFolder}\\_Build\\vscode-clang\\launcher",
             "environment": [],
-            "externalConsole": false
+            "externalConsole": false,
+            "linux": {
+                "type": "cppdbg",
+                "program": "${workspaceFolder}/_Build/vscode-clang/launcher/launcher",
+                "cwd": "${workspaceFolder}/_Build/vscode-clang/launcher",
+                "MIMode": "gdb",
+                "miDebuggerPath": "/usr/bin/gdb"
+            }
         },
         {
             "name": "Debug kyty_emulator",
@@ -25,13 +32,40 @@
             "stopAtEntry": false,
             "cwd": "${workspaceFolder}\\_Build\\vscode-clang",
             "environment": [],
-            "externalConsole": true
+            "externalConsole": true,
+            "linux": {
+                "type": "cppdbg",
+                "program": "${workspaceFolder}/_Build/vscode-clang/kyty_emulator",
+                "cwd": "${workspaceFolder}/_Build/vscode-clang",
+                "externalConsole": false,
+                "MIMode": "gdb",
+                "miDebuggerPath": "/usr/bin/gdb",
+                "setupCommands": [
+                    {
+                        "description": "Pass Kyty's internal signals to its handlers",
+                        "text": "handle SIG37 SIGSEGV SIGBUS SIGILL nostop noprint pass",
+                        "ignoreFailures": false
+                    }
+                ]
+            }
         },
         {
             "name": "Attach to running kyty_emulator",
             "type": "cppvsdbg",
             "request": "attach",
-            "processId": "${command:pickProcess}"
+            "processId": "${command:pickProcess}",
+            "linux": {
+                "type": "cppdbg",
+                "MIMode": "gdb",
+                "miDebuggerPath": "/usr/bin/gdb",
+                "setupCommands": [
+                    {
+                        "description": "Pass Kyty's internal signals to its handlers",
+                        "text": "handle SIG37 SIGSEGV SIGBUS SIGILL nostop noprint pass",
+                        "ignoreFailures": false
+                    }
+                ]
+            }
         }
     ]
 }

CMakeLists.txt

@@ -126,7 +126,7 @@ if (KYTY_LINKER STREQUAL LD AND NOT LINUX)
 	set(KYTY_LD_OPTIONS "-Wl,--image-base=0x100000000000")
 endif()
 
-project(Kyty${KYTY_PROJECT_NAME}${CMAKE_BUILD_TYPE}${KYTY_COMPILER} VERSION 0.2.2)
+project(Kyty${KYTY_PROJECT_NAME}${CMAKE_BUILD_TYPE}${KYTY_COMPILER} VERSION 0.3.0)
 
 include_directories(
 	${KYTY_THIRD_PARTY_DIR}/gtest/include
@@ -254,7 +254,7 @@ foreach(shader_name IN LISTS gpu_tiler_shader_names)
 	add_custom_command(
 		OUTPUT "${shader_header}"
 		COMMAND ${CMAKE_COMMAND} -E make_directory "${gpu_tiler_generated_dir}"
-		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.0 -Os
+		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.3 -Os
 		        "-I${gpu_tiler_shader_dir}" -o "${shader_spv}" "${shader_source}"
 		COMMAND ${CMAKE_COMMAND} -DINPUT=${shader_spv} -DOUTPUT=${shader_header}
 		        -DSYMBOL=${shader_symbol} -P "${KYTY_SOURCE_DIR}/embed_spirv.cmake"
@@ -263,6 +263,22 @@ foreach(shader_name IN LISTS gpu_tiler_shader_names)
 	)
 	list(APPEND gpu_tiler_shader_headers "${shader_header}")
 endforeach()
+
+set(fault_buffer_shader_source "${gpu_tiler_shader_dir}/fault_buffer_process.comp")
+set(fault_buffer_shader_spv "${gpu_tiler_generated_dir}/fault_buffer_process.spv")
+set(fault_buffer_shader_header "${gpu_tiler_generated_dir}/fault_buffer_process_spv.h")
+add_custom_command(
+	OUTPUT "${fault_buffer_shader_header}"
+	COMMAND ${CMAKE_COMMAND} -E make_directory "${gpu_tiler_generated_dir}"
+	COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.3 -Os
+	        -DCACHING_PAGEBITS=14 -DMAX_PAGE_FAULTS=1024
+	        -o "${fault_buffer_shader_spv}" "${fault_buffer_shader_source}"
+	COMMAND ${CMAKE_COMMAND} -DINPUT=${fault_buffer_shader_spv} -DOUTPUT=${fault_buffer_shader_header}
+	        -DSYMBOL=FAULT_BUFFER_PROCESS_SPV -P "${KYTY_SOURCE_DIR}/embed_spirv.cmake"
+	DEPENDS "${fault_buffer_shader_source}"
+	VERBATIM
+)
+list(APPEND kyty_emulator_src "${fault_buffer_shader_header}")
 list(APPEND kyty_emulator_src ${gpu_tiler_shader_headers})
 
 set(gpu_blit_generated_dir "${PROJECT_BINARY_DIR}/gpu_blit_shaders")
@@ -278,7 +294,7 @@ foreach(shader_source IN LISTS gpu_blit_shader_sources)
 	add_custom_command(
 		OUTPUT "${shader_header}"
 		COMMAND ${CMAKE_COMMAND} -E make_directory "${gpu_blit_generated_dir}"
-		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.0 -Os
+		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.3 -Os
 		        "-I${gpu_tiler_shader_dir}" -o "${shader_spv}" "${shader_source}"
 		COMMAND ${CMAKE_COMMAND} -DINPUT=${shader_spv} -DOUTPUT=${shader_header}
 		        -DSYMBOL=${shader_symbol} -P "${KYTY_SOURCE_DIR}/embed_spirv.cmake"
@@ -415,21 +431,21 @@ add_executable(shader_vertex_metadata_tests EXCLUDE_FROM_ALL
 )
 target_include_directories(shader_vertex_metadata_tests PRIVATE ${inc_headers})
 
-add_executable(shader_stage_runtime_tests EXCLUDE_FROM_ALL
-	"${KYTY_TESTS_DIR}/ShaderStageRuntimeTests.cpp"
+add_executable(resource_materialization_tests EXCLUDE_FROM_ALL
+	"${KYTY_TESTS_DIR}/ResourceMaterializationTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/guest_gpu/gpu_format.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/hostMemory.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/shaderStageRuntime.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ResourceMaterialization.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/SrtWalker.cpp"
 )
-target_link_libraries(shader_stage_runtime_tests fmt::fmt)
-target_include_directories(shader_stage_runtime_tests PRIVATE ${inc_headers})
+target_link_libraries(resource_materialization_tests fmt::fmt)
+target_include_directories(resource_materialization_tests PRIVATE ${inc_headers})
 
 add_executable(resource_tracking_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/ResourceTrackingTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/guest_gpu/gpu_format.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/hostMemory.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/shaderPixelParameter.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/SrtWalker.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ResourceTracking.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ResourceMaterialization.cpp"
@@ -483,7 +499,7 @@ set(gpu_test_ms_depth_header
 add_custom_command(
 	OUTPUT "${gpu_test_ms_depth_header}"
 	COMMAND ${CMAKE_COMMAND} -E make_directory "${gpu_test_generated_dir}"
-	COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.0 -Os
+	COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.3 -Os
 	        -o "${gpu_test_ms_depth_spv}" "${gpu_test_ms_depth_source}"
 	COMMAND ${CMAKE_COMMAND} -DINPUT=${gpu_test_ms_depth_spv}
 	        -DOUTPUT=${gpu_test_ms_depth_header} -DSYMBOL=GPU_TEST_MS_DEPTH_SPV
@@ -507,6 +523,16 @@ if(NOT KYTY_CLANG_CL)
 endif()
 
 if(BUILD_TESTING)
+	add_test(NAME emulator_user_name_cli
+	         COMMAND $<TARGET_FILE:kyty_emulator> --help --user-name 1234567890123456)
+	add_test(NAME emulator_user_name_too_long
+	         COMMAND $<TARGET_FILE:kyty_emulator> --help --user-name 12345678901234567)
+	set_tests_properties(emulator_user_name_too_long PROPERTIES WILL_FAIL TRUE)
+	add_test(NAME emulator_present_mode_cli
+	         COMMAND $<TARGET_FILE:kyty_emulator> --help --present-mode Immediate)
+	add_test(NAME emulator_present_mode_invalid
+	         COMMAND $<TARGET_FILE:kyty_emulator> --help --present-mode Invalid)
+	set_tests_properties(emulator_present_mode_invalid PROPERTIES WILL_FAIL TRUE)
 	add_test(NAME ime_dialog COMMAND $<TARGET_FILE:ime_dialog_tests>)
 	add_test(NAME shader_cfg COMMAND $<TARGET_FILE:shader_cfg_tests>)
 	add_test(NAME scalar_provenance COMMAND $<TARGET_FILE:scalar_provenance_tests>)
@@ -515,7 +541,7 @@ if(BUILD_TESTING)
 	add_test(NAME page_manager COMMAND $<TARGET_FILE:page_manager_tests>)
 	add_test(NAME bit_array COMMAND $<TARGET_FILE:bit_array_tests>)
 	add_test(NAME shader_vertex_metadata COMMAND $<TARGET_FILE:shader_vertex_metadata_tests>)
-	add_test(NAME shader_stage_runtime COMMAND $<TARGET_FILE:shader_stage_runtime_tests>)
+	add_test(NAME resource_materialization COMMAND $<TARGET_FILE:resource_materialization_tests>)
 	add_test(NAME resource_tracking COMMAND $<TARGET_FILE:resource_tracking_tests>)
 	add_test(NAME event_queue_lifetime COMMAND $<TARGET_FILE:event_queue_lifetime_tests>)
 	add_test(NAME kernel_file_system COMMAND $<TARGET_FILE:kernel_file_system_tests>)
@@ -547,6 +573,8 @@ if(BUILD_TESTING)
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --storage-sampled-only)
 	add_test(NAME texture_cache_depth_readback
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --depth-readback-only)
+	add_test(NAME compute_meta_clear_classification
+		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --compute-meta-clear-only)
 	add_test(NAME buffer_cache_dirty_gc
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --buffer-cache-gc-only)
 
@@ -569,7 +597,7 @@ if(BUILD_TESTING)
 		page_manager_tests
 		bit_array_tests
 		shader_vertex_metadata_tests
-		shader_stage_runtime_tests
+		resource_materialization_tests
 		resource_tracking_tests
 		event_queue_lifetime_tests
 		kernel_file_system_tests

README.md

@@ -220,17 +220,23 @@ cmake --install _Build/macos --prefix _Build/macos/install
 ```
 
 The build re-signs `kyty_emulator` with the JIT entitlements it needs to execute translated
-guest code; no manual signing step is required.
+guest code; no manual signing step is required. When the launcher is built, the install
+also produces `_Build/macos/install/KytyPS5.app` — double-click to launch the GUI.
+A flat `kyty_emulator` is kept for CLI usage.
 
 Vulkan comes from MoltenVK. Download `MoltenVK-macos.tar` from the
 [MoltenVK releases](https://github.com/KhronosGroup/MoltenVK/releases), then copy
-`MoltenVK/dynamic/dylib/macOS/libMoltenVK.dylib` next to `kyty_emulator` and ad-hoc sign it:
+`MoltenVK/dynamic/dylib/macOS/libMoltenVK.dylib` next to the flat `kyty_emulator`
+(and, for the bundle, into `KytyPS5.app/Contents/Frameworks/`) and ad-hoc sign it:
 
 ```bash
 codesign --force --sign - _Build/macos/install/libMoltenVK.dylib
+# For the bundle (if present):
+codesign --force --sign - _Build/macos/install/KytyPS5.app/Contents/Frameworks/libMoltenVK.dylib
+codesign --force --sign - _Build/macos/install/KytyPS5.app
 ```
 
-Release archives already include a signed `libMoltenVK.dylib`.
+Release archives already include a signed `libMoltenVK.dylib` (both flat and inside the bundle).
 
 ### Regression tests
 
@@ -274,6 +280,10 @@ To use the graphical launcher:
 ./_Build/linux/install/launcher
 ```
 
+```bash
+open _Build/macos/install/KytyPS5.app  # or double-click in Finder
+```
+
 On first launch, add one or more game folders in the global settings. The launcher searches those
 folders recursively for game directories containing `eboot.bin`. Select a detected game and run it
 from the game list.
@@ -288,12 +298,17 @@ The emulator can also be started directly with a legally obtained game directory
 ./_Build/linux/install/kyty_emulator --game "/games/ExampleGame"
 ```
 
-On macOS, point SDL at the MoltenVK library explicitly; the hardened runtime prevents it from
-being picked up from the executable's directory:
+On macOS, the adjacent flat or app-bundled `libMoltenVK.dylib` is found automatically; no
+environment variable is required:
 
 ```bash
-cd _Build/macos/install
-SDL_VULKAN_LIBRARY="$PWD/libMoltenVK.dylib" ./kyty_emulator --game "/games/ExampleGame"
+./_Build/macos/install/kyty_emulator --game "/games/ExampleGame"
+```
+
+To override the Vulkan loader, set `SDL_VULKAN_LIBRARY`:
+
+```bash
+SDL_VULKAN_LIBRARY=/path/to/libMoltenVK.dylib ./kyty_emulator --game "/games/ExampleGame"
 ```
 
 Run `kyty_emulator --help` to see the available graphics, logging, validation, profiling, and

src/common/assert.cpp

@@ -59,6 +59,7 @@ int DbgExitHandler(const char* file, int line, fmt::text_style style, std::strin
 }
 
 void DbgExit(int status) {
+	Subsystems::EmergencyShutdownActive();
 	std::fflush(nullptr);
 	std::_Exit(status);
 }

src/common/assert.h

@@ -29,11 +29,7 @@ void DbgExit(int status);
 
 } // namespace Common
 
-#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS || KYTY_PLATFORM == KYTY_PLATFORM_LINUX
 #define EXIT_HALT() (Common::DbgExit(321), 1)
-#else
-#define EXIT_HALT() (std::_Exit(321), 1)
-#endif
 
 #ifndef KYTY_FINAL
 #define EXIT_IF(x)                                                                                 \

src/common/emulatorConfig.cpp

@@ -21,6 +21,8 @@ void Shutdown() {
 
 void Load(const ConfigOptions& cfg) {
 	EXIT_IF(g_config == nullptr);
+	EXIT_IF(cfg.user_name.empty() || cfg.user_name.size() > MAX_USER_NAME_LENGTH);
+	EXIT_IF(!IsConfiguredUserIdValid(cfg.user_id));
 
 	*g_config = cfg;
 }
@@ -33,6 +35,22 @@ uint32_t GetScreenHeight() {
 	return g_config->screen_height;
 }
 
+const std::string& GetUserName() {
+	return g_config->user_name;
+}
+
+int32_t GetUserId() {
+	return g_config->user_id;
+}
+
+PresentMode GetPresentMode() {
+	return g_config->present_mode;
+}
+
+int32_t GetGpuIndex() {
+	return g_config->gpu_index;
+}
+
 bool FullscreenEnabled() {
 	return g_config->fullscreen_enabled;
 }
@@ -93,6 +111,10 @@ bool SpirvDebugPrintfEnabled() {
 	return g_config->spirv_debug_printf_enabled;
 }
 
+bool GpuAssistedValidationEnabled() {
+	return g_config->gpu_assisted_validation_enabled && g_config->vulkan_validation_enabled;
+}
+
 bool RenderDocEnabled() {
 	return g_config->renderdoc_enabled;
 }

src/common/emulatorConfig.h

@@ -3,6 +3,7 @@
 
 #include "common/common.h"
 
+#include <cstddef>
 #include <filesystem>
 #include <string>
 #include <vector>
@@ -26,14 +27,28 @@ enum class ProfilerDirection { None, Network };
 
 enum class OutputDirection { Silent, Console, File };
 
+enum class PresentMode { Fifo, Mailbox, Immediate };
+
 using Keymap = std::vector<std::string>;
 
 constexpr uint32_t DEFAULT_CONSOLE_LANGUAGE = 1;
 constexpr uint32_t MAX_CONSOLE_LANGUAGE     = 29;
+constexpr std::size_t MAX_USER_NAME_LENGTH = 16;
+constexpr int32_t DEFAULT_USER_ID           = 1000;
+
+constexpr bool IsConfiguredUserIdValid(int32_t user_id) {
+	constexpr int32_t USER_ID_EVERYONE = 0xfe;
+	constexpr int32_t USER_ID_SYSTEM   = 0xff;
+	return user_id >= 0 && user_id != USER_ID_EVERYONE && user_id != USER_ID_SYSTEM;
+}
 
 struct ConfigOptions {
 	uint32_t               screen_width                = 1280;
 	uint32_t               screen_height               = 720;
+	std::string            user_name                   = "Kyty";
+	int32_t                user_id                     = DEFAULT_USER_ID;
+	PresentMode            present_mode                = PresentMode::Fifo;
+	int32_t                gpu_index                   = -1;
 	bool                   fullscreen_enabled          = false;
 	uint32_t               vblank_frequency            = 60;
 	uint32_t               console_language            = DEFAULT_CONSOLE_LANGUAGE;
@@ -45,10 +60,11 @@ struct ConfigOptions {
 	bool                   command_buffer_dump_enabled = false;
 	std::filesystem::path  command_buffer_dump_folder  = "_Buffers";
 	bool                   graphics_debug_dump_enabled = false;
-	OutputDirection        printf_direction            = OutputDirection::Console;
+	OutputDirection        printf_direction            = OutputDirection::Silent;
 	std::filesystem::path  printf_output_file          = "_kyty.txt";
 	ProfilerDirection      profiler_direction          = ProfilerDirection::None;
 	bool                   spirv_debug_printf_enabled  = false;
+	bool                   gpu_assisted_validation_enabled = false;
 	bool                   renderdoc_enabled           = false;
 	bool                   readback_linear_images      = false;
 	bool                   playgo_hack_enabled         = false;
@@ -62,6 +78,10 @@ void Load(const ConfigOptions& cfg);
 
 uint32_t GetScreenWidth();
 uint32_t GetScreenHeight();
+const std::string& GetUserName();
+int32_t  GetUserId();
+PresentMode GetPresentMode();
+int32_t GetGpuIndex();
 bool     FullscreenEnabled();
 uint32_t GetVblankFrequency();
 uint32_t GetConsoleLanguage();
@@ -84,6 +104,8 @@ ProfilerDirection GetProfilerDirection();
 
 bool SpirvDebugPrintfEnabled();
 
+bool GpuAssistedValidationEnabled();
+
 bool RenderDocEnabled();
 bool ReadbackLinearImagesEnabled();
 bool PlayGoHackEnabled();

src/common/file.cpp

@@ -25,9 +25,6 @@
 #ifdef CopyFile
 #undef CopyFile
 #endif
-#ifdef MoveFile
-#undef MoveFile
-#endif
 #endif
 
 namespace Common {
@@ -72,7 +69,7 @@ File::File(): m_p(std::make_unique<FilePrivate>()) {
 }
 
 File::~File() {
-	EXIT_IF(m_p->f != nullptr);
+	Close();
 }
 
 File::File(const std::filesystem::path& name): m_p(std::make_unique<FilePrivate>()) {
@@ -544,14 +541,14 @@ bool File::CopyFile(const std::filesystem::path& src,
 	return SysFileCopyFile(src, dst);
 }
 
-bool File::MoveFile(const std::filesystem::path& src,
-                    const std::filesystem::path& dst) // @suppress("Member declaration not found")
+bool File::RenameFile(const std::filesystem::path& src,
+                      const std::filesystem::path& dst) // @suppress("Member declaration not found")
 {
 	if (IsFileExisting(dst)) {
 		DeleteFile(dst); // @suppress("Invalid arguments")
 	}
 
-	return SysFileMoveFile(src, dst);
+	return SysFileRenameFile(src, dst);
 }
 
 void File::RemoveReadonly(const std::filesystem::path& name) {

src/common/file.h

@@ -20,9 +20,6 @@
 #ifdef CopyFile
 #undef CopyFile
 #endif
-#ifdef MoveFile
-#undef MoveFile
-#endif
 #endif
 
 namespace Common {
@@ -107,7 +104,7 @@ public:
 	static std::vector<FindInfo> FindFiles(const std::filesystem::path& path);
 	static std::vector<DirEntry> GetDirEntries(const std::filesystem::path& path);
 	static bool CopyFile(const std::filesystem::path& src, const std::filesystem::path& dst);
-	static bool MoveFile(const std::filesystem::path& src, const std::filesystem::path& dst);
+	static bool RenameFile(const std::filesystem::path& src, const std::filesystem::path& dst);
 	static void RemoveReadonly(const std::filesystem::path& name);
 
 	KYTY_CLASS_NO_COPY(File);

src/common/hostException.cpp

@@ -5,14 +5,17 @@
 
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
 #include <windows.h> // IWYU pragma: keep
-#elif defined(__APPLE__)
+#else
+#include <algorithm>
 #include <csignal>
+#include <cstdlib>
+#include <initializer_list>
+#include <unistd.h>
+#if defined(__APPLE__)
 #include <sys/ucontext.h>
 #else
-#include <csignal>
-#include <initializer_list>
 #include <ucontext.h> // IWYU pragma: keep
-#include <unistd.h>
+#endif
 #endif
 
 // IWYU pragma: no_include <errhandlingapi.h>
@@ -32,6 +35,51 @@ static_assert(decltype(g_handler)::is_always_lock_free);
 static_assert(decltype(g_install_state)::is_always_lock_free);
 #endif
 
+#if KYTY_PLATFORM == KYTY_PLATFORM_LINUX
+
+// macOS uses the same POSIX platform setting and needs a signal stack too.
+class ThreadSignalStack {
+public:
+	ThreadSignalStack() {
+		const auto page_size = static_cast<size_t>(::getpagesize());
+		const auto stack_size =
+		    (std::max<size_t>(64 * 1024, MINSIGSTKSZ) + page_size - 1) & ~(page_size - 1);
+		if (::posix_memalign(&m_memory, page_size, stack_size) != 0) {
+			return;
+		}
+
+		stack_t stack {};
+		stack.ss_sp   = m_memory;
+		stack.ss_size = stack_size;
+		if (::sigaltstack(&stack, &m_previous) != 0) {
+			std::free(m_memory);
+			m_memory = nullptr;
+		}
+	}
+
+	~ThreadSignalStack() {
+		if (m_memory != nullptr && ::sigaltstack(&m_previous, nullptr) == 0) {
+			std::free(m_memory);
+		}
+	}
+
+	[[nodiscard]] bool IsInitialized() const { return m_memory != nullptr; }
+
+	KYTY_CLASS_NO_COPY(ThreadSignalStack)
+
+private:
+	void*   m_memory = nullptr;
+	stack_t m_previous {};
+};
+
+bool InitializeThreadSignalStack() {
+	// Keep fault handling off guest stacks, which GPU tracking can make read-only.
+	thread_local ThreadSignalStack signal_stack;
+	return signal_stack.IsInitialized();
+}
+
+#endif
+
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
 
 static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) noexcept {
@@ -253,7 +301,7 @@ bool InstallHandler(Handler handler) {
 #elif defined(__APPLE__)
 	struct sigaction sa {};
 	sa.sa_sigaction = SignalHandler;
-	sa.sa_flags     = SA_SIGINFO;
+	sa.sa_flags     = SA_SIGINFO | SA_ONSTACK;
 	sigemptyset(&sa.sa_mask);
 	// The guest signal-dispatch path (KernelRaiseException) interrupts threads with
 	// SIGUSR1; block it while a fault is being resolved so a stop-the-world request
@@ -274,8 +322,7 @@ bool InstallHandler(Handler handler) {
 	struct sigaction action {};
 	action.sa_sigaction = SignalHandler;
 	sigemptyset(&action.sa_mask);
-	// Fault resolution needs the normal thread stack.
-	action.sa_flags = SA_SIGINFO | SA_RESTART;
+	action.sa_flags = SA_SIGINFO | SA_RESTART | SA_ONSTACK;
 
 	for (const int signal_number: {SIGSEGV, SIGBUS, SIGILL}) {
 		if (::sigaction(signal_number, &action, nullptr) != 0) {

src/common/hostException.h

@@ -39,6 +39,10 @@ using Handler = bool (*)(const ExceptionInfo&);
 
 bool InstallHandler(Handler handler);
 
+#if KYTY_PLATFORM == KYTY_PLATFORM_LINUX
+bool InitializeThreadSignalStack();
+#endif
+
 } // namespace Common::HostException
 
 #endif /* KYTY_COMMON_HOST_EXCEPTION_H_ */

src/common/platform/sysFileIO.h

@@ -80,7 +80,7 @@ void SysFileFindFiles(const std::filesystem::path& path, std::vector<sys_file_fi
 // NOLINTNEXTLINE(google-runtime-references)
 void SysFileGetDents(const std::filesystem::path& path, std::vector<sys_dir_entry_t>& out);
 bool SysFileCopyFile(const std::filesystem::path& src, const std::filesystem::path& dst);
-bool SysFileMoveFile(const std::filesystem::path& src, const std::filesystem::path& dst);
+bool SysFileRenameFile(const std::filesystem::path& src, const std::filesystem::path& dst);
 void SysFileRemoveReadonly(const std::filesystem::path& name);
 
 #endif /* KYTY_COMMON_PLATFORM_SYSFILEIO_H_ */

src/common/platform/sysLinuxFileIO.cpp

@@ -693,7 +693,7 @@ bool SysFileCopyFile(const std::filesystem::path& src, const std::filesystem::pa
 	       !error;
 }
 
-bool SysFileMoveFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
+bool SysFileRenameFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
 	auto real_src = get_internal_name(src);
 	auto real_dst = get_internal_name(dst);
 

src/common/platform/sysWindowsFileIO.cpp

@@ -565,7 +565,7 @@ bool SysFileCopyFile(const std::filesystem::path& src, const std::filesystem::pa
 	return CopyFileW(src_wide.c_str(), dst_wide.c_str(), FALSE) != 0;
 }
 
-bool SysFileMoveFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
+bool SysFileRenameFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
 	auto src_wide = src.wstring();
 	auto dst_wide = dst.wstring();
 	return MoveFileW(src_wide.c_str(), dst_wide.c_str()) != 0;

src/common/slotVector.h

@@ -0,0 +1,132 @@
+#ifndef EMULATOR_SRC_COMMON_SLOTVECTOR_H_
+#define EMULATOR_SRC_COMMON_SLOTVECTOR_H_
+
+#include "common/assert.h"
+#include "common/abi.h"
+
+#include <compare>
+#include <cstdint>
+#include <deque>
+#include <functional>
+#include <limits>
+#include <optional>
+#include <utility>
+#include <vector>
+
+namespace Common {
+
+struct SlotId {
+	static constexpr uint32_t INVALID_INDEX = std::numeric_limits<uint32_t>::max();
+
+	constexpr SlotId() noexcept = default;
+	constexpr SlotId(uint32_t value) noexcept: index(value), generation(1) {}
+	constexpr SlotId(uint32_t value, uint32_t slot_generation) noexcept
+	    : index(value), generation(slot_generation) {}
+
+	[[nodiscard]] constexpr explicit operator bool() const noexcept { return index != INVALID_INDEX; }
+	constexpr auto operator<=>(const SlotId&) const noexcept = default;
+
+	uint32_t index = INVALID_INDEX;
+	uint32_t generation = 0;
+};
+
+// Stable-address slot storage for cache resources that are intentionally non-movable.
+template <typename T>
+class SlotVector {
+public:
+	SlotVector() = default;
+	KYTY_CLASS_NO_COPY(SlotVector);
+
+	[[nodiscard]] T& operator[](SlotId id) noexcept {
+		EXIT_IF(!is_allocated(id));
+		return *m_values[id.index].value;
+	}
+
+	[[nodiscard]] const T& operator[](SlotId id) const noexcept {
+		EXIT_IF(!is_allocated(id));
+		return *m_values[id.index].value;
+	}
+
+	[[nodiscard]] T* try_get(SlotId id) noexcept {
+		return is_allocated(id) ? &*m_values[id.index].value : nullptr;
+	}
+
+	[[nodiscard]] const T* try_get(SlotId id) const noexcept {
+		return is_allocated(id) ? &*m_values[id.index].value : nullptr;
+	}
+
+	[[nodiscard]] bool is_allocated(SlotId id) const noexcept {
+		return id && id.index < m_values.size() &&
+		       m_values[id.index].generation == id.generation && m_values[id.index].value.has_value();
+	}
+
+	template <typename... Args>
+	[[nodiscard]] SlotId insert(Args&&... args) {
+		uint32_t index = 0;
+		if (m_free_list.empty()) {
+			index = static_cast<uint32_t>(m_values.size());
+			m_values.emplace_back();
+			m_values.back().value.emplace(std::forward<Args>(args)...);
+		} else {
+			index = m_free_list.back();
+			m_free_list.pop_back();
+			EXIT_IF(m_values[index].value.has_value());
+			m_values[index].value.emplace(std::forward<Args>(args)...);
+		}
+		++m_size;
+		return SlotId {index, m_values[index].generation};
+	}
+
+	void erase(SlotId id) noexcept {
+		EXIT_IF(!is_allocated(id));
+		auto& slot = m_values[id.index];
+		slot.value.reset();
+		if (++slot.generation == 0) {
+			slot.generation = 1;
+		}
+		m_free_list.push_back(id.index);
+		--m_size;
+	}
+
+	[[nodiscard]] size_t size() const noexcept { return m_size; }
+	[[nodiscard]] size_t capacity() const noexcept { return m_values.size(); }
+
+	template <typename F>
+	void ForEach(F&& fn) {
+		for (uint32_t index = 0; index < m_values.size(); ++index) {
+			if (m_values[index].value) {
+				fn(SlotId {index, m_values[index].generation}, *m_values[index].value);
+			}
+		}
+	}
+
+	template <typename F>
+	void ForEach(F&& fn) const {
+		for (uint32_t index = 0; index < m_values.size(); ++index) {
+			if (m_values[index].value) {
+				fn(SlotId {index, m_values[index].generation}, *m_values[index].value);
+			}
+		}
+	}
+
+private:
+	struct Slot {
+		std::optional<T> value;
+		uint32_t         generation = 1;
+	};
+
+	std::deque<Slot>             m_values;
+	std::vector<uint32_t>        m_free_list;
+	size_t                       m_size = 0;
+};
+
+} // namespace Common
+
+template <>
+struct std::hash<Common::SlotId> {
+	[[nodiscard]] size_t operator()(Common::SlotId id) const noexcept {
+		return std::hash<uint64_t> {}((static_cast<uint64_t>(id.generation) << 32u) | id.index);
+	}
+};
+
+#endif // EMULATOR_SRC_COMMON_SLOTVECTOR_H_

src/common/stringUtils.h

@@ -316,10 +316,12 @@ inline std::string SafeCsv(std::string_view text) {
 	return std::string(text);
 }
 
-inline std::string Utf16ToUtf8(const char16_t* utf16) {
-	std::u16string                                                    input(utf16);
+inline std::string Utf16ToUtf8(std::u16string_view utf16) {
+	if (utf16.empty()) {
+		return {};
+	}
 	std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> convert;
-	return convert.to_bytes(input);
+	return convert.to_bytes(utf16.data(), utf16.data() + utf16.size());
 }
 
 inline ByteBuffer HexToBin(std::string_view text) {

src/emulator.cpp

@@ -15,6 +15,7 @@
 #include "kernel/fileSystem.h"
 #include "kernel/memory.h"
 #include "kernel/pthread.h"
+#include "kytyGitVersion.h"
 #include "libs/agc.h"
 #include "libs/audio.h"
 #include "libs/controller.h"
@@ -26,13 +27,31 @@
 
 #include <cstdlib>
 #include <filesystem>
+#include <thread>
 
 namespace Emulator {
 
 static void PrintSystemInfo() {
-	Common::SystemInfo info = Common::GetSystemInfo();
+	const Common::SystemInfo info = Common::GetSystemInfo();
 
-	LOGF("ProcessorName = %s\n", info.ProcessorName.c_str());
+#if defined(__APPLE__)
+	static constexpr auto platform_name = "macOS";
+#elif KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
+	static constexpr auto platform_name = "Windows";
+#elif KYTY_PLATFORM == KYTY_PLATFORM_LINUX
+	static constexpr auto platform_name = "Linux";
+#else
+	static constexpr auto platform_name = "Unknown";
+#endif
+
+	LOGF("Build\n"
+	     "  version: %s\n\n"
+	     "Host\n"
+	     "  os:      %s\n"
+	     "  cpu:     %s\n"
+	     "  threads: %u\n\n",
+	     KYTY_BUILD_LABEL, platform_name, info.ProcessorName.c_str(),
+	     std::thread::hardware_concurrency());
 }
 
 static void KytyClose() {
@@ -181,6 +200,10 @@ void Run(const RunOptions& options) {
 	int ok = atexit(KytyClose);
 	EXIT_NOT_IMPLEMENTED(ok != 0);
 
+	// Guest threads are still running, so skip KytyClose() and only flush emergency state.
+	ok = at_quick_exit(Common::Subsystems::EmergencyShutdownActive);
+	EXIT_NOT_IMPLEMENTED(ok != 0);
+
 	Libs::LibKernel::FileSystem::Mount(options.app0_dir, "/app0");
 	Libs::LibKernel::FileSystem::Mount(options.app0_dir, "/hostapp");
 

src/generate_version.cmake

@@ -1,5 +1,6 @@
 set(KYTY_GIT_VERSION "unknown")
 set(KYTY_GIT_HASH "unknown")
+set(KYTY_GIT_REVISION "unknown")
 if(GIT_EXECUTABLE)
 	execute_process(
 		COMMAND "${GIT_EXECUTABLE}" describe --tags --always --dirty
@@ -14,16 +15,18 @@ if(GIT_EXECUTABLE)
 	endif()
 
 	execute_process(
-		COMMAND "${GIT_EXECUTABLE}" rev-parse --short=7 HEAD
+		COMMAND "${GIT_EXECUTABLE}" rev-parse HEAD
 		WORKING_DIRECTORY "${GIT_WORKING_DIRECTORY}"
-		OUTPUT_VARIABLE KYTY_GIT_HASH
+		OUTPUT_VARIABLE KYTY_GIT_REVISION
 		OUTPUT_STRIP_TRAILING_WHITESPACE
 		RESULT_VARIABLE GIT_HASH_RESULT
 		ERROR_QUIET
 	)
 	if(NOT GIT_HASH_RESULT EQUAL 0)
 		set(KYTY_GIT_HASH "unknown")
+		set(KYTY_GIT_REVISION "unknown")
 	else()
+		string(SUBSTRING "${KYTY_GIT_REVISION}" 0 7 KYTY_GIT_HASH)
 		execute_process(
 			COMMAND "${GIT_EXECUTABLE}" diff-index --quiet HEAD --
 			WORKING_DIRECTORY "${GIT_WORKING_DIRECTORY}"

src/graphics/guest_gpu/command_processor/commandProcessor.h

@@ -7,6 +7,7 @@
 #include "graphics/host_gpu/renderer/renderContext.h"
 
 #include <cstdint>
+#include <span>
 #include <vector>
 
 namespace Libs::Graphics {
@@ -30,10 +31,9 @@ private:
 	friend class CommandProcessor;
 
 	struct BufferCursor {
-		uint32_t* next_packet         = nullptr;
-		uint32_t  remaining_dw        = 0;
-		uint32_t  total_dw            = 0;
-		uint32_t  deferred_advance_dw = 0;
+		std::span<const uint32_t> commands;
+		uint32_t                  offset_dw           = 0;
+		uint32_t                  deferred_advance_dw = 0;
 	};
 
 	std::vector<BufferCursor> m_buffer_stack;
@@ -50,7 +50,8 @@ public:
 		int64_t flip_arg  = 0;
 	};
 
-	explicit CommandProcessor(RenderContext& renderer): m_renderer(renderer) {}
+	CommandProcessor(RenderContext& renderer, int interrupt_event_id)
+	    : m_renderer(renderer), m_interrupt_event_id(interrupt_event_id) {}
 	~CommandProcessor() = default;
 
 	KYTY_CLASS_NO_COPY(CommandProcessor);
@@ -72,13 +73,9 @@ public:
 	void SetDrawIndirectArgsBaseAddress(uint64_t draw_indirect_args_base_addr);
 	void SetDispatchIndirectArgsBaseAddress(uint64_t dispatch_indirect_args_base_addr);
 	void SetNumInstances(uint32_t num_instances);
-	void DrawIndex(uint32_t index_count, const void* index_addr, uint32_t flags, uint32_t type,
-	               uint32_t instance_count = 0, const void* object_ids = nullptr,
-	               uint32_t render_target_slice_offset = 0, int32_t vertex_offset_add = 0,
-	               uint32_t first_instance = 0);
-	void DrawIndexOffset(uint32_t index_offset, uint32_t index_count, uint32_t flags);
-	void DrawIndexAuto(uint32_t index_count, uint32_t flags,
-	                   uint32_t render_target_slice_offset = 0);
+	void DrawIndex(DrawIndexArgs args);
+	void DrawIndexOffset(uint32_t index_offset, uint32_t index_count);
+	void DrawIndexAuto(DrawAutoArgs args);
 	void DrawIndirect(uint32_t data_offset, uint32_t draw_initiator, bool indexed);
 	void DrawIndirectMulti(uint32_t data_offset, uint32_t max_count_or_count,
 	                       const volatile uint32_t* count_addr, uint32_t stride_in_bytes,
@@ -103,7 +100,7 @@ public:
 	                    uint32_t mode);
 	void DispatchIndirect(uint32_t data_offset, uint32_t mode);
 	void WaitFlipDone(uint32_t video_out_handle, uint32_t display_buffer_index);
-	void TriggerEvent(uint32_t event_type, uint32_t event_index);
+	void TriggerEvent(uint32_t event_type, uint32_t event_index, uint64_t event_address = 0);
 
 	void SetUserDataMarker(HW::UserSgprType type) { m_user_data_marker = type; }
 	[[nodiscard]] HW::UserSgprType GetUserDataMarker() const { return m_user_data_marker; }
@@ -131,13 +128,14 @@ public:
 	                    const volatile void* address, uint32_t count_in_dwords);
 	[[nodiscard]] bool ShouldSkipPredicatedPackets() const { return m_predicate_skip; }
 
-	Pm4ProcessResult Process(Pm4Execution& execution, uint32_t* buffer, uint32_t size_dw);
-	void             ProcessIndirectBuffer(uint32_t* buffer, uint32_t size_dw);
+	Pm4ProcessResult Process(Pm4Execution& execution, std::span<const uint32_t> commands);
+	void             ProcessIndirectBuffer(std::span<const uint32_t> commands);
 
 	void SetFlip(const FlipInfo& flip) { m_flip = flip; }
 
 	[[nodiscard]] uint64_t GetSubmitId() const { return m_submit_id; }
 	void                   SetSubmitId(uint64_t submit_id) { m_submit_id = submit_id; }
+	[[nodiscard]] bool     IsAsyncComputeQueue() const { return m_interrupt_event_id >= 0x20; }
 
 private:
 	template <typename T>
@@ -147,14 +145,10 @@ private:
 	                      uint32_t interrupt_context_id);
 	void ProcessPm4(Pm4Execution& execution, size_t stop_depth);
 	void SuspendPm4();
-	void SubmitNonIndexedDraw(uint32_t vertex_count, uint32_t flags,
-	                          uint32_t render_target_slice_offset, uint32_t first_vertex,
-	                          uint32_t first_instance);
-
-	CommandScheduler&    GetScheduler() const { return m_renderer.GetCommandScheduler(); }
-	RenderCommandBuffer& CurrentBuffer() { return GetScheduler().Current(); }
-	void                 CheckBuffer() const { GetScheduler().CheckActive(); }
-	GpuResourceManager&  GetGpuResources() const { return m_renderer.GetGpuResources(); }
+	CommandScheduler&   GetScheduler() const { return m_renderer.GetCommandScheduler(); }
+	CommandBuffer&      CurrentBuffer() { return GetScheduler().Current(); }
+	void                CheckBuffer() const { GetScheduler().CheckActive(); }
+	GpuResourceManager& GetGpuResources() const { return m_renderer.GetGpuResources(); }
 
 	RenderContext&   m_renderer;
 	HW::Context      m_ctx;
@@ -177,9 +171,11 @@ private:
 
 	uint32_t m_const_ram[0x3000] = {0};
 
-	FlipInfo m_flip;
-	uint64_t m_submit_id      = 0;
-	bool     m_predicate_skip = false;
+	FlipInfo  m_flip;
+	const int m_interrupt_event_id;
+	uint64_t  m_submit_id                   = 0;
+	uint64_t  m_synthetic_occlusion_counter = 0;
+	bool      m_predicate_skip              = false;
 };
 
 } // namespace Libs::Graphics

src/graphics/guest_gpu/command_processor/pm4Dispatch.cpp

@@ -23,9 +23,7 @@ constexpr auto MakeContextDispatchTable() {
 	g_hw_ctx_func[Pm4::DB_STENCIL_CLEAR]             = HwCtxSetStencilClear;
 	g_hw_ctx_func[Pm4::DB_DEPTH_CLEAR]               = HwCtxSetDepthClear;
 	g_hw_ctx_func[Pm4::PA_SC_SCREEN_SCISSOR_TL]      = HwCtxSetScreenScissor;
-	g_hw_ctx_func[Pm4::DB_Z_INFO]                    = HwCtxSetDepthRenderTarget;
 	g_hw_ctx_func[Pm4::DB_STENCIL_INFO]              = HwCtxSetStencilInfo;
-	g_hw_ctx_func[Pm4::DB_HTILE_SURFACE]             = HwCtxSetDepthHtileSurface;
 	g_hw_ctx_func[Pm4::PA_SU_HARDWARE_SCREEN_OFFSET] = HwCtxSetHardwareScreenOffset;
 	g_hw_ctx_func[Pm4::PA_SC_WINDOW_OFFSET]          = HwCtxSetWindowOffset;
 	g_hw_ctx_func[Pm4::PA_SC_WINDOW_SCISSOR_TL]      = HwCtxSetWindowScissor;
@@ -35,7 +33,6 @@ constexpr auto MakeContextDispatchTable() {
 	}
 	g_hw_ctx_func[Pm4::CB_TARGET_MASK]                 = HwCtxSetRenderTargetMask;
 	g_hw_ctx_func[Pm4::PA_SC_GENERIC_SCISSOR_TL]       = HwCtxSetGenericScissor;
-	g_hw_ctx_func[Pm4::CB_BLEND_RED]                   = HwCtxSetBlendColor;
 	g_hw_ctx_func[Pm4::CB_DCC_CONTROL]                 = HwCtxSetCbDccControl;
 	g_hw_ctx_func[Pm4::DB_STENCIL_CONTROL]             = HwCtxSetStencilControl;
 	g_hw_ctx_func[Pm4::DB_STENCILREFMASK]              = HwCtxSetStencilMask;
@@ -76,16 +73,13 @@ constexpr auto MakeContextDispatchTable() {
 	g_hw_ctx_func[Pm4::PA_SC_BINNER_CNTL_0]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::PA_SC_BINNER_CNTL_1]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::PA_SC_CONSERVATIVE_RASTERIZATION_CNTL] = HwCtxSetPaScExtendedControl;
-	g_hw_ctx_func[Pm4::PA_SC_NGG_MODE_CNTL]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::DB_ALPHA_TO_MASK]                      = HwCtxSetAlphaToMask;
 	g_hw_ctx_func[Pm4::VGT_DRAW_PAYLOAD_CNTL]                 = HwCtxSetDrawPayloadControl;
 	g_hw_ctx_func[Pm4::VGT_PRIMITIVEID_RESET]                 = HwCtxSetPrimitiveIdReset;
 	g_hw_ctx_func[Pm4::PA_CL_OBJPRIM_ID_CNTL]                 = HwCtxSetObjprimIdControl;
 	g_hw_ctx_func[Pm4::VGT_SHADER_STAGES_EN]                  = HwCtxSetShaderStages;
-	g_hw_ctx_func[Pm4::PA_CL_GB_VERT_CLIP_ADJ]                = HwCtxSetGuardBands;
 
 	for (uint32_t slot = 0; slot < 8; slot++) {
-		g_hw_ctx_func[Pm4::CB_COLOR0_BASE + slot * 15] = HwCtxSetRenderTarget;
 		g_hw_ctx_func[Pm4::CB_COLOR0_INFO + slot * 15] = HwCtxSetColorInfo;
 
 		g_hw_ctx_func[Pm4::CB_BLEND0_CONTROL + slot * 1] = HwCtxSetBlendControl;
@@ -118,25 +112,21 @@ constexpr auto MakeShaderDispatchTable() {
 	}
 
 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_PS_0 + slot] = HwShSetPsUserAccumSgpr;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_PS_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 
 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_ESGS_0 + slot] = HwShSetGsUserAccumSgpr;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_ESGS_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_LSHS_0 + slot] = HwShSetHsUserAccumSgpr;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_LSHS_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 
-	for (uint32_t slot = 0; Pm4::COMPUTE_USER_ACCUM_0 + slot < Pm4::COMPUTE_USER_DATA_0; slot++) {
-		g_hw_sh_func[Pm4::COMPUTE_USER_ACCUM_0 + slot] = HwShSetCsUserAccumSgpr;
+	for (uint32_t slot = 0; slot < 4; slot++) {
+		g_hw_sh_func[Pm4::COMPUTE_USER_ACCUM_0 + slot] = HwShIgnoreUserAccumulator;
 	}
 	g_hw_sh_func[Pm4::COMPUTE_PGM_LO]                  = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_HI]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TBA_LO]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TBA_HI]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TMA_LO]                  = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_TMA_HI]                  = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC1]               = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC2]               = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_START_X]                 = HwShSetCsRegisters;
@@ -146,30 +136,16 @@ constexpr auto MakeShaderDispatchTable() {
 	g_hw_sh_func[Pm4::COMPUTE_NUM_THREAD_Y]            = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_NUM_THREAD_Z]            = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_RESOURCE_LIMITS]         = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE0]      = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE1]      = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_TMPRING_SIZE]            = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE2]      = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE3]      = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC3]               = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_SHADER_CHKSUM]           = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_DISPATCH_TUNNEL]         = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_TBA_LO_PS]            = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_PS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_PS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_TMA_LO_PS]            = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_TMA_HI_PS]            = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_PS]          = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_GS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_GS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_GS] = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_GS] = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_ESGS]        = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_HS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_HS]         = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_HS] = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_HS] = HwShIgnoreRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_LSHS]        = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_PACE_ID]                 = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_PS]  = HwShSetRegisters;
+	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_GS]  = HwShSetRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_GS] = HwShSetRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_GS] = HwShSetRegisters;
+	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_HS]  = HwShSetRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_HS] = HwShSetRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_HS] = HwShSetRegisters;
 
 	return g_hw_sh_func;
 }
@@ -181,6 +157,7 @@ constexpr auto MakeUconfigDispatchTable() {
 	g_hw_uc_func[Pm4::VGT_INDEX_TYPE]            = HwUcSetIndexType;
 	g_hw_uc_func[Pm4::VGT_OBJECT_ID]             = HwUcSetObjectId;
 	g_hw_uc_func[Pm4::TEXTURE_GRADIENT_FACTORS]  = HwUcSetTextureGradientFactors;
+	g_hw_uc_func[Pm4::TEXTURE_GRADIENT_CONTROL]  = HwUcSetTextureGradientFactors;
 	g_hw_uc_func[Pm4::IA_MULTI_VGT_PARAM]        = HwUcSetIaMultiVgtParam;
 	g_hw_uc_func[Pm4::GE_MULTI_PRIM_IB_RESET_EN] = HwUcSetMultiPrimIbReset;
 	g_hw_uc_func[Pm4::TA_CS_BC_BASE_ADDR]        = HwUcSetBorderColorTableAddr;
@@ -217,6 +194,7 @@ constexpr auto MakeOpcodeDispatchTable() {
 	g_cp_op_func[Pm4::IT_COND_EXEC]                 = CpOpCondExec;
 	g_cp_op_func[Pm4::IT_SET_PREDICATION]           = CpOpSetPredication;
 	g_cp_op_func[Pm4::IT_WRITE_DATA]                = CpOpWriteData;
+	g_cp_op_func[Pm4::IT_WAIT_REG_MEM]              = CpOpWaitRegMem32;
 	g_cp_op_func[Pm4::IT_INDIRECT_BUFFER]           = CpOpIndirectBuffer;
 	g_cp_op_func[Pm4::IT_COPY_DATA]                 = CpOpCopyData;
 	g_cp_op_func[Pm4::IT_EVENT_WRITE]               = CpOpEventWrite;
@@ -239,6 +217,7 @@ constexpr auto MakeOpcodeDispatchTable() {
 	g_cp_op_func[Pm4::IT_WAIT_ON_CE_COUNTER]        = CpOpWaitOnCeCounter;
 	g_cp_op_func[Pm4::IT_WAIT_ON_DE_COUNTER_DIFF]   = CpOpWaitOnDeCounterDiff;
 	g_cp_op_func[Pm4::IT_GET_LOD_STATS]             = CpOpGetLodStats;
+	g_cp_op_func[Pm4::IT_WAIT_REG_MEM_64]           = CpOpWaitRegMem64;
 
 	return g_cp_op_func;
 }
@@ -247,12 +226,10 @@ constexpr auto MakeCustomOpcodeDispatchTable() {
 	std::array<cp_op_parser_func_t, Pm4::R_NUM> g_cp_op_custom_func {};
 
 	g_cp_op_custom_func[Pm4::R_DISPATCH_RESET] = CpOpDispatchReset;
-	g_cp_op_custom_func[Pm4::R_WAIT_MEM_32]    = CpOpWaitRegMem32;
 	g_cp_op_custom_func[Pm4::R_WAIT_FLIP_DONE] = CpOpWaitFlipDone;
 	g_cp_op_custom_func[Pm4::R_PUSH_MARKER]    = CpOpPushMarker;
 	g_cp_op_custom_func[Pm4::R_POP_MARKER]     = CpOpPopMarker;
 	g_cp_op_custom_func[Pm4::R_ACQUIRE_MEM]    = CpOpAcquireMem;
-	g_cp_op_custom_func[Pm4::R_WAIT_MEM_64]    = CpOpWaitRegMem64;
 	g_cp_op_custom_func[Pm4::R_FLIP]           = CpOpFlip;
 	g_cp_op_custom_func[Pm4::R_RELEASE_MEM]    = CpOpReleaseMem;
 	g_cp_op_custom_func[Pm4::R_CONTEXT_STATE]  = CpOpContextState;

src/graphics/guest_gpu/command_processor/pm4Dispatch.h

@@ -44,11 +44,7 @@ uint32_t HwCtxSetBorderColorTableAddr(CommandProcessor&, uint32_t, uint32_t, con
 uint32_t HwCtxSetStencilClear(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetDepthClear(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetScreenScissor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetDepthRenderTarget(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
-                                   uint32_t);
 uint32_t HwCtxSetStencilInfo(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetDepthHtileSurface(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
-                                   uint32_t);
 uint32_t HwCtxSetHardwareScreenOffset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
                                       uint32_t);
 uint32_t HwCtxSetWindowOffset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -56,7 +52,6 @@ uint32_t HwCtxSetWindowScissor(CommandProcessor&, uint32_t, uint32_t, const uint
 uint32_t HwCtxSetClipRect(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetRenderTargetMask(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetGenericScissor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetBlendColor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetCbDccControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetStencilControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetStencilMask(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -90,8 +85,6 @@ uint32_t HwCtxSetDrawPayloadControl(CommandProcessor&, uint32_t, uint32_t, const
 uint32_t HwCtxSetPrimitiveIdReset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetObjprimIdControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetShaderStages(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetGuardBands(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwCtxSetRenderTarget(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetColorInfo(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetBlendControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetViewportScissor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -102,12 +95,10 @@ uint32_t HwShSetPsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t
 uint32_t HwShSetGsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShSetHsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShSetCsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetPsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetGsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetHsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetCsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwShIgnoreUserAccumulator(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
+                                   uint32_t);
 uint32_t HwShSetCsRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShIgnoreRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwShSetRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwUcSetPrimitiveType(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwUcSetIndexType(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwUcSetObjectId(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);

src/graphics/guest_gpu/gpu_defs.h

@@ -407,7 +407,19 @@ enum class BufferFormat : uint32_t {
 	k5_5_5_1UNorm       = 134,
 	k1_5_5_5UNorm       = 135,
 	k4_4_4_4UNorm       = 136,
+	kFmask8_S2_F1       = 156,
+	kFmask8_S4_F1       = 157,
+	kFmask8_S8_F1       = 158,
+	kFmask8_S2_F2       = 159,
+	kFmask8_S4_F2       = 160,
 	kFmask8_S4_F4       = 161,
+	kFmask16_S16_F1     = 162,
+	kFmask16_S8_F2      = 163,
+	kFmask32_S16_F2     = 164,
+	kFmask32_S8_F4      = 165,
+	kFmask32_S8_F8      = 166,
+	kFmask64_S16_F4     = 167,
+	kFmask64_S16_F8     = 168,
 	kBc1UNorm           = 169,
 	kBc1Srgb            = 170,
 	kBc2UNorm           = 171,

src/graphics/guest_gpu/gpu_format.cpp

@@ -9,79 +9,93 @@ namespace Libs::Graphics::Prospero {
 namespace {
 
 struct FormatInfo {
-	BufferFormat format;
-	uint32_t     bytes_per_element;
-	uint32_t     block_compressed_bytes_per_block;
-	uint32_t     render_target_bytes_per_element;
-	bool         supported_sampled_texture_format;
-	bool         unsigned_integer_texture_format;
+	BufferFormat       format;
+	uint32_t           bytes_per_element;
+	uint32_t           block_compressed_bytes_per_block;
+	uint32_t           render_target_bytes_per_element;
+	bool               sampled_texture;
+	bool               uint_texture;
+	bool               sint_texture = false;
 };
 
 constexpr FormatInfo kFormatInfo[] = {
-    {BufferFormat::k8UNorm, 1, 0, 1, true, false},
-    {BufferFormat::k8SNorm, 0, 0, 1, false, false},
-    {BufferFormat::k8UInt, 1, 0, 1, true, true},
-    {BufferFormat::k16UNorm, 2, 0, 2, true, false},
-    {BufferFormat::k16SNorm, 2, 0, 2, true, false},
-    {BufferFormat::k16UInt, 2, 0, 2, true, true},
-    {BufferFormat::k16SInt, 2, 0, 2, false, false},
-    {BufferFormat::k16Float, 2, 0, 2, true, false},
-    {BufferFormat::k8_8UNorm, 2, 0, 2, true, false},
-    {BufferFormat::k8_8SNorm, 2, 0, 2, true, false},
-    {BufferFormat::k8_8UInt, 2, 0, 2, true, true},
-    {BufferFormat::k8_8SInt, 2, 0, 2, false, false},
-    {BufferFormat::k32UInt, 4, 0, 4, true, true},
-    {BufferFormat::k32SInt, 4, 0, 4, false, false},
-    {BufferFormat::k32Float, 4, 0, 4, true, false},
-    {BufferFormat::k16_16UNorm, 4, 0, 4, true, false},
-    {BufferFormat::k16_16SNorm, 4, 0, 4, true, false},
-    {BufferFormat::k16_16UInt, 4, 0, 4, true, true},
-    {BufferFormat::k16_16SInt, 4, 0, 4, false, false},
-    {BufferFormat::k16_16Float, 4, 0, 4, true, false},
-    {BufferFormat::k11_11_10Float, 4, 0, 4, true, false},
-    {BufferFormat::k10_10_10_2UNorm, 4, 0, 4, true, false},
-    {BufferFormat::k10_10_10_2UInt, 4, 0, 4, true, true},
-    {BufferFormat::k8_8_8_8UNorm, 4, 0, 4, true, false},
-    {BufferFormat::k8_8_8_8SNorm, 4, 0, 4, true, false},
-    {BufferFormat::k8_8_8_8UInt, 4, 0, 4, true, true},
-    {BufferFormat::k8_8_8_8SInt, 4, 0, 4, false, false},
-    {BufferFormat::k32_32UInt, 8, 0, 8, true, true},
-    {BufferFormat::k32_32SInt, 8, 0, 8, false, false},
-    {BufferFormat::k32_32Float, 8, 0, 8, true, false},
-    {BufferFormat::k16_16_16_16UNorm, 8, 0, 8, true, false},
-    {BufferFormat::k16_16_16_16SNorm, 8, 0, 8, true, false},
-    {BufferFormat::k16_16_16_16UInt, 8, 0, 8, true, true},
-    {BufferFormat::k16_16_16_16SInt, 8, 0, 8, false, false},
-    {BufferFormat::k16_16_16_16Float, 8, 0, 8, true, false},
-    {BufferFormat::k32_32_32UInt, 12, 0, 12, true, true},
-    {BufferFormat::k32_32_32SInt, 12, 0, 12, false, false},
-    {BufferFormat::k32_32_32Float, 12, 0, 12, true, false},
-    {BufferFormat::k32_32_32_32UInt, 16, 0, 16, true, true},
-    {BufferFormat::k32_32_32_32SInt, 16, 0, 16, false, false},
-    {BufferFormat::k32_32_32_32Float, 16, 0, 16, true, false},
-    {BufferFormat::k8Srgb, 1, 0, 0, true, false},
-    {BufferFormat::k8_8Srgb, 2, 0, 0, true, false},
-    {BufferFormat::k8_8_8_8Srgb, 4, 0, 4, true, false},
-    {BufferFormat::k9_9_9_5Float, 4, 0, 0, true, false},
-    {BufferFormat::k5_6_5UNorm, 2, 0, 2, true, false},
-    {BufferFormat::k5_5_5_1UNorm, 2, 0, 2, true, false},
-    {BufferFormat::k1_5_5_5UNorm, 0, 0, 2, false, false},
-    {BufferFormat::k4_4_4_4UNorm, 2, 0, 2, true, false},
-    {BufferFormat::kFmask8_S4_F4, 1, 0, 1, true, false},
-    {BufferFormat::kBc1UNorm, 0, 8, 0, true, false},
-    {BufferFormat::kBc1Srgb, 0, 8, 0, true, false},
-    {BufferFormat::kBc2UNorm, 0, 16, 0, true, false},
-    {BufferFormat::kBc2Srgb, 0, 16, 0, true, false},
-    {BufferFormat::kBc3UNorm, 0, 16, 0, true, false},
-    {BufferFormat::kBc3Srgb, 0, 16, 0, true, false},
-    {BufferFormat::kBc4UNorm, 0, 8, 0, true, false},
-    {BufferFormat::kBc4SNorm, 0, 8, 0, true, false},
-    {BufferFormat::kBc5UNorm, 0, 16, 0, true, false},
-    {BufferFormat::kBc5SNorm, 0, 16, 0, true, false},
-    {BufferFormat::kBc6UFloat, 0, 16, 0, true, false},
-    {BufferFormat::kBc6SFloat, 0, 16, 0, true, false},
-    {BufferFormat::kBc7UNorm, 0, 16, 0, true, false},
-    {BufferFormat::kBc7Srgb, 0, 16, 0, true, false},
+	{BufferFormat::k8UNorm, 1, 0, 1, true, false},
+	{BufferFormat::k8SNorm, 0, 0, 1, false, false},
+	{BufferFormat::k8UInt, 1, 0, 1, true, true},
+	{BufferFormat::k16UNorm, 2, 0, 2, true, false},
+	{BufferFormat::k16SNorm, 2, 0, 2, true, false},
+	{BufferFormat::k16UInt, 2, 0, 2, true, true},
+	{BufferFormat::k16SInt, 2, 0, 2, true, false, true},
+	{BufferFormat::k16Float, 2, 0, 2, true, false},
+	{BufferFormat::k8_8UNorm, 2, 0, 2, true, false},
+	{BufferFormat::k8_8SNorm, 2, 0, 2, true, false},
+	{BufferFormat::k8_8UInt, 2, 0, 2, true, true},
+	{BufferFormat::k8_8SInt, 2, 0, 2, true, false, true},
+	{BufferFormat::k32UInt, 4, 0, 4, true, true},
+	{BufferFormat::k32SInt, 4, 0, 4, true, false, true},
+	{BufferFormat::k32Float, 4, 0, 4, true, false},
+	{BufferFormat::k16_16UNorm, 4, 0, 4, true, false},
+	{BufferFormat::k16_16SNorm, 4, 0, 4, true, false},
+	{BufferFormat::k16_16UInt, 4, 0, 4, true, true},
+	{BufferFormat::k16_16SInt, 4, 0, 4, true, false, true},
+	{BufferFormat::k16_16Float, 4, 0, 4, true, false},
+	{BufferFormat::k11_11_10UInt, 4, 0, 4, true, true},
+	{BufferFormat::k11_11_10Float, 4, 0, 4, true, false},
+	{BufferFormat::k10_10_10_2UNorm, 4, 0, 4, true, false},
+	{BufferFormat::k10_10_10_2UInt, 4, 0, 4, true, true},
+	{BufferFormat::k8_8_8_8UNorm, 4, 0, 4, true, false},
+	{BufferFormat::k8_8_8_8SNorm, 4, 0, 4, true, false},
+	{BufferFormat::k8_8_8_8UInt, 4, 0, 4, true, true},
+	{BufferFormat::k8_8_8_8SInt, 4, 0, 4, true, false, true},
+	{BufferFormat::k32_32UInt, 8, 0, 8, true, true},
+	{BufferFormat::k32_32SInt, 8, 0, 8, true, false, true},
+	{BufferFormat::k32_32Float, 8, 0, 8, true, false},
+	{BufferFormat::k16_16_16_16UNorm, 8, 0, 8, true, false},
+	{BufferFormat::k16_16_16_16SNorm, 8, 0, 8, true, false},
+	{BufferFormat::k16_16_16_16UInt, 8, 0, 8, true, true},
+	{BufferFormat::k16_16_16_16SInt, 8, 0, 8, true, false, true},
+	{BufferFormat::k16_16_16_16Float, 8, 0, 8, true, false},
+	{BufferFormat::k32_32_32UInt, 12, 0, 12, true, true},
+	{BufferFormat::k32_32_32SInt, 12, 0, 12, true, false, true},
+	{BufferFormat::k32_32_32Float, 12, 0, 12, true, false},
+	{BufferFormat::k32_32_32_32UInt, 16, 0, 16, true, true},
+	{BufferFormat::k32_32_32_32SInt, 16, 0, 16, true, false, true},
+	{BufferFormat::k32_32_32_32Float, 16, 0, 16, true, false},
+	{BufferFormat::k8Srgb, 1, 0, 0, true, false},
+	{BufferFormat::k8_8Srgb, 2, 0, 0, true, false},
+	{BufferFormat::k8_8_8_8Srgb, 4, 0, 4, true, false},
+	{BufferFormat::k9_9_9_5Float, 4, 0, 0, true, false},
+	{BufferFormat::k5_6_5UNorm, 2, 0, 2, true, false},
+	{BufferFormat::k5_5_5_1UNorm, 2, 0, 2, true, false},
+	{BufferFormat::k1_5_5_5UNorm, 0, 0, 2, false, false},
+	{BufferFormat::k4_4_4_4UNorm, 2, 0, 2, true, false},
+	{BufferFormat::kFmask8_S2_F1, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S4_F1, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S8_F1, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S2_F2, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S4_F2, 1, 0, 1, true, false},
+	{BufferFormat::kFmask8_S4_F4, 1, 0, 1, true, false},
+	{BufferFormat::kFmask16_S16_F1, 2, 0, 2, true, false},
+	{BufferFormat::kFmask16_S8_F2, 2, 0, 2, true, false},
+	{BufferFormat::kFmask32_S16_F2, 4, 0, 4, true, false},
+	{BufferFormat::kFmask32_S8_F4, 4, 0, 4, true, false},
+	{BufferFormat::kFmask32_S8_F8, 4, 0, 4, true, false},
+	{BufferFormat::kFmask64_S16_F4, 8, 0, 8, true, false},
+	{BufferFormat::kFmask64_S16_F8, 8, 0, 8, true, false},
+	{BufferFormat::kBc1UNorm, 0, 8, 0, true, false},
+	{BufferFormat::kBc1Srgb, 0, 8, 0, true, false},
+	{BufferFormat::kBc2UNorm, 0, 16, 0, true, false},
+	{BufferFormat::kBc2Srgb, 0, 16, 0, true, false},
+	{BufferFormat::kBc3UNorm, 0, 16, 0, true, false},
+	{BufferFormat::kBc3Srgb, 0, 16, 0, true, false},
+	{BufferFormat::kBc4UNorm, 0, 8, 0, true, false},
+	{BufferFormat::kBc4SNorm, 0, 8, 0, true, false},
+	{BufferFormat::kBc5UNorm, 0, 16, 0, true, false},
+	{BufferFormat::kBc5SNorm, 0, 16, 0, true, false},
+	{BufferFormat::kBc6UFloat, 0, 16, 0, true, false},
+	{BufferFormat::kBc6SFloat, 0, 16, 0, true, false},
+	{BufferFormat::kBc7UNorm, 0, 16, 0, true, false},
+	{BufferFormat::kBc7Srgb, 0, 16, 0, true, false},
 };
 
 constexpr uint8_t RenderTargetTypeBit(ChannelType type) {
@@ -201,18 +215,23 @@ uint32_t RenderTargetBytesPerElement(BufferFormat format) {
 	return info != nullptr ? info->render_target_bytes_per_element : 0;
 }
 
-bool IsSupportedTextureFormat(BufferFormat format) {
-	const auto* info = FindFormatInfo(format);
-	return info != nullptr && info->supported_sampled_texture_format;
-}
-
-bool IsUintTextureFormat(BufferFormat format) {
-	const auto* info = FindFormatInfo(format);
-	return info != nullptr && info->unsigned_integer_texture_format;
-}
-
 bool IsFmaskTextureFormat(BufferFormat format) {
-	return format == BufferFormat::kFmask8_S4_F4;
+	return format >= BufferFormat::kFmask8_S2_F1 && format <= BufferFormat::kFmask64_S16_F8;
+}
+
+TextureNumericClass SampledTextureNumericClass(BufferFormat format) {
+	const auto* info = FindFormatInfo(format);
+	if (info == nullptr || !info->sampled_texture) {
+		return TextureNumericClass::Unsupported;
+	}
+	if (info->sint_texture) {
+		return TextureNumericClass::Sint;
+	}
+	return info->uint_texture ? TextureNumericClass::Uint : TextureNumericClass::Float;
+}
+
+BufferFormat RemapTextureFormat(BufferFormat format) {
+	return format == BufferFormat::k11_11_10UInt ? BufferFormat::k32UInt : format;
 }
 
 } // namespace Libs::Graphics::Prospero

src/graphics/guest_gpu/gpu_format.h

@@ -37,13 +37,15 @@ struct RenderTargetFormatEncoding {
 	}
 };
 
+enum class TextureNumericClass { Unsupported, Float, Uint, Sint };
+
 RenderTargetFormatEncoding ResolveRenderTargetFormat(ChannelLayout layout, ChannelType type);
 uint32_t                   NumBytesPerElement(BufferFormat format);
 uint32_t                   BlockCompressedBytesPerBlock(BufferFormat format);
 uint32_t                   RenderTargetBytesPerElement(BufferFormat format);
-bool                       IsSupportedTextureFormat(BufferFormat format);
-bool                       IsUintTextureFormat(BufferFormat format);
 bool                       IsFmaskTextureFormat(BufferFormat format);
+TextureNumericClass        SampledTextureNumericClass(BufferFormat format);
+BufferFormat               RemapTextureFormat(BufferFormat format);
 
 } // namespace Libs::Graphics::Prospero
 

src/graphics/guest_gpu/graphicsRun.cpp

@@ -13,7 +13,6 @@
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/renderer/sync.h"
-#include "graphics/presentation/renderDoc.h"
 #include "graphics/presentation/videoOut.h"
 #include "graphics/presentation/window.h"
 #include "graphics/shader/shader.h"
@@ -27,6 +26,7 @@
 #include <cstdio>
 #include <deque>
 #include <memory>
+#include <mutex>
 #include <semaphore>
 #include <thread>
 #include <vector>
@@ -39,6 +39,21 @@ static thread_local bool              g_gpu_mutex_owned   = false;
 static thread_local bool              g_gpu_thread        = false;
 static thread_local GuestGpu*         g_gpu_state         = nullptr;
 
+struct DrawIndirectArgs {
+	uint32_t vertex_count_per_instance;
+	uint32_t instance_count;
+	uint32_t start_vertex_location;
+	uint32_t start_instance_location;
+};
+
+struct DrawIndexedIndirectArgs {
+	uint32_t index_count_per_instance;
+	uint32_t instance_count;
+	uint32_t start_index_location;
+	uint32_t base_vertex_location;
+	uint32_t start_instance_location;
+};
+
 class GpuMutexLock final {
 public:
 	explicit GpuMutexLock(Common::Mutex& mutex): m_mutex(mutex) {
@@ -65,17 +80,10 @@ static bool GraphicsRunDebugDumpEnabled() {
 	       Config::GetPrintfDirection() != Config::OutputDirection::Silent;
 }
 
-GuestGpu::OwnedCmdBuffer::OwnedCmdBuffer(const uint32_t* data, uint32_t count) {
-	EXIT_IF(data == nullptr && count != 0);
-	if (count != 0) {
-		m_words.assign(data, data + count);
-	}
-}
-
 GuestGpu::GuestGpu(RenderContext& renderer): m_renderer(renderer) {
 	EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
 	GraphicsInitJmpTables();
-	m_gfx_cp = std::make_unique<CommandProcessor>(renderer);
+	m_gfx_cp = std::make_unique<CommandProcessor>(renderer, 0);
 	m_thread = std::jthread(ThreadRun, this);
 }
 
@@ -147,39 +155,33 @@ void GuestGpu::SendCommandSync(Common::UniqueFunction<void>&& command) {
 	done.acquire();
 }
 
-void GuestGpu::Submit(uint32_t* cmd_draw_buffer, uint32_t num_draw_dw, uint32_t* cmd_const_buffer,
-                      uint32_t num_const_dw, bool trigger_agc_interrupt_on_done) {
-	if (num_draw_dw == 0) {
+void GuestGpu::Submit(std::span<const uint32_t> draw_commands,
+                      std::span<const uint32_t> constant_commands) {
+	if (draw_commands.empty()) {
 		return;
 	}
-	EXIT_IF(cmd_draw_buffer == nullptr);
 	GpuMutexLock lock(m_submission_mutex);
 	Submission   submission;
-	submission.type                          = SubmissionType::Graphics;
-	submission.queue_id                      = 0;
-	submission.commands                      = OwnedCmdBuffer(cmd_draw_buffer, num_draw_dw);
-	submission.constant_commands             = OwnedCmdBuffer(cmd_const_buffer, num_const_dw);
-	submission.trigger_agc_interrupt_on_done = trigger_agc_interrupt_on_done;
-	submission.reset_processor               = m_graphics_done;
-	m_graphics_done                          = false;
+	submission.type              = SubmissionType::Graphics;
+	submission.queue_id          = 0;
+	submission.commands          = draw_commands;
+	submission.constant_commands = constant_commands;
+	submission.reset_processor   = m_graphics_done;
+	m_graphics_done              = false;
 	Enqueue(std::move(submission));
 }
 
-void GuestGpu::SubmitCompute(uint32_t queue, uint32_t* cmd_buffer, uint32_t num_dw,
-                             bool trigger_agc_interrupt_on_done) {
-	EXIT_IF(cmd_buffer == nullptr || num_dw == 0);
+void GuestGpu::SubmitCompute(uint32_t queue, std::span<const uint32_t> commands) {
+	EXIT_IF(commands.empty());
 	GpuMutexLock lock(m_submission_mutex);
 
-	constexpr uint32_t compute_queue_base = 0x20u;
-	EXIT_NOT_IMPLEMENTED(queue < compute_queue_base ||
-	                     queue >= compute_queue_base + ComputeQueueCount);
+	EXIT_NOT_IMPLEMENTED(queue < ComputeQueueBase || queue >= ComputeQueueBase + ComputeQueueCount);
 
-	const auto compute_queue = queue - compute_queue_base;
+	const auto compute_queue = queue - ComputeQueueBase;
 	Submission submission;
-	submission.type                          = SubmissionType::Compute;
-	submission.queue_id                      = 1 + compute_queue;
-	submission.commands                      = OwnedCmdBuffer(cmd_buffer, num_dw);
-	submission.trigger_agc_interrupt_on_done = trigger_agc_interrupt_on_done;
+	submission.type     = SubmissionType::Compute;
+	submission.queue_id = 1 + compute_queue;
+	submission.commands = commands;
 	Enqueue(std::move(submission));
 }
 
@@ -199,12 +201,6 @@ void GuestGpu::Done() {
 	if (!IsGpuThread()) {
 		WaitForIdle();
 	}
-	if (RenderDocCaptureInProgress()) {
-		SendCommandSync([this] {
-			Common::LockGuard render_lock(m_renderer.GetMutex());
-			RenderDocEndCapture();
-		});
-	}
 	m_graphics_done = true;
 	m_done_num++;
 }
@@ -220,7 +216,7 @@ CommandProcessor& GuestGpu::GetProcessor(uint32_t queue_id) {
 	}
 	auto& processor = m_compute_cp[queue_id - 1];
 	if (processor == nullptr) {
-		processor = std::make_unique<CommandProcessor>(m_renderer);
+		processor = std::make_unique<CommandProcessor>(m_renderer, ComputeQueueBase + queue_id - 1);
 	}
 	return *processor;
 }
@@ -273,8 +269,7 @@ void CommandProcessor::BufferFlush() {
 }
 
 void CommandProcessor::BufferFlushAndWait() {
-	auto& submitted = GetScheduler().FlushAndGetSubmitted();
-	submitted.WaitForFenceOnly();
+	GetScheduler().FlushAndWait();
 }
 
 void CommandProcessor::BufferWait() {
@@ -359,10 +354,8 @@ template void CommandProcessor::WaitRegMem<uint64_t>(uint32_t, const uint64_t*,
 
 void CommandProcessor::WriteData(uint32_t* dst, const uint32_t* src, uint32_t dw_num,
                                  uint32_t write_control) {
-	const uint32_t dst_sel      = ((write_control >> 30u) & 0x1u) | ((write_control >> 7u) & 0x1eu);
-	const uint32_t cache_policy = (write_control >> 25u) & 0x3u;
-	const uint32_t increment    = (write_control >> 16u) & 0x1u;
-	const uint32_t write_confirm = (write_control >> 20u) & 0x1u;
+	const uint32_t dst_sel = ((write_control >> 30u) & 0x1u) | ((write_control >> 7u) & 0x1eu);
+	const bool     write_one_address = ((write_control >> 16u) & 0x1u) != 0;
 
 	switch (dst_sel) {
 		case 0:
@@ -372,16 +365,17 @@ void CommandProcessor::WriteData(uint32_t* dst, const uint32_t* src, uint32_t dw
 		case 6: break;
 		default: EXIT("unsupported writeData destination selector 0x%02" PRIx32 "\n", dst_sel);
 	}
-	EXIT_NOT_IMPLEMENTED(increment != 0);
-
-	if (cache_policy > 3 || write_confirm > 1) {
-		LOGF("\t warning: unexpected write_data control 0x%08" PRIx32 "\n", write_control);
-	}
 	if (dw_num == 0) {
 		return;
 	}
 
-	memcpy(dst, src, static_cast<size_t>(dw_num) * sizeof(uint32_t));
+	if (write_one_address) {
+		for (uint32_t i = 0; i < dw_num; i++) {
+			dst[0] = src[i];
+		}
+	} else {
+		memcpy(dst, src, static_cast<size_t>(dw_num) * sizeof(uint32_t));
+	}
 }
 
 void CommandProcessor::WriteReferenceClock(uint64_t dst_address, uint32_t num_bytes) {
@@ -566,10 +560,6 @@ void GuestGpu::ThreadRun(void* data) {
 
 bool GuestGpu::Process(Submission& submission) {
 	const bool first_slice = !submission.started;
-	if (first_slice && RenderDocCaptureRequested()) {
-		Common::LockGuard render_lock(m_renderer.GetMutex());
-		RenderDocStartCapture();
-	}
 	auto& cp = GetProcessor(submission.queue_id);
 
 	if (first_slice && submission.reset_processor) {
@@ -589,21 +579,20 @@ bool GuestGpu::Process(Submission& submission) {
 	switch (submission.type) {
 		case SubmissionType::Graphics: {
 			bool progressed = false;
-			submission.constant_complete |= submission.constant_commands.Empty();
+			submission.constant_complete |= submission.constant_commands.empty();
 			for (;;) {
 				bool round_progress = false;
 				if (!submission.constant_complete) {
 					submission.constant_complete =
-					    cp.Process(
-					        submission.constant_execution, submission.constant_commands.Data(),
-					        submission.constant_commands.Size()) == Pm4ProcessResult::Complete;
+					    cp.Process(submission.constant_execution, submission.constant_commands) ==
+					    Pm4ProcessResult::Complete;
 					round_progress |= submission.constant_execution.MadeProgress();
 				}
 				cp.SetCeComplete(submission.constant_complete);
 				if (!submission.command_complete) {
 					submission.command_complete =
-					    cp.Process(submission.command_execution, submission.commands.Data(),
-					               submission.commands.Size()) == Pm4ProcessResult::Complete;
+					    cp.Process(submission.command_execution, submission.commands) ==
+					    Pm4ProcessResult::Complete;
 					round_progress |= submission.command_execution.MadeProgress();
 				}
 				progressed |= round_progress;
@@ -620,14 +609,11 @@ bool GuestGpu::Process(Submission& submission) {
 			} else if (complete) {
 				m_renderer.GetGpuResources().RunGarbageCollector();
 			}
-			if (complete && submission.trigger_agc_interrupt_on_done) {
-				m_renderer.TriggerEopEvent(0);
-			}
 			break;
 		}
 		case SubmissionType::Compute: {
-			const auto      num_dw                  = submission.commands.Size();
-			auto*           buffer                  = submission.commands.Data();
+			const auto      num_dw = static_cast<uint32_t>(submission.commands.size());
+			const auto*     buffer = submission.commands.data();
 			static uint32_t compute_batch_log_count = 0;
 			if (first_slice && num_dw <= 128 && compute_batch_log_count++ < 32) {
 				LOGF("compute direct batch: data=0x%016" PRIx64 ", num_dw=%" PRIu32 "\n",
@@ -639,7 +625,7 @@ bool GuestGpu::Process(Submission& submission) {
 			if (first_slice) {
 				GraphicsDbgDumpDcb("cc", num_dw, buffer);
 			}
-			complete = cp.Process(submission.command_execution, buffer, num_dw) ==
+			complete = cp.Process(submission.command_execution, submission.commands) ==
 			           Pm4ProcessResult::Complete;
 			if (submission.command_execution.MadeProgress()) {
 				if (complete) {
@@ -649,9 +635,6 @@ bool GuestGpu::Process(Submission& submission) {
 			} else if (complete) {
 				m_renderer.GetGpuResources().RunGarbageCollector();
 			}
-			if (complete && submission.trigger_agc_interrupt_on_done) {
-				Sync::TriggerAgcUserInterrupt();
-			}
 			break;
 		}
 		case SubmissionType::FlipPreparation:
@@ -663,12 +646,13 @@ bool GuestGpu::Process(Submission& submission) {
 	return complete;
 }
 
-Pm4ProcessResult CommandProcessor::Process(Pm4Execution& execution, uint32_t* buffer,
-                                           uint32_t size_dw) {
+Pm4ProcessResult CommandProcessor::Process(Pm4Execution&             execution,
+                                           std::span<const uint32_t> commands) {
 	KYTY_PROFILER_BLOCK("CommandProcessor::Process");
 	EXIT_IF(g_current_execution != nullptr);
-	if (execution.m_buffer_stack.empty() && size_dw != 0) {
-		execution.m_buffer_stack.push_back({buffer, size_dw, size_dw, 0});
+	EXIT_IF(commands.size() > UINT32_MAX);
+	if (execution.m_buffer_stack.empty() && !commands.empty()) {
+		execution.m_buffer_stack.push_back({commands});
 	}
 	execution.m_suspended     = false;
 	execution.m_made_progress = false;
@@ -693,14 +677,14 @@ Pm4ProcessResult CommandProcessor::Process(Pm4Execution& execution, uint32_t* bu
 	                                        : Pm4ProcessResult::Blocked;
 }
 
-void CommandProcessor::ProcessIndirectBuffer(uint32_t* buffer, uint32_t size_dw) {
+void CommandProcessor::ProcessIndirectBuffer(std::span<const uint32_t> commands) {
 	EXIT_IF(g_current_execution == nullptr);
-	if (size_dw == 0) {
+	if (commands.empty()) {
 		return;
 	}
 	auto&      execution  = *g_current_execution;
 	const auto stop_depth = execution.m_buffer_stack.size();
-	execution.m_buffer_stack.push_back({buffer, size_dw, size_dw, 0});
+	execution.m_buffer_stack.push_back({commands});
 	ProcessPm4(execution, stop_depth);
 }
 
@@ -716,29 +700,28 @@ void CommandProcessor::ProcessPm4(Pm4Execution& execution, size_t stop_depth) {
 		}
 		const auto buffer_index = execution.m_buffer_stack.size() - 1;
 		auto&      cursor       = execution.m_buffer_stack[buffer_index];
+		EXIT_IF(cursor.offset_dw > cursor.commands.size());
 		if (cursor.deferred_advance_dw != 0) {
-			EXIT_IF(cursor.deferred_advance_dw > cursor.remaining_dw);
-			cursor.next_packet += cursor.deferred_advance_dw;
-			cursor.remaining_dw -= cursor.deferred_advance_dw;
+			EXIT_IF(cursor.deferred_advance_dw > cursor.commands.size() - cursor.offset_dw);
+			cursor.offset_dw += cursor.deferred_advance_dw;
 			cursor.deferred_advance_dw = 0;
 			execution.m_made_progress  = true;
 			continue;
 		}
-		if (cursor.remaining_dw == 0) {
+		if (cursor.offset_dw == cursor.commands.size()) {
 			execution.m_buffer_stack.pop_back();
 			continue;
 		}
 
-		auto* const packet        = cursor.next_packet;
-		const auto  remaining_dw  = cursor.remaining_dw;
-		const auto  total_dw      = cursor.total_dw;
-		const auto  packet_header = packet[0];
-		const auto  opcode        = (packet_header >> 8u) & 0xffu;
+		const auto* const packet        = cursor.commands.data() + cursor.offset_dw;
+		const auto        total_dw      = static_cast<uint32_t>(cursor.commands.size());
+		const auto        remaining_dw  = total_dw - cursor.offset_dw;
+		const auto        packet_header = packet[0];
+		const auto        opcode        = (packet_header >> 8u) & 0xffu;
 		EXIT_NOT_IMPLEMENTED(remaining_dw > total_dw);
 
 		if (packet_header == 0x80000000u) {
-			cursor.next_packet++;
-			cursor.remaining_dw--;
+			cursor.offset_dw++;
 			execution.m_made_progress = true;
 			continue;
 		}
@@ -773,8 +756,7 @@ void CommandProcessor::ProcessPm4(Pm4Execution& execution, size_t stop_depth) {
 					     dst, val, packet[1], packet[2]);
 				}
 			}
-			cursor.next_packet += packet_dw;
-			cursor.remaining_dw -= packet_dw;
+			cursor.offset_dw += packet_dw;
 			execution.m_made_progress = true;
 			continue;
 		}
@@ -807,8 +789,7 @@ void CommandProcessor::ProcessPm4(Pm4Execution& execution, size_t stop_depth) {
 			return;
 		}
 		EXIT_IF(execution.m_buffer_stack.size() != buffer_index + 1);
-		execution.m_buffer_stack[buffer_index].next_packet += packet_dw;
-		execution.m_buffer_stack[buffer_index].remaining_dw -= packet_dw;
+		execution.m_buffer_stack[buffer_index].offset_dw += packet_dw;
 		execution.m_made_progress = true;
 	}
 }
@@ -876,36 +857,21 @@ void CommandProcessor::SetPredication(uint32_t condition, uint32_t op, uint32_t
 	}
 }
 
-void CommandProcessor::DrawIndex(uint32_t index_count, const void* index_addr, uint32_t flags,
-                                 uint32_t type, uint32_t instance_count, const void* object_ids,
-                                 uint32_t render_target_slice_offset, int32_t vertex_offset_add,
-                                 uint32_t first_instance) {
+void CommandProcessor::DrawIndex(DrawIndexArgs args) {
 	CheckBuffer();
 
-	if (instance_count == 0) {
-		instance_count = m_num_instances;
+	args.index_type_and_size = m_index_type_and_size;
+	if (args.instance_count == 0) {
+		args.instance_count = m_num_instances;
 	}
-	if (object_ids != nullptr) {
-		LOGF("\t draw indexed multi-instanced objectIds = 0x%016" PRIx64 "\n",
-		     reinterpret_cast<uint64_t>(object_ids));
+	if (args.base_vertex != 0 || args.first_instance != 0) {
+		LOGF("\t draw indexed offsets: base_vertex = %" PRId32 ", first_instance = %" PRIu32 "\n",
+		     args.base_vertex, args.first_instance);
 	}
-	if (render_target_slice_offset != 0) {
-		LOGF("\t draw render target slice offset = %" PRIu32 "\n", render_target_slice_offset);
-	}
-	if (vertex_offset_add != 0 || first_instance != 0) {
-		LOGF("\t draw indexed offsets: vertex_offset_add = %" PRId32 ", first_instance = %" PRIu32
-		     "\n",
-		     vertex_offset_add, first_instance);
-	}
-	m_renderer.GetRenderExecutor().DrawIndex(
-	    m_submit_id, CurrentBuffer(), m_index_type_and_size, index_count, index_addr, flags, type,
-	    instance_count, render_target_slice_offset, vertex_offset_add, first_instance);
+	m_renderer.GetRenderExecutor().DrawIndex(m_submit_id, CurrentBuffer(), args);
 }
 
-void CommandProcessor::DrawIndexOffset(uint32_t index_offset, uint32_t index_count,
-                                       uint32_t flags) {
-	CheckBuffer();
-
+void CommandProcessor::DrawIndexOffset(uint32_t index_offset, uint32_t index_count) {
 	uint64_t index_size = 0;
 	switch (m_index_type_and_size) {
 		case 0: index_size = 2; break;
@@ -917,25 +883,10 @@ void CommandProcessor::DrawIndexOffset(uint32_t index_offset, uint32_t index_cou
 	auto* index_addr = reinterpret_cast<const void*>(
 	    m_index_base_addr + static_cast<uint64_t>(index_offset) * index_size);
 
-	m_renderer.GetRenderExecutor().DrawIndex(m_submit_id, CurrentBuffer(), m_index_type_and_size,
-	                                         index_count, index_addr, flags, 1, m_num_instances);
+	DrawIndex({.index_count = index_count, .index_addr = index_addr});
 }
 
 void CommandProcessor::DrawIndirect(uint32_t data_offset, uint32_t draw_initiator, bool indexed) {
-	struct DrawIndirectArgs {
-		uint32_t vertex_count_per_instance;
-		uint32_t instance_count;
-		uint32_t start_vertex_location;
-		uint32_t start_instance_location;
-	};
-	struct DrawIndexedIndirectArgs {
-		uint32_t index_count_per_instance;
-		uint32_t instance_count;
-		uint32_t start_index_location;
-		uint32_t base_vertex_location;
-		uint32_t start_instance_location;
-	};
-
 	EXIT_NOT_IMPLEMENTED((draw_initiator & ~0x20u) != 2u);
 	EXIT_NOT_IMPLEMENTED(m_draw_indirect_args_base_addr == 0);
 
@@ -957,8 +908,11 @@ void CommandProcessor::DrawIndirect(uint32_t data_offset, uint32_t draw_initiato
 			}
 		}
 		m_num_instances = args.instance_count;
-		SubmitNonIndexedDraw(args.vertex_count_per_instance, 0, 0, args.start_vertex_location,
-		                     args.start_instance_location);
+		DrawIndexAuto({.vertex_count   = args.vertex_count_per_instance,
+		               .instance_count = args.instance_count,
+		               .first_vertex   = args.start_vertex_location,
+		               .first_instance = args.start_instance_location,
+		               .offset_source  = DrawOffsetSource::IndirectArgs});
 		return;
 	}
 
@@ -999,28 +953,18 @@ void CommandProcessor::DrawIndirect(uint32_t data_offset, uint32_t draw_initiato
 	}
 
 	m_num_instances = args.instance_count;
-	DrawIndex(index_count, index_addr, 0, 1, args.instance_count, nullptr, 0,
-	          static_cast<int32_t>(args.base_vertex_location), args.start_instance_location);
+	DrawIndex({.index_count    = index_count,
+	           .index_addr     = index_addr,
+	           .instance_count = args.instance_count,
+	           .base_vertex    = static_cast<int32_t>(args.base_vertex_location),
+	           .first_instance = args.start_instance_location,
+	           .offset_source  = DrawOffsetSource::IndirectArgs});
 }
 
 void CommandProcessor::DrawIndirectMulti(uint32_t data_offset, uint32_t max_count_or_count,
                                          const volatile uint32_t* count_addr,
                                          uint32_t stride_in_bytes, uint32_t draw_initiator,
                                          bool indexed) {
-	struct DrawIndirectArgs {
-		uint32_t vertex_count_per_instance;
-		uint32_t instance_count;
-		uint32_t start_vertex_location;
-		uint32_t start_instance_location;
-	};
-	struct DrawIndexedIndirectArgs {
-		uint32_t index_count_per_instance;
-		uint32_t instance_count;
-		uint32_t start_index_location;
-		uint32_t base_vertex_location;
-		uint32_t start_instance_location;
-	};
-
 	EXIT_NOT_IMPLEMENTED((draw_initiator & ~0x20u) != 2u);
 	EXIT_NOT_IMPLEMENTED(m_draw_indirect_args_base_addr == 0);
 
@@ -1057,8 +1001,11 @@ void CommandProcessor::DrawIndirectMulti(uint32_t data_offset, uint32_t max_coun
 				}
 			}
 			m_num_instances = args->instance_count;
-			SubmitNonIndexedDraw(args->vertex_count_per_instance, 0, 0, args->start_vertex_location,
-			                     args->start_instance_location);
+			DrawIndexAuto({.vertex_count   = args->vertex_count_per_instance,
+			               .instance_count = args->instance_count,
+			               .first_vertex   = args->start_vertex_location,
+			               .first_instance = args->start_instance_location,
+			               .offset_source  = DrawOffsetSource::IndirectArgs});
 			continue;
 		}
 
@@ -1100,13 +1047,19 @@ void CommandProcessor::DrawIndirectMulti(uint32_t data_offset, uint32_t max_coun
 		}
 
 		m_num_instances = args->instance_count;
-		DrawIndex(index_count, index_addr, 0, 1, args->instance_count, nullptr, 0,
-		          static_cast<int32_t>(args->base_vertex_location), args->start_instance_location);
+		DrawIndex({.index_count    = index_count,
+		           .index_addr     = index_addr,
+		           .instance_count = args->instance_count,
+		           .base_vertex    = static_cast<int32_t>(args->base_vertex_location),
+		           .first_instance = args->start_instance_location,
+		           .offset_source  = DrawOffsetSource::IndirectArgs});
 	}
 }
 
 void CommandProcessor::DispatchDirect(uint32_t thread_group_x, uint32_t thread_group_y,
                                       uint32_t thread_group_z, uint32_t mode) {
+	m_sh_ctx.SetCsWaveSize(Pm4::ComputeWaveSize(mode));
+
 	uint32_t frame_num = 0;
 	// uint32_t local_x   = 1;
 	// uint32_t local_y   = 1;
@@ -1190,19 +1143,13 @@ void CommandProcessor::DispatchIndirect(uint32_t data_offset, uint32_t mode) {
 	DispatchDirect(args->thread_group_x, args->thread_group_y, args->thread_group_z, mode);
 }
 
-void CommandProcessor::DrawIndexAuto(uint32_t index_count, uint32_t flags,
-                                     uint32_t render_target_slice_offset) {
-	SubmitNonIndexedDraw(index_count, flags, render_target_slice_offset, 0, 0);
-}
-
-void CommandProcessor::SubmitNonIndexedDraw(uint32_t vertex_count, uint32_t flags,
-                                            uint32_t render_target_slice_offset,
-                                            uint32_t first_vertex, uint32_t first_instance) {
+void CommandProcessor::DrawIndexAuto(DrawAutoArgs args) {
 	CheckBuffer();
 
-	m_renderer.GetRenderExecutor().DrawAuto(m_submit_id, CurrentBuffer(), vertex_count, flags,
-	                                        render_target_slice_offset, m_num_instances,
-	                                        first_vertex, first_instance);
+	if (args.instance_count == 0) {
+		args.instance_count = m_num_instances;
+	}
+	m_renderer.GetRenderExecutor().DrawAuto(m_submit_id, CurrentBuffer(), args);
 }
 
 void CommandProcessor::WaitFlipDone(uint32_t video_out_handle, uint32_t display_buffer_index) {
@@ -1249,7 +1196,14 @@ void CommandProcessor::WriteAtEndOfPipe(uint32_t cache_policy, uint32_t event_wr
 	switch (interrupt_selector) {
 		case 0x00:
 		case 0x03: with_interrupt = false; break;
-		case 0x01: Sync::TriggerEopEventAtEndOfPipe(CurrentBuffer(), interrupt_context_id); return;
+		case 0x01:
+			if (!IsAsyncComputeQueue()) {
+				Sync::TriggerEopEventAtEndOfPipe(CurrentBuffer(), m_interrupt_event_id,
+				                                 interrupt_context_id);
+				return;
+			}
+			with_interrupt = true;
+			break;
 		case 0x02: with_interrupt = true; break;
 		default: EXIT("unknown interrupt_selector\n");
 	}
@@ -1262,10 +1216,11 @@ void CommandProcessor::WriteAtEndOfPipe(uint32_t cache_policy, uint32_t event_wr
 		if (with_interrupt) {
 			if (with_writeback) {
 				Sync::WriteAtEndOfPipeWithInterruptWriteBack32(m_submit_id, CurrentBuffer(), dst,
-				                                               data, interrupt_context_id);
+				                                               data, m_interrupt_event_id,
+				                                               interrupt_context_id);
 			} else {
 				Sync::WriteAtEndOfPipeWithInterrupt32(m_submit_id, CurrentBuffer(), dst, data,
-				                                      interrupt_context_id);
+				                                      m_interrupt_event_id, interrupt_context_id);
 			}
 		} else if (with_writeback) {
 			Sync::WriteAtEndOfPipeWithWriteBack32(m_submit_id, CurrentBuffer(), dst, data);
@@ -1280,12 +1235,12 @@ void CommandProcessor::WriteAtEndOfPipe(uint32_t cache_policy, uint32_t event_wr
 				if (eop_event_type == 0x2f && cache_action == 0x00 && event_index == 0x06) {
 					auto* dst = static_cast<uint32_t*>(dst_gpu_addr);
 					SynchronizeGpu();
-					Sync::ReadGds(m_renderer.GetBufferCache().GetGdsBuffer(), dst, value & 0xffffu,
+					Sync::ReadGds(*m_renderer.GetBufferCache().GetGdsBuffer(), dst, value & 0xffffu,
 					              value >> 16u);
 					Sync::WriteAtEndOfPipeGds32(m_submit_id, CurrentBuffer(), dst, value & 0xffffu,
 					                            value >> 16u);
 					if (with_interrupt) {
-						m_renderer.TriggerEopEvent(interrupt_context_id);
+						m_renderer.TriggerInterrupt(m_interrupt_event_id, interrupt_context_id);
 					}
 					return;
 				}
@@ -1311,10 +1266,12 @@ void CommandProcessor::WriteAtEndOfPipe(uint32_t cache_policy, uint32_t event_wr
 					if (with_interrupt) {
 						if (with_writeback) {
 							Sync::WriteAtEndOfPipeWithInterruptWriteBack64(
-							    m_submit_id, CurrentBuffer(), dst, value, interrupt_context_id);
+							    m_submit_id, CurrentBuffer(), dst, value, m_interrupt_event_id,
+							    interrupt_context_id);
 						} else {
 							Sync::WriteAtEndOfPipeWithInterrupt64(m_submit_id, CurrentBuffer(), dst,
-							                                      value, interrupt_context_id);
+							                                      value, m_interrupt_event_id,
+							                                      interrupt_context_id);
 						}
 					} else if (with_writeback) {
 						Sync::WriteAtEndOfPipeWithWriteBack64(m_submit_id, CurrentBuffer(), dst,
@@ -1393,26 +1350,35 @@ void CommandProcessor::WriteAtEndOfPipe(uint32_t cache_policy, uint32_t event_wr
 		case 0x04:
 			if constexpr (sizeof(T) == sizeof(uint64_t)) {
 				const auto clock = Sync::ReadReferenceClock();
-				std::memcpy(dst_gpu_addr, &clock, sizeof(clock));
+				auto*      dst   = static_cast<uint64_t*>(dst_gpu_addr);
+				std::memcpy(dst, &clock, sizeof(clock));
 				switch (cache_action) {
 					case 0x00:
-						if (((eop_event_type == 0x04 && event_index == 0x05) ||
-						     (eop_event_type == 0x28 && event_index == 0x00)) &&
-						    !with_interrupt) {
-							Sync::WriteAtEndOfPipeClockCounter(m_submit_id, CurrentBuffer(),
-							                                   static_cast<uint64_t*>(dst_gpu_addr),
-							                                   clock);
+						if ((eop_event_type == 0x04 && event_index == 0x05) ||
+						    (eop_event_type == 0x28 && event_index == 0x00)) {
+							if (with_interrupt) {
+								Sync::WriteAtEndOfPipeWithInterrupt64(
+								    m_submit_id, CurrentBuffer(), dst, clock, m_interrupt_event_id,
+								    interrupt_context_id);
+							} else {
+								Sync::WriteAtEndOfPipeClockCounter(m_submit_id, CurrentBuffer(),
+								                                   dst, clock);
+							}
 							return;
 						}
 						break;
 					case 0x38:
-						if (((eop_event_type == 0x04 &&
-						      (event_index == 0x00 || event_index == 0x05)) ||
-						     (eop_event_type == 0x28 && event_index == 0x00)) &&
-						    !with_interrupt) {
-							Sync::WriteAtEndOfPipeClockCounterWithWriteBack(
-							    m_submit_id, CurrentBuffer(), static_cast<uint64_t*>(dst_gpu_addr),
-							    clock);
+						if ((eop_event_type == 0x04 &&
+						     (event_index == 0x00 || event_index == 0x05)) ||
+						    (eop_event_type == 0x28 && event_index == 0x00)) {
+							if (with_interrupt) {
+								Sync::WriteAtEndOfPipeWithInterruptWriteBack64(
+								    m_submit_id, CurrentBuffer(), dst, clock, m_interrupt_event_id,
+								    interrupt_context_id);
+							} else {
+								Sync::WriteAtEndOfPipeClockCounterWithWriteBack(
+								    m_submit_id, CurrentBuffer(), dst, clock);
+							}
 							return;
 						}
 						break;
@@ -1469,15 +1435,17 @@ void CommandProcessor::EmitGlobalBarrier() {
 void CommandProcessor::TriggerEopEventAtEndOfPipe(uint32_t interrupt_context_id) {
 	CheckBuffer();
 
-	Sync::TriggerEopEventAtEndOfPipe(CurrentBuffer(), interrupt_context_id);
+	Sync::TriggerEopEventAtEndOfPipe(CurrentBuffer(), m_interrupt_event_id, interrupt_context_id);
 }
 
-void CommandProcessor::TriggerEvent(uint32_t event_type, uint32_t event_index) {
+void CommandProcessor::TriggerEvent(uint32_t event_type, uint32_t event_index,
+                                    uint64_t event_address) {
 	if (GraphicsRunDebugDumpEnabled()) {
 		LOGF("CommandProcessor::TriggerEvent()\n"
 		     "\t event_type  = 0x%08" PRIx32 "\n"
-		     "\t event_index = 0x%08" PRIx32 "\n",
-		     event_type, event_index);
+		     "\t event_index = 0x%08" PRIx32 "\n"
+		     "\t address     = 0x%016" PRIx64 "\n",
+		     event_type, event_index, event_address);
 	}
 
 	const auto valid_cache_event_index = event_index == 0x00000000 || event_index == 0x00000007;
@@ -1514,12 +1482,36 @@ void CommandProcessor::TriggerEvent(uint32_t event_type, uint32_t event_index) {
 		case 0x0000001a:
 		case 0x0000001b:
 		case 0x00000038:
-		case 0x00000039:
 		case 0x0000003a:
 			LOGF("\t temporary: ignoring unsupported event_write type 0x%08" PRIx32
 			     ", index 0x%08" PRIx32 "\n",
 			     event_type, event_index);
 			break;
+		case 0x00000039: {
+			if (event_index != 0x00000001 || event_address == 0 || (event_address & 0x7u) != 0) {
+				EXIT("invalid occlusion-counter dump: index=0x%08" PRIx32 ", address=0x%016" PRIx64
+				     "\n",
+				     event_index, event_address);
+			}
+			static std::once_flag warning_once;
+			std::call_once(warning_once, [] {
+				std::printf("Warning: game uses occlusion queries, which are currently treated as "
+				            "always visible; GPU usage may be higher and FPS may be lower.\n");
+			});
+
+			// Until host occlusion queries are implemented, publish an always-visible result. The
+			// PS5 layout contains one interleaved begin/end pair per DB, and bit 63 marks a result
+			// ready.
+			constexpr uint64_t ready_bit    = 1ull << 63u;
+			constexpr uint64_t counter_mask = ready_bit - 1u;
+			auto*              results      = reinterpret_cast<volatile uint64_t*>(event_address);
+			const auto         value        = ready_bit | m_synthetic_occlusion_counter;
+			for (uint32_t db = 0; db < 16u; db++) {
+				results[db * 2u] = value;
+			}
+			m_synthetic_occlusion_counter = (m_synthetic_occlusion_counter + 1u) & counter_mask;
+			break;
+		}
 		default:
 			EXIT("unknown event type: 0x%08" PRIx32 ", 0x%08" PRIx32 "\n", event_type, event_index);
 	}
@@ -1582,7 +1574,7 @@ void CommandProcessor::FlipWithInterrupt(uint32_t eop_event_type, uint32_t cache
 	                                         m_flip.flip_arg);
 	Sync::WriteAtEndOfPipeWithInterruptWriteBackFlip32(
 	    m_submit_id, command, static_cast<uint32_t*>(dst_gpu_addr), value, m_flip.handle,
-	    m_flip.index, m_flip.flip_mode, m_flip.flip_arg, request);
+	    m_flip.index, m_flip.flip_mode, m_flip.flip_arg, request, m_interrupt_event_id);
 	GetScheduler().Flush();
 }
 
@@ -1603,7 +1595,7 @@ void CommandProcessor::PrepareCpuFlip(uint64_t request_id) {
 }
 
 void CommandProcessor::SynchronizeGpu() {
-	GetScheduler().FinishCurrent();
+	GetScheduler().Finish();
 }
 
 bool GuestGpu::IsGpuThread() noexcept {

src/graphics/guest_gpu/graphicsRun.h

@@ -13,8 +13,8 @@
 #include <deque>
 #include <memory>
 #include <mutex>
+#include <span>
 #include <thread>
-#include <vector>
 
 namespace Libs::Graphics {
 
@@ -31,12 +31,12 @@ public:
 	void               SendCommand(Common::UniqueFunction<void>&& command);
 	void               SendCommandSync(Common::UniqueFunction<void>&& command);
 
-	void Submit(uint32_t* draw_commands, uint32_t draw_size_dw, uint32_t* constant_commands,
-	            uint32_t constant_size_dw, bool trigger_agc_interrupt_on_done = false);
-	void SubmitCompute(uint32_t queue, uint32_t* commands, uint32_t size_dw,
-	                   bool trigger_agc_interrupt_on_done = false);
-	void SubmitFlipPreparation(uint64_t request_id);
-	void Done();
+	// Submitted command memory is borrowed and must remain valid until GPU execution completes.
+	void              Submit(std::span<const uint32_t> draw_commands,
+	                         std::span<const uint32_t> constant_commands);
+	void              SubmitCompute(uint32_t queue, std::span<const uint32_t> commands);
+	void              SubmitFlipPreparation(uint64_t request_id);
+	void              Done();
 	[[nodiscard]] int GetFrameNum() const;
 
 	[[nodiscard]] static bool IsGpuThread() noexcept;
@@ -45,39 +45,24 @@ private:
 	static constexpr uint32_t ComputePipeCount     = 7;
 	static constexpr uint32_t QueuesPerComputePipe = 8;
 	static constexpr uint32_t ComputeQueueCount    = ComputePipeCount * QueuesPerComputePipe;
+	static constexpr uint32_t ComputeQueueBase     = 0x20;
 	static constexpr uint32_t QueueCount           = 1 + ComputeQueueCount;
 
-	class OwnedCmdBuffer {
-	public:
-		OwnedCmdBuffer() = default;
-		explicit OwnedCmdBuffer(const uint32_t* data, uint32_t count);
-
-		[[nodiscard]] bool     Empty() const noexcept { return m_words.empty(); }
-		[[nodiscard]] uint32_t Size() const noexcept {
-			return static_cast<uint32_t>(m_words.size());
-		}
-		[[nodiscard]] uint32_t* Data() noexcept { return m_words.data(); }
-
-	private:
-		std::vector<uint32_t> m_words;
-	};
-
 	enum class SubmissionType { Graphics, Compute, FlipPreparation };
 
 	struct Submission {
-		SubmissionType type     = SubmissionType::Graphics;
-		uint32_t       queue_id = 0;
-		OwnedCmdBuffer commands;
-		OwnedCmdBuffer constant_commands;
-		Pm4Execution   command_execution;
-		Pm4Execution   constant_execution;
-		bool           trigger_agc_interrupt_on_done = false;
-		bool           reset_processor               = false;
-		bool           started                       = false;
-		bool           command_complete              = false;
-		bool           constant_complete             = false;
-		bool           blocked                       = false;
-		uint64_t       flip_request_id               = 0;
+		SubmissionType            type     = SubmissionType::Graphics;
+		uint32_t                  queue_id = 0;
+		std::span<const uint32_t> commands;
+		std::span<const uint32_t> constant_commands;
+		Pm4Execution              command_execution;
+		Pm4Execution              constant_execution;
+		bool                      reset_processor   = false;
+		bool                      started           = false;
+		bool                      command_complete  = false;
+		bool                      constant_complete = false;
+		bool                      blocked           = false;
+		uint64_t                  flip_request_id   = 0;
 	};
 
 	void              Enqueue(Submission submission);

src/graphics/guest_gpu/hardwareContext.h

@@ -11,15 +11,6 @@ struct ColorBase {
 	uint64_t addr = 0;
 };
 
-struct ColorPitch {
-	uint32_t pitch_div8_minus1       = 0;
-	uint32_t fmask_pitch_div8_minus1 = 0;
-};
-
-struct ColorSlice {
-	uint32_t slice_div64_minus1 = 0;
-};
-
 struct ColorView {
 	uint32_t base_array_slice_index = 0;
 	uint32_t last_array_slice_index = 0;
@@ -36,20 +27,15 @@ struct ColorInfo {
 	bool                    blend_clamp                    = false;
 	bool                    blend_bypass                   = false;
 	bool                    round_mode                     = false;
-	uint32_t                cmask_is_linear                = 0;
-	uint32_t                cmask_addr_type                = 0;
-	bool                    alt_tile_mode                  = false;
 	Prospero::ChannelLayout format                         = Prospero::ChannelLayout::kInvalid;
 	Prospero::ChannelType   channel_type                   = Prospero::ChannelType::kUNorm;
 	Prospero::ChannelOrder  channel_order                  = Prospero::ChannelOrder::kStandard;
 };
 
 struct ColorAttrib {
-	bool               force_dest_alpha_to_one = false;
-	Prospero::TileMode tile_mode               = Prospero::TileMode::kLinear;
-	Prospero::TileMode fmask_tile_mode         = Prospero::TileMode::kLinear;
-	uint32_t           num_samples             = 0;
-	uint32_t           num_fragments           = 0;
+	bool     force_dest_alpha_to_one = false;
+	uint32_t num_samples             = 0;
+	uint32_t num_fragments           = 0;
 };
 
 struct ColorAttrib2 {
@@ -59,41 +45,33 @@ struct ColorAttrib2 {
 };
 
 struct ColorAttrib3 {
-	uint32_t           depth              = 0;
-	Prospero::TileMode tile_mode          = Prospero::TileMode::kLinear;
-	uint32_t           dimension          = 0;
-	bool               cmask_pipe_aligned = false;
-	bool               dcc_pipe_aligned   = false;
+	uint32_t           depth                        = 0;
+	Prospero::TileMode tile_mode                    = Prospero::TileMode::kRenderTarget;
+	uint32_t           dimension                    = 0;
+	bool               metadata_pipe_aligned        = false;
+	bool               write_vrs_rate_hint_to_cmask = false;
 };
 
 struct ColorDccControl {
-	uint32_t max_uncompressed_block_size    = 0;
-	uint32_t max_compressed_block_size      = 0;
-	uint32_t min_compressed_block_size      = 0;
-	uint32_t color_transform                = 0;
-	bool     dcc_clear_key_enable           = false;
-	bool     overwrite_combiner_disable     = false;
-	bool     independent_64b_blocks         = false;
-	bool     independent_128b_blocks        = false;
-	bool     data_write_on_dcc_clear_to_reg = false;
+	enum class IndependentBlockSize : uint8_t { Disabled, Bytes64, Bytes128 };
+
+	uint32_t             max_uncompressed_block_size    = 2;
+	uint32_t             max_compressed_block_size      = 2;
+	uint32_t             color_transform                = 0;
+	bool                 dcc_clear_key_enable           = false;
+	bool                 overwrite_combiner_disable     = false;
+	IndependentBlockSize independent_block_size         = IndependentBlockSize::Disabled;
+	bool                 data_write_on_dcc_clear_to_reg = false;
 };
 
 struct ColorCmask {
 	uint64_t addr = 0;
 };
 
-struct ColorCmaskSlice {
-	uint32_t slice_minus1 = 0;
-};
-
 struct ColorFmask {
 	uint64_t addr = 0;
 };
 
-struct ColorFmaskSlice {
-	uint32_t slice_minus1 = 0;
-};
-
 struct ColorClearWord0 {
 	uint32_t word0 = 0;
 };
@@ -106,15 +84,8 @@ struct ColorDccAddr {
 	uint64_t addr = 0;
 };
 
-struct ColorSize {
-	uint32_t width  = 0;
-	uint32_t height = 0;
-};
-
 struct RenderTarget {
 	ColorBase       base;
-	ColorPitch      pitch;
-	ColorSlice      slice;
 	ColorView       view;
 	ColorInfo       info;
 	ColorAttrib     attrib;
@@ -122,13 +93,10 @@ struct RenderTarget {
 	ColorAttrib3    attrib3;
 	ColorDccControl dcc;
 	ColorCmask      cmask;
-	ColorCmaskSlice cmask_slice;
 	ColorFmask      fmask;
-	ColorFmaskSlice fmask_slice;
 	ColorClearWord0 clear_word0;
 	ColorClearWord1 clear_word1;
 	ColorDccAddr    dcc_addr;
-	ColorSize       size;
 };
 
 struct DepthZInfo {
@@ -136,7 +104,7 @@ struct DepthZInfo {
 	uint32_t                                    num_samples = 0;
 	Prospero::TextureCompatiblePlaneCompression texture_compatibility =
 	    Prospero::TextureCompatiblePlaneCompression::kDisable;
-	Prospero::ZCompareBase z_compare_base     = Prospero::ZCompareBase::kZMin;
+	Prospero::ZCompareBase z_compare_base     = Prospero::ZCompareBase::kZMax;
 	bool                   htile_acceleration = false;
 	bool                   expclear_enabled   = false;
 	bool                   partially_resident = false;
@@ -171,7 +139,7 @@ struct DepthStencilInfo {
 	Prospero::TextureCompatibleStencil texture_compatibility =
 	    Prospero::TextureCompatibleStencil::kDisable;
 	bool expclear_enabled       = false;
-	bool htile_stencil_disabled = false;
+	bool htile_stencil_disabled = true;
 	bool partially_resident     = false;
 
 	[[nodiscard]] static DepthStencilInfo Decode(uint32_t value) {
@@ -194,16 +162,6 @@ struct DepthStencilInfo {
 	}
 };
 
-struct DepthRenderTargetDepthInfo {
-	uint32_t addr5_swizzle_mask = 0;
-	uint32_t array_mode         = 0;
-	uint32_t pipe_config        = 0;
-	uint32_t bank_width         = 0;
-	uint32_t bank_height        = 0;
-	uint32_t macro_tile_aspect  = 0;
-	uint32_t num_banks          = 0;
-};
-
 struct DepthDepthView {
 	uint32_t slice_start           = 0;
 	uint32_t slice_max             = 0;
@@ -218,36 +176,18 @@ struct DepthDepthSizeXY {
 	bool     valid = false;
 };
 
-struct DepthRenderTargetHTileSurface {
-	uint32_t linear                  = 0;
-	uint32_t full_cache              = 0;
-	uint32_t htile_uses_preload_win  = 0;
-	uint32_t preload                 = 0;
-	uint32_t prefetch_width          = 0;
-	uint32_t prefetch_height         = 0;
-	uint32_t dst_outside_zero_to_one = 0;
-};
-
 struct DepthRenderTarget {
-	DepthZInfo                    z_info;
-	DepthStencilInfo              stencil_info;
-	DepthDepthView                depth_view;
-	DepthDepthSizeXY              size;
-	DepthRenderTargetDepthInfo    depth_info;
-	DepthRenderTargetHTileSurface htile_surface;
+	DepthZInfo       z_info;
+	DepthStencilInfo stencil_info;
+	DepthDepthView   depth_view;
+	DepthDepthSizeXY size;
 
 	uint64_t z_read_base_addr        = 0;
 	uint64_t stencil_read_base_addr  = 0;
 	uint64_t z_write_base_addr       = 0;
 	uint64_t stencil_write_base_addr = 0;
-	uint32_t pitch_div8_minus1       = 0;
-	uint32_t height_div8_minus1      = 0;
-	bool     pitch_height_valid      = false;
-	uint32_t slice_div64_minus1      = 0;
 	uint64_t htile_data_base_addr    = 0;
-	uint32_t width                   = 0;
-	uint32_t height                  = 0;
-	bool     width_height_valid      = false;
+	uint8_t  shading_rate_encoding   = 0;
 };
 
 struct RenderControl {
@@ -256,6 +196,8 @@ struct RenderControl {
 	bool    resummarize_enable       = false;
 	bool    stencil_compress_disable = false;
 	bool    depth_compress_disable   = false;
+	bool    copy_depth_to_color      = false;
+	bool    copy_stencil_to_color    = false;
 	bool    copy_centroid            = false;
 	uint8_t copy_sample              = 0;
 };
@@ -299,16 +241,14 @@ struct ClipControl {
 };
 
 struct DepthControl {
-	bool    stencil_enable                     = false;
-	bool    z_enable                           = false;
-	bool    z_write_enable                     = false;
-	bool    depth_bounds_enable                = false;
-	uint8_t zfunc                              = 0;
-	bool    backface_enable                    = false;
-	uint8_t stencilfunc                        = 0;
-	uint8_t stencilfunc_bf                     = 0;
-	bool    color_writes_on_depth_fail_enable  = false;
-	bool    color_writes_on_depth_pass_disable = false;
+	bool    stencil_enable      = false;
+	bool    z_enable            = false;
+	bool    z_write_enable      = false;
+	bool    depth_bounds_enable = false;
+	uint8_t zfunc               = 0;
+	bool    backface_enable     = false;
+	uint8_t stencilfunc         = 0;
+	uint8_t stencilfunc_bf      = 0;
 };
 
 struct StencilControl {
@@ -422,16 +362,16 @@ struct AaConfig {
 struct Viewport {
 	float zmin                                  = 0.0f;
 	float zmax                                  = 0.0f;
-	float xscale                                = 0.0f;
+	float xscale                                = 1.0f;
 	float xoffset                               = 0.0f;
-	float yscale                                = 0.0f;
+	float yscale                                = 1.0f;
 	float yoffset                               = 0.0f;
-	float zscale                                = 0.0f;
+	float zscale                                = 1.0f;
 	float zoffset                               = 0.0f;
 	int   viewport_scissor_left                 = 0;
 	int   viewport_scissor_top                  = 0;
-	int   viewport_scissor_right                = 0;
-	int   viewport_scissor_bottom               = 0;
+	int   viewport_scissor_right                = 16384;
+	int   viewport_scissor_bottom               = 16384;
 	bool  viewport_scissor_window_offset_enable = false;
 };
 
@@ -440,26 +380,26 @@ struct ScreenViewport {
 	uint32_t transform_control                    = 1087;
 	int      screen_scissor_left                  = 0;
 	int      screen_scissor_top                   = 0;
-	int      screen_scissor_right                 = 0;
-	int      screen_scissor_bottom                = 0;
+	int      screen_scissor_right                 = 16384;
+	int      screen_scissor_bottom                = 16384;
 	int      window_scissor_left                  = 0;
 	int      window_scissor_top                   = 0;
-	int      window_scissor_right                 = 0;
-	int      window_scissor_bottom                = 0;
+	int      window_scissor_right                 = 16384;
+	int      window_scissor_bottom                = 16384;
 	bool     window_scissor_window_offset_enable  = false;
 	int      generic_scissor_left                 = 0;
 	int      generic_scissor_top                  = 0;
-	int      generic_scissor_right                = 0;
-	int      generic_scissor_bottom               = 0;
+	int      generic_scissor_right                = 16384;
+	int      generic_scissor_bottom               = 16384;
 	bool     generic_scissor_window_offset_enable = false;
 	int      window_offset_x                      = 0;
 	int      window_offset_y                      = 0;
 	uint32_t hw_offset_x                          = 0;
 	uint32_t hw_offset_y                          = 0;
-	float    guard_band_horz_clip                 = 0.0f;
-	float    guard_band_vert_clip                 = 0.0f;
-	float    guard_band_horz_discard              = 0.0f;
-	float    guard_band_vert_discard              = 0.0f;
+	float    guard_band_horz_clip                 = 1.0f;
+	float    guard_band_vert_clip                 = 1.0f;
+	float    guard_band_horz_discard              = 1.0f;
+	float    guard_band_vert_discard              = 1.0f;
 	uint16_t clip_rect_rule                       = 0xffffu;
 	int      clip_rect_left[4]                    = {};
 	int      clip_rect_top[4]                     = {};
@@ -468,27 +408,6 @@ struct ScreenViewport {
 	bool     clip_rect_window_offset_enable[4]    = {};
 };
 
-struct VsShaderResource1 {
-	uint8_t vgprs                    = 0;
-	uint8_t sgprs                    = 0;
-	uint8_t priority                 = 0;
-	uint8_t float_mode               = 0;
-	bool    dx10_clamp               = false;
-	bool    ieee_mode                = false;
-	uint8_t vgpr_component_count     = 0;
-	bool    cu_group_enable          = false;
-	bool    require_forward_progress = false;
-	bool    fp16_overflow            = false;
-};
-
-struct VsShaderResource2 {
-	bool    scratch_en        = false;
-	uint8_t user_sgpr         = 0;
-	bool    offchip_lds       = false;
-	bool    streamout_enabled = false;
-	uint8_t shared_vgprs      = 0;
-};
-
 struct HsShaderResource1 {
 	uint8_t vgprs                     = 0;
 	uint8_t priority                  = 0;
@@ -511,7 +430,6 @@ struct HsShaderResource2 {
 
 struct PsShaderResource1 {
 	uint8_t vgprs                    = 0;
-	uint8_t sgprs                    = 0;
 	uint8_t priority                 = 0;
 	uint8_t float_mode               = 0;
 	bool    dx10_clamp               = false;
@@ -535,7 +453,6 @@ struct PsStageRegisters {
 	uint64_t          data_addr = 0;
 	PsShaderResource1 rsrc1;
 	PsShaderResource2 rsrc2;
-	uint64_t          chksum = 0;
 };
 
 struct CsStageRegisters {
@@ -545,12 +462,13 @@ struct CsStageRegisters {
 	uint32_t num_thread_y              = 0;
 	uint32_t num_thread_z              = 0;
 	uint8_t  vgprs                     = 0;
-	uint8_t  sgprs                     = 0;
+	uint8_t  priority                  = 0;
 	uint8_t  float_mode                = 0;
 	bool     dx10_clamp                = false;
+	bool     debug_mode                = false;
 	bool     ieee_mode                 = false;
+	bool     require_forward_progress  = false;
 	bool     fp16_overflow             = false;
-	bool     bulky                     = false;
 	bool     threadgroup_configuration = false;
 	uint8_t  wave_size                 = 64;
 	bool     scratch_en                = false;
@@ -561,6 +479,7 @@ struct CsStageRegisters {
 	bool     tg_size_en                = false;
 	uint8_t  tidig_comp_cnt            = 0;
 	uint16_t lds_size                  = 0;
+	uint8_t  shared_vgprs              = 0;
 };
 
 struct EsStageRegisters {
@@ -568,21 +487,17 @@ struct EsStageRegisters {
 };
 
 struct LsStageRegisters {
-	uint64_t          data_addr = 0;
-	VsShaderResource1 rsrc1;
-	VsShaderResource2 rsrc2;
+	uint64_t data_addr = 0;
 };
 
 struct HsStageRegisters {
 	uint64_t          data_addr = 0;
 	HsShaderResource1 rsrc1;
 	HsShaderResource2 rsrc2;
-	uint64_t          chksum = 0;
 };
 
 struct GsShaderResource1 {
 	uint8_t vgprs                     = 0;
-	uint8_t sgprs                     = 0;
 	uint8_t priority                  = 0;
 	uint8_t float_mode                = 0;
 	bool    dx10_clamp                = false;
@@ -605,10 +520,10 @@ struct GsShaderResource2 {
 };
 
 struct GsStageRegisters {
-	uint64_t          data_addr = 0;
+	uint64_t          data_addr      = 0;
+	uint64_t          user_data_addr = 0;
 	GsShaderResource1 rsrc1;
 	GsShaderResource2 rsrc2;
-	uint64_t          chksum = 0;
 };
 
 struct ShaderRegisters {
@@ -664,8 +579,7 @@ struct ShaderRegisters {
 enum class UserSgprType { Unknown, Region, Vsharp };
 
 struct UserSgprInfo {
-	// Changed from 16 to 64
-	static constexpr int SGPRS_MAX = 64;
+	static constexpr int SGPRS_MAX = 32;
 
 	uint32_t     value[SGPRS_MAX] = {0};
 	UserSgprType type[SGPRS_MAX]  = {};
@@ -712,12 +626,6 @@ public:
 	void Reset() { *this = Context(); }
 
 	void SetColorBase(uint32_t slot, const ColorBase& base) { m_render_targets[slot].base = base; }
-	void SetColorPitch(uint32_t slot, const ColorPitch& pitch) {
-		m_render_targets[slot].pitch = pitch;
-	}
-	void SetColorSlice(uint32_t slot, const ColorSlice& slice) {
-		m_render_targets[slot].slice = slice;
-	}
 	void SetColorView(uint32_t slot, const ColorView& view) { m_render_targets[slot].view = view; }
 	void SetColorInfo(uint32_t slot, const ColorInfo& info) { m_render_targets[slot].info = info; }
 	void SetColorAttrib(uint32_t slot, const ColorAttrib& attrib) {
@@ -735,15 +643,9 @@ public:
 	void SetColorCmask(uint32_t slot, const ColorCmask& cmask) {
 		m_render_targets[slot].cmask = cmask;
 	}
-	void SetColorCmaskSlice(uint32_t slot, const ColorCmaskSlice& cmask_slice) {
-		m_render_targets[slot].cmask_slice = cmask_slice;
-	}
 	void SetColorFmask(uint32_t slot, const ColorFmask& fmask) {
 		m_render_targets[slot].fmask = fmask;
 	}
-	void SetColorFmaskSlice(uint32_t slot, const ColorFmaskSlice& fmask_slice) {
-		m_render_targets[slot].fmask_slice = fmask_slice;
-	}
 	void SetColorClearWord0(uint32_t slot, const ColorClearWord0& clear_word0) {
 		m_render_targets[slot].clear_word0 = clear_word0;
 	}
@@ -753,7 +655,6 @@ public:
 	void SetColorDccAddr(uint32_t slot, const ColorDccAddr& dcc_addr) {
 		m_render_targets[slot].dcc_addr = dcc_addr;
 	}
-	void SetColorSize(uint32_t slot, const ColorSize& size) { m_render_targets[slot].size = size; }
 	[[nodiscard]] const RenderTarget& GetRenderTarget(uint32_t slot) const {
 		return m_render_targets[slot];
 	}
@@ -1072,23 +973,25 @@ public:
 
 	void SetEsShaderBase(uint64_t addr) { m_vs.es_regs.data_addr = addr; }
 	void SetLsShaderBase(uint64_t addr) { m_vs.ls_regs.data_addr = addr; }
-	void SetLsShaderResource1(const VsShaderResource1& rsrc1) { m_vs.ls_regs.rsrc1 = rsrc1; }
-	void SetLsShaderResource2(const VsShaderResource2& rsrc2) { m_vs.ls_regs.rsrc2 = rsrc2; }
 	void SetHsShaderBase(uint64_t addr) { m_vs.hs_regs.data_addr = addr; }
 	void SetHsShaderResource1(const HsShaderResource1& rsrc1) { m_vs.hs_regs.rsrc1 = rsrc1; }
 	void SetHsShaderResource2(const HsShaderResource2& rsrc2) { m_vs.hs_regs.rsrc2 = rsrc2; }
-	void SetHsShaderChksum(uint32_t value) { m_vs.hs_regs.chksum = value; }
 	void SetGsShaderBase(uint64_t addr) { m_vs.gs_regs.data_addr = addr; }
+	void SetGsUserDataAddress(uint32_t word, uint32_t value) {
+		const auto shift   = word * 32u;
+		auto&      address = m_vs.gs_regs.user_data_addr;
+		address = (address & ~(uint64_t {0xffffffffu} << shift)) |
+		          (static_cast<uint64_t>(value) << shift);
+	}
 	void SetGsShaderResource1(const GsShaderResource1& rsrc1) { m_vs.gs_regs.rsrc1 = rsrc1; }
 	void SetGsShaderResource2(const GsShaderResource2& rsrc2) { m_vs.gs_regs.rsrc2 = rsrc2; }
-	void SetGsShaderChksum(uint32_t value) { m_vs.gs_regs.chksum = value; }
 
 	void SetPsShaderBase(uint64_t addr) { m_ps.ps_regs.data_addr = addr; }
 	void SetPsShaderResource1(const PsShaderResource1& rsrc1) { m_ps.ps_regs.rsrc1 = rsrc1; }
 	void SetPsShaderResource2(const PsShaderResource2& rsrc2) { m_ps.ps_regs.rsrc2 = rsrc2; }
-	void SetPsShaderChksum(uint32_t value) { m_ps.ps_regs.chksum = value; }
 
 	void SetCsShader(const CsStageRegisters& cs_regs) { m_cs.cs_regs = cs_regs; }
+	void SetCsWaveSize(uint8_t wave_size) { m_cs.cs_regs.wave_size = wave_size; }
 
 	void SetPsUserSgpr(uint32_t id, uint32_t value, UserSgprType type) {
 		m_ps.ps_user_sgpr.value[id] = value;

src/graphics/guest_gpu/pm4.cpp

@@ -21,12 +21,10 @@ constexpr auto MakeRegisterNames() {
 	names[R_DRAW_RESET]     = "R_DRAW_RESET";
 	names[R_WAIT_FLIP_DONE] = "R_WAIT_FLIP_DONE";
 	names[R_DISPATCH_RESET] = "R_DISPATCH_RESET";
-	names[R_WAIT_MEM_32]    = "R_WAIT_MEM_32";
 	names[R_PUSH_MARKER]    = "R_PUSH_MARKER";
 	names[R_POP_MARKER]     = "R_POP_MARKER";
 	names[R_ACQUIRE_MEM]    = "R_ACQUIRE_MEM";
 	names[R_WRITE_DATA]     = "R_WRITE_DATA";
-	names[R_WAIT_MEM_64]    = "R_WAIT_MEM_64";
 	names[R_FLIP]           = "R_FLIP";
 	names[R_RELEASE_MEM]    = "R_RELEASE_MEM";
 	names[R_DMA_DATA]       = "R_DMA_DATA";
@@ -62,6 +60,7 @@ constexpr auto MakeOpcodeNames() {
 	names[IT_WRITE_DATA]                = "IT_WRITE_DATA";
 	names[IT_MEM_SEMAPHORE]             = "IT_MEM_SEMAPHORE";
 	names[IT_DRAW_INDEX_INDIRECT_MULTI] = "IT_DRAW_INDEX_INDIRECT_MULTI";
+	names[IT_WAIT_REG_MEM]               = "IT_WAIT_REG_MEM";
 	names[IT_INDIRECT_BUFFER]           = "IT_INDIRECT_BUFFER";
 	names[IT_COPY_DATA]                 = "IT_COPY_DATA";
 	names[IT_CP_DMA]                    = "IT_CP_DMA";
@@ -91,6 +90,7 @@ constexpr auto MakeOpcodeNames() {
 	names[IT_DISPATCH_DRAW_PREAMBLE]    = "IT_DISPATCH_DRAW_PREAMBLE";
 	names[IT_DISPATCH_DRAW]             = "IT_DISPATCH_DRAW";
 	names[IT_GET_LOD_STATS]             = "IT_GET_LOD_STATS";
+	names[IT_WAIT_REG_MEM_64]           = "IT_WAIT_REG_MEM_64";
 	names[IT_SET_CONTEXT_REG_INDIRECT]  = "IT_SET_CONTEXT_REG_INDIRECT";
 	return names;
 }
@@ -100,7 +100,7 @@ constexpr auto g_opcode_names   = MakeOpcodeNames();
 
 } // namespace
 
-void DumpPm4PacketStream(Common::File* file, uint32_t* cmd_buffer, uint32_t start_dw,
+void DumpPm4PacketStream(Common::File* file, const uint32_t* cmd_buffer, uint32_t start_dw,
                          uint32_t num_dw) {
 	// db_dump();
 
@@ -143,9 +143,9 @@ void DumpPm4PacketStream(Common::File* file, uint32_t* cmd_buffer, uint32_t star
 				if ((op == IT_SET_CONTEXT_REG_INDIRECT || op == IT_SET_SH_REG_INDIRECT ||
 				     op == IT_SET_UCONFIG_REG_INDIRECT) &&
 				    len == 4) {
-					auto* indirect_buffer =
-					    reinterpret_cast<uint32_t*>((static_cast<uint64_t>(cmd[0]) & 0xfffffffcu) |
-					                                (static_cast<uint64_t>(cmd[1]) << 32u));
+					auto* indirect_buffer = reinterpret_cast<const uint32_t*>(
+					    (static_cast<uint64_t>(cmd[0]) & 0xfffffffcu) |
+					    (static_cast<uint64_t>(cmd[1]) << 32u));
 					const uint32_t indirect_num_regs = cmd[3] & 0x3fffu;
 
 					for (uint32_t i = 0; indirect_buffer != nullptr && i < indirect_num_regs;

src/graphics/guest_gpu/pm4.h

@@ -42,6 +42,7 @@ constexpr uint32_t IT_WRITE_DATA                = 0x37;
 constexpr uint32_t IT_MEM_SEMAPHORE             = 0x39;
 constexpr uint32_t IT_DRAW_INDEX_INDIRECT_MULTI = 0x38;
 constexpr uint32_t IT_DISPATCH_DRAW_PREAMBLE    = 0x3A;
+constexpr uint32_t IT_WAIT_REG_MEM              = 0x3C;
 constexpr uint32_t IT_INDIRECT_BUFFER           = 0x3F;
 constexpr uint32_t IT_COPY_DATA                 = 0x40;
 constexpr uint32_t IT_CP_DMA                    = 0x41;
@@ -70,6 +71,7 @@ constexpr uint32_t IT_WAIT_ON_CE_COUNTER        = 0x86;
 constexpr uint32_t IT_WAIT_ON_DE_COUNTER_DIFF   = 0x88;
 constexpr uint32_t IT_DISPATCH_DRAW             = 0x8D;
 constexpr uint32_t IT_GET_LOD_STATS             = 0x8E;
+constexpr uint32_t IT_WAIT_REG_MEM_64           = 0x93;
 constexpr uint32_t IT_SET_CONTEXT_REG_INDIRECT  = 0x9F;
 
 /* Custom codes. Implemented via IT_NOP */
@@ -78,12 +80,10 @@ constexpr uint32_t R_ZERO           = 0x00;
 constexpr uint32_t R_DRAW_RESET     = 0x05;
 constexpr uint32_t R_WAIT_FLIP_DONE = 0x06;
 constexpr uint32_t R_DISPATCH_RESET = 0x09;
-constexpr uint32_t R_WAIT_MEM_32    = 0x0A;
 constexpr uint32_t R_PUSH_MARKER    = 0x0B;
 constexpr uint32_t R_POP_MARKER     = 0x0C;
 constexpr uint32_t R_ACQUIRE_MEM    = 0x14;
 constexpr uint32_t R_WRITE_DATA     = 0x15;
-constexpr uint32_t R_WAIT_MEM_64    = 0x16;
 constexpr uint32_t R_FLIP           = 0x17;
 constexpr uint32_t R_RELEASE_MEM    = 0x18;
 constexpr uint32_t R_DMA_DATA       = 0x19;
@@ -98,6 +98,10 @@ constexpr uint32_t DB_RENDER_CONTROL_DEPTH_CLEAR_ENABLE_SHIFT       = 0;
 constexpr uint32_t DB_RENDER_CONTROL_DEPTH_CLEAR_ENABLE_MASK        = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_CLEAR_ENABLE_SHIFT     = 1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_CLEAR_ENABLE_MASK      = 0x1;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_DEPTH_TO_COLOR_SHIFT      = 2;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_DEPTH_TO_COLOR_MASK       = 0x1;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_STENCIL_TO_COLOR_SHIFT    = 3;
+constexpr uint32_t DB_RENDER_CONTROL_COPY_STENCIL_TO_COLOR_MASK     = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_RESUMMARIZE_ENABLE_SHIFT       = 4;
 constexpr uint32_t DB_RENDER_CONTROL_RESUMMARIZE_ENABLE_MASK        = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_COMPRESS_DISABLE_SHIFT = 5;
@@ -163,22 +167,6 @@ constexpr uint32_t PA_SC_SCREEN_SCISSOR_BR_BR_Y_MASK  = 0xFFFF;
 
 constexpr uint32_t DB_DFSM_CONTROL = 0xE;
 
-constexpr uint32_t DB_DEPTH_INFO                          = 0xF;
-constexpr uint32_t DB_DEPTH_INFO_ADDR5_SWIZZLE_MASK_SHIFT = 0;
-constexpr uint32_t DB_DEPTH_INFO_ADDR5_SWIZZLE_MASK_MASK  = 0xF;
-constexpr uint32_t DB_DEPTH_INFO_ARRAY_MODE_SHIFT         = 4;
-constexpr uint32_t DB_DEPTH_INFO_ARRAY_MODE_MASK          = 0xF;
-constexpr uint32_t DB_DEPTH_INFO_PIPE_CONFIG_SHIFT        = 8;
-constexpr uint32_t DB_DEPTH_INFO_PIPE_CONFIG_MASK         = 0x1F;
-constexpr uint32_t DB_DEPTH_INFO_BANK_WIDTH_SHIFT         = 13;
-constexpr uint32_t DB_DEPTH_INFO_BANK_WIDTH_MASK          = 0x3;
-constexpr uint32_t DB_DEPTH_INFO_BANK_HEIGHT_SHIFT        = 15;
-constexpr uint32_t DB_DEPTH_INFO_BANK_HEIGHT_MASK         = 0x3;
-constexpr uint32_t DB_DEPTH_INFO_MACRO_TILE_ASPECT_SHIFT  = 17;
-constexpr uint32_t DB_DEPTH_INFO_MACRO_TILE_ASPECT_MASK   = 0x3;
-constexpr uint32_t DB_DEPTH_INFO_NUM_BANKS_SHIFT          = 19;
-constexpr uint32_t DB_DEPTH_INFO_NUM_BANKS_MASK           = 0x3;
-
 constexpr uint32_t DB_Z_INFO                               = 0x10;
 constexpr uint32_t DB_Z_INFO_FORMAT_SHIFT                  = 0;
 constexpr uint32_t DB_Z_INFO_FORMAT_MASK                   = 0x3;
@@ -222,16 +210,6 @@ constexpr uint32_t DB_STENCIL_READ_BASE  = 0x13;
 constexpr uint32_t DB_Z_WRITE_BASE       = 0x14;
 constexpr uint32_t DB_STENCIL_WRITE_BASE = 0x15;
 
-constexpr uint32_t DB_DEPTH_SIZE                       = 0x16;
-constexpr uint32_t DB_DEPTH_SIZE_PITCH_TILE_MAX_SHIFT  = 0;
-constexpr uint32_t DB_DEPTH_SIZE_PITCH_TILE_MAX_MASK   = 0x7FF;
-constexpr uint32_t DB_DEPTH_SIZE_HEIGHT_TILE_MAX_SHIFT = 11;
-constexpr uint32_t DB_DEPTH_SIZE_HEIGHT_TILE_MAX_MASK  = 0x7FF;
-
-constexpr uint32_t DB_DEPTH_SLICE                      = 0x17;
-constexpr uint32_t DB_DEPTH_SLICE_SLICE_TILE_MAX_SHIFT = 0;
-constexpr uint32_t DB_DEPTH_SLICE_SLICE_TILE_MAX_MASK  = 0x3FFFFF;
-
 constexpr uint32_t DB_Z_READ_BASE_HI                  = 0x1A;
 constexpr uint32_t DB_STENCIL_READ_BASE_HI            = 0x1B;
 constexpr uint32_t DB_Z_WRITE_BASE_HI                 = 0x1C;
@@ -317,8 +295,8 @@ constexpr uint32_t FSR_RECURSIONS1              = 0xF2;
 constexpr uint32_t VGT_MULTI_PRIM_IB_RESET_INDX = 0x103;
 constexpr uint32_t CB_RMI_GL2_CACHE_CONTROL     = 0x104;
 constexpr uint32_t CB_BLEND_RED                 = 0x105;
-constexpr uint32_t CB_BLEND_GREEN               = 0x106;
-constexpr uint32_t CB_BLEND_BLUE                = 0x107;
+constexpr uint32_t CB_BLEND_BLUE                = 0x106;
+constexpr uint32_t CB_BLEND_GREEN               = 0x107;
 constexpr uint32_t CB_BLEND_ALPHA               = 0x108;
 constexpr uint32_t CB_DCC_CONTROL               = 0x109;
 
@@ -387,9 +365,6 @@ constexpr uint32_t SPI_SHADER_POS_FORMAT  = 0x1C3;
 constexpr uint32_t SPI_SHADER_Z_FORMAT    = 0x1C4;
 constexpr uint32_t SPI_SHADER_COL_FORMAT  = 0x1C5;
 
-// Indirect Cx descriptor selector for the 32-entry PS input-control register bank.
-constexpr uint32_t CX_PS_SHADER_USAGE_BASE = 0x10000000u;
-
 constexpr uint32_t CB_BLEND0_CONTROL                            = 0x1E0;
 constexpr uint32_t CB_BLEND0_CONTROL_COLOR_SRCBLEND_SHIFT       = 0;
 constexpr uint32_t CB_BLEND0_CONTROL_COLOR_SRCBLEND_MASK        = 0x1F;
@@ -427,11 +402,6 @@ constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_SHIFT                        = 8
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_MASK                         = 0x7;
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_BF_SHIFT                     = 20;
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_BF_MASK                      = 0x7;
-constexpr uint32_t DB_DEPTH_CONTROL_ENABLE_COLOR_WRITES_ON_DEPTH_FAIL_SHIFT  = 30;
-constexpr uint32_t DB_DEPTH_CONTROL_ENABLE_COLOR_WRITES_ON_DEPTH_FAIL_MASK   = 0x1;
-constexpr uint32_t DB_DEPTH_CONTROL_DISABLE_COLOR_WRITES_ON_DEPTH_PASS_SHIFT = 31;
-constexpr uint32_t DB_DEPTH_CONTROL_DISABLE_COLOR_WRITES_ON_DEPTH_PASS_MASK  = 0x1;
-
 constexpr uint32_t DB_EQAA                                  = 0x201;
 constexpr uint32_t DB_EQAA_MAX_ANCHOR_SAMPLES_SHIFT         = 0;
 constexpr uint32_t DB_EQAA_MAX_ANCHOR_SAMPLES_MASK          = 0x7;
@@ -563,21 +533,9 @@ constexpr uint32_t VGT_DRAW_PAYLOAD_CNTL  = 0x2A6;
 constexpr uint32_t VGT_ESGS_RING_ITEMSIZE = 0x2AB;
 constexpr uint32_t VGT_REUSE_OFF          = 0x2AD;
 
-constexpr uint32_t DB_HTILE_SURFACE                               = 0x2AF;
-constexpr uint32_t DB_HTILE_SURFACE_LINEAR_SHIFT                  = 0;
-constexpr uint32_t DB_HTILE_SURFACE_LINEAR_MASK                   = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_FULL_CACHE_SHIFT              = 1;
-constexpr uint32_t DB_HTILE_SURFACE_FULL_CACHE_MASK               = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_HTILE_USES_PRELOAD_WIN_SHIFT  = 2;
-constexpr uint32_t DB_HTILE_SURFACE_HTILE_USES_PRELOAD_WIN_MASK   = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_PRELOAD_SHIFT                 = 3;
-constexpr uint32_t DB_HTILE_SURFACE_PRELOAD_MASK                  = 0x1;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_WIDTH_SHIFT          = 4;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_WIDTH_MASK           = 0x3F;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_HEIGHT_SHIFT         = 10;
-constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_HEIGHT_MASK          = 0x3F;
-constexpr uint32_t DB_HTILE_SURFACE_DST_OUTSIDE_ZERO_TO_ONE_SHIFT = 16;
-constexpr uint32_t DB_HTILE_SURFACE_DST_OUTSIDE_ZERO_TO_ONE_MASK  = 0x1;
+constexpr uint32_t DB_SHADING_RATE_ENCODING                             = 0x2AF;
+constexpr uint32_t DB_SHADING_RATE_ENCODING_SHADING_RATE_ENCODING_SHIFT = 19;
+constexpr uint32_t DB_SHADING_RATE_ENCODING_SHADING_RATE_ENCODING_MASK  = 0x3;
 
 constexpr uint32_t DB_SRESULTS_COMPARE_STATE0                          = 0x2B0;
 constexpr uint32_t DB_SRESULTS_COMPARE_STATE1                          = 0x2B1;
@@ -624,7 +582,6 @@ constexpr uint32_t PA_SC_SHADER_CONTROL                  = 0x310;
 constexpr uint32_t PA_SC_BINNER_CNTL_0                   = 0x311;
 constexpr uint32_t PA_SC_BINNER_CNTL_1                   = 0x312;
 constexpr uint32_t PA_SC_CONSERVATIVE_RASTERIZATION_CNTL = 0x313;
-constexpr uint32_t PA_SC_NGG_MODE_CNTL                   = 0x314;
 constexpr uint32_t CB_COLOR0_BASE                        = 0x318;
 
 constexpr uint32_t CB_COLOR0_VIEW                   = 0x31B;
@@ -652,24 +609,13 @@ constexpr uint32_t CB_COLOR0_INFO_BLEND_BYPASS_SHIFT              = 16;
 constexpr uint32_t CB_COLOR0_INFO_BLEND_BYPASS_MASK               = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_ROUND_MODE_SHIFT                = 18;
 constexpr uint32_t CB_COLOR0_INFO_ROUND_MODE_MASK                 = 0x1;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_IS_LINEAR_SHIFT           = 19;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_IS_LINEAR_MASK            = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESSION_DISABLE_SHIFT = 26;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESSION_DISABLE_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESS_1FRAG_ONLY_SHIFT = 27;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESS_1FRAG_ONLY_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_DCC_ENABLE_SHIFT                = 28;
 constexpr uint32_t CB_COLOR0_INFO_DCC_ENABLE_MASK                 = 0x1;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_ADDR_TYPE_SHIFT           = 29;
-constexpr uint32_t CB_COLOR0_INFO_CMASK_ADDR_TYPE_MASK            = 0x3;
-constexpr uint32_t CB_COLOR0_INFO_ALT_TILE_MODE_SHIFT             = 31;
-constexpr uint32_t CB_COLOR0_INFO_ALT_TILE_MODE_MASK              = 0x1;
-
 constexpr uint32_t CB_COLOR0_ATTRIB                             = 0x31D;
-constexpr uint32_t CB_COLOR0_ATTRIB_TILE_MODE_INDEX_SHIFT       = 0;
-constexpr uint32_t CB_COLOR0_ATTRIB_TILE_MODE_INDEX_MASK        = 0x1F;
-constexpr uint32_t CB_COLOR0_ATTRIB_FMASK_TILE_MODE_INDEX_SHIFT = 5;
-constexpr uint32_t CB_COLOR0_ATTRIB_FMASK_TILE_MODE_INDEX_MASK  = 0x1F;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_SAMPLES_SHIFT           = 12;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_SAMPLES_MASK            = 0x7;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_FRAGMENTS_SHIFT         = 15;
@@ -684,8 +630,6 @@ constexpr uint32_t CB_COLOR0_DCC_CONTROL_KEY_CLEAR_ENABLE_SHIFT
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_KEY_CLEAR_ENABLE_MASK                  = 0x1;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_UNCOMPRESSED_BLOCK_SIZE_SHIFT      = 2;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_UNCOMPRESSED_BLOCK_SIZE_MASK       = 0x3;
-constexpr uint32_t CB_COLOR0_DCC_CONTROL_MIN_COMPRESSED_BLOCK_SIZE_SHIFT        = 4;
-constexpr uint32_t CB_COLOR0_DCC_CONTROL_MIN_COMPRESSED_BLOCK_SIZE_MASK         = 0x1;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_COMPRESSED_BLOCK_SIZE_SHIFT        = 5;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_COMPRESSED_BLOCK_SIZE_MASK         = 0x3;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_COLOR_TRANSFORM_SHIFT                  = 7;
@@ -742,31 +686,25 @@ constexpr uint32_t CB_COLOR0_ATTRIB3_CMASK_PIPE_ALIGNED_SHIFT = 26;
 constexpr uint32_t CB_COLOR0_ATTRIB3_CMASK_PIPE_ALIGNED_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_ATTRIB3_DCC_PIPE_ALIGNED_SHIFT   = 30;
 constexpr uint32_t CB_COLOR0_ATTRIB3_DCC_PIPE_ALIGNED_MASK    = 0x1;
+constexpr uint32_t CB_COLOR0_ATTRIB3_WRITE_VRS_RATE_HINT_TO_CMASK_SHIFT = 31;
+constexpr uint32_t CB_COLOR0_ATTRIB3_WRITE_VRS_RATE_HINT_TO_CMASK_MASK  = 0x1;
 
 constexpr uint32_t CB_COLOR7_ATTRIB3 = 0x3BF;
 
-/* Fake codes. Don't exist on real HW */
-
-constexpr uint32_t CX_NOP = 0x800003FF;
+constexpr uint32_t CX_NOP = 0xDB;
 
 constexpr uint32_t CX_NUM = 0x3FF + 1;
 
 /* Shader registers */
 
-constexpr uint32_t SPI_SHADER_TBA_LO_PS     = 0x0;
-constexpr uint32_t SPI_SHADER_PGM_RSRC4_PS  = 0x1;
-constexpr uint32_t SPI_SHADER_TMA_LO_PS     = 0x2;
-constexpr uint32_t SPI_SHADER_TMA_HI_PS     = 0x3;
-constexpr uint32_t SPI_SHADER_PGM_CHKSUM_PS = 0x6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC3_PS  = 0x7;
-constexpr uint32_t SPI_SHADER_PGM_LO_PS     = 0x8;
-constexpr uint32_t SPI_SHADER_PGM_HI_PS     = 0x9;
+constexpr uint32_t SPI_SHADER_PACE_ID_PS            = 0x6;
+constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_PS    = 0x7;
+constexpr uint32_t SPI_SHADER_PGM_LO_PS              = 0x8;
+constexpr uint32_t SPI_SHADER_PGM_HI_PS              = 0x9;
 
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS                        = 0xA;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_VGPRS_SHIFT            = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_VGPRS_MASK             = 0x3F;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_SGPRS_SHIFT            = 6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_SGPRS_MASK             = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_PRIORITY_SHIFT         = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_PRIORITY_MASK          = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_FLOAT_MODE_SHIFT       = 12;
@@ -803,14 +741,11 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SHARED_VGPR_CNT_MASK         = 0xF;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_0  = 0xC;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_15 = 0x1B;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_31 = 0x2B;
-constexpr uint32_t SPI_SHADER_REQ_CTRL_PS     = 0x30;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_PS_0 = 0x32;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_PS_3 = 0x35;
 
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_VGPRS_SHIFT           = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_VGPRS_MASK            = 0x3F;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_SGPRS_SHIFT           = 6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_SGPRS_MASK            = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_PRIORITY_SHIFT        = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_PRIORITY_MASK         = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_FLOAT_MODE_SHIFT      = 12;
@@ -841,19 +776,20 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_USER_SGPR_MSB_MASK    = 0x1;
 constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_SHARED_VGPR_CNT_SHIFT = 28;
 constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_SHARED_VGPR_CNT_MASK  = 0xF;
 
-constexpr uint32_t SPI_SHADER_PGM_CHKSUM_GS        = 0x80;
+constexpr uint32_t SPI_SHADER_PACE_ID_GS           = 0x80;
+// AGC compiler records use this numeric tag as opaque metadata. It is not a shader checksum
+// register and intentionally has no checksum-state consumer in HardwareContext.
+constexpr uint32_t SPI_SHADER_PGM_CHKSUM_GS        = SPI_SHADER_PACE_ID_GS;
 constexpr uint32_t SPI_SHADER_PGM_RSRC4_GS         = 0x81;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_LO_GS = 0x82;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_HI_GS = 0x83;
-constexpr uint32_t SPI_SHADER_PGM_RSRC3_GS         = 0x87;
+constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_GS  = 0x87;
 constexpr uint32_t SPI_SHADER_PGM_LO_GS            = 0x88;
 constexpr uint32_t SPI_SHADER_PGM_HI_GS            = 0x89;
 
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS                        = 0x8A;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_VGPRS_SHIFT            = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_VGPRS_MASK             = 0x3F;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_SGPRS_SHIFT            = 6;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_SGPRS_MASK             = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_PRIORITY_SHIFT         = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_PRIORITY_MASK          = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_FLOAT_MODE_SHIFT       = 12;
@@ -894,22 +830,16 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_GS_SHARED_VGPR_CNT_MASK   = 0xF;
 constexpr uint32_t SPI_SHADER_USER_DATA_GS_0                      = 0x8C;
 constexpr uint32_t SPI_SHADER_USER_DATA_GS_15                     = 0x9B;
 constexpr uint32_t SPI_SHADER_USER_DATA_GS_31                     = 0xAB;
-constexpr uint32_t SPI_SHADER_REQ_CTRL_ESGS                       = 0xB0;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_ESGS_0                   = 0xB2;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_ESGS_3                   = 0xB5;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_ES_GS                     = 0xBC;
 constexpr uint32_t SPI_SHADER_PGM_LO_ES                           = 0xC8;
 constexpr uint32_t SPI_SHADER_PGM_HI_ES                           = 0xC9;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_ES                        = 0xCA;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_ES                        = 0xCB;
-constexpr uint32_t SPI_SHADER_USER_DATA_ES_0                      = 0xCC;
-constexpr uint32_t SPI_SHADER_USER_DATA_ES_15                     = 0xDB;
-constexpr uint32_t SPI_SHADER_USER_DATA_ES_31                     = 0xEB;
+// Private AGC registers used by shader binaries submitted through SET_SH_REG_INDIRECT.
 constexpr uint32_t SPI_SHADER_PGM_CHKSUM_HS                       = 0x100;
 constexpr uint32_t SPI_SHADER_PGM_RSRC4_HS                        = 0x101;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_LO_HS                = 0x102;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_HI_HS                = 0x103;
-constexpr uint32_t SPI_SHADER_PGM_RSRC3_HS                        = 0x107;
+constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_HS                 = 0x107;
 constexpr uint32_t SPI_SHADER_PGM_LO_HS                           = 0x108;
 constexpr uint32_t SPI_SHADER_PGM_HI_HS                           = 0x109;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS                        = 0x10A;
@@ -947,13 +877,10 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_SHARED_VGPR_CNT_MASK   = 0xF;
 constexpr uint32_t SPI_SHADER_USER_DATA_HS_0                      = 0x10C;
 constexpr uint32_t SPI_SHADER_USER_DATA_HS_15                     = 0x11B;
 constexpr uint32_t SPI_SHADER_USER_DATA_HS_31                     = 0x12B;
-constexpr uint32_t SPI_SHADER_REQ_CTRL_LSHS                       = 0x130;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_LSHS_0                   = 0x132;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_LSHS_3                   = 0x135;
 constexpr uint32_t SPI_SHADER_PGM_LO_LS                           = 0x148;
 constexpr uint32_t SPI_SHADER_PGM_HI_LS                           = 0x149;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_LS                        = 0x14A;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_LS                        = 0x14B;
 constexpr uint32_t COMPUTE_START_X                                = 0x204;
 constexpr uint32_t COMPUTE_START_Y                                = 0x205;
 constexpr uint32_t COMPUTE_START_Z                                = 0x206;
@@ -962,30 +889,36 @@ constexpr uint32_t COMPUTE_NUM_THREAD_Y                           = 0x208;
 constexpr uint32_t COMPUTE_NUM_THREAD_Z                           = 0x209;
 constexpr uint32_t COMPUTE_PGM_LO                                 = 0x20C;
 constexpr uint32_t COMPUTE_PGM_HI                                 = 0x20D;
-constexpr uint32_t COMPUTE_TBA_LO                                 = 0x20E;
-constexpr uint32_t COMPUTE_TBA_HI                                 = 0x20F;
-constexpr uint32_t COMPUTE_TMA_LO                                 = 0x210;
-constexpr uint32_t COMPUTE_TMA_HI                                 = 0x211;
 
 constexpr uint32_t COMPUTE_PGM_RSRC1                  = 0x212;
 constexpr uint32_t COMPUTE_PGM_RSRC1_VGPRS_SHIFT      = 0;
 constexpr uint32_t COMPUTE_PGM_RSRC1_VGPRS_MASK       = 0x3F;
-constexpr uint32_t COMPUTE_PGM_RSRC1_SGPRS_SHIFT      = 6;
-constexpr uint32_t COMPUTE_PGM_RSRC1_SGPRS_MASK       = 0xF;
+constexpr uint32_t COMPUTE_PGM_RSRC1_PRIORITY_SHIFT   = 10;
+constexpr uint32_t COMPUTE_PGM_RSRC1_PRIORITY_MASK    = 0x3;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FLOAT_MODE_SHIFT = 12;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FLOAT_MODE_MASK  = 0xFF;
 constexpr uint32_t COMPUTE_PGM_RSRC1_DX10_CLAMP_SHIFT = 21;
 constexpr uint32_t COMPUTE_PGM_RSRC1_DX10_CLAMP_MASK  = 0x1;
+constexpr uint32_t COMPUTE_PGM_RSRC1_DEBUG_MODE_SHIFT = 22;
+constexpr uint32_t COMPUTE_PGM_RSRC1_DEBUG_MODE_MASK  = 0x1;
 constexpr uint32_t COMPUTE_PGM_RSRC1_IEEE_MODE_SHIFT  = 23;
 constexpr uint32_t COMPUTE_PGM_RSRC1_IEEE_MODE_MASK   = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_BULKY_SHIFT      = 24;
-constexpr uint32_t COMPUTE_PGM_RSRC1_BULKY_MASK       = 0x1;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FP16_OVFL_SHIFT  = 26;
 constexpr uint32_t COMPUTE_PGM_RSRC1_FP16_OVFL_MASK   = 0x1;
 constexpr uint32_t COMPUTE_PGM_RSRC1_WGP_MODE_SHIFT   = 29;
 constexpr uint32_t COMPUTE_PGM_RSRC1_WGP_MODE_MASK    = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_W32_EN_SHIFT     = 30;
-constexpr uint32_t COMPUTE_PGM_RSRC1_W32_EN_MASK      = 0x1;
+constexpr uint32_t COMPUTE_PGM_RSRC1_FWD_PROGRESS_SHIFT = 31;
+constexpr uint32_t COMPUTE_PGM_RSRC1_FWD_PROGRESS_MASK  = 0x1;
+
+constexpr uint32_t COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_SHIFT = 15;
+constexpr uint32_t COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_MASK  = 0x1;
+
+[[nodiscard]] constexpr uint8_t ComputeWaveSize(uint32_t dispatch_initiator) {
+	return ((dispatch_initiator >> COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_SHIFT) &
+	        COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_MASK) != 0u
+	           ? 32u
+	           : 64u;
+}
 
 constexpr uint32_t COMPUTE_PGM_RSRC2                      = 0x213;
 constexpr uint32_t COMPUTE_PGM_RSRC2_SCRATCH_EN_SHIFT     = 0;
@@ -1006,21 +939,16 @@ constexpr uint32_t COMPUTE_PGM_RSRC2_LDS_SIZE_SHIFT       = 15;
 constexpr uint32_t COMPUTE_PGM_RSRC2_LDS_SIZE_MASK        = 0x1FF;
 
 constexpr uint32_t COMPUTE_RESOURCE_LIMITS    = 0x215;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE0 = 0x216;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE1 = 0x217;
 constexpr uint32_t COMPUTE_TMPRING_SIZE       = 0x218;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE2 = 0x219;
-constexpr uint32_t COMPUTE_DESTINATION_EN_SE3 = 0x21A;
 constexpr uint32_t COMPUTE_USER_ACCUM_0       = 0x224;
 constexpr uint32_t COMPUTE_PGM_RSRC3          = 0x228;
-constexpr uint32_t COMPUTE_SHADER_CHKSUM      = 0x22A;
+constexpr uint32_t COMPUTE_PGM_RSRC3_SHARED_VGPRS_SHIFT = 0;
+constexpr uint32_t COMPUTE_PGM_RSRC3_SHARED_VGPRS_MASK  = 0xF;
+constexpr uint32_t COMPUTE_PACE_ID             = 0x22A;
 constexpr uint32_t COMPUTE_USER_DATA_0        = 0x240;
 constexpr uint32_t COMPUTE_USER_DATA_15       = 0x24F;
-constexpr uint32_t COMPUTE_DISPATCH_TUNNEL    = 0x27D;
 
-/* Fake codes. Don't exist on real HW */
-
-constexpr uint32_t SH_NOP = 0x800002FF;
+constexpr uint32_t SH_NOP = 0x280;
 
 constexpr uint32_t SH_NUM = 0x2FF + 1;
 
@@ -1037,6 +965,7 @@ constexpr uint32_t GE_INDX_OFFSET            = 0x24A;
 constexpr uint32_t GE_MULTI_PRIM_IB_RESET_EN = 0x24B;
 constexpr uint32_t VGT_HS_OFFCHIP_PARAM      = 0x24F;
 constexpr uint32_t VGT_TF_MEMORY_BASE        = 0x250;
+constexpr uint32_t IA_MULTI_VGT_PARAM        = 0x258;
 
 constexpr uint32_t GE_CNTL                     = 0x25B;
 constexpr uint32_t GE_CNTL_PRIM_GRP_SIZE_SHIFT = 0;
@@ -1048,7 +977,7 @@ constexpr uint32_t GE_USER_VGPR1  = 0x25C;
 constexpr uint32_t GE_USER_VGPR2  = 0x25D;
 constexpr uint32_t GE_USER_VGPR3  = 0x25E;
 constexpr uint32_t GE_STEREO_CNTL = 0x25F;
-constexpr uint32_t GE_PC_ALLOC    = 0x260;
+constexpr uint32_t UC_PARAMETER_OVERSUBSCRIPTION = 0x260;
 
 constexpr uint32_t GE_USER_VGPR_EN                     = 0x262;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR1_SHIFT = 0;
@@ -1058,32 +987,19 @@ constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR2_MASK  = 0x1;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR3_SHIFT = 2;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR3_MASK  = 0x1;
 
-constexpr uint32_t IA_MULTI_VGT_PARAM       = 0x258;
 constexpr uint32_t TA_CS_BC_BASE_ADDR       = 0x380;
 constexpr uint32_t TA_CS_BC_BASE_ADDR_HI    = 0x381;
-constexpr uint32_t TEXTURE_GRADIENT_FACTORS = 0x382;
+constexpr uint32_t TEXTURE_GRADIENT_FACTORS = 0x383;
+constexpr uint32_t TEXTURE_GRADIENT_CONTROL = 0x384;
 constexpr uint32_t GDS_OA_CNTL              = 0x41D;
 constexpr uint32_t GDS_OA_COUNTER           = 0x41E;
 constexpr uint32_t GDS_OA_ADDRESS           = 0x41F;
 
-/* Fake codes. Don't exist on real HW */
-
-constexpr uint32_t FSR_EXTEND_SUBPIXEL_ROUNDING = 0x80003FF4;
-constexpr uint32_t FSR_ALPHA_VALUE0             = 0x80003FF5;
-constexpr uint32_t FSR_ALPHA_VALUE1             = 0x80003FF6;
-constexpr uint32_t FSR_CONTROL_POINT0           = 0x80003FF7;
-constexpr uint32_t FSR_CONTROL_POINT1           = 0x80003FF8;
-constexpr uint32_t FSR_CONTROL_POINT2           = 0x80003FF9;
-constexpr uint32_t FSR_CONTROL_POINT3           = 0x80003FFA;
-constexpr uint32_t FSR_WINDOW0                  = 0x80003FFB;
-constexpr uint32_t FSR_WINDOW1                  = 0x80003FFC;
-constexpr uint32_t TEXTURE_GRADIENT_CONTROL     = 0x80003FFD;
-constexpr uint32_t MEMORY_MAPPING_MASK          = 0x80003FFE;
-constexpr uint32_t UC_NOP                       = 0x80003FFF;
+constexpr uint32_t UC_NOP = 0xA2;
 
 constexpr uint32_t UC_NUM = 0x3FFF + 1;
 
-void DumpPm4PacketStream(Common::File* file, uint32_t* cmd_buffer, uint32_t start_dw,
+void DumpPm4PacketStream(Common::File* file, const uint32_t* cmd_buffer, uint32_t start_dw,
                          uint32_t num_dw);
 
 } // namespace Libs::Graphics::Pm4

src/graphics/guest_gpu/tile.cpp

@@ -1221,6 +1221,31 @@ bool TileGetRenderTargetSize(uint32_t width, uint32_t height, uint32_t pitch,
 	return true;
 }
 
+bool TileGetDccSize(uint32_t width, uint32_t height, uint32_t slices,
+                    uint32_t bytes_per_element, uint32_t levels, Prospero::TileMode tile,
+                    TileSizeAlign& total_size, uint32_t num_fragments_log2) {
+	total_size = {};
+	if (width == 0 || height == 0 || slices == 0 || levels != 1 || num_fragments_log2 != 0 ||
+	    !std::has_single_bit(bytes_per_element) || bytes_per_element > 16 ||
+	    (tile != Prospero::TileMode::kRenderTarget && tile != Prospero::TileMode::kDepth)) {
+		return false;
+	}
+	// Gen5 color metadata uses 4 KiB blocks, with one byte per 256 bytes of color data.
+	// Metadata stays thin for volume surfaces, so every depth slice has its own block raster.
+	const uint32_t coverage_bits = 20u - std::countr_zero(bytes_per_element);
+	const uint32_t block_width   = 1u << ((coverage_bits + 1u) / 2u);
+	const uint32_t block_height  = 1u << (coverage_bits / 2u);
+	const uint64_t blocks_x = (static_cast<uint64_t>(width) + block_width - 1u) / block_width;
+	const uint64_t blocks_y = (static_cast<uint64_t>(height) + block_height - 1u) / block_height;
+	const uint64_t blocks   = blocks_x * blocks_y;
+	if (blocks > UINT32_MAX / 4096u / slices) {
+		return false;
+	}
+	total_size.size  = static_cast<uint32_t>(blocks * slices * 4096u);
+	total_size.align = 4096;
+	return true;
+}
+
 bool TileGetRenderTargetMipLayout(uint32_t width, uint32_t height, uint32_t pitch,
                                   uint32_t bytes_per_element, uint32_t levels,
                                   TileSizeAlign& total_size, TileSizeOffset* level_sizes,

src/graphics/guest_gpu/tile.h

@@ -122,6 +122,9 @@ uint32_t TileGetDepthPitch(uint32_t width, uint32_t bytes_per_element,
 bool     TileGetRenderTargetSize(uint32_t width, uint32_t height, uint32_t pitch,
                                  uint32_t bytes_per_element, TileSizeAlign& total_size,
                                  uint32_t num_fragments_log2 = 0);
+bool     TileGetDccSize(uint32_t width, uint32_t height, uint32_t slices,
+                        uint32_t bytes_per_element, uint32_t levels, Prospero::TileMode tile,
+                        TileSizeAlign& total_size, uint32_t num_fragments_log2 = 0);
 bool     TileGetRenderTargetMipLayout(uint32_t width, uint32_t height, uint32_t pitch,
                                       uint32_t bytes_per_element, uint32_t levels,
                                       TileSizeAlign& total_size, TileSizeOffset* level_sizes,

src/graphics/host_gpu/graphicContext.h

@@ -3,32 +3,102 @@
 
 #include "common/abi.h"
 #include "common/common.h"
+#include "common/threads.h"
 #include "graphics/host_gpu/vulkanCommon.h" // IWYU pragma: export
-#include "graphics/host_gpu/vulkanInstance.h"
 
+#include <map>
+#include <mutex>
+#include <tuple>
 #include <vector>
 #include <vk_mem_alloc.h>
 
 namespace Libs::Graphics {
 
-class CommandBuffer;
 struct VulkanBuffer;
 struct VulkanImage;
 struct VulkanMemory;
 
-struct GraphicContext: public VulkanInstance {
+inline constexpr uint32_t VULKAN_TARGET_API_VERSION = VK_API_VERSION_1_3;
+
+struct GraphicContext {
+	vk::Instance                       instance                              = nullptr;
+	vk::DebugUtilsMessengerEXT         debug_messenger                       = nullptr;
+	vk::PhysicalDevice                 physical_device                       = nullptr;
+	vk::PhysicalDeviceProperties       physical_device_properties            = {};
+	vk::PhysicalDeviceMemoryProperties physical_device_memory_properties     = {};
+	vk::Device                         device                                = nullptr;
+	VmaAllocator                       allocator                             = nullptr;
+	bool                               memory_budget_ext_enabled             = false;
+	bool                               rt_extensions_enabled                 = false;
+	bool                               compute_subgroup_size_control_enabled = false;
+	bool                               sample_rate_shading_enabled           = false;
+	bool                               attachment_feedback_loop_enabled      = false;
+	bool                                      mesh_shader_enabled                   = false;
+	vk::PhysicalDeviceMeshShaderPropertiesEXT mesh_shader_properties                = {};
+	uint32_t                           subgroup_size                         = 0;
+	uint32_t                           min_subgroup_size                     = 0;
+	uint32_t                           max_subgroup_size                     = 0;
+	uint32_t                           max_push_descriptors                  = 0;
+	vk::ShaderStageFlags               required_subgroup_size_stages         = {};
+	Common::Mutex                      queue_mutex;
+	uint32_t                           queue_family = static_cast<uint32_t>(-1);
+	vk::Queue                          queue        = nullptr;
+
+	[[nodiscard]] const vk::PhysicalDeviceProperties& GetPhysicalDeviceProperties() const {
+		return physical_device_properties;
+	}
+
+	[[nodiscard]] const vk::PhysicalDeviceMemoryProperties&
+	GetPhysicalDeviceMemoryProperties() const {
+		return physical_device_memory_properties;
+	}
+
+	[[nodiscard]] vk::FormatProperties GetFormatProperties(vk::Format format) const {
+		std::scoped_lock lock(m_format_properties_mutex);
+		auto [it, inserted] = m_format_properties.try_emplace(format);
+		if (inserted) {
+			physical_device.getFormatProperties(format, &it->second);
+		}
+		return it->second;
+	}
+
+	[[nodiscard]] vk::Result GetImageFormatProperties(vk::Format format, vk::ImageType type,
+	                                                  vk::ImageTiling            tiling,
+	                                                  vk::ImageUsageFlags        usage,
+	                                                  vk::ImageCreateFlags       flags,
+	                                                  vk::ImageFormatProperties* properties) const {
+		using Key = std::tuple<vk::Format, vk::ImageType, vk::ImageTiling, vk::ImageUsageFlags,
+		                       vk::ImageCreateFlags>;
+		std::scoped_lock lock(m_image_format_properties_mutex);
+		auto [it, inserted] =
+		    m_image_format_properties.try_emplace(Key {format, type, tiling, usage, flags});
+		if (inserted) {
+			it->second.first = physical_device.getImageFormatProperties(format, type, tiling, usage,
+			                                                            flags, &it->second.second);
+		}
+		if (properties != nullptr) {
+			*properties = it->second.second;
+		}
+		return it->second.first;
+	}
+
+	[[nodiscard]] bool SupportsComputeWave64() const noexcept {
+		return subgroup_size == 64u || compute_subgroup_size_control_enabled;
+	}
+
+	[[nodiscard]] vk::DeviceSize StorageMinAlignment() const {
+		const auto alignment = physical_device_properties.limits.minStorageBufferOffsetAlignment;
+		return alignment != 0 ? alignment : 1;
+	}
+
 	[[nodiscard]] bool CreateAllocator();
 	void               DestroyAllocator();
 	void               LogMemoryBudget() const;
 	[[nodiscard]] bool CanReportMemoryUsage() const noexcept { return memory_budget_ext_enabled; }
 	[[nodiscard]] uint64_t GetDeviceMemoryUsage() const;
 	[[nodiscard]] uint64_t GetTotalMemoryBudget() const;
-	void                   CreateBuffer(uint64_t size, VulkanBuffer& buffer);
-	void                   DeleteBuffer(VulkanBuffer& buffer);
 	[[nodiscard]] bool     CreateImage(const vk::ImageCreateInfo& info, VulkanImage& image);
 	void                   DeleteImage(VulkanImage& image);
-	void                   MapMemory(VulkanMemory& memory, void*& data);
-	void                   UnmapMemory(VulkanMemory& memory);
 	void                   AppendHardwareRayTracingDeviceExtensions(
 	    const std::vector<vk::ExtensionProperties>& available_extensions,
 	    std::vector<const char*>&                   device_extensions);
@@ -36,18 +106,23 @@ struct GraphicContext: public VulkanInstance {
 
 	uint32_t screen_width  = 0;
 	uint32_t screen_height = 0;
+
+private:
+	mutable std::mutex                                 m_format_properties_mutex;
+	mutable std::map<vk::Format, vk::FormatProperties> m_format_properties;
+	mutable std::mutex                                 m_image_format_properties_mutex;
+	mutable std::map<std::tuple<vk::Format, vk::ImageType, vk::ImageTiling, vk::ImageUsageFlags,
+	                            vk::ImageCreateFlags>,
+	                 std::pair<vk::Result, vk::ImageFormatProperties>>
+	    m_image_format_properties;
 };
 
 struct VulkanMemory {
 	vk::MemoryRequirements  requirements       = {};
 	vk::MemoryPropertyFlags property           = {};
 	vk::MemoryPropertyFlags preferred_property = {};
-	vk::DeviceMemory        memory             = nullptr;
 	VmaAllocation           allocation         = nullptr;
-	VmaAllocationInfo       allocation_info    = {};
-	vk::DeviceSize          offset             = 0;
 	uint32_t                type               = 0;
-	uint64_t                unique_id          = 0;
 };
 
 struct VulkanImageState {

src/graphics/host_gpu/memoryTracker.cpp

@@ -18,15 +18,12 @@ MemoryTracker::~MemoryTracker() = default;
 #if KYTY_BUILD == KYTY_BUILD_DEBUG
 void MemoryTracker::ValidateGpuDirtyPages(const RangeSet& dirty, uint64_t vaddr, uint64_t size,
                                           const char* operation) const noexcept {
-	if (vaddr == 0 || size == 0 || size > UINT64_MAX - vaddr ||
-	    (vaddr & (TRACKER_PAGE_SIZE - 1)) != 0 || (size & (TRACKER_PAGE_SIZE - 1)) != 0) {
+	if (!GuestRange {vaddr, size}.Valid() || (vaddr & (TRACKER_PAGE_SIZE - 1)) != 0 ||
+	    (size & (TRACKER_PAGE_SIZE - 1)) != 0) {
 		EXIT("MemoryTracker: invalid dirty-page validation range\n");
 	}
 	for (auto page = vaddr; page < vaddr + size; page += TRACKER_PAGE_SIZE) {
-		bool found = false;
-		dirty.ForEachIntersection(page, TRACKER_PAGE_SIZE,
-		                          [&found](RangeSet::Range) { found = true; });
-		if (!found) {
+		if (!dirty.Intersects(page, TRACKER_PAGE_SIZE)) {
 			EXIT("MemoryTracker: GPU-dirty tracker page has no dirty bytes, operation=%s "
 			     "addr=0x%016" PRIx64 "\n",
 			     operation, page);
@@ -37,15 +34,10 @@ void MemoryTracker::ValidateGpuDirtyPages(const RangeSet& dirty, uint64_t vaddr,
 void MemoryTracker::ValidateGpuDirtyOwnership(const RangeSet& dirty, uint64_t vaddr, uint64_t size,
                                               const char* operation) {
 	ValidateRange(vaddr, size);
-	if (vaddr + size > UINT64_MAX - (TRACKER_PAGE_SIZE - 1)) {
-		EXIT("MemoryTracker: dirty ownership range alignment overflow\n");
-	}
 	const auto begin = vaddr & ~(TRACKER_PAGE_SIZE - 1);
 	const auto end   = (vaddr + size + TRACKER_PAGE_SIZE - 1) & ~(TRACKER_PAGE_SIZE - 1);
 	for (auto page = begin; page < end; page += TRACKER_PAGE_SIZE) {
-		bool has_dirty_bytes = false;
-		dirty.ForEachIntersection(page, TRACKER_PAGE_SIZE,
-		                          [&has_dirty_bytes](RangeSet::Range) { has_dirty_bytes = true; });
+		const bool has_dirty_bytes = dirty.Intersects(page, TRACKER_PAGE_SIZE);
 		if (IsRegionGpuModified(page, TRACKER_PAGE_SIZE) != has_dirty_bytes) {
 			EXIT("MemoryTracker: tracker and byte ownership disagree, operation=%s "
 			     "addr=0x%016" PRIx64 "\n",
@@ -56,8 +48,7 @@ void MemoryTracker::ValidateGpuDirtyOwnership(const RangeSet& dirty, uint64_t va
 #endif
 
 void MemoryTracker::ValidateRange(uint64_t vaddr, uint64_t size) {
-	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
-	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		EXIT("invalid memory tracker range\n");
 	}
 }
@@ -117,7 +108,8 @@ void MemoryTracker::UnmarkRegionAsGpuModified(uint64_t vaddr, uint64_t size) {
 	});
 }
 
-void MemoryTracker::UntrackMemoryImpl(uint64_t vaddr, uint64_t size) {
+void MemoryTracker::UntrackMemory(uint64_t vaddr, uint64_t size) {
+	CheckNotInUploadCallback();
 	std::vector<RegionManager*> managers;
 	managers.reserve((vaddr % TRACKER_REGION_SIZE + size + TRACKER_REGION_SIZE - 1) /
 	                 TRACKER_REGION_SIZE);
@@ -138,12 +130,6 @@ void MemoryTracker::UntrackMemoryImpl(uint64_t vaddr, uint64_t size) {
 	Iterate<false>(vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 		manager->ChangeState<DirtySource::Cpu, true>(manager->GetCpuAddr() + offset, bytes);
 	});
-	locks.clear();
-}
-
-void MemoryTracker::UntrackMemory(uint64_t vaddr, uint64_t size) {
-	CheckNotInUploadCallback();
-	UntrackMemoryImpl(vaddr, size);
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/memoryTracker.h

@@ -34,7 +34,6 @@ public:
 	void InvalidateRegion(uint64_t vaddr, uint64_t size, Flush&& on_flush) noexcept {
 		static_assert(std::is_invocable_v<Flush&>);
 		CheckNotInUploadCallback();
-		ValidateRange(vaddr, size);
 
 		Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 			const bool should_flush = [&] {
@@ -171,7 +170,6 @@ private:
 	}
 
 	static void    ValidateRange(uint64_t vaddr, uint64_t size);
-	void           UntrackMemoryImpl(uint64_t vaddr, uint64_t size);
 	RegionManager* GetOrCreateRegion(uint64_t index);
 
 	std::unique_ptr<std::atomic<RegionManager*>[]> m_regions;

src/graphics/host_gpu/pageManager.cpp

@@ -110,7 +110,7 @@ private:
 };
 
 void ValidateRange(uint64_t vaddr, uint64_t size) {
-	if (vaddr == 0 || size == 0 || vaddr >= ADDRESS_SIZE || size > ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		Fatal("invalid range vaddr=0x%016" PRIx64 ", size=0x%016" PRIx64, vaddr, size);
 	}
 }
@@ -371,8 +371,4 @@ template void PageManager::UpdatePageWatchersForRegion<true, false>(uint64_t, Re
 template void PageManager::UpdatePageWatchersForRegion<false, true>(uint64_t, RegionBits&);
 template void PageManager::UpdatePageWatchersForRegion<false, false>(uint64_t, RegionBits&);
 
-void PageManager::OnGpuMap(uint64_t, uint64_t) {}
-
-void PageManager::OnGpuUnmap(uint64_t, uint64_t) {}
-
 } // namespace Libs::Graphics

src/graphics/host_gpu/pageManager.h

@@ -24,8 +24,6 @@ public:
 	void UpdatePageWatchers(uint64_t vaddr, uint64_t size);
 	template <bool track, bool is_read = false>
 	void UpdatePageWatchersForRegion(uint64_t base_addr, RegionBits& mask);
-	void OnGpuMap(uint64_t vaddr, uint64_t size);
-	void OnGpuUnmap(uint64_t vaddr, uint64_t size);
 
 private:
 	struct Impl;

src/graphics/host_gpu/rangeSet.h

@@ -53,12 +53,30 @@ public:
 		}
 	}
 
+	void Clear() { m_ranges.clear(); }
+
+	template <typename Func>
+	void ForEach(Func&& func) const {
+		for (const auto& [begin, end]: m_ranges) {
+			func(begin, end);
+		}
+	}
+
 	[[nodiscard]] std::vector<Range> Intersections(uint64_t address, uint64_t size) const {
 		std::vector<Range> result;
 		ForEachIntersection(address, size, [&result](Range range) { result.push_back(range); });
 		return result;
 	}
 
+	[[nodiscard]] bool Intersects(uint64_t address, uint64_t size) const {
+		const auto end = End(address, size);
+		auto       it  = m_ranges.lower_bound(address);
+		if (it != m_ranges.begin() && std::prev(it)->second > address) {
+			return true;
+		}
+		return it != m_ranges.end() && it->first < end;
+	}
+
 	[[nodiscard]] bool Contains(uint64_t address, uint64_t size) const {
 		const auto end = End(address, size);
 		auto       it  = m_ranges.upper_bound(address);
@@ -89,7 +107,7 @@ public:
 
 private:
 	static uint64_t End(uint64_t address, uint64_t size) {
-		if (address == 0 || size == 0 || size > UINT64_MAX - address) {
+		if (size == 0 || size > UINT64_MAX - address) {
 			EXIT("invalid range-set address or size\n");
 		}
 		return address + size;

src/graphics/host_gpu/regionDefinitions.h

@@ -4,6 +4,8 @@
 #include "common/bitArray.h"
 #include "common/common.h"
 
+#include <compare>
+
 namespace Libs::Graphics {
 
 constexpr uint64_t TRACKER_PAGE_SIZE    = 4ull * 1024ull;
@@ -11,6 +13,20 @@ constexpr uint64_t TRACKER_REGION_SIZE  = 4ull * 1024ull * 1024ull;
 constexpr uint64_t TRACKER_ADDRESS_SIZE = 1ull << 40u;
 constexpr size_t   TRACKER_REGION_PAGES = TRACKER_REGION_SIZE / TRACKER_PAGE_SIZE;
 
+struct GuestRange {
+	uint64_t address = 0;
+	uint64_t size    = 0;
+
+	[[nodiscard]] constexpr bool Empty() const noexcept { return address == 0 && size == 0; }
+	[[nodiscard]] constexpr bool Valid() const noexcept {
+		return address != 0 && size != 0 && address < TRACKER_ADDRESS_SIZE &&
+		       size <= TRACKER_ADDRESS_SIZE - address;
+	}
+	[[nodiscard]] constexpr bool     ValidOrEmpty() const noexcept { return Empty() || Valid(); }
+	[[nodiscard]] constexpr uint64_t End() const noexcept { return address + size; }
+	auto                             operator<=>(const GuestRange&) const = default;
+};
+
 enum class DirtySource { Cpu, Gpu };
 using RegionBits = Common::BitArray<TRACKER_REGION_PAGES>;
 static_assert(sizeof(RegionBits) == TRACKER_REGION_PAGES / 8);

src/graphics/host_gpu/renderer/cache/bufferCache.h

@@ -3,12 +3,15 @@
 
 #include "common/abi.h"
 #include "common/common.h"
+#include "common/lruCache.h"
+#include "common/slotVector.h"
 #include "graphics/host_gpu/memoryTracker.h"
 #include "graphics/host_gpu/rangeSet.h"
+#include "graphics/host_gpu/renderer/cache/faultManager.h"
+#include "graphics/host_gpu/renderer/cache/multiLevelPageTable.h"
 #include "graphics/host_gpu/renderer/cache/streamBuffer.h"
 
 #include <map>
-#include <memory>
 #include <span>
 #include <utility>
 #include <vector>
@@ -16,93 +19,105 @@
 namespace Libs::Graphics {
 
 struct GraphicContext;
-class CommandBuffer;
 class CommandScheduler;
 class TextureCache;
 
-struct BufferBinding {
-	std::shared_ptr<void> owner;
-	vk::Buffer            buffer = nullptr;
-	uint64_t              offset = 0;
-};
-
-struct ImageBufferSource {
-	Buffer*  buffer = nullptr;
-	uint64_t offset = 0;
-};
+using BufferId = Common::SlotId;
+inline constexpr BufferId NULL_BUFFER_ID {0};
 
 class BufferCache {
 public:
-	static constexpr uint64_t CACHING_PAGE_SIZE = 16ull * 1024ull;
-	static constexpr uint64_t GetBufferOffset(uint64_t vaddr) {
-		return vaddr & (CACHING_PAGE_SIZE - 1);
-	}
+	static constexpr uint32_t CACHING_PAGEBITS  = 14;
+	static constexpr uint64_t CACHING_PAGESIZE  = uint64_t {1} << CACHING_PAGEBITS;
+	static constexpr uint64_t CACHING_NUMPAGES  = uint64_t {1} << (40 - CACHING_PAGEBITS);
+	static constexpr uint64_t BDA_PAGETABLE_SIZE =
+	    CACHING_NUMPAGES * sizeof(vk::DeviceAddress);
 
 	BufferCache(GraphicContext& graphics, CommandScheduler& scheduler, PageManager& page_manager,
 	            TextureCache& texture_cache);
 	~BufferCache();
 	KYTY_CLASS_NO_COPY(BufferCache);
 
-	void                        InvalidateMemory(uint64_t vaddr, uint64_t size);
-	void                        ReadMemory(uint64_t vaddr, uint64_t size, bool is_write = false);
-	void                        UnmapMemory(uint64_t vaddr, uint64_t size);
-	[[nodiscard]] BufferBinding ObtainBuffer(CommandBuffer& command, uint64_t vaddr, uint64_t size,
-	                                         bool is_written = false, bool is_read = true,
-	                                         bool is_formatted = false);
-	[[nodiscard]] StreamBuffer& GetUtilityBuffer(MemoryUsage usage) noexcept;
-	[[nodiscard]] Buffer&       GetGdsBuffer() noexcept { return m_gds_buffer; }
-	[[nodiscard]] const Buffer& GetGdsBuffer() const noexcept { return m_gds_buffer; }
-	[[nodiscard]] BufferBinding UploadTransient(const void* data, uint64_t size,
-	                                            uint64_t alignment);
-	[[nodiscard]] std::shared_ptr<Buffer> ObtainNullBuffer();
-	[[nodiscard]] ImageBufferSource       ObtainBufferForImage(uint64_t vaddr, uint64_t size);
-	void FillBuffer(uint64_t vaddr, uint64_t size, uint32_t value, bool is_gds = false);
-	void CopyBuffer(uint64_t dst_vaddr, uint64_t src_vaddr, uint64_t size, bool dst_gds = false,
-	                bool src_gds = false);
+	void                   InvalidateMemory(uint64_t vaddr, uint64_t size);
+	void                   ReadMemory(uint64_t vaddr, uint64_t size, bool is_write = false);
+	[[nodiscard]] Buffer&  GetBuffer(BufferId id) { return m_slot_buffers[id]; }
+	[[nodiscard]] BufferId FindBuffer(uint64_t vaddr, uint64_t size);
+	[[nodiscard]] std::pair<Buffer*, uint64_t> ObtainBuffer(uint64_t vaddr, uint64_t size,
+	                                                        bool     is_written,
+	                                                        bool     is_texel_buffer = false,
+	                                                        BufferId id              = {});
+	[[nodiscard]] StreamBuffer&                GetUtilityBuffer(MemoryUsage usage) noexcept {
+		switch (usage) {
+			case MemoryUsage::Upload: return m_staging_buffer;
+			case MemoryUsage::Stream: return m_stream_buffer;
+			case MemoryUsage::Download: return m_download_buffer;
+			case MemoryUsage::DeviceLocal: return m_device_buffer;
+		}
+		EXIT("BufferCache: invalid utility-buffer usage\n");
+	}
+	[[nodiscard]] const Buffer* GetGdsBuffer() const noexcept { return &m_gds_buffer; }
+	[[nodiscard]] Buffer* GetBdaPageTableBuffer() noexcept { return &m_bda_pagetable_buffer; }
+	[[nodiscard]] Buffer* GetFaultBuffer() noexcept { return m_fault_manager.GetFaultBuffer(); }
+	[[nodiscard]] std::pair<Buffer*, uint64_t> ObtainBufferForImage(uint64_t vaddr, uint64_t size);
+	void FillBuffer(uint64_t vaddr, uint64_t size, uint32_t value, bool is_gds);
+	void CopyBuffer(uint64_t dst_vaddr, uint64_t src_vaddr, uint64_t size, bool dst_gds,
+	                bool src_gds);
 	// Cache-index and exact dirty-range queries require GPU-thread serialization.
 	[[nodiscard]] bool IsRegionRegistered(uint64_t vaddr, uint64_t size);
 	[[nodiscard]] bool HasGpuDirtyBytes(uint64_t vaddr, uint64_t size);
 	[[nodiscard]] bool IsRegionCpuModified(uint64_t vaddr, uint64_t size);
 	[[nodiscard]] bool IsRegionGpuModified(uint64_t vaddr, uint64_t size);
+	void               ProcessFaultBuffer();
+	void               SynchronizeBuffersInRange(uint64_t vaddr, uint64_t size);
 	void               RunGarbageCollector();
 
 private:
 	friend struct BufferCacheTestAccess;
 
-	struct CacheRange {
-		uint64_t address = 0;
-		uint64_t size    = 0;
+	using BufferMap = std::map<uint64_t, BufferId>;
+	struct OverlapResult {
+		BufferMap::iterator first;
+		BufferMap::iterator last;
+		uint64_t            begin;
+		uint64_t            end;
+		bool                has_stream_leap;
 	};
-	struct CachedBuffer;
+
 	struct DownloadCopy;
-	struct DownloadRange;
-	struct RetiredBuffer;
+	using PageTable = MultiLevelPageTable<BufferId, CACHING_PAGEBITS, 40, 16>;
+	static_assert(CACHING_PAGESIZE == (uint64_t {1} << PageTable::kPageBits));
 	static constexpr uint64_t               DOWNLOAD_ALIGNMENT = 64;
-	[[nodiscard]] static uint64_t           AlignDown(uint64_t value) noexcept;
-	[[nodiscard]] static uint64_t           AlignUp(uint64_t value);
 	[[nodiscard]] static constexpr uint64_t AlignDownload(uint64_t size) noexcept {
 		return (size + DOWNLOAD_ALIGNMENT - 1) & ~(DOWNLOAD_ALIGNMENT - 1);
 	}
 	[[nodiscard]] static std::pair<uint64_t, uint64_t> DownloadEnvelope(const DownloadCopy& copy);
-	[[nodiscard]] static bool ResolveOverlap(CacheRange& merged, CacheRange candidate) noexcept;
-	void Upload(CommandBuffer& command, Buffer& destination, uint64_t destination_offset,
-	            const void* source, uint64_t size);
-	[[nodiscard]] CachedBuffer& GetOrCreateBuffer(CommandBuffer& command, uint64_t vaddr,
-	                                              uint64_t size);
-	[[nodiscard]] bool SynchronizeBuffer(CommandBuffer& command, Buffer& buffer, uint64_t vaddr,
-	                                     uint64_t size, bool is_written, bool is_texel_buffer);
+	void WriteDataBuffer(Buffer& buffer, uint64_t address, const void* source, uint64_t size);
+	void TouchBuffer(const Buffer& buffer);
+	[[nodiscard]] OverlapResult ResolveOverlaps(uint64_t vaddr, uint64_t size);
+	void JoinOverlap(BufferId new_id, BufferId overlap_id, bool accumulate_stream_score);
+	[[nodiscard]] BufferId CreateBuffer(uint64_t vaddr, uint64_t size);
+	void                   Register(BufferId id);
+	void Unregister(BufferId id);
+	template <bool insert>
+	void ChangeRegister(BufferId id);
+	void DeleteBuffer(BufferId id);
+	[[nodiscard]] bool SynchronizeBuffer(Buffer& buffer, uint64_t vaddr, uint64_t size,
+	                                     bool is_written, bool is_texel_buffer);
+	[[nodiscard]] vk::Buffer UploadCopies(Buffer& buffer, std::span<vk::BufferCopy> copies,
+	                                      uint64_t total_size);
 	[[nodiscard]] bool SynchronizeBufferFromImage(Buffer& buffer, uint64_t vaddr, uint64_t size);
-	[[nodiscard]] std::vector<DownloadRange> RecordDownloads(std::span<const DownloadCopy> copies);
-	void PublishDownloads(std::span<const DownloadRange> downloads);
-	void QueueGarbageDownload(std::span<const DownloadCopy> copies, RetiredBuffer retire);
-	void WriteHostMemory(uint64_t vaddr, std::span<const uint8_t> data);
+	void DownloadBufferMemory(std::span<const DownloadCopy> copies);
 	void ReadMemoryOnGpu(uint64_t vaddr, uint64_t size, bool is_write);
 
 	GraphicContext&                                   m_graphics;
 	CommandScheduler&                                 m_scheduler;
+	FaultManager                                      m_fault_manager;
 	Buffer                                            m_gds_buffer;
-	std::shared_ptr<Buffer>                           m_null_buffer;
-	std::map<uint64_t, std::unique_ptr<CachedBuffer>> m_buffers;
+	Buffer                                            m_bda_pagetable_buffer;
+	Common::SlotVector<Buffer>                        m_slot_buffers;
+	Common::LeastRecentlyUsedCache<BufferId, uint64_t> m_lru_cache;
+	BufferMap                                         m_buffers;
+	PageTable                                         m_page_table;
 	RangeSet                                          m_gpu_modified_ranges;
 	MemoryTracker                                     m_memory_tracker;
 	StreamBuffer                                      m_staging_buffer;
@@ -110,7 +125,7 @@ private:
 	StreamBuffer                                      m_download_buffer;
 	StreamBuffer                                      m_device_buffer;
 	TextureCache&                                     m_texture_cache;
-	uint64_t                                          m_total_used_memory = 0;
+	uint64_t                                          m_total_used_memory  = 0;
 	uint64_t m_trigger_gc_memory  = 1ull * 1024 * 1024 * 1024;
 	uint64_t m_critical_gc_memory = 2ull * 1024 * 1024 * 1024;
 	uint64_t m_gc_tick            = 0;

src/graphics/host_gpu/renderer/cache/faultManager.cpp

@@ -0,0 +1,156 @@
+#include "graphics/host_gpu/renderer/cache/faultManager.h"
+
+#include "common/assert.h"
+#include "common/logging/log.h"
+#include "gpu_tiler_shaders/fault_buffer_process_spv.h"
+#include "graphics/host_gpu/graphicContext.h"
+#include "graphics/host_gpu/renderer/cache/bufferCache.h"
+#include "graphics/host_gpu/renderer/commandScheduler.h"
+#include "graphics/host_gpu/vulkanCommon.h"
+
+#include <bit>
+#include <cinttypes>
+#include <cstring>
+#include <limits>
+
+namespace Libs::Graphics {
+
+namespace {
+
+constexpr size_t MaxPageFaults    = 1024;
+constexpr size_t PageFaultAreaSize = MaxPageFaults * sizeof(uint64_t);
+
+} // namespace
+
+FaultManager::FaultManager(GraphicContext& graphics, CommandScheduler& scheduler,
+                           BufferCache& buffer_cache)
+    : m_graphics(graphics), m_scheduler(scheduler), m_buffer_cache(buffer_cache),
+      m_fault_buffer(graphics, scheduler, MemoryUsage::DeviceLocal, 0, AllFlags,
+                     BufferCache::CACHING_NUMPAGES / 8),
+      m_download_buffer(graphics, scheduler, MemoryUsage::Download, 0, AllFlags,
+                        MaxPendingFaults * PageFaultAreaSize) {
+	SetVulkanObjectNameF(m_graphics.device, m_fault_buffer.Handle(), "Fault Buffer");
+
+	const vk::DescriptorSetLayoutBinding bindings[] {
+	    {0, vk::DescriptorType::eStorageBuffer, 1, vk::ShaderStageFlagBits::eCompute, nullptr},
+	    {1, vk::DescriptorType::eStorageBuffer, 1, vk::ShaderStageFlagBits::eCompute, nullptr},
+	};
+	vk::DescriptorSetLayoutCreateInfo layout_info {};
+	layout_info.flags        = vk::DescriptorSetLayoutCreateFlagBits::ePushDescriptorKHR;
+	layout_info.bindingCount = std::size(bindings);
+	layout_info.pBindings    = bindings;
+	RequireVulkanSuccess(
+	    m_graphics.device.createDescriptorSetLayout(&layout_info, nullptr,
+	                                                &m_fault_process_desc_layout),
+	    "create fault-buffer descriptor layout");
+
+	vk::ShaderModuleCreateInfo module_info {};
+	module_info.codeSize = std::size(FAULT_BUFFER_PROCESS_SPV) * sizeof(uint32_t);
+	module_info.pCode    = FAULT_BUFFER_PROCESS_SPV;
+	vk::ShaderModule module = nullptr;
+	RequireVulkanSuccess(m_graphics.device.createShaderModule(&module_info, nullptr, &module),
+	                     "create fault-buffer shader module");
+
+	vk::PipelineLayoutCreateInfo pipeline_layout_info {};
+	pipeline_layout_info.setLayoutCount = 1;
+	pipeline_layout_info.pSetLayouts    = &m_fault_process_desc_layout;
+	RequireVulkanSuccess(
+	    m_graphics.device.createPipelineLayout(&pipeline_layout_info, nullptr,
+	                                           &m_fault_process_pipeline_layout),
+	    "create fault-buffer pipeline layout");
+
+	vk::PipelineShaderStageCreateInfo stage {};
+	stage.stage  = vk::ShaderStageFlagBits::eCompute;
+	stage.module = module;
+	stage.pName  = "main";
+	vk::ComputePipelineCreateInfo pipeline_info {};
+	pipeline_info.stage  = stage;
+	pipeline_info.layout = m_fault_process_pipeline_layout;
+	const auto result = m_graphics.device.createComputePipelines(
+	    nullptr, 1, &pipeline_info, nullptr, &m_fault_process_pipeline);
+	m_graphics.device.destroyShaderModule(module, nullptr);
+	RequireVulkanSuccess(result, "create fault-buffer pipeline");
+	SetVulkanObjectNameF(m_graphics.device, m_fault_process_pipeline, "Fault Buffer Parser");
+}
+
+FaultManager::~FaultManager() {
+	m_graphics.device.destroyPipeline(m_fault_process_pipeline, nullptr);
+	m_graphics.device.destroyPipelineLayout(m_fault_process_pipeline_layout, nullptr);
+	m_graphics.device.destroyDescriptorSetLayout(m_fault_process_desc_layout, nullptr);
+}
+
+void FaultManager::ProcessFaultBuffer() {
+	if (const auto wait_tick = m_fault_areas[m_current_area]; wait_tick != 0) {
+		m_scheduler.Wait(wait_tick);
+		m_scheduler.PopPendingOperations();
+	}
+
+	const auto offset = m_current_area * PageFaultAreaSize;
+	auto*      mapped = m_download_buffer.Mapped().data() + offset;
+	std::memset(mapped, 0, PageFaultAreaSize);
+	m_download_buffer.Flush(offset, PageFaultAreaSize);
+
+	vk::BufferMemoryBarrier2 pre_barrier {};
+	pre_barrier.srcStageMask  = vk::PipelineStageFlagBits2::eAllCommands;
+	pre_barrier.srcAccessMask = vk::AccessFlagBits2::eShaderWrite;
+	pre_barrier.dstStageMask  = vk::PipelineStageFlagBits2::eComputeShader;
+	pre_barrier.dstAccessMask = vk::AccessFlagBits2::eShaderRead;
+	pre_barrier.buffer        = m_fault_buffer.Handle();
+	pre_barrier.offset        = 0;
+	pre_barrier.size           = m_fault_buffer.Size();
+	auto post_barrier         = pre_barrier;
+	post_barrier.srcStageMask  = vk::PipelineStageFlagBits2::eComputeShader;
+	post_barrier.srcAccessMask = vk::AccessFlagBits2::eShaderWrite;
+	post_barrier.dstStageMask  = vk::PipelineStageFlagBits2::eAllCommands;
+	post_barrier.dstAccessMask = vk::AccessFlagBits2::eShaderWrite;
+
+	const vk::DescriptorBufferInfo infos[] {
+	    {m_fault_buffer.Handle(), 0, m_fault_buffer.Size()},
+	    {m_download_buffer.Handle(), offset, PageFaultAreaSize},
+	};
+	std::array<vk::WriteDescriptorSet, 2> writes {};
+	for (uint32_t index = 0; index < writes.size(); ++index) {
+		writes[index].dstBinding      = index;
+		writes[index].descriptorCount = 1;
+		writes[index].descriptorType  = vk::DescriptorType::eStorageBuffer;
+		writes[index].pBufferInfo     = &infos[index];
+	}
+
+	m_scheduler.EndRendering();
+	auto command = m_scheduler.Current().Handle();
+	vk::DependencyInfo dependency {};
+	dependency.dependencyFlags          = vk::DependencyFlagBits::eByRegion;
+	dependency.bufferMemoryBarrierCount = 1;
+	dependency.pBufferMemoryBarriers    = &pre_barrier;
+	command.pipelineBarrier2(dependency);
+	command.bindPipeline(vk::PipelineBindPoint::eCompute, m_fault_process_pipeline);
+	command.pushDescriptorSetKHR(vk::PipelineBindPoint::eCompute,
+	                             m_fault_process_pipeline_layout, 0, writes);
+	const auto num_threads    = BufferCache::CACHING_NUMPAGES / 32;
+	const auto num_workgroups = (num_threads + 63) / 64;
+	command.dispatch(static_cast<uint32_t>(num_workgroups), 1, 1);
+	dependency.pBufferMemoryBarriers = &post_barrier;
+	command.pipelineBarrier2(dependency);
+
+	const auto area = m_current_area;
+	m_scheduler.DeferOperation([this, mapped, offset, area] {
+		m_download_buffer.Invalidate(offset, PageFaultAreaSize);
+		RangeSet    fault_ranges;
+		const auto* faults = std::bit_cast<const uint64_t*>(mapped);
+		const auto  count  = static_cast<uint32_t>(faults[0]);
+		for (uint32_t index = 1; index <= count; ++index) {
+			fault_ranges.Add(faults[index], BufferCache::CACHING_PAGESIZE);
+			LOGF("Accessed non-GPU cached memory at 0x%016" PRIx64 "\n", faults[index]);
+		}
+		fault_ranges.ForEach([this](uint64_t start, uint64_t end) {
+			EXIT_IF(end - start > std::numeric_limits<uint32_t>::max());
+			(void)m_buffer_cache.FindBuffer(start, end - start);
+		});
+		m_fault_areas[area] = 0;
+	});
+
+	m_fault_areas[m_current_area++] = m_scheduler.CurrentTick();
+	m_current_area %= MaxPendingFaults;
+}
+
+} // namespace Libs::Graphics

src/graphics/host_gpu/renderer/cache/faultManager.h

@@ -0,0 +1,40 @@
+#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_CACHE_FAULTMANAGER_H_
+#define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_CACHE_FAULTMANAGER_H_
+
+#include "common/abi.h"
+#include "graphics/host_gpu/renderer/cache/streamBuffer.h"
+
+#include <array>
+#include <cstdint>
+
+namespace Libs::Graphics {
+
+class BufferCache;
+
+class FaultManager {
+	static constexpr size_t MaxPendingFaults = 8;
+
+public:
+	FaultManager(GraphicContext& graphics, CommandScheduler& scheduler, BufferCache& buffer_cache);
+	~FaultManager();
+	KYTY_CLASS_NO_COPY(FaultManager);
+
+	[[nodiscard]] Buffer* GetFaultBuffer() noexcept { return &m_fault_buffer; }
+	void                  ProcessFaultBuffer();
+
+private:
+	GraphicContext&                            m_graphics;
+	CommandScheduler&                          m_scheduler;
+	BufferCache&                               m_buffer_cache;
+	Buffer                                     m_fault_buffer;
+	Buffer                                     m_download_buffer;
+	std::array<uint64_t, MaxPendingFaults>      m_fault_areas {};
+	uint32_t                                   m_current_area = 0;
+	vk::DescriptorSetLayout                    m_fault_process_desc_layout = nullptr;
+	vk::Pipeline                               m_fault_process_pipeline = nullptr;
+	vk::PipelineLayout                         m_fault_process_pipeline_layout = nullptr;
+};
+
+} // namespace Libs::Graphics
+
+#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_CACHE_FAULTMANAGER_H_

src/graphics/host_gpu/renderer/cache/gpuResourceManager.cpp

@@ -12,7 +12,9 @@ GpuResourceManager::GpuResourceManager(GraphicContext& graphics, CommandSchedule
 GpuResourceManager::~GpuResourceManager() = default;
 
 bool GpuResourceManager::HandleFault(PageFaultAccess access, uint64_t fault_vaddr) noexcept {
-	constexpr uint64_t fault_size = 8;
+	// The host reports the faulting byte, not the instruction's access width. Both caches
+	// resolve its page; guessing a width can cross the end of a valid guest mapping.
+	constexpr uint64_t fault_size = 1;
 	if (!IsMapped(fault_vaddr, fault_size)) {
 		return false;
 	}
@@ -35,8 +37,7 @@ bool GpuResourceManager::InvalidateMemory(uint64_t vaddr, uint64_t size) {
 }
 
 bool GpuResourceManager::IsMapped(uint64_t vaddr, uint64_t size) const noexcept {
-	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
-	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+	if (!GuestRange {vaddr, size}.Valid()) {
 		return false;
 	}
 	std::shared_lock lock(m_mapped_ranges_mutex);
@@ -48,7 +49,6 @@ void GpuResourceManager::MapMemory(uint64_t vaddr, uint64_t size) {
 		std::lock_guard lock(m_mapped_ranges_mutex);
 		m_mapped_ranges.Add(vaddr, size);
 	}
-	m_page_manager.OnGpuMap(vaddr, size);
 }
 
 void GpuResourceManager::UnmapMemory(uint64_t vaddr, uint64_t size) {
@@ -60,12 +60,11 @@ void GpuResourceManager::UnmapMemory(uint64_t vaddr, uint64_t size) {
 	const auto unmap = [this, vaddr, size] {
 		if (m_scheduler.Active()) {
 			const auto tick = m_scheduler.CurrentTick();
-			m_scheduler.FinishCurrent();
+			m_scheduler.Finish();
 			m_scheduler.WaitPriorityOperations(tick);
 		}
-		m_buffer_cache.UnmapMemory(vaddr, size);
+		m_buffer_cache.InvalidateMemory(vaddr, size);
 		m_texture_cache.UnmapMemory(vaddr, size);
-		m_page_manager.OnGpuUnmap(vaddr, size);
 		std::lock_guard lock(m_mapped_ranges_mutex);
 		m_mapped_ranges.Subtract(vaddr, size);
 	};
@@ -76,7 +75,19 @@ void GpuResourceManager::UnmapMemory(uint64_t vaddr, uint64_t size) {
 	m_gpu->SendCommandSync(unmap);
 }
 
+void GpuResourceManager::PrepareBda() {
+	std::shared_lock lock(m_mapped_ranges_mutex);
+	m_mapped_ranges.ForEach([this](uint64_t start, uint64_t end) {
+		m_buffer_cache.SynchronizeBuffersInRange(start, end - start);
+	});
+	m_fault_process_pending = true;
+}
+
 void GpuResourceManager::RunGarbageCollector() {
+	if (m_fault_process_pending) {
+		m_fault_process_pending = false;
+		m_buffer_cache.ProcessFaultBuffer();
+	}
 	m_texture_cache.ProcessDownloadImages();
 	m_texture_cache.RunGarbageCollector();
 	m_buffer_cache.RunGarbageCollector();

src/graphics/host_gpu/renderer/cache/gpuResourceManager.h

@@ -30,6 +30,7 @@ public:
 	[[nodiscard]] bool IsMapped(uint64_t vaddr, uint64_t size) const noexcept;
 	void               MapMemory(uint64_t vaddr, uint64_t size);
 	void               UnmapMemory(uint64_t vaddr, uint64_t size);
+	void               PrepareBda();
 	void               RunGarbageCollector();
 
 private:
@@ -40,6 +41,7 @@ private:
 	mutable std::shared_mutex m_mapped_ranges_mutex;
 	RangeSet                  m_mapped_ranges;
 	GuestGpu*                 m_gpu = nullptr;
+	bool                      m_fault_process_pending = false;
 };
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/cache/multiLevelPageTable.h

@@ -256,17 +256,6 @@ private:
 	size_type                              m_inline_size = 0;
 };
 
-// Removes only the requested owner. Other owners in the same page entry remain intact.
-template <typename Container, typename Value>
-[[nodiscard]] bool EraseExact(Container& owners, const Value& owner) {
-	const auto it = std::find(owners.begin(), owners.end(), owner);
-	if (it == owners.end()) {
-		return false;
-	}
-	owners.erase(it);
-	return true;
-}
-
 } // namespace Libs::Graphics
 
 #endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_MULTILEVELPAGETABLE_H_

src/graphics/host_gpu/renderer/cache/samplerCache.cpp

@@ -116,9 +116,6 @@ vk::Sampler SamplerCache::GetSampler(const ShaderSamplerResource& r) {
 		default: EXIT("unknown border color: %d", static_cast<int>(r.BorderColorType()));
 	}
 
-	sampler_info.sType     = vk::StructureType::eSamplerCreateInfo;
-	sampler_info.pNext     = nullptr;
-	sampler_info.flags     = {};
 	sampler_info.magFilter = to_vk_filter(mag_filter);
 	sampler_info.minFilter = to_vk_filter(min_filter);
 	sampler_info.mipmapMode =

src/graphics/host_gpu/renderer/cache/streamBuffer.cpp

@@ -64,17 +64,20 @@ constexpr size_t WATCHES_RESERVE_CHUNK   = 0x1000;
 Buffer::Buffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsage usage,
                uint64_t cpu_address, vk::BufferUsageFlags flags, uint64_t size)
     : m_graphics(&graphics), m_scheduler(&scheduler), m_usage(usage), m_cpu_address(cpu_address),
-      m_size(size), m_buffer(std::make_unique<VulkanBuffer>()) {
+      m_buffer(std::make_unique<VulkanBuffer>()) {
 	KYTY_PROFILER_FUNCTION();
 	EXIT_IF(graphics.allocator == nullptr || size == 0);
 
 	vk::BufferCreateInfo buffer_info {};
 	buffer_info.size        = size;
 	buffer_info.usage       = flags;
-	buffer_info.sharingMode = vk::SharingMode::eExclusive;
 
+	const bool with_bda = bool(flags & vk::BufferUsageFlagBits::eShaderDeviceAddress);
+	const VmaAllocationCreateFlags bda_flag =
+	    with_bda ? VMA_ALLOCATION_CREATE_DEDICATED_MEMORY_BIT : 0;
 	VmaAllocationCreateInfo allocation_info {};
-	allocation_info.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | AllocationFlags(usage);
+	allocation_info.flags =
+	    VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | bda_flag | AllocationFlags(usage);
 	allocation_info.usage = AllocationUsage(usage);
 	allocation_info.preferredFlags = usage == MemoryUsage::DeviceLocal
 	                                     ? VkMemoryPropertyFlags {}
@@ -93,17 +96,18 @@ Buffer::Buffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsag
 	m_buffer->buffer                 = native_buffer;
 	m_buffer->usage                  = flags;
 	m_buffer->buffer_size            = size;
-	m_buffer->memory.allocation_info = allocation_result;
-	m_buffer->memory.memory          = allocation_result.deviceMemory;
-	m_buffer->memory.offset          = allocation_result.offset;
 	m_buffer->memory.type            = allocation_result.memoryType;
-	m_buffer->memory.unique_id       = VulkanNextMemoryUniqueId();
 	graphics.device.getBufferMemoryRequirements(m_buffer->buffer, &m_buffer->memory.requirements);
+	if (static_cast<bool>(flags & vk::BufferUsageFlagBits::eShaderDeviceAddress)) {
+		vk::BufferDeviceAddressInfo address_info {};
+		address_info.buffer = m_buffer->buffer;
+		m_device_address    = graphics.device.getBufferAddress(address_info);
+		EXIT_IF(m_device_address == 0);
+	}
 
 	VkMemoryPropertyFlags properties = 0;
 	vmaGetAllocationMemoryProperties(graphics.allocator, m_buffer->memory.allocation, &properties);
 	m_buffer->memory.property = vk::MemoryPropertyFlags(properties);
-	m_is_coherent             = (properties & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) != 0;
 	if (allocation_result.pMappedData != nullptr) {
 		m_mapped = {static_cast<uint8_t*>(allocation_result.pMappedData),
 		            static_cast<size_t>(size)};
@@ -122,34 +126,55 @@ vk::Buffer Buffer::Handle() const noexcept {
 	return m_buffer->buffer;
 }
 
+uint64_t Buffer::Size() const noexcept {
+	return m_buffer->buffer_size;
+}
+
+bool Buffer::IsCoherent() const noexcept {
+	return bool(m_buffer->memory.property & vk::MemoryPropertyFlagBits::eHostCoherent);
+}
+
+vk::DeviceAddress Buffer::BufferDeviceAddress() const noexcept {
+	EXIT_IF(m_device_address == 0);
+	return m_device_address;
+}
+
 bool Buffer::IsInBounds(uint64_t address, uint64_t size) const noexcept {
-	return address >= m_cpu_address && size <= m_size && address - m_cpu_address <= m_size - size;
+	return address >= m_cpu_address && size <= Size() && address - m_cpu_address <= Size() - size;
 }
 
 void Buffer::Write(uint64_t offset, const void* source, uint64_t size) {
-	EXIT_IF(source == nullptr || m_mapped.empty() || offset > m_size || size > m_size - offset);
+	EXIT_IF(source == nullptr || m_mapped.empty() || offset > Size() || size > Size() - offset);
 	std::memcpy(m_mapped.data() + offset, source, static_cast<size_t>(size));
 	Flush(offset, size);
 }
 
 void Buffer::Flush(uint64_t offset, uint64_t size) {
-	EXIT_IF(m_mapped.empty() || offset > m_size || size > m_size - offset);
-	if (!m_is_coherent && size != 0) {
+	EXIT_IF(m_mapped.empty() || offset > Size() || size > Size() - offset);
+	if (!IsCoherent() && size != 0) {
 		const auto result =
 		    vmaFlushAllocation(m_graphics->allocator, m_buffer->memory.allocation, offset, size);
 		EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
 	}
 }
 
+void Buffer::Invalidate(uint64_t offset, uint64_t size) {
+	EXIT_IF(m_usage != MemoryUsage::Download || offset > Size() || size > Size() - offset);
+	if (!IsCoherent() && size != 0) {
+		const auto result =
+		    vmaInvalidateAllocation(m_graphics->allocator, m_buffer->memory.allocation, offset, size);
+		EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
+	}
+}
+
 vk::BufferMemoryBarrier Buffer::Barrier(uint64_t offset, uint64_t size, vk::AccessFlags source,
                                         vk::AccessFlags destination) const {
-	if (Handle() == nullptr || size == 0 || offset > m_size || size > m_size - offset) {
+	if (Handle() == nullptr || size == 0 || offset > Size() || size > Size() - offset) {
 		EXIT("Buffer: invalid DMA barrier, handle=%p offset=0x%016" PRIx64 " size=0x%016" PRIx64
 		     " capacity=0x%016" PRIx64 "\n",
-		     static_cast<const void*>(Handle()), offset, size, m_size);
+		     static_cast<const void*>(Handle()), offset, size, Size());
 	}
 	vk::BufferMemoryBarrier barrier {};
-	barrier.sType               = vk::StructureType::eBufferMemoryBarrier;
 	barrier.srcAccessMask       = source;
 	barrier.dstAccessMask       = destination;
 	barrier.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
@@ -164,8 +189,8 @@ void Buffer::CopyFrom(CommandBuffer& command, const Buffer& source, uint64_t sou
                       uint64_t destination_offset, uint64_t size, vk::AccessFlags source_before,
                       vk::AccessFlags destination_before, vk::AccessFlags source_after,
                       vk::AccessFlags destination_after) {
-	if (size == 0 || source_offset > source.m_size || size > source.m_size - source_offset ||
-	    destination_offset > m_size || size > m_size - destination_offset) {
+	if (size == 0 || source_offset > source.Size() || size > source.Size() - source_offset ||
+	    destination_offset > Size() || size > Size() - destination_offset) {
 		EXIT("Buffer: invalid copy range\n");
 	}
 	if (source.Handle() == Handle() && source_offset < destination_offset + size &&
@@ -222,10 +247,8 @@ void Buffer::Fill(uint64_t offset, uint64_t size, uint32_t value) {
 
 StreamBuffer::StreamBuffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsage usage,
                            uint64_t size)
-    : Buffer(graphics, scheduler, usage, 0, AllFlags, size) {
-	ReserveWatches(m_current_watches, WATCHES_INITIAL_RESERVE);
-	ReserveWatches(m_previous_watches, WATCHES_INITIAL_RESERVE);
-}
+    : Buffer(graphics, scheduler, usage, 0, AllFlags, size),
+      m_current_watches(WATCHES_INITIAL_RESERVE), m_previous_watches(WATCHES_INITIAL_RESERVE) {}
 
 bool StreamBuffer::NormalizeReservation(bool coherent, uint64_t atom, uint64_t& size,
                                         uint64_t& alignment) {
@@ -290,10 +313,8 @@ std::pair<uint8_t*, uint64_t> StreamBuffer::Map(uint64_t size, uint64_t alignmen
 }
 
 void StreamBuffer::Commit() {
-	if (!IsCoherent() && Usage() != MemoryUsage::Download && m_mapped_size != 0) {
-		const auto result = vmaFlushAllocation(
-		    Graphics().allocator, NativeBuffer().memory.allocation, m_offset, m_mapped_size);
-		EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
+	if (Usage() != MemoryUsage::Download && m_mapped_size != 0) {
+		Flush(m_offset, m_mapped_size);
 	}
 
 	m_offset += m_mapped_size;
@@ -303,23 +324,13 @@ void StreamBuffer::Commit() {
 		return;
 	}
 	if (m_current_watch_cursor + 1 >= m_current_watches.size()) {
-		ReserveWatches(m_current_watches, WATCHES_RESERVE_CHUNK);
+		m_current_watches.resize(m_current_watches.size() + WATCHES_RESERVE_CHUNK);
 	}
 	auto& watch       = m_current_watches[m_current_watch_cursor++];
 	watch.upper_bound = m_offset;
 	watch.tick        = tick;
 }
 
-void StreamBuffer::Invalidate(uint64_t offset, uint64_t size) {
-	EXIT_IF(Usage() != MemoryUsage::Download || offset > Size() || size > Size() - offset);
-	if (IsCoherent() || size == 0) {
-		return;
-	}
-	const auto result = vmaInvalidateAllocation(Graphics().allocator,
-	                                            NativeBuffer().memory.allocation, offset, size);
-	EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
-}
-
 uint64_t StreamBuffer::Copy(const void* source, uint64_t size, uint64_t alignment) {
 	EXIT_IF(source == nullptr);
 	const auto [data, offset] = Map(size, alignment);
@@ -329,10 +340,6 @@ uint64_t StreamBuffer::Copy(const void* source, uint64_t size, uint64_t alignmen
 	return offset;
 }
 
-void StreamBuffer::ReserveWatches(std::vector<Watch>& watches, size_t grow_size) {
-	watches.resize(watches.size() + grow_size);
-}
-
 bool StreamBuffer::WaitPendingOperations(const std::vector<Watch>& watches,
                                          std::optional<size_t>     invalidation_mark,
                                          uint64_t requested_upper_bound, bool allow_wait,

src/graphics/host_gpu/renderer/cache/streamBuffer.h

@@ -43,17 +43,21 @@ public:
 	KYTY_CLASS_NO_COPY(Buffer);
 
 	[[nodiscard]] vk::Buffer         Handle() const noexcept;
-	[[nodiscard]] uint64_t           Size() const noexcept { return m_size; }
+	[[nodiscard]] uint64_t           Size() const noexcept;
 	[[nodiscard]] std::span<uint8_t> Mapped() const noexcept { return m_mapped; }
-	[[nodiscard]] bool               IsCoherent() const noexcept { return m_is_coherent; }
+	[[nodiscard]] bool               IsCoherent() const noexcept;
 	[[nodiscard]] MemoryUsage        Usage() const noexcept { return m_usage; }
 	[[nodiscard]] uint64_t           CpuAddress() const noexcept { return m_cpu_address; }
+	[[nodiscard]] vk::DeviceAddress BufferDeviceAddress() const noexcept;
 	[[nodiscard]] uint64_t           Offset(uint64_t address) const noexcept {
 		return address - m_cpu_address;
 	}
 	[[nodiscard]] bool IsInBounds(uint64_t address, uint64_t size) const noexcept;
+	void               IncreaseStreamScore(int score) noexcept { stream_score += score; }
+	[[nodiscard]] int  StreamScore() const noexcept { return stream_score; }
 	void               Write(uint64_t offset, const void* source, uint64_t size);
 	void               Flush(uint64_t offset, uint64_t size);
+	void               Invalidate(uint64_t offset, uint64_t size);
 	void CopyFrom(CommandBuffer& command, const Buffer& source, uint64_t source_offset,
 	              uint64_t destination_offset, uint64_t size,
 	              vk::AccessFlags source_before      = vk::AccessFlagBits::eMemoryWrite,
@@ -65,10 +69,14 @@ public:
 	                                                   vk::AccessFlagBits::eMemoryWrite);
 	void Fill(uint64_t offset, uint64_t size, uint32_t value);
 
+	// BufferCache state lives directly on the resource.
+	bool   is_deleted   = false;
+	int    stream_score = 0;
+	size_t lru_id       = 0;
+
 protected:
 	[[nodiscard]] GraphicContext&   Graphics() const noexcept { return *m_graphics; }
 	[[nodiscard]] CommandScheduler& Scheduler() const noexcept { return *m_scheduler; }
-	[[nodiscard]] VulkanBuffer&     NativeBuffer() noexcept { return *m_buffer; }
 
 private:
 	[[nodiscard]] vk::BufferMemoryBarrier Barrier(uint64_t offset, uint64_t size,
@@ -79,10 +87,9 @@ private:
 	CommandScheduler*             m_scheduler   = nullptr;
 	MemoryUsage                   m_usage       = MemoryUsage::DeviceLocal;
 	uint64_t                      m_cpu_address = 0;
-	uint64_t                      m_size        = 0;
+	vk::DeviceAddress             m_device_address = 0;
 	std::unique_ptr<VulkanBuffer> m_buffer;
 	std::span<uint8_t>            m_mapped;
-	bool                          m_is_coherent = false;
 };
 
 class StreamBuffer final: public Buffer {
@@ -93,9 +100,6 @@ public:
 	[[nodiscard]] std::pair<uint8_t*, uint64_t> Map(uint64_t size, uint64_t alignment = 0,
 	                                                bool allow_wait = true);
 	void                                        Commit();
-	// Download mappings become visible to the CPU only after their GPU completion tick is free.
-	// Call this from the scheduler's deferred completion operation before reading Mapped().
-	void                   Invalidate(uint64_t offset, uint64_t size);
 	[[nodiscard]] uint64_t Copy(const void* source, uint64_t size, uint64_t alignment = 0);
 
 private:
@@ -106,7 +110,6 @@ private:
 		uint64_t upper_bound = 0;
 	};
 
-	void                      ReserveWatches(std::vector<Watch>& watches, size_t grow_size);
 	[[nodiscard]] static bool NormalizeReservation(bool coherent, uint64_t atom, uint64_t& size,
 	                                               uint64_t& alignment);
 	[[nodiscard]] bool        WaitPendingOperations(const std::vector<Watch>& watches,

src/graphics/host_gpu/renderer/cache/textureCache.h

@@ -4,17 +4,17 @@
 #include "common/abi.h"
 #include "common/common.h"
 #include "common/lruCache.h"
+#include "common/slotVector.h"
 #include "graphics/host_gpu/pageManager.h"
 #include "graphics/host_gpu/regionManager.h"
 #include "graphics/host_gpu/renderer/cache/multiLevelPageTable.h"
 #include "graphics/host_gpu/renderer/image/blitHelper.h"
 #include "graphics/host_gpu/renderer/image/image.h"
+#include "graphics/host_gpu/renderer/image/tiler.h"
 
-#include <compare>
 #include <map>
-#include <memory>
-#include <set>
-#include <utility>
+#include <unordered_map>
+#include <unordered_set>
 #include <vector>
 
 namespace Libs::Graphics {
@@ -25,8 +25,6 @@ class BufferCache;
 class CommandBuffer;
 class CommandScheduler;
 class RenderExecutor;
-class StreamBuffer;
-class TileManager;
 struct TextureCacheTestAccess;
 
 class TextureCache {
@@ -39,12 +37,6 @@ public:
 		BindingType   type = BindingType::Texture;
 	};
 
-	struct RegionInfo {
-		bool image_pages     = false;
-		bool image_bytes     = false;
-		bool gpu_image_bytes = false;
-	};
-
 	TextureCache(GraphicContext& graphics, CommandScheduler& scheduler, PageManager& page_manager,
 	             BufferCache& buffer_cache);
 	~TextureCache();
@@ -57,20 +49,25 @@ public:
 	[[nodiscard]] vk::ImageView FindTexture(ImageId id, const ImageDesc& desc);
 	[[nodiscard]] vk::ImageView FindRenderTarget(ImageId id, const ImageDesc& desc);
 	[[nodiscard]] vk::ImageView FindDepthTarget(ImageId id, const ImageDesc& desc);
-	[[nodiscard]] Image&        GetImage(ImageId id);
-	[[nodiscard]] const Image&  GetImage(ImageId id) const;
-	void                        MarkGpuWritten(ImageId id);
+	[[nodiscard]] Image&        GetImage(ImageId id) {
+		auto& image = m_slot_images[id];
+		TouchImage(image);
+		return image;
+	}
+	void MarkGpuWritten(ImageId id);
 
 	[[nodiscard]] bool ClearImageFromBuffer(CommandBuffer& command, uint64_t address, uint64_t size,
 	                                        uint32_t packed_clear);
 	void               InvalidateMemory(uint64_t address, uint64_t size);
-	[[nodiscard]] bool InvalidateMemoryFromGPU(uint64_t address, uint64_t size,
-	                                           bool formatted_buffer_write = false);
-	[[nodiscard]] RegionInfo QueryRegion(uint64_t address, uint64_t size);
+	void               InvalidateMemoryFromGPU(uint64_t address, uint64_t size);
+	[[nodiscard]] bool IsRegionGpuModified(uint64_t address, uint64_t size);
 
 	[[nodiscard]] bool IsMeta(uint64_t address);
-	[[nodiscard]] bool IsMetaCleared(uint64_t address, uint32_t slice);
+	[[nodiscard]] bool IsMetaCleared(uint64_t address, uint32_t slice,
+	                                 uint32_t* fill_value = nullptr);
 	[[nodiscard]] bool ClearMeta(uint64_t address);
+	// Record deferred DCC state while the original guest dispatch writes the metadata.
+	void               TrackDccFill(uint64_t address, uint64_t size, uint32_t fill_value);
 	[[nodiscard]] bool TouchMeta(uint64_t address, uint32_t slice, bool is_clear);
 
 	void UnmapMemory(uint64_t address, uint64_t size);
@@ -79,16 +76,21 @@ public:
 
 private:
 	enum class TransferDirection { Upload, Download };
-	struct ColorTransferPlan;
+	struct TextureTransferPlan;
 	struct DownloadPlan;
 
-	struct Slot {
-		std::shared_ptr<Image> image;
-		uint32_t               generation = 1;
-	};
-
 	struct MetaDataInfo {
+		// A guest metadata-fill dispatch may initialize DCC before its render target is bound.
+		// PendingDcc retains that exact fill until an image binding classifies the address,
+		// without exposing an unconfirmed buffer address to the normal metadata heuristics.
+		// Keep all surface metadata in one entry so CMask/FMask can be
+		// registered beside HTile and DCC without introducing parallel tracking paths.
+		enum class Type : uint8_t { PendingDcc, CMask, FMask, HTile, Dcc };
+
+		Type     type       = Type::PendingDcc;
 		uint32_t clear_mask = 0;
+		uint32_t fill_value = 0xffffffffu;
+		uint64_t fill_size  = 0;
 	};
 
 	struct OverlapResult {
@@ -100,26 +102,21 @@ private:
 	using ImageIds       = InlinePageOwnerList<ImageId, 16>;
 	using ImagePageTable = MultiLevelPageTable<ImageIds, 20, 40, 10>;
 
-	[[nodiscard]] Image&                 ResolveImage(ImageId id);
-	[[nodiscard]] const Image&           ResolveImage(ImageId id) const;
-	[[nodiscard]] std::shared_ptr<Image> ResolveOwner(ImageId id) const;
-	[[nodiscard]] ImageId                InsertImage(const ImageInfo& info);
-	[[nodiscard]] ImageId                GetNullImage(const ImageDesc& desc);
-	void                                 RegisterImage(ImageId id);
-	void                                 UnregisterImage(ImageId id);
-	void                                 DeleteImage(ImageId id);
-	void                                 FreeImage(ImageId id);
-	void                                 RetainImage(CommandBuffer& command, ImageId id);
-	void                                 TouchImage(Image& image);
-	void                                 TrackImage(ImageId id);
-	void                                 TrackImageHead(ImageId id);
-	void                                 TrackImageTail(ImageId id);
-	void                                 UntrackImage(ImageId id);
-	void                                 UntrackImageHead(ImageId id);
-	void                                 UntrackImageTail(ImageId id);
-	void                                 MarkAsMaybeDirty(ImageId id, Image& image);
-	void                                 TrackImageDownload(ImageId id);
-	void                                 TrackImageDownloadLocked(ImageId id, Image& image);
+	[[nodiscard]] ImageId     InsertImage(const ImageInfo& info);
+	[[nodiscard]] ImageId     GetNullImage(const ImageDesc& desc);
+	void                      RegisterImage(ImageId id);
+	void                      UnregisterImage(ImageId id);
+	void                      DeleteImage(ImageId id);
+	void                      FreeImage(ImageId id);
+	void                      TouchImage(Image& image);
+	void                      TrackImage(ImageId id);
+	void                      TrackImageHead(ImageId id);
+	void                      TrackImageTail(ImageId id);
+	void                      UntrackImage(ImageId id);
+	void                      UntrackImageHead(ImageId id);
+	void                      UntrackImageTail(ImageId id);
+	void                      MarkAsMaybeDirty(ImageId id, Image& image);
+	void                      TrackImageDownload(ImageId id, Image& image);
 	[[nodiscard]] static bool SameBacking(const ImageInfo& cached, const ImageInfo& requested,
 	                                      bool exact_format);
 	[[nodiscard]] static BindingType UploadBinding(const Image& image);
@@ -133,46 +130,43 @@ private:
 	[[nodiscard]] ImageId       ResolveDepthOverlap(const ImageInfo& requested, BindingType binding,
 	                                                ImageId cached);
 	[[nodiscard]] ImageId       ExpandImage(const ImageInfo& info, ImageId source);
-	void                        RefreshImage(ImageId id, const ImageDesc& desc);
-	void                        InitializeImage(ImageId id, const ImageDesc& desc);
-	[[nodiscard]] ColorTransferPlan BuildColorTransfer(const Image& image, BindingType binding,
-	                                                   TransferDirection direction) const;
-	[[nodiscard]] DownloadPlan      BuildDownload(const Image& image) const;
-	void UploadImage(Image& image, const ImageDesc& desc, Buffer& source, uint64_t source_offset);
+	void                        RefreshImage(ImageId id);
+	void                        PrepareDccClear(ImageId id, const ImageDesc& desc);
+	void                        InitializeImage(ImageId id);
+	[[nodiscard]] TextureTransferPlan
+	BuildTextureTransfer(const Image& image, BindingType binding, TransferDirection direction) const;
+	[[nodiscard]] DownloadPlan BuildDownload(const Image& image) const;
+	void UploadImage(Image& image, Buffer& source, uint64_t source_offset);
 	void DownloadImageData(Image& image, Buffer& destination, uint64_t destination_offset,
 	                       uint64_t destination_size, DownloadPlan plan);
 	void DownloadDepth(Image& image, Buffer& destination, uint64_t destination_offset);
 	void CommitGpuWrite(Image& image);
+	// Caller holds m_lock. Volume layer ranges select depth slices.
+	void ClearImage(CommandBuffer& command, ImageId id, const vk::ImageSubresourceRange& range,
+	                const vk::ClearValue& clear);
 	void PrepareImageCopy(Image& image);
 	void RefreshCopySource(ImageId id);
 	[[nodiscard]] bool CopyD16(Image& destination, Image& source);
 	void               CopyImage(ImageId destination, ImageId source);
 	void               AssociateStencil(ImageId depth, GuestRange stencil);
-	void               AssociateStencilLocked(ImageId depth, GuestRange stencil);
 	void CopyImageMip(ImageId destination, ImageId source, uint32_t mip, uint32_t layer);
 	void ValidateImageDesc(const ImageDesc& desc) const;
 
-	void InvalidateCpuAliases(uint64_t address, uint64_t size);
-	void ClearGpuModified(ImageId id);
-
-	void                                        DownloadImage(ImageId id);
-	[[nodiscard]] bool                          TryDownloadImage(ImageId id);
-	[[nodiscard]] std::pair<uint8_t*, uint64_t> MapDownload(uint64_t size, uint64_t alignment);
-	void QueueDownload(GuestRange range, StreamBuffer& download, uint8_t* mapped, uint64_t offset);
+	void               InvalidateCpuAliases(uint64_t address, uint64_t size);
+	[[nodiscard]] bool TryDownloadImage(ImageId id);
 
 	GraphicContext&                                   m_graphics;
 	CommandScheduler&                                 m_scheduler;
 	TrackingSpinLock                                  m_lock;
 	PageManager&                                      m_page_manager;
 	BlitHelper                                        m_blit_helper;
-	std::unique_ptr<TileManager>                      m_tiler;
+	TileManager                                       m_tiler;
 	BufferCache&                                      m_buffer_cache;
-	std::vector<Slot>                                 m_slots;
-	std::vector<uint32_t>                             m_free_slots;
+	Common::SlotVector<Image>                         m_slot_images;
 	ImagePageTable                                    m_image_page_table;
-	std::map<vk::Format, ImageId>                     m_null_images;
+	std::unordered_map<vk::Format, ImageId>           m_null_images;
 	Common::LeastRecentlyUsedCache<ImageId, uint64_t> m_lru_cache;
-	std::set<ImageId>                                 m_download_images;
+	std::unordered_set<ImageId>                       m_download_images;
 	std::map<uint64_t, MetaDataInfo>                  m_surface_metas;
 	uint64_t                                          m_total_used_memory  = 0;
 	uint64_t                                          m_trigger_gc_memory  = 0;

src/graphics/host_gpu/renderer/colorRenderTarget.cpp

@@ -4,25 +4,44 @@
 #include "common/logging/log.h"
 #include "common/profiler.h"
 #include "graphics/guest_gpu/gpu_defs.h"
+#include "graphics/guest_gpu/gpu_format.h"
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/guest_gpu/tile.h"
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/host_gpu/renderer/debug.h"
 #include "graphics/host_gpu/renderer/image/textureCommon.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 
 #include <algorithm>
+#include <array>
 #include <atomic>
 
 namespace Libs::Graphics {
 
 static std::atomic<uint32_t> g_render_color_log_count = 0;
 
+static bool DccAlphaOnMsb(const HW::ColorInfo& info) {
+	switch (info.format) {
+		case Prospero::ChannelLayout::k10_10_10_2:
+		case Prospero::ChannelLayout::k10_10_10_2Float:
+		case Prospero::ChannelLayout::k5_5_5_1: return true;
+		case Prospero::ChannelLayout::k2_10_10_10:
+		case Prospero::ChannelLayout::k1_5_5_5: return false;
+		default: break;
+	}
+	const auto components =
+	    Prospero::ResolveRenderTargetFormat(info.format, info.channel_type).components;
+	if (components == 1) {
+		return info.channel_order != Prospero::ChannelOrder::kStandard;
+	}
+	return components == 3 || info.channel_order == Prospero::ChannelOrder::kStandard ||
+	       info.channel_order == Prospero::ChannelOrder::kAlt;
+}
+
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
-void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandBuffer& buffer,
+void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, CommandBuffer& buffer,
                                               RenderColorInfo& r,
                                               uint32_t         render_target_slice_offset,
                                               uint32_t render_target_slot, bool ignore_target_mask,
@@ -38,8 +57,8 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		mask = 0x0f;
 	}
 
-	r.target_slot    = rt_slot;
-	r.export_mapping = {};
+	r             = {};
+	r.target_slot = rt_slot;
 
 	if (rt.base.addr == 0 || mask == 0) {
 		if (graphics_debug_dump_enabled()) {
@@ -54,21 +73,6 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 			}
 		}
 
-		// No color output
-		r.type               = RenderColorType::NoColorOutput;
-		r.desc               = {};
-		r.base_addr          = 0;
-		r.image_id           = {};
-		r.image_view         = nullptr;
-		r.format             = vk::Format::eUndefined;
-		r.extent             = {};
-		r.base_mip_level     = 0;
-		r.base_array_layer   = 0;
-		r.buffer_size        = 0;
-		r.samples            = 1;
-		r.export_mapping     = {};
-		r.color_clear_enable = false;
-		r.color_clear_value  = {};
 		return;
 	}
 	const auto samples = render_sample_count(rt.attrib.num_fragments);
@@ -76,8 +80,41 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		EXIT("unsupported render-target sample configuration: samples=%u fragments=%u\n",
 		     rt.attrib.num_samples, rt.attrib.num_fragments);
 	}
+	const uint32_t levels = rt.attrib2.num_mip_levels + 1u;
+	if (levels == 0 || levels > 16 || rt.view.current_mip_level >= levels) {
+		EXIT("unsupported render-target mip range: current=%u levels=%u\n",
+		     rt.view.current_mip_level, levels);
+	}
+	static constexpr std::array image_types {Prospero::ImageType::kColor1D,
+	                                         Prospero::ImageType::kColor2D,
+	                                         Prospero::ImageType::kColor3D};
+	if (rt.attrib3.dimension >= image_types.size()) {
+		EXIT("unsupported render-target dimension: %u\n", rt.attrib3.dimension);
+	}
+	const auto image_type = image_types[rt.attrib3.dimension];
+	const bool is_1d      = image_type == Prospero::ImageType::kColor1D;
+	const bool volume     = image_type == Prospero::ImageType::kColor3D;
+	if (is_1d && rt.attrib2.height != 0) {
+		EXIT("1D render target has nonzero height: %u\n", rt.attrib2.height);
+	}
+	if (!volume && rt.attrib3.depth != 0) {
+		EXIT("non-3D render target has nonzero depth: %u\n", rt.attrib3.depth);
+	}
+	if (is_1d && samples != 1) {
+		EXIT("multisampled 1D render targets are unsupported\n");
+	}
+	if (volume && samples != 1) {
+		EXIT("multisampled 3D render targets are unsupported\n");
+	}
+	const uint32_t depth = volume ? rt.attrib3.depth + 1u : 1u;
+	// For volumes, CB_COLOR_VIEW bounds exported slices; ATTRIB3 defines storage depth.
+	// The host attachment contains only the selected slices that exist in this mip.
+	const uint32_t last_layer = volume
+	                                ? std::min(rt.view.last_array_slice_index,
+	                                           std::max(depth >> rt.view.current_mip_level, 1u) - 1u)
+	                                : rt.view.last_array_slice_index;
 	const auto view = ResolveTargetViewInfo(
-	    rt.view.base_array_slice_index, rt.view.last_array_slice_index, render_target_slice_offset);
+	    rt.view.base_array_slice_index, last_layer, render_target_slice_offset);
 	switch (view.type) {
 		case TargetViewType::Image2D:
 		case TargetViewType::Image2DArray: break;
@@ -86,12 +123,6 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 			     rt.view.base_array_slice_index, rt.view.last_array_slice_index,
 			     render_target_slice_offset);
 	}
-	r.base_array_layer    = view.base_layer;
-	const uint32_t levels = rt.attrib2.num_mip_levels + 1u;
-	if (levels == 0 || levels > 16 || rt.view.current_mip_level >= levels) {
-		EXIT("unsupported render-target mip range: current=%u levels=%u\n",
-		     rt.view.current_mip_level, levels);
-	}
 	if (graphics_debug_dump_enabled()) {
 		static std::atomic_uint log_count = 0;
 		const auto              log_id    = log_count.fetch_add(1, std::memory_order_relaxed);
@@ -107,30 +138,16 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		}
 	}
 
-	// CB_COLOR_CONTROL describes the color-buffer operation / ROP3 logic op.
-	// ROP3 Copy is the normal color write path, not a render-target clear.
-	// SRGB clear words are still encoded as normalized component values.
+	// Color-control state selects the color-buffer operation and logical blend operation.
+	// The normal copy operation is a regular color write, not an attachment clear.
+	// Nonlinear clear values are still stored as normalized components.
 	// Fast color clears are metadata driven and must be handled explicitly when
 	// that metadata path is implemented; render-pass load must preserve contents.
-	r.color_clear_enable = false;
-	r.color_clear_value  = {};
-
 	uint32_t   width  = 0;
 	uint32_t   height = 0;
 	uint32_t   pitch  = 0;
 	uint64_t   size   = 0;
 	bool       tile   = false;
-	const bool volume = rt.attrib3.dimension == 2;
-	if (rt.attrib3.dimension != 1 && !volume) {
-		EXIT("unsupported render-target dimension: %u\n", rt.attrib3.dimension);
-	}
-	if (!volume && rt.attrib3.depth != 0) {
-		EXIT("2D render target has nonzero depth: %u\n", rt.attrib3.depth);
-	}
-	if (volume && samples != 1) {
-		EXIT("multisampled 3D render targets are unsupported\n");
-	}
-	const uint32_t depth        = volume ? rt.attrib3.depth + 1u : 1u;
 	const bool     standard4    = rt.attrib3.tile_mode == Prospero::TileMode::kStandard4KB;
 	const bool     standard64   = rt.attrib3.tile_mode == Prospero::TileMode::kStandard64KB;
 	const bool     depth_tile   = rt.attrib3.tile_mode == Prospero::TileMode::kDepth;
@@ -169,13 +186,11 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	if (texture_tile &&
 	    (!TileGetTextureBlockLayout(transfer_format, rt.attrib3.tile_mode, volume,
 	                                texture_tile_layout) ||
-	     rt.pitch.pitch_div8_minus1 != 0 ||
 	     (rt.base.addr & (texture_tile_layout.block.block_size - 1u)) != 0 ||
 	     rt.info.fmask_compression_enable || rt.info.fmask_data_compression_disable ||
 	     rt.info.fmask_one_frag_mode || rt.info.cmask_fast_clear_enable ||
-	     rt.info.dcc_compression_enable || rt.info.cmask_is_linear != 0 ||
-	     rt.info.cmask_addr_type != 0 || rt.info.alt_tile_mode || rt.cmask.addr != 0 ||
-	     rt.fmask.addr != 0 || rt.dcc_addr.addr != 0 || rt.dcc.data_write_on_dcc_clear_to_reg)) {
+	     rt.info.dcc_compression_enable || rt.cmask.addr != 0 || rt.fmask.addr != 0 ||
+	     rt.dcc_addr.addr != 0 || rt.dcc.data_write_on_dcc_clear_to_reg)) {
 		EXIT("unsupported texture-tiled render target: addr=0x%016" PRIx64 " tile=%u"
 		     " dimension=%u depth=%u levels=%u layer=%u/%u samples=%u fragments=%u bpe=%u"
 		     " cmask=0x%016" PRIx64 " fmask=0x%016" PRIx64 " dcc=0x%016" PRIx64 "\n",
@@ -184,18 +199,15 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		     rt.attrib.num_fragments, bytes_per_element, rt.cmask.addr, rt.fmask.addr,
 		     rt.dcc_addr.addr);
 	}
-	if ((standard64 || depth_tile) && (rt.attrib3.dimension != 1 || rt.attrib3.depth != 0 ||
-	                                   view.base_layer != 0 || view.image_layers != 1)) {
+	if ((standard64 || depth_tile) &&
+	    (rt.attrib3.dimension != 1 || rt.attrib3.depth != 0 ||
+	     (depth_tile && (view.base_layer != 0 || view.image_layers != 1)))) {
 		EXIT("unsupported 64KB texture-tiled render-target view: dimension=%u depth=%u"
 		     " layer=%u/%u\n",
 		     rt.attrib3.dimension, rt.attrib3.depth, view.base_layer, view.image_layers);
 	}
-	if (rt.pitch.pitch_div8_minus1 != 0) {
-		pitch = (rt.pitch.pitch_div8_minus1 + 1u) << 3u;
-	} else if (tile) {
-		if (volume) {
-			pitch = TileGetTexturePitch(transfer_format, width, rt.attrib3.tile_mode);
-		} else if (texture_tile) {
+	if (tile) {
+		if (volume || texture_tile) {
 			pitch = TileGetTexturePitch(transfer_format, width, rt.attrib3.tile_mode);
 		} else {
 			pitch = TileGetRenderTargetPitch(width, bytes_per_element, rt.attrib.num_fragments);
@@ -258,12 +270,6 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		mip_sizes[0]  = {static_cast<uint32_t>(size), 0, 0, 0, 0, 0};
 		mip_padded[0] = {pitch, height};
 	}
-	if (rt.slice.slice_div64_minus1 != 0 &&
-	    (static_cast<uint64_t>(rt.slice.slice_div64_minus1) + 1u) * 64u != size) {
-		EXIT("render-target slice span mismatch: encoded=0x%016" PRIx64 " derived=0x%016" PRIx64
-		     "\n",
-		     (static_cast<uint64_t>(rt.slice.slice_div64_minus1) + 1u) * 64u, size);
-	}
 	if (size == 0 || (!volume && size > UINT64_MAX / view.image_layers)) {
 		EXIT("render-target memory footprint is invalid\n");
 	}
@@ -273,18 +279,12 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	if (backing_size == 0) {
 		EXIT("render-target backing is empty\n");
 	}
-	if (backing_size > TRACKER_ADDRESS_SIZE - rt.base.addr) {
+	if (!GuestRange {rt.base.addr, backing_size}.Valid()) {
 		EXIT("render-target backing range is invalid\n");
 	}
 
 	const vk::Extent2D view_extent = {std::max(width >> rt.view.current_mip_level, 1u),
 	                                  std::max(height >> rt.view.current_mip_level, 1u)};
-	const uint32_t     view_depth  = std::max(depth >> rt.view.current_mip_level, 1u);
-	if (volume &&
-	    (view.base_layer >= view_depth || view.layer_count > view_depth - view.base_layer)) {
-		EXIT("3D render-target view exceeds mip depth: base=%u count=%u depth=%u mip=%u\n",
-		     view.base_layer, view.layer_count, view_depth, rt.view.current_mip_level);
-	}
 
 	auto decision_log_id = g_render_color_log_count.fetch_add(1);
 	if (decision_log_id < 128) {
@@ -302,13 +302,24 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	desc.info.data         = {rt.base.addr, backing_size};
 	desc.info.pixel_format = target_format.format;
 	desc.info.guest_format = transfer_format;
-	desc.info.type         = volume ? Prospero::ImageType::kColor3D : Prospero::ImageType::kColor2D;
+	desc.info.type         = image_type;
 	desc.info.extent       = {width, height, depth};
 	desc.info.resources    = {levels, volume ? 1u : view.image_layers};
 	desc.info.pitch        = pitch;
 	desc.info.bytes_per_block = bytes_per_element;
 	desc.info.samples         = samples;
 	desc.info.tile_mode       = rt.attrib3.tile_mode;
+	const bool has_dcc        = rt.info.dcc_compression_enable && rt.dcc_addr.addr != 0;
+	if (has_dcc) {
+		TileSizeAlign metadata_size {};
+		(void)TileGetDccSize(width, height, volume ? depth : view.image_layers, bytes_per_element,
+		                     levels, rt.attrib3.tile_mode, metadata_size, rt.attrib.num_fragments);
+		desc.info.metadata.kind                     = ImageMetadataKind::Dcc;
+		desc.info.metadata.range                    = {rt.dcc_addr.addr, metadata_size.size};
+		desc.info.metadata.dcc_clear_word           = rt.clear_word0.word0;
+		desc.info.metadata.dcc_clear_register_valid = true;
+		desc.info.metadata.dcc_alpha_msb            = DccAlphaOnMsb(rt.info);
+	}
 	for (uint32_t level = 0; level < levels; level++) {
 		if (volume) {
 			const auto& mip             = volume_layout.mips[level];
@@ -334,8 +345,13 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		};
 	}
 	desc.view_info.format = target_format.format;
-	desc.view_info.type =
-	    view.layer_count == 1 ? vk::ImageViewType::e2D : vk::ImageViewType::e2DArray;
+	if (is_1d) {
+		desc.view_info.type =
+		    view.layer_count == 1 ? vk::ImageViewType::e1D : vk::ImageViewType::e1DArray;
+	} else {
+		desc.view_info.type =
+		    view.layer_count == 1 ? vk::ImageViewType::e2D : vk::ImageViewType::e2DArray;
+	}
 	desc.view_info.aspect      = vk::ImageAspectFlagBits::eColor;
 	desc.view_info.base_level  = rt.view.current_mip_level;
 	desc.view_info.level_count = 1;
@@ -344,18 +360,10 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	desc.view_info.usage       = vk::ImageUsageFlagBits::eColorAttachment;
 	auto& texture_cache        = m_context.GetTextureCache();
 	r.desc                     = std::move(desc);
+	r.guest_mip_level          = rt.view.current_mip_level;
+	r.guest_array_layer        = view.base_layer;
 	r.image_id                 = texture_cache.FindImage(r.desc, exact_format);
-	r.type                     = RenderColorType::RenderTexture;
-	r.base_addr                = rt.base.addr;
-	r.image_view               = nullptr;
-	r.format                   = r.desc.view_info.format;
-	r.extent                   = view_extent;
-	r.base_mip_level           = rt.view.current_mip_level;
-	r.buffer_size              = backing_size;
-	r.samples                  = samples;
 	r.export_mapping           = target_format.export_mapping;
-	r.color_clear_enable       = false;
-	r.color_clear_value        = {};
 	BindRenderTarget(r.image_id);
 }
 

src/graphics/host_gpu/renderer/colorRenderTarget.h

@@ -6,33 +6,24 @@
 #include "graphics/host_gpu/renderer/renderTarget.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 
+#include <algorithm>
 #include <cstdint>
 
 namespace Libs::Graphics {
 
-class RenderCommandBuffer;
-
-enum class RenderColorType {
-	NoColorOutput,
-	RenderTexture,
-};
-
 struct RenderColorInfo {
-	RenderColorType                 type = RenderColorType::NoColorOutput;
+	// Discovery keeps guest image information but can remap the view into a larger cache image.
 	TextureCache::ImageDesc         desc;
 	ImageId                         image_id;
-	vk::ImageView                   image_view       = nullptr;
-	vk::Format                      format           = vk::Format::eUndefined;
-	vk::Extent2D                    extent           = {};
-	uint32_t                        base_mip_level   = 0;
-	uint32_t                        base_array_layer = 0;
-	uint64_t                        base_addr        = 0;
-	uint64_t                        buffer_size      = 0;
 	uint32_t                        target_slot      = 0;
-	uint32_t                        samples          = 1;
+	uint32_t                        guest_mip_level   = 0;
+	uint32_t                        guest_array_layer = 0;
 	Prospero::ColorComponentMapping export_mapping;
-	bool                            color_clear_enable = false;
-	vk::ClearColorValue             color_clear_value {};
+
+	[[nodiscard]] vk::Extent2D Extent() const {
+		return {std::max(desc.info.extent.width >> guest_mip_level, 1u),
+		        std::max(desc.info.extent.height >> guest_mip_level, 1u)};
+	}
 };
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/commandScheduler.cpp

@@ -1,96 +1,92 @@
 #include "graphics/host_gpu/renderer/commandScheduler.h"
 
 #include "common/assert.h"
+#include "common/logging/log.h"
 #include "graphics/host_gpu/graphicContext.h"
 
 #include <algorithm>
+#include <cstdio>
+#include <optional>
 
 namespace Libs::Graphics {
 
 static thread_local CommandScheduler* g_deferred_callback_scheduler = nullptr;
 
-void CommandSlot::Reset() {
-	EXIT_IF(buffer == nullptr);
-	const auto result = buffer.reset(vk::CommandBufferResetFlagBits::eReleaseResources);
-	if (result != vk::Result::eSuccess) {
-		EXIT("failed to reset Vulkan command buffer: %s (%d)\n", VulkanToString(result).c_str(),
-		     static_cast<int>(result));
-	}
+namespace {
+
+void ReportVulkanFatal(const char* what, vk::Result result, uint64_t tick, uint32_t debug_op,
+                       uint64_t debug_submit, uint32_t arg0, uint32_t arg1, uint32_t arg2,
+                       uint32_t arg3, uint64_t arg4) {
+	LOGF("%s failed: %s (%d), tick=%" PRIu64 " debug_op=%u debug_submit=%" PRIu64
+	     " args=%u,%u,%u,%u,0x%016" PRIx64 "\n",
+	     what, vk::to_string(result).c_str(), static_cast<int>(result), tick, debug_op,
+	     debug_submit, arg0, arg1, arg2, arg3, arg4);
+	std::printf("%s failed: %s (%d), tick=%" PRIu64 " debug_op=%u debug_submit=%" PRIu64
+	            " args=%u,%u,%u,%u,0x%016" PRIx64 "\n",
+	            what, vk::to_string(result).c_str(), static_cast<int>(result), tick, debug_op,
+	            debug_submit, arg0, arg1, arg2, arg3, arg4);
+	std::fflush(stdout);
 }
 
-CommandScheduler::CommandPool::~CommandPool() {
-	Destroy();
-}
-
-void CommandScheduler::CommandPool::Create(GraphicContext& graphics) {
-	EXIT_IF(m_pool != nullptr || m_graphics != nullptr ||
-	        graphics.queue_family == static_cast<uint32_t>(-1));
-	m_graphics = &graphics;
+} // namespace
 
+CommandScheduler::CommandPool::CommandPool(GraphicContext& graphics, MasterSemaphore& master)
+    : m_graphics(graphics), m_master(master) {
+	EXIT_IF(graphics.queue_family == static_cast<uint32_t>(-1));
 	vk::CommandPoolCreateInfo create {};
-	create.sType            = vk::StructureType::eCommandPoolCreateInfo;
 	create.queueFamilyIndex = graphics.queue_family;
-	create.flags            = vk::CommandPoolCreateFlagBits::eResetCommandBuffer;
+	create.flags            = vk::CommandPoolCreateFlagBits::eTransient |
+	                          vk::CommandPoolCreateFlagBits::eResetCommandBuffer;
 	const auto result       = graphics.device.createCommandPool(&create, nullptr, &m_pool);
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess || m_pool == nullptr);
 }
 
-CommandSlot* CommandScheduler::CommandPool::CreateSlot() {
-	EXIT_IF(m_graphics == nullptr);
-	auto& graphics = *m_graphics;
+CommandScheduler::CommandPool::~CommandPool() {
+	m_graphics.device.destroyCommandPool(m_pool, nullptr);
+}
+
+size_t CommandScheduler::CommandPool::Grow() {
+	const auto first = m_ticks.size();
+	m_ticks.resize(first + GrowStep);
+	m_buffers.resize(first + GrowStep);
 
 	vk::CommandBufferAllocateInfo allocate {};
-	allocate.sType              = vk::StructureType::eCommandBufferAllocateInfo;
 	allocate.commandPool        = m_pool;
 	allocate.level              = vk::CommandBufferLevel::ePrimary;
-	allocate.commandBufferCount = 1;
-	vk::CommandBuffer buffer    = nullptr;
-	EXIT_IF(graphics.device.allocateCommandBuffers(&allocate, &buffer) != vk::Result::eSuccess);
-
-	vk::FenceCreateInfo fence_create {};
-	fence_create.sType = vk::StructureType::eFenceCreateInfo;
-	fence_create.flags = vk::FenceCreateFlagBits::eSignaled;
-	vk::Fence fence    = nullptr;
-	if (graphics.device.createFence(&fence_create, nullptr, &fence) != vk::Result::eSuccess) {
-		graphics.device.freeCommandBuffers(m_pool, 1, &buffer);
-		EXIT("failed to create command-buffer fence\n");
-	}
-
-	auto& slot      = m_slots.emplace_back();
-	slot.pool_mutex = &m_mutex;
-	slot.id         = static_cast<uint32_t>(m_slots.size() - 1);
-	slot.buffer     = buffer;
-	slot.fence      = fence;
-	return &slot;
+	allocate.commandBufferCount = static_cast<uint32_t>(GrowStep);
+	EXIT_IF(m_graphics.device.allocateCommandBuffers(&allocate, m_buffers.data() + first) !=
+	        vk::Result::eSuccess);
+	return first;
 }
 
-CommandSlot* CommandScheduler::CommandPool::Allocate(GraphicContext& graphics) {
-	Common::LockGuard lock(m_mutex);
-	if (m_pool == nullptr) {
-		Create(graphics);
-	}
-	EXIT_IF(m_graphics != &graphics);
-	auto  found = std::ranges::find_if(m_slots, [](const auto& slot) { return !slot.busy; });
-	auto* slot  = found != m_slots.end() ? &*found : CreateSlot();
-	slot->busy  = true;
-	slot->Reset();
-	return slot;
-}
+vk::CommandBuffer CommandScheduler::CommandPool::Commit() {
+	auto       gpu_tick = m_master.KnownGpuTick();
+	const auto search   = [this, &gpu_tick](size_t begin, size_t end) -> std::optional<size_t> {
+		for (size_t index = begin; index < end; ++index) {
+			if (gpu_tick >= m_ticks[index]) {
+				m_ticks[index] = m_master.CurrentTick();
+				return index;
+			}
+		}
+		return std::nullopt;
+	};
 
-void CommandScheduler::CommandPool::Destroy() {
-	Common::LockGuard lock(m_mutex);
-	if (m_pool == nullptr) {
-		return;
+	auto found = search(m_hint, m_ticks.size());
+	if (!found) {
+		m_master.Refresh();
+		gpu_tick = m_master.KnownGpuTick();
+		found    = search(m_hint, m_ticks.size());
 	}
-	EXIT_IF(std::ranges::any_of(m_slots, [](const auto& slot) { return slot.busy; }));
-	EXIT_IF(m_graphics == nullptr);
-	for (const auto& slot: m_slots) {
-		m_graphics->device.destroyFence(slot.fence, nullptr);
+	if (!found) {
+		found = search(0, m_hint);
 	}
-	m_graphics->device.destroyCommandPool(m_pool, nullptr);
-	m_slots.clear();
-	m_pool     = nullptr;
-	m_graphics = nullptr;
+	if (!found) {
+		found           = Grow();
+		m_ticks[*found] = m_master.CurrentTick();
+	}
+
+	m_hint = (*found + 1) % m_ticks.size();
+	return m_buffers[*found];
 }
 
 bool CommandScheduler::InDeferredOperation() noexcept {
@@ -99,10 +95,8 @@ bool CommandScheduler::InDeferredOperation() noexcept {
 
 CommandScheduler::CommandScheduler(RenderContext& context, GraphicContext& graphics)
     : m_master(graphics), m_context(context), m_graphics(graphics),
-      m_priority_thread([this](std::stop_token stop) { PriorityOperationsThread(stop); }) {
-	m_buffers.reserve(CommandBufferGrowStep);
-	m_buffer_ticks.reserve(CommandBufferGrowStep);
-}
+      m_command_pool(graphics, m_master), m_command(*this),
+      m_priority_thread([this](std::stop_token stop) { PriorityOperationsThread(stop); }) {}
 
 CommandScheduler::~CommandScheduler() {
 	Shutdown();
@@ -128,9 +122,11 @@ void CommandScheduler::Shutdown() {
 		}
 		m_operation_state = OperationState::Draining;
 	}
-	if (Active() && m_recording) {
-		Finish();
+	if (!m_command.IsInvalid()) {
+		Submit();
 	}
+	m_master.Wait(CurrentTick() - 1);
+	PopPendingOperations();
 	DrainPriorityOperations();
 	m_priority_thread.request_stop();
 	m_operation_available.notify_all();
@@ -152,18 +148,10 @@ void CommandScheduler::Begin(HW::Context& registers, HW::UserConfig& user_config
 		std::lock_guard lock(m_operation_mutex);
 		EXIT_IF(m_operation_state != OperationState::Open);
 	}
-	m_registers   = &registers;
-	m_user_config = &user_config;
-	m_shaders     = &shaders;
+	m_command.Bind(registers, user_config, shaders);
 
-	if (!Active()) {
-		m_current = static_cast<int>(GrowCommandBuffers());
-	}
-
-	BindCurrent();
-	if (!m_recording) {
-		Current().Begin();
-		m_recording = true;
+	if (m_command.IsInvalid()) {
+		BeginNext();
 	}
 }
 
@@ -172,7 +160,7 @@ void CommandScheduler::BeginRendering(const RenderState& state) {
 }
 
 void CommandScheduler::EndRendering() {
-	if (Active() && m_recording) {
+	if (Active() && !m_command.IsInvalid()) {
 		Current().EndRendering();
 	}
 }
@@ -183,55 +171,40 @@ void CommandScheduler::Flush() {
 }
 
 void CommandScheduler::Flush(SubmitInfo& submit) {
-	SubmitCurrent(submit);
+	Submit(submit);
 	BeginNext();
 }
 
-CommandBuffer& CommandScheduler::FlushAndGetSubmitted() {
-	SubmitInfo submit;
-	auto&      submitted = SubmitCurrent(submit);
+void CommandScheduler::FlushAndWait() {
+	const auto tick = Submit();
+	m_master.Wait(tick);
 	BeginNext();
-	return submitted;
 }
 
 void CommandScheduler::Finish() {
 	CheckActive();
-	const auto tick = CurrentTick();
-	if (m_recording) {
-		SubmitInfo submit;
-		SubmitCurrent(submit);
+	if (!m_command.IsInvalid()) {
+		Submit();
 	}
-	for (auto& buffer: m_buffers) {
-		buffer->WaitForFenceAndReset();
-	}
-	m_master.Wait(tick);
+	m_master.Wait(CurrentTick() - 1);
+	BeginNext();
 	PopPendingOperations();
-	BindCurrent();
-	Current().Begin();
-	m_recording = true;
-}
-
-void CommandScheduler::FinishCurrent() {
-	SubmitInfo submit;
-	auto&      submitted = SubmitCurrent(submit);
-	submitted.WaitForFenceAndReset();
-	m_master.Refresh();
-	PopPendingOperations();
-	submitted.Begin();
-	m_recording = true;
 }
 
 void CommandScheduler::Wait(uint64_t tick) {
-	CheckActive();
 	EXIT_IF(tick > CurrentTick());
-	if (tick >= CurrentTick()) {
+	if (tick == CurrentTick()) {
+		CheckActive();
 		// A stream-buffer wrap can wait while a draw is being prepared through a reference to
-		// Current(). Recycle the same command object so that reference remains valid.
-		FinishCurrent();
-		return;
+		// Current(). The wrapper stays stable while its pooled Vulkan buffer is retired. Deferred
+		// resources are released only at the next GPU operation boundary.
+		const auto submitted_tick = Submit();
+		EXIT_IF(submitted_tick != tick);
+		m_master.Wait(tick);
+		BeginNext();
+	} else {
+		m_master.Wait(tick);
 	}
-	m_master.Wait(tick);
-	PopPendingOperations();
 }
 
 void CommandScheduler::PopPendingOperations() {
@@ -243,17 +216,18 @@ void CommandScheduler::PopPendingOperations(bool refresh_gpu_tick) {
 		m_master.Refresh();
 	}
 	for (;;) {
-		Common::UniqueFunction<void> callback;
+		PendingOperation operation;
 		{
 			std::lock_guard lock(m_operation_mutex);
 			if (m_pending_operations.empty() ||
 			    !m_master.IsFree(m_pending_operations.front().tick)) {
 				return;
 			}
-			callback = std::move(m_pending_operations.front().callback);
+			operation = std::move(m_pending_operations.front());
 			m_pending_operations.pop();
 		}
-		RunOperation(std::move(callback));
+		WaitPriorityOperations(operation.tick);
+		RunOperation(std::move(operation.callback));
 	}
 }
 
@@ -359,86 +333,73 @@ bool CommandScheduler::IsFree(uint64_t tick) {
 	return m_master.IsFree(tick);
 }
 
-CommandSlot* CommandScheduler::AllocateCommandBuffer() {
-	return m_command_pool.Allocate(m_graphics);
-}
-
-uint64_t CommandScheduler::NextSubmitSequence() noexcept {
-	return m_submit_sequence.fetch_add(1, std::memory_order_relaxed) + 1;
-}
-
 void CommandScheduler::CheckActive() const {
-	EXIT_IF(!Active() || static_cast<size_t>(m_current) >= m_buffers.size());
+	EXIT_IF(!Active());
 }
 
-RenderCommandBuffer& CommandScheduler::Current() const {
+CommandBuffer& CommandScheduler::Current() {
 	CheckActive();
-	EXIT_IF(m_buffers[m_current] == nullptr);
-	return *m_buffers[m_current];
+	return m_command;
 }
 
-void CommandScheduler::BindCurrent() const {
-	EXIT_IF(m_registers == nullptr || m_user_config == nullptr || m_shaders == nullptr);
-	Current().Bind(*m_registers, *m_user_config, *m_shaders);
+CommandBuffer& CommandScheduler::BeginCommand() {
+	EXIT_IF(!m_command.IsInvalid());
+	m_command.m_buffer = m_command_pool.Commit();
+	m_command.Begin();
+	return m_command;
 }
 
-CommandBuffer& CommandScheduler::SubmitCurrent(SubmitInfo& submit) {
-	CheckActive();
-	EXIT_IF(!m_recording);
-	auto& submitted = Current();
-	submitted.End();
-	const auto signal_tick = m_master.NextTick();
-	submit.AddSignal(m_master.Handle(), signal_tick);
-	submitted.Execute(submit);
-	m_buffer_ticks[static_cast<size_t>(m_current)] = signal_tick;
-	m_recording                                    = false;
-	return submitted;
-}
+uint64_t CommandScheduler::Submit(SubmitInfo submit) {
+	EXIT_IF(m_command.IsInvalid());
+	EXIT_IF(submit.num_wait_semaphores > SubmitInfo::MaxSemaphores ||
+	        submit.num_signal_semaphores >= SubmitInfo::MaxSemaphores);
 
-int CommandScheduler::FindReusableBuffer(uint64_t gpu_tick) const {
-	EXIT_IF(m_buffer_ticks.size() != m_buffers.size());
-	const auto buffer_count = m_buffers.size();
-	for (size_t offset = 1; offset <= buffer_count; ++offset) {
-		const auto candidate = (static_cast<size_t>(m_current) + offset) % buffer_count;
-		if (gpu_tick >= m_buffer_ticks[candidate]) {
-			return static_cast<int>(candidate);
-		}
+	m_command.End();
+	const auto buffer   = m_command.m_buffer;
+	auto&      graphics = m_graphics;
+	EXIT_IF(graphics.queue == nullptr);
+
+	vk::Result result;
+	uint64_t   tick;
+	{
+		Common::LockGuard lock(graphics.queue_mutex);
+		tick = m_master.NextTick();
+		submit.AddSignal(m_master.Handle(), tick);
+
+		vk::TimelineSemaphoreSubmitInfo timeline_info {};
+		timeline_info.waitSemaphoreValueCount   = submit.num_wait_semaphores;
+		timeline_info.pWaitSemaphoreValues      = submit.wait_ticks.data();
+		timeline_info.signalSemaphoreValueCount = submit.num_signal_semaphores;
+		timeline_info.pSignalSemaphoreValues    = submit.signal_ticks.data();
+
+		vk::SubmitInfo submit_info {};
+		submit_info.pNext                = &timeline_info;
+		submit_info.waitSemaphoreCount   = submit.num_wait_semaphores;
+		submit_info.pWaitSemaphores      = submit.wait_semaphores.data();
+		submit_info.pWaitDstStageMask    = submit.wait_stages.data();
+		submit_info.commandBufferCount   = 1;
+		submit_info.pCommandBuffers      = &buffer;
+		submit_info.signalSemaphoreCount = submit.num_signal_semaphores;
+		submit_info.pSignalSemaphores    = submit.signal_semaphores.data();
+
+		result = graphics.queue.submit(1, &submit_info, nullptr);
 	}
-	return -1;
-}
 
-size_t CommandScheduler::GrowCommandBuffers() {
-	const auto first = m_buffers.size();
-	const auto end   = first + CommandBufferGrowStep;
-	m_buffers.reserve(end);
-	m_buffer_ticks.reserve(end);
-	for (size_t i = first; i < end; ++i) {
-		m_buffers.emplace_back(std::make_unique<RenderCommandBuffer>(*this));
-		m_buffer_ticks.push_back(0);
+	if (result != vk::Result::eSuccess) {
+		ReportVulkanFatal("vkQueueSubmit", result, tick, m_command.m_debug_op,
+		                  m_command.m_debug_submit_id, m_command.m_debug_arg0,
+		                  m_command.m_debug_arg1, m_command.m_debug_arg2, m_command.m_debug_arg3,
+		                  m_command.m_debug_arg4);
 	}
-	return first;
+	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
+
+	m_command.m_buffer = nullptr;
+	return tick;
 }
 
 void CommandScheduler::BeginNext() {
-	EXIT_IF(m_recording);
-
-	auto candidate = FindReusableBuffer(m_master.KnownGpuTick());
-	bool refreshed = false;
-	if (candidate < 0) {
-		m_master.Refresh();
-		refreshed = true;
-		candidate = FindReusableBuffer(m_master.KnownGpuTick());
-	}
-	if (candidate < 0) {
-		candidate = static_cast<int>(GrowCommandBuffers());
-	}
-
-	m_current = candidate;
-	Current().WaitForFenceAndReset();
-	PopPendingOperations(!refreshed);
-	BindCurrent();
-	Current().Begin();
-	m_recording = true;
+	CheckActive();
+	BeginCommand();
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/commandScheduler.h

@@ -6,11 +6,7 @@
 #include "graphics/host_gpu/renderer/masterSemaphore.h"
 #include "graphics/host_gpu/renderer/render.h"
 
-#include <array>
-#include <atomic>
 #include <condition_variable>
-#include <deque>
-#include <memory>
 #include <mutex>
 
 #include <queue>
@@ -20,16 +16,6 @@
 
 namespace Libs::Graphics {
 
-struct CommandSlot {
-	Common::Mutex*    pool_mutex = nullptr;
-	uint32_t          id         = 0;
-	vk::CommandBuffer buffer     = nullptr;
-	vk::Fence         fence      = nullptr;
-	bool              busy       = false;
-
-	void Reset();
-};
-
 class CommandScheduler {
 public:
 	CommandScheduler(RenderContext& context, GraphicContext& graphics);
@@ -41,9 +27,10 @@ public:
 	void           EndRendering();
 	void           Flush();
 	void           Flush(SubmitInfo& submit);
-	CommandBuffer& FlushAndGetSubmitted();
+	void           FlushAndWait();
 	void           Finish();
-	void           FinishCurrent();
+	CommandBuffer& BeginCommand();
+	uint64_t       Submit(SubmitInfo submit = {});
 	// Deferred callbacks can observe an externally owned drain, but cannot initiate shutdown:
 	// the priority runner cannot join itself.
 	void                      Shutdown();
@@ -55,34 +42,35 @@ public:
 	void                      DeferPriorityOperation(Common::UniqueFunction<void>&& operation);
 	[[nodiscard]] static bool InDeferredOperation() noexcept;
 
-	[[nodiscard]] bool            Active() const noexcept { return m_current >= 0; }
-	void                          CheckActive() const;
-	RenderCommandBuffer&          Current() const;
-	[[nodiscard]] uint64_t        CurrentTick() const noexcept { return m_master.CurrentTick(); }
-	[[nodiscard]] bool            IsFree(uint64_t tick);
-	[[nodiscard]] RenderContext&  Context() const noexcept { return m_context; }
-	[[nodiscard]] GraphicContext& Graphics() const noexcept { return m_graphics; }
+	[[nodiscard]] bool Active() const noexcept { return m_command.m_registers != nullptr; }
+	void                           CheckActive() const;
+	CommandBuffer&                 Current();
+	[[nodiscard]] uint64_t         CurrentTick() const noexcept { return m_master.CurrentTick(); }
+	[[nodiscard]] bool             IsFree(uint64_t tick);
+	[[nodiscard]] MasterSemaphore& GetMasterSemaphore() noexcept { return m_master; }
+	[[nodiscard]] RenderContext&   Context() const noexcept { return m_context; }
+	[[nodiscard]] GraphicContext&  Graphics() const noexcept { return m_graphics; }
 
 private:
-	static constexpr size_t CommandBufferGrowStep = 4;
-
 	class CommandPool {
 	public:
-		CommandPool() = default;
+		CommandPool(GraphicContext& graphics, MasterSemaphore& master);
 		~CommandPool();
 		KYTY_CLASS_NO_COPY(CommandPool);
 
-		CommandSlot* Allocate(GraphicContext& graphics);
+		vk::CommandBuffer Commit();
 
 	private:
-		void         Create(GraphicContext& graphics);
-		CommandSlot* CreateSlot();
-		void         Destroy();
+		static constexpr size_t GrowStep = 4;
 
-		GraphicContext*         m_graphics = nullptr;
-		Common::Mutex           m_mutex;
-		vk::CommandPool         m_pool = nullptr;
-		std::deque<CommandSlot> m_slots;
+		size_t Grow();
+
+		GraphicContext&                m_graphics;
+		MasterSemaphore&               m_master;
+		vk::CommandPool                m_pool = nullptr;
+		std::vector<vk::CommandBuffer> m_buffers;
+		std::vector<uint64_t>          m_ticks;
+		size_t                         m_hint = 0;
 	};
 
 	enum class OperationState { Open, Draining, Closed };
@@ -92,39 +80,24 @@ private:
 		uint64_t                     tick = 0;
 	};
 
-	void                       BindCurrent() const;
-	CommandBuffer&             SubmitCurrent(SubmitInfo& submit);
-	void                       BeginNext();
-	void                       PopPendingOperations(bool refresh_gpu_tick);
-	[[nodiscard]] int          FindReusableBuffer(uint64_t gpu_tick) const;
-	[[nodiscard]] size_t       GrowCommandBuffers();
-	void                       PriorityOperationsThread(std::stop_token stop);
-	void                       RunOperation(Common::UniqueFunction<void>&& operation);
-	[[nodiscard]] CommandSlot* AllocateCommandBuffer();
-	[[nodiscard]] uint64_t     NextSubmitSequence() noexcept;
+	void BeginNext();
+	void PopPendingOperations(bool refresh_gpu_tick);
+	void PriorityOperationsThread(std::stop_token stop);
+	void RunOperation(Common::UniqueFunction<void>&& operation);
 
-	MasterSemaphore                                   m_master;
-	RenderContext&                                    m_context;
-	GraphicContext&                                   m_graphics;
-	CommandPool                                       m_command_pool;
-	std::vector<std::unique_ptr<RenderCommandBuffer>> m_buffers;
-	std::vector<uint64_t>                             m_buffer_ticks;
-	std::queue<PendingOperation>                      m_pending_operations;
-	std::queue<PendingOperation>                      m_priority_operations;
-	std::mutex                                        m_operation_mutex;
-	std::condition_variable                           m_operation_available;
-	std::jthread                                      m_priority_thread;
-	bool                                              m_priority_active      = false;
-	uint64_t                                          m_priority_active_tick = 0;
-	OperationState                                    m_operation_state      = OperationState::Open;
-	int                                               m_current              = -1;
-	bool                                              m_recording            = false;
-	HW::Context*                                      m_registers            = nullptr;
-	HW::UserConfig*                                   m_user_config          = nullptr;
-	HW::Shader*                                       m_shaders              = nullptr;
-	std::atomic<uint64_t>                             m_submit_sequence      = 0;
-
-	friend class CommandBuffer;
+	MasterSemaphore              m_master;
+	RenderContext&               m_context;
+	GraphicContext&              m_graphics;
+	CommandPool                  m_command_pool;
+	CommandBuffer                m_command;
+	std::queue<PendingOperation> m_pending_operations;
+	std::queue<PendingOperation> m_priority_operations;
+	std::mutex                   m_operation_mutex;
+	std::condition_variable      m_operation_available;
+	std::jthread                 m_priority_thread;
+	bool                         m_priority_active      = false;
+	uint64_t                     m_priority_active_tick = 0;
+	OperationState               m_operation_state      = OperationState::Open;
 };
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/context.cpp

@@ -1,7 +1,5 @@
 #include "common/assert.h"
 #include "common/common.h"
-#include "common/emulatorConfig.h"
-#include "common/logging/log.h"
 #include "common/profiler.h"
 #include "common/threads.h"
 #include "graphics/host_gpu/graphicContext.h"
@@ -9,7 +7,6 @@
 #include "graphics/host_gpu/renderer/debug.h"
 #include "graphics/host_gpu/renderer/depthRenderTarget.h"
 #include "graphics/host_gpu/renderer/image/imageView.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vma.h"
@@ -18,99 +15,26 @@
 #include <algorithm>
 #include <bit>
 #include <cstring>
-#include <memory>
 namespace Libs::Graphics {
 
-FenceResourceRetainer::~FenceResourceRetainer() {
-	if (!m_resources.empty()) {
-		EXIT("fence resource retainer destroyed before release\n");
-	}
-}
-
-void FenceResourceRetainer::Retain(std::shared_ptr<void> resource) {
-	if (resource == nullptr) {
-		EXIT("cannot retain a null fence resource\n");
-	}
-	if (std::ranges::none_of(m_resources, [&resource](const auto& retained) {
-		    return retained.get() == resource.get();
-	    })) {
-		m_resources.push_back(std::move(resource));
-	}
-}
-
-void FenceResourceRetainer::ReleaseAfterFence() noexcept {
-	m_resources.clear();
-}
-
 CommandBuffer::CommandBuffer(CommandScheduler& scheduler)
-    : m_context(scheduler.Context()), m_scheduler(scheduler), m_graphics(scheduler.Graphics()),
-      m_slot(scheduler.AllocateCommandBuffer()) {}
-
-CommandBuffer::~CommandBuffer() {
-	Release();
-}
+    : m_context(scheduler.Context()), m_graphics(scheduler.Graphics()) {}
 
 bool CommandBuffer::IsInvalid() const {
-	return m_slot == nullptr;
+	return m_buffer == nullptr;
 }
 
 vk::CommandBuffer CommandBuffer::Handle() const {
 	EXIT_IF(IsInvalid());
-
-	const auto handle = m_slot->buffer;
-	EXIT_IF(handle == nullptr);
-	return handle;
+	return m_buffer;
 }
 
-void CommandBuffer::Release() {
-	EXIT_IF(IsInvalid());
-
-	Common::LockGuard lock(*m_slot->pool_mutex);
-
-	WaitForFence();
-
-	m_slot->busy = false;
-	m_slot->Reset();
-	ReleaseResourcesAfterFence();
-	m_slot = nullptr;
-
-	EXIT_NOT_IMPLEMENTED(!IsInvalid());
-}
-
-void CommandBuffer::RetireBufferAfterFence(std::unique_ptr<VulkanBuffer> buffer) {
-	if (IsInvalid() || m_execute || buffer == nullptr || buffer->buffer == nullptr) {
-		EXIT("cannot retire a buffer on an invalid or submitted command buffer\n");
-	}
-	m_retired_buffers.push_back(std::move(buffer));
-}
-
-void CommandBuffer::RetainResourceUntilFence(std::shared_ptr<void> resource) {
-	if (IsInvalid() || m_execute) {
-		EXIT("cannot retain a resource on an invalid or submitted command buffer\n");
-	}
-	m_fence_resources.Retain(std::move(resource));
-}
-
-void CommandBuffer::RecycleDescriptorAfterFence(VulkanDescriptorSet& set) {
-	m_descriptor_sets_after_fence.push_back(&set);
-}
-
-void CommandBuffer::RecycleDescriptorsAfterFence() {
-	for (auto* set: m_descriptor_sets_after_fence) {
-		m_context.GetDescriptorCache().Recycle(*set);
-	}
-	m_descriptor_sets_after_fence.clear();
-}
-
-void CommandBuffer::Begin() const {
-	EXIT_IF(m_rendering);
+void CommandBuffer::Begin() {
+	EXIT_IF(m_rendering || IsInvalid());
 	auto buffer = Handle();
 
 	vk::CommandBufferBeginInfo begin_info {};
-	begin_info.sType            = vk::StructureType::eCommandBufferBeginInfo;
-	begin_info.pNext            = nullptr;
-	begin_info.flags            = {};
-	begin_info.pInheritanceInfo = nullptr;
+	begin_info.flags = vk::CommandBufferUsageFlagBits::eOneTimeSubmit;
 
 	auto result = buffer.begin(&begin_info);
 
@@ -137,125 +61,6 @@ void CommandBuffer::SetDebugInfo(uint32_t op, uint64_t submit_id, uint32_t arg0,
 	m_debug_arg4      = arg4;
 }
 
-void CommandBuffer::Execute(const SubmitInfo& submit) {
-	EXIT_IF(IsInvalid());
-	EXIT_IF(m_execute);
-	EXIT_IF(submit.num_wait_semaphores > SubmitInfo::MaxSemaphores ||
-	        submit.num_signal_semaphores > SubmitInfo::MaxSemaphores);
-
-	auto buffer = Handle();
-	auto fence  = m_slot->fence;
-
-	vk::TimelineSemaphoreSubmitInfo timeline_info {};
-	timeline_info.sType                     = vk::StructureType::eTimelineSemaphoreSubmitInfo;
-	timeline_info.waitSemaphoreValueCount   = submit.num_wait_semaphores;
-	timeline_info.pWaitSemaphoreValues      = submit.wait_ticks.data();
-	timeline_info.signalSemaphoreValueCount = submit.num_signal_semaphores;
-	timeline_info.pSignalSemaphoreValues    = submit.signal_ticks.data();
-
-	vk::SubmitInfo submit_info {};
-	submit_info.sType                = vk::StructureType::eSubmitInfo;
-	submit_info.pNext                = &timeline_info;
-	submit_info.waitSemaphoreCount   = submit.num_wait_semaphores;
-	submit_info.pWaitSemaphores      = submit.wait_semaphores.data();
-	submit_info.pWaitDstStageMask    = submit.wait_stages.data();
-	submit_info.commandBufferCount   = 1;
-	submit_info.pCommandBuffers      = &buffer;
-	submit_info.signalSemaphoreCount = submit.num_signal_semaphores;
-	submit_info.pSignalSemaphores    = submit.signal_semaphores.data();
-
-	auto& graphics = m_graphics;
-	EXIT_IF(graphics.queue == nullptr);
-
-	auto result = graphics.device.resetFences(1, &fence);
-	if (result != vk::Result::eSuccess) {
-		LOGF("vkResetFences failed before submit: %s (%d)\n", VulkanToString(result).c_str(),
-		     static_cast<int>(result));
-	}
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
-
-	if (Config::GraphicsDebugDumpEnabled()) {
-		LOGF("vkQueueSubmit begin: slot=%u waits=%u signals=%u debug_op=%u debug_submit=%" PRIu64
-		     " args=%u,%u,%u,%u,0x%016" PRIx64 "\n",
-		     m_slot->id, submit.num_wait_semaphores, submit.num_signal_semaphores, m_debug_op,
-		     m_debug_submit_id, m_debug_arg0, m_debug_arg1, m_debug_arg2, m_debug_arg3,
-		     m_debug_arg4);
-	}
-
-	{
-		Common::LockGuard lock(graphics.queue_mutex);
-		m_submit_seq = m_scheduler.NextSubmitSequence();
-		result       = graphics.queue.submit(1, &submit_info, fence);
-	}
-
-	m_execute      = true;
-	m_fence_waited = false;
-
-	if (result != vk::Result::eSuccess) {
-		LOGF("vkQueueSubmit failed: %s (%d), slot=%u submit_seq=%" PRIu64
-		     " debug_op=%u debug_submit=%" PRIu64 " args=%u,%u,%u,%u,0x%016" PRIx64 "\n",
-		     VulkanToString(result).c_str(), static_cast<int>(result), m_slot->id, m_submit_seq,
-		     m_debug_op, m_debug_submit_id, m_debug_arg0, m_debug_arg1, m_debug_arg2, m_debug_arg3,
-		     m_debug_arg4);
-	}
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
-}
-
-void CommandBuffer::WaitForFence() {
-	FinalizeFence(false);
-}
-
-void CommandBuffer::WaitForFenceOnly() {
-	EXIT_IF(IsInvalid());
-	if (!m_execute || m_fence_waited) {
-		return;
-	}
-	auto device = m_graphics.device;
-	auto result = device.waitForFences(1, &m_slot->fence, VK_TRUE, UINT64_MAX);
-	if (result != vk::Result::eSuccess) {
-		LOGF("vkWaitForFences failed: %s (%d), slot=%u submit_seq=%" PRIu64
-		     " debug_op=%u debug_submit=%" PRIu64 " args=%u,%u,%u,%u,0x%016" PRIx64 "\n",
-		     VulkanToString(result).c_str(), static_cast<int>(result), m_slot->id, m_submit_seq,
-		     m_debug_op, m_debug_submit_id, m_debug_arg0, m_debug_arg1, m_debug_arg2, m_debug_arg3,
-		     m_debug_arg4);
-	}
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
-	m_fence_waited = true;
-}
-
-void CommandBuffer::WaitForFenceAndReset() {
-	FinalizeFence(true);
-}
-
-void CommandBuffer::FinalizeFence(bool reset_recording) {
-	const bool was_executed = m_execute;
-	WaitForFenceOnly();
-	if (was_executed) {
-		m_execute      = false;
-		m_fence_waited = false;
-		if (reset_recording) {
-			Common::LockGuard lock(*m_slot->pool_mutex);
-			m_slot->Reset();
-		}
-	}
-	if (was_executed) {
-		ReleaseResourcesAfterFence();
-	}
-	DeleteBuffersAfterFence();
-}
-
-void CommandBuffer::ReleaseResourcesAfterFence() {
-	RecycleDescriptorsAfterFence();
-	m_fence_resources.ReleaseAfterFence();
-}
-
-void CommandBuffer::DeleteBuffersAfterFence() {
-	for (const auto& buffer: m_retired_buffers) {
-		m_graphics.DeleteBuffer(*buffer);
-	}
-	m_retired_buffers.clear();
-}
-
 void CommandBuffer::BeginRendering(const RenderState& state) const {
 	EXIT_IF(state.width == 0 || state.height == 0 || state.num_layers == 0 ||
 	        state.num_color_attachments > RENDER_COLOR_ATTACHMENTS_MAX);
@@ -267,7 +72,6 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {
 	std::array<vk::RenderingAttachmentInfo, RENDER_COLOR_ATTACHMENTS_MAX> colors {};
 	for (uint32_t i = 0; i < state.num_color_attachments; i++) {
 		const auto& attachment = state.color_attachments[i];
-		colors[i].sType        = vk::StructureType::eRenderingAttachmentInfo;
 		colors[i].imageView    = attachment.image_view;
 		colors[i].imageLayout  = attachment.image_layout;
 		colors[i].loadOp =
@@ -278,7 +82,6 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {
 
 	const auto&                 depth_stencil = state.depth_stencil_attachment;
 	vk::RenderingAttachmentInfo depth {};
-	depth.sType       = vk::StructureType::eRenderingAttachmentInfo;
 	depth.imageView   = depth_stencil.image_view;
 	depth.imageLayout = depth_stencil.image_layout;
 	depth.loadOp =
@@ -287,7 +90,6 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {
 	depth.clearValue.depthStencil.depth = std::bit_cast<float>(depth_stencil.clear_value[0]);
 
 	vk::RenderingAttachmentInfo stencil {};
-	stencil.sType       = vk::StructureType::eRenderingAttachmentInfo;
 	stencil.imageView   = depth_stencil.image_view;
 	stencil.imageLayout = depth_stencil.image_layout;
 	stencil.loadOp =
@@ -296,7 +98,6 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {
 	stencil.clearValue.depthStencil.stencil = depth_stencil.clear_value[1];
 
 	vk::RenderingInfo rendering {};
-	rendering.sType                = vk::StructureType::eRenderingInfo;
 	rendering.renderArea.extent    = {state.width, state.height};
 	rendering.layerCount           = state.num_layers;
 	rendering.colorAttachmentCount = state.num_color_attachments;

src/graphics/host_gpu/renderer/debug.cpp

@@ -17,8 +17,6 @@
 
 namespace Libs::Graphics {
 
-static std::atomic<uint32_t> g_scissor_default_log_count = 0;
-
 uint32_t render_target_mask_slot(uint32_t mask, uint32_t slot) {
 	return (mask >> (slot * 4u)) & 0x0fu;
 }
@@ -27,7 +25,7 @@ static bool RenderTargetMaskHasMrt(uint32_t mask) {
 	return (mask & ~0x0fu) != 0;
 }
 
-static bool RenderTargetMaskHasBoundMrt(const RenderCommandBuffer& buffer) {
+static bool RenderTargetMaskHasBoundMrt(const CommandBuffer& buffer) {
 	const auto& hw   = buffer.GetRegisters();
 	const auto  mask = hw.GetRenderTargetMask();
 
@@ -45,7 +43,7 @@ static bool RenderTargetMaskHasBoundMrt(const RenderCommandBuffer& buffer) {
 	return bound_targets > 1;
 }
 
-uint32_t render_target_first_bound_slot(const RenderCommandBuffer& buffer) {
+uint32_t render_target_first_bound_slot(const CommandBuffer& buffer) {
 	const auto& hw   = buffer.GetRegisters();
 	const auto  mask = hw.GetRenderTargetMask();
 	for (uint32_t i = 0; i < 8; i++) {
@@ -84,17 +82,8 @@ void uc_print(const char* func, const HW::UserConfig& uc) {
 }
 
 void uc_check(const HW::UserConfig& uc) {
-	const auto& ge_cntl = uc.GetGeControl();
 	const auto& user_en = uc.GetGeUserVgprEn();
 
-	if (ge_cntl.primitive_group_size > 0x0040) {
-		LOGF("\t warning: unsupported GE_CNTL primitive_group_size = 0x%04" PRIx16 "\n",
-		     ge_cntl.primitive_group_size);
-	}
-	if (ge_cntl.vertex_group_size > 0x0040) {
-		LOGF("\t warning: unsupported GE_CNTL vertex_group_size = 0x%04" PRIx16 "\n",
-		     ge_cntl.vertex_group_size);
-	}
 	EXIT_NOT_IMPLEMENTED(user_en.vgpr1 != false);
 	EXIT_NOT_IMPLEMENTED(user_en.vgpr2 != false);
 	EXIT_NOT_IMPLEMENTED(user_en.vgpr3 != false);
@@ -106,17 +95,11 @@ void sh_print(const char* func, const HW::Shader& /*uc*/) {
 
 std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt) {
 	std::vector<std::string> dst;
-	dst.reserve(53);
+	dst.reserve(40);
 
 	dst.push_back(fmt::format("{}\n", func));
 
 	dst.push_back(fmt::format("\t base.addr                       = 0x{:016x}\n", rt.base.addr));
-	dst.push_back(
-	    fmt::format("\t pitch.pitch_div8_minus1         = 0x{:08x}\n", rt.pitch.pitch_div8_minus1));
-	dst.push_back(fmt::format("\t pitch.fmask_pitch_div8_minus1   = 0x{:08x}\n",
-	                          rt.pitch.fmask_pitch_div8_minus1));
-	dst.push_back(fmt::format("\t slice.slice_div64_minus1        = 0x{:08x}\n",
-	                          rt.slice.slice_div64_minus1));
 	dst.push_back(fmt::format("\t view.base_array_slice_index     = 0x{:08x}\n",
 	                          rt.view.base_array_slice_index));
 	dst.push_back(fmt::format("\t view.last_array_slice_index     = 0x{:08x}\n",
@@ -137,12 +120,6 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	                          rt.info.cmask_fast_clear_enable ? "true" : "false"));
 	dst.push_back(fmt::format("\t info.dcc_compression_enable     = {}\n",
 	                          rt.info.dcc_compression_enable ? "true" : "false"));
-	dst.push_back(
-	    fmt::format("\t info.cmask_is_linear            = 0x{:08x}\n", rt.info.cmask_is_linear));
-	dst.push_back(
-	    fmt::format("\t info.cmask_addr_type            = 0x{:08x}\n", rt.info.cmask_addr_type));
-	dst.push_back(fmt::format("\t info.alt_tile_mode              = {}\n",
-	                          rt.info.alt_tile_mode ? "true" : "false"));
 	dst.push_back(fmt::format("\t info.format                     = 0x{:08x}\n",
 	                          static_cast<uint32_t>(rt.info.format)));
 	dst.push_back(fmt::format("\t info.channel_type               = 0x{:08x}\n",
@@ -157,10 +134,6 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	                          rt.info.round_mode ? "true" : "false"));
 	dst.push_back(fmt::format("\t attrib.force_dest_alpha_to_one  = {}\n",
 	                          rt.attrib.force_dest_alpha_to_one ? "true" : "false"));
-	dst.push_back(fmt::format("\t attrib.tile_mode                = 0x{:08x}\n",
-	                          static_cast<uint32_t>(rt.attrib.tile_mode)));
-	dst.push_back(fmt::format("\t attrib.fmask_tile_mode          = 0x{:08x}\n",
-	                          static_cast<uint32_t>(rt.attrib.fmask_tile_mode)));
 	dst.push_back(
 	    fmt::format("\t attrib.num_samples              = 0x{:08x}\n", rt.attrib.num_samples));
 	dst.push_back(
@@ -175,42 +148,32 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	                          static_cast<uint32_t>(rt.attrib3.tile_mode)));
 	dst.push_back(
 	    fmt::format("\t attrib3.dimension               = 0x{:08x}\n", rt.attrib3.dimension));
-	dst.push_back(fmt::format("\t attrib3.cmask_pipe_aligned      = {}\n",
-	                          rt.attrib3.cmask_pipe_aligned ? "true" : "false"));
-	dst.push_back(fmt::format("\t attrib3.dcc_pipe_aligned        = {}\n",
-	                          rt.attrib3.dcc_pipe_aligned ? "true" : "false"));
+	dst.push_back(fmt::format("\t attrib3.metadata_pipe_aligned   = {}\n",
+	                          rt.attrib3.metadata_pipe_aligned ? "true" : "false"));
+	dst.push_back(fmt::format("\t attrib3.write_vrs_rate_hint_to_cmask = {}\n",
+	                          rt.attrib3.write_vrs_rate_hint_to_cmask ? "true" : "false"));
 	dst.push_back(fmt::format("\t dcc.max_uncompressed_block_size = 0x{:08x}\n",
 	                          rt.dcc.max_uncompressed_block_size));
 	dst.push_back(fmt::format("\t dcc.max_compressed_block_size   = 0x{:08x}\n",
 	                          rt.dcc.max_compressed_block_size));
-	dst.push_back(fmt::format("\t dcc.min_compressed_block_size   = 0x{:08x}\n",
-	                          rt.dcc.min_compressed_block_size));
 	dst.push_back(
 	    fmt::format("\t dcc.color_transform             = 0x{:08x}\n", rt.dcc.color_transform));
 	dst.push_back(fmt::format("\t dcc.overwrite_combiner_disable  = {}\n",
 	                          rt.dcc.overwrite_combiner_disable ? "true" : "false"));
-	dst.push_back(fmt::format("\t dcc.independent_64b_blocks      = {}\n",
-	                          rt.dcc.independent_64b_blocks ? "true" : "false"));
-	dst.push_back(fmt::format("\t dcc.independent_128b_blocks     = {}\n",
-	                          rt.dcc.independent_128b_blocks ? "true" : "false"));
+	dst.push_back(fmt::format("\t dcc.independent_block_size      = 0x{:02x}\n",
+	                          static_cast<uint8_t>(rt.dcc.independent_block_size)));
 	dst.push_back(fmt::format("\t data_write_on_dcc_clear_to_reg  = {}\n",
 	                          rt.dcc.data_write_on_dcc_clear_to_reg ? "true" : "false"));
 	dst.push_back(fmt::format("\t dcc.dcc_clear_key_enable        = {}\n",
 	                          rt.dcc.dcc_clear_key_enable ? "true" : "false"));
 	dst.push_back(fmt::format("\t cmask.addr                      = 0x{:016x}\n", rt.cmask.addr));
-	dst.push_back(fmt::format("\t cmask_slice.slice_minus1        = 0x{:08x}\n",
-	                          rt.cmask_slice.slice_minus1));
 	dst.push_back(fmt::format("\t fmask.addr                      = 0x{:016x}\n", rt.fmask.addr));
-	dst.push_back(fmt::format("\t fmask_slice.slice_minus1        = 0x{:08x}\n",
-	                          rt.fmask_slice.slice_minus1));
 	dst.push_back(
 	    fmt::format("\t clear_word0.word0               = 0x{:08x}\n", rt.clear_word0.word0));
 	dst.push_back(
 	    fmt::format("\t clear_word1.word1               = 0x{:08x}\n", rt.clear_word1.word1));
 	dst.push_back(
 	    fmt::format("\t dcc_addr.addr                   = 0x{:016x}\n", rt.dcc_addr.addr));
-	dst.push_back(fmt::format("\t size.width                      = 0x{:08x}\n", rt.size.width));
-	dst.push_back(fmt::format("\t size.height                     = 0x{:08x}\n", rt.size.height));
 
 	return dst;
 }
@@ -240,13 +203,8 @@ static bool RenderIsColorDimension(uint32_t dimension) {
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
 static void RtCheck(const HW::RenderTarget& rt) {
 	if (rt.base.addr != 0) {
-		// bool render_to_texture = (rt.attrib.tile_mode == 0x0d);
 		//  EXIT_NOT_IMPLEMENTED(rt.base_addr == 0);
 
-		EXIT_NOT_IMPLEMENTED(rt.pitch.pitch_div8_minus1 != 0);
-		EXIT_NOT_IMPLEMENTED(rt.pitch.fmask_pitch_div8_minus1 != 0);
-		EXIT_NOT_IMPLEMENTED(rt.slice.slice_div64_minus1 != 0);
-
 		EXIT_NOT_IMPLEMENTED(rt.view.base_array_slice_index > rt.view.last_array_slice_index);
 		if (rt.view.base_array_slice_index != 0x00000000 ||
 		    rt.view.last_array_slice_index != 0x00000000) {
@@ -270,16 +228,14 @@ static void RtCheck(const HW::RenderTarget& rt) {
 			EXIT_NOT_IMPLEMENTED(rt.attrib.num_samples == 0 && rt.attrib.num_fragments == 0);
 			static bool logged = false;
 			if (!logged) {
-				LOGF("RenderTarget: temporary: ignoring PS5 FMASK metadata for single-sample MSAA "
-				     "fallback, fmask=0x%016" PRIx64 "\n",
+				LOGF("RenderTarget: using native Vulkan MSAA without guest FMASK metadata, "
+				     "fmask=0x%016" PRIx64 "\n",
 				     rt.fmask.addr);
 				logged = true;
 			}
 		}
 
-		// EXIT_NOT_IMPLEMENTED(rt.info.fmask_compression_mode != 0x00000000);
 		EXIT_NOT_IMPLEMENTED(rt.info.fmask_data_compression_disable != false);
-		EXIT_NOT_IMPLEMENTED(rt.info.fmask_one_frag_mode != false);
 
 		if (rt.info.cmask_fast_clear_enable || rt.info.dcc_compression_enable) {
 			static bool logged = false;
@@ -292,22 +248,6 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
-		if (rt.info.cmask_is_linear != 0x00000000 || rt.info.cmask_addr_type != 0x00000000) {
-			static bool logged = false;
-			if (!logged) {
-				LOGF("RenderTarget: temporary: ignoring PS5 CMASK metadata fields "
-				     "is_linear=%u addr_type=%u\n",
-				     rt.info.cmask_is_linear, rt.info.cmask_addr_type);
-				logged = true;
-			}
-		}
-		if (rt.info.alt_tile_mode) {
-			static bool logged = false;
-			if (!logged) {
-				LOGF("RenderTarget: temporary: ignoring alternate tile mode flag\n");
-				logged = true;
-			}
-		}
 		if (rt.info.blend_bypass) {
 			static bool logged = false;
 			if (!logged) {
@@ -334,8 +274,6 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
-		// EXIT_NOT_IMPLEMENTED(rt.tile_mode != 0x0000000a);
-		// EXIT_NOT_IMPLEMENTED(rt.fmask_tile_mode != 0x0000000a);
 		if (rt.attrib.num_samples != 0x00000000 || rt.attrib.num_fragments != 0x00000000) {
 			static bool logged = false;
 			if (!logged) {
@@ -370,34 +308,22 @@ static void RtCheck(const HW::RenderTarget& rt) {
 		if (!RenderIsColorDimension(rt.attrib3.dimension)) {
 			EXIT("unknown PS5 render-target dimension: 0x%08" PRIx32 "\n", rt.attrib3.dimension);
 		}
-		if (!rt.attrib3.cmask_pipe_aligned) {
+		if (!rt.attrib3.metadata_pipe_aligned) {
 			static bool logged = false;
 			if (!logged) {
-				LOGF("RenderTarget: temporary: accepting unaligned PS5 CMASK pipe flag\n");
-				logged = true;
-			}
-		}
-		if (!rt.attrib3.dcc_pipe_aligned) {
-			static bool logged = false;
-			if (!logged) {
-				LOGF("RenderTarget: temporary: accepting unaligned PS5 DCC pipe flag\n");
+				LOGF("RenderTarget: temporary: accepting unaligned PS5 metadata pipe flag\n");
 				logged = true;
 			}
 		}
+		EXIT_NOT_IMPLEMENTED(rt.attrib3.write_vrs_rate_hint_to_cmask);
 
 		// EXIT_NOT_IMPLEMENTED(rt.dcc_max_uncompressed_block_size != 0x00000002);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.max_compressed_block_size != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(rt.dcc.min_compressed_block_size != 0x00000000);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.color_transform != 0x00000000);
 		EXIT_NOT_IMPLEMENTED(rt.dcc.overwrite_combiner_disable != false);
-		// EXIT_NOT_IMPLEMENTED(rt.dcc.force_independent_blocks != false);
-		// EXIT_NOT_IMPLEMENTED(rt.dcc.independent_128b_blocks != false);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.data_write_on_dcc_clear_to_reg != false);
 		EXIT_NOT_IMPLEMENTED(rt.dcc.dcc_clear_key_enable != false);
-		if (rt.cmask.addr != 0x0000000000000000 || rt.cmask_slice.slice_minus1 != 0x00000000 ||
-		    rt.fmask.addr != 0x0000000000000000 ||
-		    (rt.fmask_slice.slice_minus1 != 0x00000000 &&
-		     rt.fmask_slice.slice_minus1 != rt.slice.slice_div64_minus1) ||
+		if (rt.cmask.addr != 0x0000000000000000 || rt.fmask.addr != 0x0000000000000000 ||
 		    rt.dcc_addr.addr != 0x0000000000000000) {
 			static bool logged = false;
 			if (!logged) {
@@ -407,9 +333,6 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
-
-		EXIT_NOT_IMPLEMENTED(rt.size.width != 0);
-		EXIT_NOT_IMPLEMENTED(rt.size.height != 0);
 	}
 }
 
@@ -429,37 +352,20 @@ static void ZPrint(const char* func, const HW::DepthRenderTarget& z) {
 	     "\t stencil_info.htile_stencil_disabled   = %s\n"
 	     "\t stencil_info.expclear_enabled         = %s\n"
 	     "\t stencil_info.partially_resident       = %s\n"
-	     "\t depth_info.addr5_swizzle_mask         = 0x%08" PRIx32 "\n"
-	     "\t depth_info.array_mode                 = 0x%08" PRIx32 "\n"
-	     "\t depth_info.pipe_config                = 0x%08" PRIx32 "\n"
-	     "\t depth_info.bank_width                 = 0x%08" PRIx32 "\n"
-	     "\t depth_info.bank_height                = 0x%08" PRIx32 "\n"
-	     "\t depth_info.macro_tile_aspect          = 0x%08" PRIx32 "\n"
-	     "\t depth_info.num_banks                  = 0x%08" PRIx32 "\n"
 	     "\t depth_view.slice_start                = 0x%08" PRIx32 "\n"
 	     "\t depth_view.slice_max                  = 0x%08" PRIx32 "\n"
 	     "\t depth_view.current_mip_level          = 0x%02" PRIx8 "\n"
 	     "\t depth_view.depth_write_disable        = %s\n"
 	     "\t depth_view.stencil_write_disable      = %s\n"
-	     "\t htile_surface.linear                  = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.full_cache              = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.htile_uses_preload_win  = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.preload                 = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.prefetch_width          = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.prefetch_height         = 0x%08" PRIx32 "\n"
-	     "\t htile_surface.dst_outside_zero_to_one = 0x%08" PRIx32 "\n"
 	     "\t z_read_base_addr                      = 0x%016" PRIx64 "\n"
 	     "\t stencil_read_base_addr                = 0x%016" PRIx64 "\n"
 	     "\t z_write_base_addr                     = 0x%016" PRIx64 "\n"
 	     "\t stencil_write_base_addr               = 0x%016" PRIx64 "\n"
-	     "\t pitch_div8_minus1                     = 0x%08" PRIx32 "\n"
-	     "\t height_div8_minus1                    = 0x%08" PRIx32 "\n"
-	     "\t slice_div64_minus1                    = 0x%08" PRIx32 "\n"
 	     "\t htile_data_base_addr                  = 0x%016" PRIx64 "\n"
-	     "\t width                                 = 0x%08" PRIx32 "\n"
-	     "\t height                                = 0x%08" PRIx32 "\n"
+	     "\t shading_rate_encoding                 = 0x%02" PRIx8 "\n"
 	     "\t size.x_max                            = 0x%04" PRIx16 "\n"
-	     "\t size.y_max                            = 0x%04" PRIx16 "\n",
+	     "\t size.y_max                            = 0x%04" PRIx16 "\n"
+	     "\t size.valid                            = %s\n",
 	     static_cast<uint32_t>(z.z_info.format), z.z_info.num_samples,
 	     static_cast<uint32_t>(z.z_info.texture_compatibility),
 	     z.z_info.htile_acceleration ? "true" : "false",
@@ -470,123 +376,13 @@ static void ZPrint(const char* func, const HW::DepthRenderTarget& z) {
 	     static_cast<uint32_t>(z.stencil_info.texture_compatibility),
 	     z.stencil_info.htile_stencil_disabled ? "true" : "false",
 	     z.stencil_info.expclear_enabled ? "true" : "false",
-	     z.stencil_info.partially_resident ? "true" : "false", z.depth_info.addr5_swizzle_mask,
-	     z.depth_info.array_mode, z.depth_info.pipe_config, z.depth_info.bank_width,
-	     z.depth_info.bank_height, z.depth_info.macro_tile_aspect, z.depth_info.num_banks,
-	     z.depth_view.slice_start, z.depth_view.slice_max, z.depth_view.current_mip_level,
+	     z.stencil_info.partially_resident ? "true" : "false", z.depth_view.slice_start,
+	     z.depth_view.slice_max, z.depth_view.current_mip_level,
 	     z.depth_view.depth_write_disable ? "true" : "false",
-	     z.depth_view.stencil_write_disable ? "true" : "false", z.htile_surface.linear,
-	     z.htile_surface.full_cache, z.htile_surface.htile_uses_preload_win,
-	     z.htile_surface.preload, z.htile_surface.prefetch_width, z.htile_surface.prefetch_height,
-	     z.htile_surface.dst_outside_zero_to_one, z.z_read_base_addr, z.stencil_read_base_addr,
-	     z.z_write_base_addr, z.stencil_write_base_addr, z.pitch_div8_minus1, z.height_div8_minus1,
-	     z.slice_div64_minus1, z.htile_data_base_addr, z.width, z.height, z.size.x_max,
-	     z.size.y_max);
-}
-
-// NOLINTNEXTLINE(readability-function-cognitive-complexity)
-static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
-                   const HW::RenderControl& rc) {
-	const bool depth_active =
-	    dc.z_enable || dc.z_write_enable || dc.depth_bounds_enable || rc.depth_clear_enable;
-	const bool stencil_active = dc.stencil_enable || rc.stencil_clear_enable;
-	if (!depth_active && !stencil_active) {
-		return;
-	}
-
-	EXIT_NOT_IMPLEMENTED(!z.z_info.HasValidTextureCompatibility());
-	EXIT_NOT_IMPLEMENTED(!z.stencil_info.HasValidTextureCompatibility());
-	if (z.z_info.format == Prospero::DepthFormat::kInvalid) {
-		EXIT_NOT_IMPLEMENTED(z.z_info.format != Prospero::DepthFormat::kInvalid);
-		EXIT_NOT_IMPLEMENTED(z.z_info.num_samples != 0);
-		EXIT_NOT_IMPLEMENTED(z.z_info.htile_acceleration != false);
-		EXIT_NOT_IMPLEMENTED(z.z_info.expclear_enabled != false);
-		EXIT_NOT_IMPLEMENTED(z.z_info.partially_resident != false);
-		EXIT_NOT_IMPLEMENTED(z.z_info.max_mip_level != 0);
-	} else {
-		EXIT_NOT_IMPLEMENTED(z.z_info.format != Prospero::DepthFormat::kZ16 &&
-		                     z.z_info.format != Prospero::DepthFormat::kZ32F);
-		if (z.z_info.num_samples != 0x00000000) {
-			static bool logged = false;
-			if (!logged) {
-				LOGF("DepthTarget: using native num_samples=0x%08" PRIx32 "\n",
-				     z.z_info.num_samples);
-				logged = true;
-			}
-		}
-		EXIT_NOT_IMPLEMENTED(z.z_info.expclear_enabled != false);
-		EXIT_NOT_IMPLEMENTED(z.z_info.partially_resident != false);
-		EXIT_NOT_IMPLEMENTED(z.z_info.max_mip_level != 0);
-	}
-
-	if (z.stencil_info.format == Prospero::StencilFormat::kInvalid) {
-		EXIT_NOT_IMPLEMENTED(z.stencil_info.expclear_enabled != false);
-		EXIT_NOT_IMPLEMENTED(z.stencil_info.partially_resident != false);
-	} else {
-		EXIT_NOT_IMPLEMENTED(z.stencil_info.format != Prospero::StencilFormat::k8UInt);
-		EXIT_NOT_IMPLEMENTED(z.stencil_info.expclear_enabled != false);
-		EXIT_NOT_IMPLEMENTED(z.stencil_info.partially_resident != false);
-	}
-
-	if (z.z_info.format != Prospero::DepthFormat::kInvalid ||
-	    z.stencil_info.format != Prospero::StencilFormat::kInvalid) {
-
-		EXIT_NOT_IMPLEMENTED(z.depth_info.addr5_swizzle_mask != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.array_mode != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.pipe_config != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.bank_width != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.bank_height != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.macro_tile_aspect != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.depth_info.num_banks != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.linear != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.full_cache != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.htile_uses_preload_win != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.preload != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.prefetch_width != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.prefetch_height != 0x00000000);
-		EXIT_NOT_IMPLEMENTED(z.htile_surface.dst_outside_zero_to_one != 0x00000000);
-
-		if (z.depth_view.current_mip_level != 0x00000000) {
-			static std::atomic<uint32_t> log_count {0};
-			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {
-				LOGF("DepthTarget: temporary: ignoring PS5 current mip level=0x%08" PRIx32 "\n",
-				     z.depth_view.current_mip_level);
-			}
-		}
-		if (z.depth_view.depth_write_disable || z.depth_view.stencil_write_disable) {
-			static std::atomic<uint32_t> log_count {0};
-			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {
-				LOGF("DepthTarget: honoring write disable depth=%s, stencil=%s\n",
-				     z.depth_view.depth_write_disable ? "true" : "false",
-				     z.depth_view.stencil_write_disable ? "true" : "false");
-			}
-		}
-		EXIT_NOT_IMPLEMENTED(z.z_read_base_addr != z.z_write_base_addr);
-		EXIT_NOT_IMPLEMENTED(z.stencil_read_base_addr != z.stencil_write_base_addr);
-		EXIT_NOT_IMPLEMENTED(z.z_write_base_addr == 0);
-		if (z.stencil_info.format != Prospero::StencilFormat::kInvalid &&
-		    z.stencil_write_base_addr == 0) {
-			LOGF("\t warning: stencil format is set without a stencil base address, continuing "
-			     "without stencil attachment\n");
-		}
-		// EXIT_NOT_IMPLEMENTED(z.pitch_div8_minus1 != 0x000000ff);
-		// EXIT_NOT_IMPLEMENTED(z.height_div8_minus1 != 0x0000008f);
-		// EXIT_NOT_IMPLEMENTED(z.slice_div64_minus1 != 0x00008fff);
-		// EXIT_NOT_IMPLEMENTED(z.htile_data_base_addr == 0);
-		// EXIT_NOT_IMPLEMENTED(z.width != 0x00000780);
-		// EXIT_NOT_IMPLEMENTED(z.height != 0x00000438);
-
-		EXIT_NOT_IMPLEMENTED(z.width != 0);
-		EXIT_NOT_IMPLEMENTED(z.height != 0);
-		if (z.size.x_max == 0 || z.size.y_max == 0) {
-			static std::atomic<uint32_t> log_count {0};
-			if (log_count.fetch_add(1) < 16) {
-				LOGF("DepthTarget: temporary: inferring PS5 default depth extent x_max=0x%04" PRIx16
-				     ", y_max=0x%04" PRIx16 "\n",
-				     z.size.x_max, z.size.y_max);
-			}
-		}
-	}
+	     z.depth_view.stencil_write_disable ? "true" : "false", z.z_read_base_addr,
+	     z.stencil_read_base_addr, z.z_write_base_addr, z.stencil_write_base_addr,
+	     z.htile_data_base_addr, z.shading_rate_encoding, z.size.x_max, z.size.y_max,
+	     z.size.valid ? "true" : "false");
 }
 
 static void ClipPrint(const char* func, const HW::ClipControl& c) {
@@ -617,7 +413,7 @@ static void ClipCheck(const HW::ClipControl& c) {
 	// dx_linear_attr_clip_enable preserves linear (noperspective) attributes at clip-generated
 	// vertices, which Vulkan provides as part of clipping and interpolation.
 	EXIT_NOT_IMPLEMENTED(c.user_clip_planes != 0 || c.user_clip_plane_mode != 0 ||
-	                     c.vertex_kill_any || c.user_clip_plane_negate_y || c.clip_disable ||
+	                     c.vertex_kill_any || c.user_clip_plane_negate_y ||
 	                     c.user_clip_plane_cull_only || c.cull_on_clipping_error_disable ||
 	                     c.force_viewport_index_from_vs_enable);
 }
@@ -673,16 +469,6 @@ static void McCheck(const HW::ModeControl& c) {
 	// EXIT_NOT_IMPLEMENTED(c.cull_front != false);
 	// EXIT_NOT_IMPLEMENTED(c.cull_back != false);
 	// EXIT_NOT_IMPLEMENTED(c.face != false);
-	if (c.poly_mode != 0) {
-		static bool logged = false;
-		if (!logged) {
-			LOGF("\t temporary: PA_SU_SC_MODE_CNTL.POLY_MODE is not fully implemented; continuing "
-			     "with filled polygons\n");
-			logged = true;
-		}
-	}
-	EXIT_NOT_IMPLEMENTED(c.polymode_front_ptype != 0 && c.polymode_front_ptype != 2);
-	EXIT_NOT_IMPLEMENTED(c.polymode_back_ptype != 0 && c.polymode_back_ptype != 2);
 	if (c.vtx_window_offset_enable) {
 		static bool logged = false;
 		if (!logged) {
@@ -768,8 +554,6 @@ static void DPrint(const char* func, const HW::DepthControl& c, const HW::Stenci
 	     "\t backface_enable      = %s\n"
 	     "\t stencilfunc          = %" PRIu8 "\n"
 	     "\t stencilfunc_bf       = %" PRIu8 "\n"
-	     "\t color_writes_on_depth_fail_enable  = %s\n"
-	     "\t color_writes_on_depth_pass_disable = %s\n"
 	     "\t stencil_fail         = %" PRIu8 "\n"
 	     "\t stencil_zpass        = %" PRIu8 "\n"
 	     "\t stencil_zfail        = %" PRIu8 "\n"
@@ -786,12 +570,11 @@ static void DPrint(const char* func, const HW::DepthControl& c, const HW::Stenci
 	     "\t stencil_opval_bf     = %" PRIu8 "\n",
 	     c.stencil_enable ? "true" : "false", c.z_enable ? "true" : "false",
 	     c.z_write_enable ? "true" : "false", c.depth_bounds_enable ? "true" : "false", c.zfunc,
-	     c.backface_enable ? "true" : "false", c.stencilfunc, c.stencilfunc_bf,
-	     c.color_writes_on_depth_fail_enable ? "true" : "false",
-	     c.color_writes_on_depth_pass_disable ? "true" : "false", s.stencil_fail, s.stencil_zpass,
-	     s.stencil_zfail, s.stencil_fail_bf, s.stencil_zpass_bf, s.stencil_zfail_bf,
-	     sm.stencil_testval, sm.stencil_mask, sm.stencil_writemask, sm.stencil_opval,
-	     sm.stencil_testval_bf, sm.stencil_mask_bf, sm.stencil_writemask_bf, sm.stencil_opval_bf);
+	     c.backface_enable ? "true" : "false", c.stencilfunc, c.stencilfunc_bf, s.stencil_fail,
+	     s.stencil_zpass, s.stencil_zfail, s.stencil_fail_bf, s.stencil_zpass_bf,
+	     s.stencil_zfail_bf, sm.stencil_testval, sm.stencil_mask, sm.stencil_writemask,
+	     sm.stencil_opval, sm.stencil_testval_bf, sm.stencil_mask_bf, sm.stencil_writemask_bf,
+	     sm.stencil_opval_bf);
 }
 
 static void DCheck(const HW::DepthControl& c, const HW::StencilControl& s,
@@ -803,8 +586,6 @@ static void DCheck(const HW::DepthControl& c, const HW::StencilControl& s,
 	// Back-face stencil state is handled separately when enabled.
 	// EXIT_NOT_IMPLEMENTED(c.stencilfunc != 0);
 	// EXIT_NOT_IMPLEMENTED(c.stencilfunc_bf != 0);
-	EXIT_NOT_IMPLEMENTED(c.color_writes_on_depth_fail_enable != false);
-	EXIT_NOT_IMPLEMENTED(c.color_writes_on_depth_pass_disable != false);
 	// EXIT_NOT_IMPLEMENTED(s.stencil_fail != 0);
 	// EXIT_NOT_IMPLEMENTED(s.stencil_zpass != 0);
 	// EXIT_NOT_IMPLEMENTED(s.stencil_zfail != 0);
@@ -836,7 +617,12 @@ static void EqaaPrint(const char* func, const HW::EqaaControl& c) {
 	     c.static_anchor_associations ? "true" : "false");
 }
 
-static void EqaaCheck(const HW::EqaaControl& c) {
+static void EqaaCheck(const HW::EqaaControl& c, const HW::AaConfig& cf) {
+	// The EQAA controls only take effect while multisampling is on, so a single-sample
+	// target leaves them inert and has nothing to warn about.
+	if (cf.msaa_num_samples == 0) {
+		return;
+	}
 	if (c.max_anchor_samples != 0 || c.ps_iter_samples != 0 || c.mask_export_num_samples != 0 ||
 	    c.alpha_to_mask_num_samples != 0 || c.high_quality_intersections ||
 	    c.incoherent_eqaa_reads || c.interpolate_comp_z || c.static_anchor_associations) {
@@ -888,12 +674,10 @@ void LogDrawPhase(const char* draw_name, const char* phase) {
 	}
 }
 
-void LogPipelineTrace(const char* phase, uint32_t vs_hash0, uint32_t vs_crc32, uint32_t ps_hash0,
-                      uint32_t ps_crc32) {
+void LogPipelineTrace(const char* phase, uint64_t vertex_program_id, uint64_t pixel_program_id) {
 	if (graphics_debug_dump_enabled()) {
-		LOGF("PipelineTrace: %s VS=0x%08" PRIx32 "/0x%08" PRIx32 " PS=0x%08" PRIx32 "/0x%08" PRIx32
-		     "\n",
-		     phase, vs_hash0, vs_crc32, ps_hash0, ps_crc32);
+		LOGF("PipelineTrace: %s VS=%" PRIu64 " PS=%" PRIu64 "\n", phase, vertex_program_id,
+		     pixel_program_id);
 	}
 }
 
@@ -974,15 +758,6 @@ static void VpCheck(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc
 	// EXIT_NOT_IMPLEMENTED(smc.vport_scissor_enable);
 	EXIT_NOT_IMPLEMENTED(smc.line_stipple_enable);
 
-	if (vp.viewports[0].zmin > 0.000000 || vp.viewports[0].zmax != 1.000000) {
-		static bool logged = false;
-		if (!logged) {
-			LOGF("\t warning: non-default viewport depth clamp zmin = %f, zmax = %f; using "
-			     "viewport scale/offset for Vulkan depth range\n",
-			     vp.viewports[0].zmin, vp.viewports[0].zmax);
-			logged = true;
-		}
-	}
 	// EXIT_NOT_IMPLEMENTED(vp.viewports[0].xscale != 960.000000);
 	// EXIT_NOT_IMPLEMENTED(vp.viewports[0].xoffset != 960.000000);
 	// EXIT_NOT_IMPLEMENTED(vp.viewports[0].yscale != -540.000000);
@@ -1002,7 +777,7 @@ static void VpCheck(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc
 	// EXIT_NOT_IMPLEMENTED(fabsf(vp.guard_band_horz_clip - 33.133327f) > 0.001f);
 	// EXIT_NOT_IMPLEMENTED(fabsf(vp.guard_band_vert_clip - 59.629623f) > 0.001f);
 
-	if (vp.guard_band_horz_discard != 0.0f || vp.guard_band_vert_discard != 0.0f) {
+	if (vp.guard_band_horz_discard != 1.0f || vp.guard_band_vert_discard != 1.0f) {
 		static std::atomic<uint32_t> log_count {0};
 		if (log_count.fetch_add(1) < 16) {
 			LOGF("\t warning: unsupported PS5 guard band discard = %f, %f, continuing\n",
@@ -1022,10 +797,6 @@ static bool ScissorRectValid(const ScissorRect& r) {
 	return r.right > r.left && r.bottom > r.top;
 }
 
-static bool ScissorRectSet(const ScissorRect& r) {
-	return r.left != 0 || r.top != 0 || r.right != 0 || r.bottom != 0;
-}
-
 static ScissorRect ScissorRectOffset(ScissorRect r, int x, int y) {
 	r.left += x;
 	r.right += x;
@@ -1035,8 +806,8 @@ static ScissorRect ScissorRectOffset(ScissorRect r, int x, int y) {
 }
 
 static ScissorRect ScissorRectIntersect(const ScissorRect& a, const ScissorRect& b) {
-	return {a.left > b.left ? a.left : b.left, a.top > b.top ? a.top : b.top,
-	        a.right < b.right ? a.right : b.right, a.bottom < b.bottom ? a.bottom : b.bottom};
+	return {std::max(a.left, b.left), std::max(a.top, b.top),
+	        std::min(a.right, b.right), std::min(a.bottom, b.bottom)};
 }
 
 static ScissorRect ScissorRectClamp(ScissorRect r, uint32_t width, uint32_t height) {
@@ -1083,49 +854,26 @@ static bool ScissorClipRuleToIntersectionMask(uint16_t rule, uint8_t* mask) {
 }
 
 ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc,
-                               vk::Extent2D extent) {
-	ScissorRect screen {vp.screen_scissor_left, vp.screen_scissor_top, vp.screen_scissor_right,
-	                    vp.screen_scissor_bottom};
-	ScissorRect final = screen;
-
-	if (!ScissorRectSet(screen)) {
-		final = {0, 0, static_cast<int>(extent.width), static_cast<int>(extent.height)};
-
-		auto log_id = g_scissor_default_log_count.fetch_add(1);
-		if (log_id < 32) {
-			LOGF("temporary: default unset screen scissor to framebuffer extent %ux%u\n",
-			     extent.width, extent.height);
+                               vk::Extent2D extent, uint32_t viewport_index) {
+	EXIT_IF(viewport_index >= std::size(vp.viewports));
+	ScissorRect final {vp.screen_scissor_left, vp.screen_scissor_top, vp.screen_scissor_right,
+	                   vp.screen_scissor_bottom};
+	const auto intersect = [&](ScissorRect rect, bool window_offset) {
+		if (window_offset) {
+			rect = ScissorRectOffset(rect, vp.window_offset_x, vp.window_offset_y);
 		}
-	}
+		final = ScissorRectIntersect(final, rect);
+	};
+	intersect({vp.window_scissor_left, vp.window_scissor_top, vp.window_scissor_right,
+	           vp.window_scissor_bottom}, vp.window_scissor_window_offset_enable);
+	intersect({vp.generic_scissor_left, vp.generic_scissor_top, vp.generic_scissor_right,
+	           vp.generic_scissor_bottom}, vp.generic_scissor_window_offset_enable);
 
-	ScissorRect window {vp.window_scissor_left, vp.window_scissor_top, vp.window_scissor_right,
-	                    vp.window_scissor_bottom};
-	if (ScissorRectSet(window)) {
-		if (vp.window_scissor_window_offset_enable) {
-			window = ScissorRectOffset(window, vp.window_offset_x, vp.window_offset_y);
-		}
-		final = ScissorRectIntersect(final, window);
-	}
-
-	ScissorRect generic {vp.generic_scissor_left, vp.generic_scissor_top, vp.generic_scissor_right,
-	                     vp.generic_scissor_bottom};
-	if (ScissorRectSet(generic)) {
-		if (vp.generic_scissor_window_offset_enable) {
-			generic = ScissorRectOffset(generic, vp.window_offset_x, vp.window_offset_y);
-		}
-		final = ScissorRectIntersect(final, generic);
-	}
-
-	const auto& viewport = vp.viewports[0];
-	ScissorRect viewport_scissor {viewport.viewport_scissor_left, viewport.viewport_scissor_top,
-	                              viewport.viewport_scissor_right,
-	                              viewport.viewport_scissor_bottom};
-	if (smc.vport_scissor_enable && ScissorRectSet(viewport_scissor)) {
-		if (viewport.viewport_scissor_window_offset_enable) {
-			viewport_scissor =
-			    ScissorRectOffset(viewport_scissor, vp.window_offset_x, vp.window_offset_y);
-		}
-		final = ScissorRectIntersect(final, viewport_scissor);
+	const auto& viewport = vp.viewports[viewport_index];
+	if (smc.vport_scissor_enable) {
+		intersect({viewport.viewport_scissor_left, viewport.viewport_scissor_top,
+		           viewport.viewport_scissor_right, viewport.viewport_scissor_bottom},
+		          viewport.viewport_scissor_window_offset_enable);
 	}
 
 	if (vp.clip_rect_rule == 0) {
@@ -1138,12 +886,8 @@ ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeC
 					continue;
 				}
 
-				ScissorRect clip {vp.clip_rect_left[i], vp.clip_rect_top[i], vp.clip_rect_right[i],
-				                  vp.clip_rect_bottom[i]};
-				if (vp.clip_rect_window_offset_enable[i]) {
-					clip = ScissorRectOffset(clip, vp.window_offset_x, vp.window_offset_y);
-				}
-				final = ScissorRectIntersect(final, clip);
+				intersect({vp.clip_rect_left[i], vp.clip_rect_top[i], vp.clip_rect_right[i],
+				           vp.clip_rect_bottom[i]}, vp.clip_rect_window_offset_enable[i]);
 			}
 		} else {
 			static std::atomic<uint32_t> log_count {0};
@@ -1157,14 +901,13 @@ ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeC
 	return ScissorRectClamp(final, extent.width, extent.height);
 }
 
-void hw_check(const RenderCommandBuffer& buffer) {
+void hw_check(const CommandBuffer& buffer) {
 	const auto& hw      = buffer.GetRegisters();
 	const auto  rt_slot = render_target_first_bound_slot(buffer);
 	const auto& rt      = hw.GetRenderTarget(rt_slot);
 	const auto& bc      = hw.GetBlendControl(rt_slot);
 	const auto& bclr    = hw.GetBlendColor();
 	const auto& vp      = hw.GetScreenViewport();
-	const auto& z       = hw.GetDepthRenderTarget();
 	const auto& c       = hw.GetClipControl();
 	const auto& rc      = hw.GetRenderControl();
 	const auto& d       = hw.GetDepthControl();
@@ -1190,8 +933,6 @@ void hw_check(const RenderCommandBuffer& buffer) {
 	RtCheck(rt);
 	log_phase("vp");
 	VpCheck(vp, smc);
-	log_phase("z");
-	ZCheck(z, d, rc);
 	log_phase("clip");
 	ClipCheck(c);
 	log_phase("rc");
@@ -1203,12 +944,12 @@ void hw_check(const RenderCommandBuffer& buffer) {
 	log_phase("blend");
 	BcCheck(bc, bclr, cc);
 	log_phase("eqaa");
-	EqaaCheck(eqaa);
+	EqaaCheck(eqaa, ac);
 	log_phase("aa");
 	AaCheck(aa, ac);
 	log_phase("done");
 
-	if (RenderTargetMaskHasBoundMrt(buffer)) {
+	if (graphics_debug_dump_enabled() && RenderTargetMaskHasBoundMrt(buffer)) {
 		LOGF("MRT render target mask: 0x%08" PRIx32 "\n", hw.GetRenderTargetMask());
 		for (uint32_t i = 0; i < 8; i++) {
 			const auto& mrt = hw.GetRenderTarget(i);
@@ -1230,7 +971,7 @@ void hw_check(const RenderCommandBuffer& buffer) {
 	// EXIT_NOT_IMPLEMENTED(hw.GetStencilClearValue() != 0);
 }
 
-void hw_print(const RenderCommandBuffer& buffer) {
+void hw_print(const CommandBuffer& buffer) {
 	const auto& hw      = buffer.GetRegisters();
 	const auto  rt_slot = render_target_first_bound_slot(buffer);
 	const auto& rt      = hw.GetRenderTarget(rt_slot);

src/graphics/host_gpu/renderer/debug.h

@@ -9,7 +9,7 @@
 
 namespace Libs::Graphics {
 
-class RenderCommandBuffer;
+class CommandBuffer;
 
 namespace HW {
 class Context;
@@ -28,18 +28,18 @@ struct ScissorRect {
 };
 
 uint32_t                 render_target_mask_slot(uint32_t mask, uint32_t slot);
-uint32_t                 render_target_first_bound_slot(const RenderCommandBuffer& buffer);
+uint32_t                 render_target_first_bound_slot(const CommandBuffer& buffer);
 bool                     graphics_debug_dump_enabled();
 void                     uc_print(const char* func, const HW::UserConfig& uc);
 void                     uc_check(const HW::UserConfig& uc);
 void                     sh_print(const char* func, const HW::Shader& uc);
 std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt);
 bool                     RenderIsColorTileModeLinear(Prospero::TileMode tile_mode);
-void                     hw_print(const RenderCommandBuffer& buffer);
-void                     hw_check(const RenderCommandBuffer& buffer);
+void                     hw_print(const CommandBuffer& buffer);
+void                     hw_check(const CommandBuffer& buffer);
 void                     LogDrawPhase(const char* draw_name, const char* phase);
 ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc,
-                               vk::Extent2D extent);
+                               vk::Extent2D extent, uint32_t viewport_index);
 
 } // namespace Libs::Graphics
 

src/graphics/host_gpu/renderer/depthRenderTarget.cpp

@@ -13,7 +13,6 @@
 #include "graphics/host_gpu/renderer/debug.h"
 #include "graphics/host_gpu/renderer/image/imageView.h"
 #include "graphics/host_gpu/renderer/image/textureCommon.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vulkanCommon.h"
@@ -71,8 +70,8 @@ static bool UsesStencilOpValue(uint8_t fail, uint8_t pass, uint8_t depth_fail) {
 	return fail == replace_op || pass == replace_op || depth_fail == replace_op;
 }
 
-[[nodiscard]] static vk::Format ResolveHostDepthAttachmentFormat(const RenderCommandBuffer& buffer,
-                                                                 const DepthFormatPolicy&   policy,
+[[nodiscard]] static vk::Format ResolveHostDepthAttachmentFormat(const CommandBuffer&     buffer,
+                                                                 const DepthFormatPolicy& policy,
                                                                  bool     has_stencil,
                                                                  uint32_t samples) {
 	auto&      graphics         = buffer.GetGraphics();
@@ -98,7 +97,7 @@ static bool UsesStencilOpValue(uint8_t fail, uint8_t pass, uint8_t depth_fail) {
 }
 
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
-void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandBuffer& buffer,
+void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer& buffer,
                                               RenderDepthInfo& r) {
 	KYTY_PROFILER_FUNCTION();
 	(void)submit_id;
@@ -109,9 +108,10 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	const auto& sc          = hw.GetStencilControl();
 	const auto& sm          = hw.GetStencilMask();
 	const bool  has_stencil = z.stencil_info.format != Prospero::StencilFormat::kInvalid;
-	const bool  depth_active =
-	    dc.z_enable || dc.z_write_enable || dc.depth_bounds_enable || rc.depth_clear_enable;
-	const bool stencil_active = has_stencil && (dc.stencil_enable || rc.stencil_clear_enable);
+	const bool depth_active = dc.z_enable || dc.depth_bounds_enable ||
+	                          rc.depth_clear_enable || rc.copy_depth_to_color;
+	const bool stencil_active =
+	    has_stencil && (dc.stencil_enable || rc.stencil_clear_enable || rc.copy_stencil_to_color);
 	if (!depth_active && !stencil_active) {
 		return;
 	}
@@ -128,22 +128,14 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	    !z.stencil_info.expclear_enabled && !z.stencil_info.partially_resident &&
 	    z.depth_view.slice_start == 0 && z.depth_view.slice_max == 0 &&
 	    z.depth_view.current_mip_level == 0 && !z.depth_view.depth_write_disable &&
-	    !z.depth_view.stencil_write_disable && z.depth_info.addr5_swizzle_mask == 0 &&
-	    z.depth_info.array_mode == 0 && z.depth_info.pipe_config == 0 &&
-	    z.depth_info.bank_width == 0 && z.depth_info.bank_height == 0 &&
-	    z.depth_info.macro_tile_aspect == 0 && z.depth_info.num_banks == 0 &&
-	    z.htile_surface.linear == 0 && z.htile_surface.full_cache == 0 &&
-	    z.htile_surface.htile_uses_preload_win == 0 && z.htile_surface.preload == 0 &&
-	    z.htile_surface.prefetch_width == 0 && z.htile_surface.prefetch_height == 0 &&
-	    z.htile_surface.dst_outside_zero_to_one == 0 && z.z_read_base_addr == 0 &&
-	    z.z_write_base_addr == 0 && z.stencil_read_base_addr == 0 &&
-	    z.stencil_write_base_addr == 0 && z.htile_data_base_addr == 0 &&
+	    !z.depth_view.stencil_write_disable && z.z_read_base_addr == 0 && z.z_write_base_addr == 0 &&
+	    z.stencil_read_base_addr == 0 && z.stencil_write_base_addr == 0 &&
+	    z.htile_data_base_addr == 0 &&
 	    // DB_DEPTH_SIZE_XY is independent state and may remain programmed after the attachment
 	    // formats and addresses are unbound. A zero encoding is the valid 1x1 value, so its
 	    // presence alone must not manufacture a depth attachment.
-	    !z.z_info.htile_acceleration && !z.width_height_valid && !z.pitch_height_valid &&
-	    z.size.x_max == 0 && z.size.y_max == 0 && z.pitch_div8_minus1 == 0 &&
-	    z.height_div8_minus1 == 0 && z.slice_div64_minus1 == 0 && z.width == 0 && z.height == 0;
+	    !z.z_info.htile_acceleration && z.shading_rate_encoding == 0 && z.size.x_max == 0 &&
+	    z.size.y_max == 0;
 	if (attachment_unbound) {
 		static std::atomic_bool logged = false;
 		if (!logged.exchange(true, std::memory_order_relaxed)) {
@@ -152,6 +144,8 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 		return;
 	}
 	const bool has_htile = z.z_info.htile_acceleration;
+	const bool unsupported_shading_rate_encoding =
+	    z.shading_rate_encoding > 1 || (z.shading_rate_encoding != 0 && !has_htile);
 	const auto samples   = render_sample_count(z.z_info.num_samples);
 	if (samples == 0) {
 		DepthFatal("unsupported depth fragment count: %u", z.z_info.num_samples);
@@ -166,32 +160,26 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 			DepthFatal("invalid depth view: base=%u last=%u", z.depth_view.slice_start,
 			           z.depth_view.slice_max);
 	}
-	if (rc.copy_centroid || rc.copy_sample != 0 || z.z_info.expclear_enabled ||
+	if (rc.copy_depth_to_color || rc.copy_stencil_to_color || rc.copy_centroid ||
+	    rc.copy_sample != 0 || z.z_info.expclear_enabled ||
 	    z.stencil_info.expclear_enabled || z.z_info.partially_resident ||
 	    z.stencil_info.partially_resident || z.z_info.max_mip_level != 0 ||
-	    z.depth_view.current_mip_level != 0 || z.depth_info.addr5_swizzle_mask != 0 ||
-	    z.depth_info.array_mode != 0 || z.depth_info.pipe_config != 0 ||
-	    z.depth_info.bank_width != 0 || z.depth_info.bank_height != 0 ||
-	    z.depth_info.macro_tile_aspect != 0 || z.depth_info.num_banks != 0 ||
-	    z.htile_surface.linear != 0 || z.htile_surface.full_cache != 0 ||
-	    z.htile_surface.htile_uses_preload_win != 0 || z.htile_surface.preload != 0 ||
-	    z.htile_surface.prefetch_width != 0 || z.htile_surface.prefetch_height != 0 ||
-	    z.htile_surface.dst_outside_zero_to_one != 0 || z.z_read_base_addr == 0 ||
-	    z.z_write_base_addr != z.z_read_base_addr || (z.z_read_base_addr & 0xffffu) != 0 ||
+	    z.depth_view.current_mip_level != 0 || unsupported_shading_rate_encoding ||
+	    z.z_read_base_addr == 0 ||
+	    (!z.depth_view.depth_write_disable && z.z_write_base_addr != z.z_read_base_addr) ||
+	    (z.z_read_base_addr & 0xffffu) != 0 ||
 	    dc.zfunc > static_cast<uint8_t>(vk::CompareOp::eAlways)) {
 		DepthFatal("unsupported depth register state");
 	}
 	if (has_stencil) {
 		if (z.stencil_info.format != Prospero::StencilFormat::k8UInt || !htile_stencil_compat ||
 		    z.stencil_read_base_addr == 0 ||
-		    z.stencil_write_base_addr != z.stencil_read_base_addr ||
-		    (z.stencil_read_base_addr & 0xffffu) != 0 || z.depth_view.stencil_write_disable) {
+		    (!z.depth_view.stencil_write_disable &&
+		     z.stencil_write_base_addr != z.stencil_read_base_addr) ||
+		    (z.stencil_read_base_addr & 0xffffu) != 0) {
 			DepthFatal("unsupported stencil attachment state");
 		}
-	} else if (z.stencil_read_base_addr != 0 || z.stencil_write_base_addr != 0 ||
-	           !htile_stencil_compat ||
-	           z.stencil_info.texture_compatibility !=
-	               Prospero::TextureCompatibleStencil::kDisable) {
+	} else if (z.stencil_read_base_addr != 0 || z.stencil_write_base_addr != 0) {
 		DepthFatal("stencil state without an active stencil attachment");
 	}
 	if (has_htile) {
@@ -202,37 +190,27 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 			DepthFatal("HTile clear tracking supports at most 32 slices");
 		}
 	}
-	const bool size_xy_valid = z.size.valid;
-	const bool wh_valid      = z.width_height_valid && z.width != 0 && z.height != 0;
-	if (!size_xy_valid && !wh_valid) {
+	if (!z.size.valid) {
 		DepthFatal("missing depth extent");
 	}
-	const uint32_t width  = size_xy_valid ? static_cast<uint32_t>(z.size.x_max) + 1u : z.width;
-	const uint32_t height = size_xy_valid ? static_cast<uint32_t>(z.size.y_max) + 1u : z.height;
-	if (width > 16384 || height > 16384 ||
-	    (size_xy_valid && wh_valid && (width != z.width || height != z.height)) ||
-	    (!z.pitch_height_valid &&
-	     (z.pitch_div8_minus1 != 0 || z.height_div8_minus1 != 0 || z.slice_div64_minus1 != 0))) {
-		DepthFatal("inconsistent depth extent or encoded layout");
+	const uint32_t width  = static_cast<uint32_t>(z.size.x_max) + 1u;
+	const uint32_t height = static_cast<uint32_t>(z.size.y_max) + 1u;
+	if (width > 16384 || height > 16384) {
+		DepthFatal("invalid depth extent");
 	}
 	const auto* policy = FindDepthFormatPolicy(z.z_info.format);
 	if (policy == nullptr) {
 		DepthFatal("unsupported depth/stencil format pair");
 	}
 	const auto ideal_format = DepthAttachmentFormat(*policy, has_stencil);
-	r.format = ResolveHostDepthAttachmentFormat(buffer, *policy, has_stencil, samples);
-	if (r.format == vk::Format::eUndefined) {
+	const auto format = ResolveHostDepthAttachmentFormat(buffer, *policy, has_stencil, samples);
+	if (format == vk::Format::eUndefined) {
 		DepthFatal("no host depth/stencil format supports required usage for %s",
-		           VulkanToString(ideal_format).c_str());
+		           vk::to_string(ideal_format).c_str());
 	}
 	const auto     guest_format = policy->guest_format;
 	const uint32_t bytes        = policy->bytes_per_element;
 	const auto     pitch        = TileGetDepthPitch(width, bytes, z.z_info.num_samples);
-	if (z.pitch_height_valid && ((static_cast<uint64_t>(z.pitch_div8_minus1) + 1u) * 8u != pitch ||
-	                             (static_cast<uint64_t>(z.height_div8_minus1) + 1u) * 8u !=
-	                                 ((static_cast<uint64_t>(height) + 7u) & ~7ull))) {
-		DepthFatal("encoded depth pitch or height mismatch");
-	}
 	TileSizeAlign depth_size {};
 	TileSizeAlign stencil_size {};
 	TileSizeAlign htile_size {};
@@ -243,14 +221,6 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	    (has_htile != (htile_size.align == 32768 && htile_size.size != 0))) {
 		DepthFatal("unsupported depth/stencil/HTile footprint");
 	}
-	if (z.pitch_height_valid &&
-	    (static_cast<uint64_t>(z.slice_div64_minus1) + 1u) * 64u != depth_size.size) {
-		DepthFatal("depth footprint mismatch: extent=%ux%u pitch=%u expected=0x%016" PRIx64
-		           " align=0x%016" PRIx64 " encoded_valid=%u encoded=0x%016" PRIx64,
-		           width, height, pitch, depth_size.size, depth_size.align,
-		           z.pitch_height_valid ? 1u : 0u,
-		           (static_cast<uint64_t>(z.slice_div64_minus1) + 1u) * 64u);
-	}
 	if (depth_size.size > UINT64_MAX / view.image_layers ||
 	    stencil_size.size > UINT64_MAX / view.image_layers ||
 	    htile_size.size > UINT64_MAX / view.image_layers) {
@@ -259,45 +229,40 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	const auto depth_backing_size   = depth_size.size * view.image_layers;
 	const auto stencil_backing_size = stencil_size.size * view.image_layers;
 	const auto htile_backing_size   = htile_size.size * view.image_layers;
-	if (depth_backing_size > TRACKER_ADDRESS_SIZE - z.z_read_base_addr ||
-	    (has_stencil && stencil_backing_size > TRACKER_ADDRESS_SIZE - z.stencil_read_base_addr) ||
-	    (has_htile && htile_backing_size > TRACKER_ADDRESS_SIZE - z.htile_data_base_addr)) {
+	if (!GuestRange {z.z_read_base_addr, depth_backing_size}.Valid() ||
+	    (has_stencil && !GuestRange {z.stencil_read_base_addr, stencil_backing_size}.Valid()) ||
+	    (has_htile && !GuestRange {z.htile_data_base_addr, htile_backing_size}.Valid())) {
 		DepthFatal("layered depth backing range is invalid");
 	}
-	r.htile                   = has_htile;
-	r.width                   = width;
-	r.height                  = height;
-	r.samples                 = samples;
-	r.depth_buffer_size       = depth_backing_size;
-	r.depth_buffer_vaddr      = z.z_read_base_addr;
-	r.stencil_buffer_size     = has_stencil ? stencil_backing_size : 0;
-	r.stencil_buffer_vaddr    = has_stencil ? z.stencil_read_base_addr : 0;
-	r.htile_buffer_size       = has_htile ? htile_backing_size : 0;
-	r.htile_buffer_vaddr      = has_htile ? z.htile_data_base_addr : 0;
 	r.depth_clear_enable      = rc.depth_clear_enable;
 	r.depth_meta_clear_enable = false;
 	r.depth_load_clear_enable = r.depth_clear_enable;
 	r.depth_clear_value       = hw.GetDepthClearValue();
 	r.depth_test_enable       = dc.z_enable;
-	r.depth_write_enable      = dc.z_write_enable && !z.depth_view.depth_write_disable;
+	r.depth_write_enable      = r.depth_test_enable && dc.z_write_enable &&
+	                            !z.depth_view.depth_write_disable && !r.depth_clear_enable;
 	r.depth_compare_op        = static_cast<vk::CompareOp>(dc.zfunc);
 
 	r.depth_bounds_test_enable = dc.depth_bounds_enable;
 	r.depth_min_bounds         = hw.GetDepthBoundsMin();
 	r.depth_max_bounds         = hw.GetDepthBoundsMax();
 
-	r.stencil_clear_enable = has_stencil && rc.stencil_clear_enable;
-	r.stencil_clear_value  = hw.GetStencilClearValue();
-	r.stencil_test_enable  = has_stencil && dc.stencil_enable;
+	r.stencil_clear_enable =
+	    has_stencil && rc.stencil_clear_enable && !z.depth_view.stencil_write_disable;
+	r.stencil_clear_value = hw.GetStencilClearValue();
+	r.stencil_test_enable = has_stencil && dc.stencil_enable;
 	if (r.stencil_test_enable) {
-		const uint8_t front_write_mask = rc.stencil_clear_enable ? 0 : sm.stencil_writemask;
-		const uint8_t back_write_mask  = rc.stencil_clear_enable ? 0 : sm.stencil_writemask_bf;
+		const bool stencil_ops_disabled =
+		    rc.stencil_clear_enable || z.depth_view.stencil_write_disable;
+		const uint8_t front_write_mask = stencil_ops_disabled ? 0 : sm.stencil_writemask;
+		const uint8_t back_write_mask  = stencil_ops_disabled ? 0 : sm.stencil_writemask_bf;
 		if (dc.stencilfunc > static_cast<uint8_t>(vk::CompareOp::eAlways) ||
 		    (dc.backface_enable &&
 		     dc.stencilfunc_bf > static_cast<uint8_t>(vk::CompareOp::eAlways)) ||
-		    (UsesStencilOpValue(sc.stencil_fail, sc.stencil_zpass, sc.stencil_zfail) &&
+		    (front_write_mask != 0 &&
+		     UsesStencilOpValue(sc.stencil_fail, sc.stencil_zpass, sc.stencil_zfail) &&
 		     sm.stencil_opval != sm.stencil_testval) ||
-		    (dc.backface_enable &&
+		    (dc.backface_enable && back_write_mask != 0 &&
 		     UsesStencilOpValue(sc.stencil_fail_bf, sc.stencil_zpass_bf, sc.stencil_zfail_bf) &&
 		     sm.stencil_opval_bf != sm.stencil_testval_bf)) {
 			DepthFatal("unsupported stencil compare or replacement state");
@@ -320,18 +285,12 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 			r.stencil_dynamic_back = r.stencil_dynamic_front;
 		}
 	}
-	r.vaddr_num = has_stencil ? 2 : 1;
-	r.vaddr[0]  = r.depth_buffer_vaddr;
-	r.size[0]   = r.depth_buffer_size;
-	if (has_stencil) {
-		r.vaddr[1] = r.stencil_buffer_vaddr;
-		r.size[1]  = r.stencil_buffer_size;
-	}
 	TextureCache::ImageDesc desc {};
 	desc.type                 = TextureCache::BindingType::DepthTarget;
-	desc.info.data            = {r.depth_buffer_vaddr, r.depth_buffer_size};
-	desc.info.stencil         = {r.stencil_buffer_vaddr, r.stencil_buffer_size};
-	desc.info.pixel_format    = r.format;
+	desc.info.data            = {z.z_read_base_addr, depth_backing_size};
+	desc.info.stencil =
+	    has_stencil ? GuestRange {z.stencil_read_base_addr, stencil_backing_size} : GuestRange {};
+	desc.info.pixel_format    = format;
 	desc.info.guest_format    = guest_format;
 	desc.info.type            = Prospero::ImageType::kColor2D;
 	desc.info.extent          = {width, height, 1};
@@ -340,15 +299,16 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	desc.info.bytes_per_block = bytes;
 	desc.info.samples         = samples;
 	desc.info.tile_mode       = Prospero::TileMode::kDepth;
-	desc.info.mip_layout[0]   = {0, r.depth_buffer_size, pitch, height};
-	desc.info.metadata.range  = {r.htile_buffer_vaddr, r.htile_buffer_size};
+	desc.info.mip_layout[0]   = {0, depth_backing_size, pitch, height};
+	desc.info.metadata.range =
+	    has_htile ? GuestRange {z.htile_data_base_addr, htile_backing_size} : GuestRange {};
 	desc.info.metadata.kind   = has_htile ? ImageMetadataKind::Htile : ImageMetadataKind::None;
 	desc.info.metadata.stencil_compressed =
 	    has_stencil && has_htile && !z.stencil_info.htile_stencil_disabled;
-	desc.view_info.format = r.format;
+	desc.view_info.format = format;
 	desc.view_info.type =
 	    view.layer_count == 1 ? vk::ImageViewType::e2D : vk::ImageViewType::e2DArray;
-	desc.view_info.aspect      = ImageViewOps::DepthAspectMask(r.format);
+	desc.view_info.aspect      = ImageViewOps::DepthAspectMask(format);
 	desc.view_info.base_level  = 0;
 	desc.view_info.level_count = 1;
 	desc.view_info.base_layer  = view.base_layer;
@@ -357,11 +317,11 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	r.desc                     = std::move(desc);
 	auto& cache                = m_context.GetTextureCache();
 	r.image_id                 = cache.FindImage(r.desc);
-	r.image_view               = nullptr;
 	BindRenderTarget(r.image_id);
 }
 
 vk::ImageAspectFlags RenderDepthInfo::AttachmentWriteAspects() const {
+	const auto format = desc.view_info.format;
 	if (format == vk::Format::eUndefined) {
 		return {};
 	}
@@ -369,7 +329,7 @@ vk::ImageAspectFlags RenderDepthInfo::AttachmentWriteAspects() const {
 	const auto           available = ImageViewOps::DepthAspectMask(format);
 	vk::ImageAspectFlags writes {};
 	if ((available & vk::ImageAspectFlagBits::eDepth) &&
-	    (depth_load_clear_enable || (depth_test_enable && depth_write_enable))) {
+	    (depth_load_clear_enable || depth_write_enable)) {
 		writes |= vk::ImageAspectFlagBits::eDepth;
 	}
 	if (!(available & vk::ImageAspectFlagBits::eStencil)) {

src/graphics/host_gpu/renderer/depthRenderTarget.h

@@ -11,34 +11,22 @@
 
 namespace Libs::Graphics {
 
-class RenderCommandBuffer;
-
+// Without a stencil plane, Hi-Stencil fields are inactive. An active plane is compatible when
+// Hi-Stencil is disabled or HTile backing is present.
 inline constexpr bool depth_htile_stencil_acceleration_compatible(bool has_stencil, bool has_htile,
                                                                   bool htile_stencil_disabled) {
-	return htile_stencil_disabled || (has_stencil && has_htile);
+	return !has_stencil || htile_stencil_disabled || has_htile;
 }
 
 struct RenderDepthInfo {
+	// Discovery keeps guest image information but can remap the view into a larger cache image.
 	TextureCache::ImageDesc     desc;
-	vk::Format                  format                   = vk::Format::eUndefined;
-	uint32_t                    width                    = 0;
-	uint32_t                    height                   = 0;
-	uint32_t                    samples                  = 1;
-	bool                        htile                    = false;
-	uint64_t                    depth_buffer_size        = 0;
-	uint64_t                    depth_buffer_vaddr       = 0;
-	uint64_t                    depth_tile_swizzle       = 0;
-	uint64_t                    stencil_buffer_size      = 0;
-	uint64_t                    stencil_buffer_vaddr     = 0;
-	uint64_t                    stencil_tile_swizzle     = 0;
-	uint64_t                    htile_buffer_size        = 0;
-	uint64_t                    htile_buffer_vaddr       = 0;
-	uint64_t                    htile_tile_swizzle       = 0;
 	bool                        depth_clear_enable       = false;
 	bool                        depth_load_clear_enable  = false;
 	bool                        depth_meta_clear_enable  = false;
 	float                       depth_clear_value        = 0.0f;
 	bool                        depth_test_enable        = false;
+	// Effective draw writes; discovery applies test, target-write and clear controls.
 	bool                        depth_write_enable       = false;
 	vk::CompareOp               depth_compare_op         = vk::CompareOp::eNever;
 	bool                        depth_bounds_test_enable = false;
@@ -52,24 +40,17 @@ struct RenderDepthInfo {
 	PipelineStencilDynamicState stencil_dynamic_front;
 	PipelineStencilDynamicState stencil_dynamic_back;
 	ImageId                     image_id;
-	vk::ImageView               image_view = nullptr;
-	uint64_t                    vaddr[3]   = {};
-	uint64_t                    size[3]    = {};
-	int                         vaddr_num  = 0;
 
 	[[nodiscard]] vk::ImageAspectFlags AttachmentWriteAspects() const;
 };
 
-inline bool depth_attachment_read_only(const RenderDepthInfo& depth) {
-	return !depth.AttachmentWriteAspects();
-}
-
 inline vk::ImageLayout depth_attachment_layout(const RenderDepthInfo& depth) {
-	const auto available     = ImageViewOps::DepthAspectMask(depth.format);
+	const auto available     = ImageViewOps::DepthAspectMask(depth.desc.view_info.format);
 	const auto writes        = depth.AttachmentWriteAspects();
 	const bool has_depth     = static_cast<bool>(available & vk::ImageAspectFlagBits::eDepth);
 	const bool has_stencil   = static_cast<bool>(available & vk::ImageAspectFlagBits::eStencil);
-	const bool depth_write   = static_cast<bool>(writes & vk::ImageAspectFlagBits::eDepth);
+	// LoadOp clears are separate from guest depth writes.
+	const bool depth_write   = depth.depth_write_enable;
 	const bool stencil_write = static_cast<bool>(writes & vk::ImageAspectFlagBits::eStencil);
 	if (!has_stencil) {
 		return depth_write ? vk::ImageLayout::eDepthAttachmentOptimal

src/graphics/host_gpu/renderer/image/blitHelper.cpp

@@ -23,7 +23,6 @@ BlitHelper::BlitHelper(GraphicContext& graphics, CommandScheduler& scheduler)
 	texture_binding.stageFlags      = vk::ShaderStageFlagBits::eFragment;
 
 	vk::DescriptorSetLayoutCreateInfo descriptor_info {};
-	descriptor_info.sType        = vk::StructureType::eDescriptorSetLayoutCreateInfo;
 	descriptor_info.flags        = vk::DescriptorSetLayoutCreateFlagBits::ePushDescriptorKHR;
 	descriptor_info.bindingCount = 1;
 	descriptor_info.pBindings    = &texture_binding;
@@ -32,7 +31,6 @@ BlitHelper::BlitHelper(GraphicContext& graphics, CommandScheduler& scheduler)
 	                     "create BlitHelper descriptor layout");
 
 	vk::PipelineLayoutCreateInfo layout_info {};
-	layout_info.sType          = vk::StructureType::ePipelineLayoutCreateInfo;
 	layout_info.setLayoutCount = 1;
 	layout_info.pSetLayouts    = &m_descriptor_layout;
 	RequireVulkanSuccess(
@@ -65,7 +63,6 @@ BlitHelper::~BlitHelper() {
 vk::ShaderModule BlitHelper::CreateShader(const uint32_t* code, size_t words) const {
 	EXIT_IF(code == nullptr || words == 0);
 	vk::ShaderModuleCreateInfo create {};
-	create.sType            = vk::StructureType::eShaderModuleCreateInfo;
 	create.codeSize         = words * sizeof(uint32_t);
 	create.pCode            = code;
 	vk::ShaderModule module = nullptr;
@@ -84,51 +81,37 @@ vk::Pipeline BlitHelper::GetPipeline(PipelineKey key) {
 	EXIT_IF(samples == vk::SampleCountFlagBits {} || key.format == vk::Format::eUndefined);
 
 	std::array<vk::PipelineShaderStageCreateInfo, 2> stages {};
-	stages[0].sType  = vk::StructureType::ePipelineShaderStageCreateInfo;
 	stages[0].stage  = vk::ShaderStageFlagBits::eVertex;
 	stages[0].module = m_vertex_shader;
 	stages[0].pName  = "main";
-	stages[1].sType  = vk::StructureType::ePipelineShaderStageCreateInfo;
 	stages[1].stage  = vk::ShaderStageFlagBits::eFragment;
 	stages[1].module = m_fragment_shader;
 	stages[1].pName  = "main";
 
 	vk::PipelineVertexInputStateCreateInfo vertex_input {};
-	vertex_input.sType = vk::StructureType::ePipelineVertexInputStateCreateInfo;
 	vk::PipelineInputAssemblyStateCreateInfo input_assembly {};
-	input_assembly.sType    = vk::StructureType::ePipelineInputAssemblyStateCreateInfo;
 	input_assembly.topology = vk::PrimitiveTopology::eTriangleList;
 	vk::PipelineViewportStateCreateInfo viewport {};
-	viewport.sType         = vk::StructureType::ePipelineViewportStateCreateInfo;
 	viewport.viewportCount = 1;
 	viewport.scissorCount  = 1;
 	vk::PipelineRasterizationStateCreateInfo rasterization {};
-	rasterization.sType       = vk::StructureType::ePipelineRasterizationStateCreateInfo;
-	rasterization.polygonMode = vk::PolygonMode::eFill;
-	rasterization.cullMode    = vk::CullModeFlagBits::eNone;
-	rasterization.lineWidth   = 1.0f;
+	rasterization.lineWidth = 1.0f;
 	vk::PipelineMultisampleStateCreateInfo multisample {};
-	multisample.sType                = vk::StructureType::ePipelineMultisampleStateCreateInfo;
 	multisample.rasterizationSamples = samples;
 	vk::PipelineDepthStencilStateCreateInfo depth {};
-	depth.sType            = vk::StructureType::ePipelineDepthStencilStateCreateInfo;
 	depth.depthTestEnable  = VK_TRUE;
 	depth.depthWriteEnable = VK_TRUE;
 	depth.depthCompareOp   = vk::CompareOp::eAlways;
 	vk::PipelineColorBlendStateCreateInfo color_blend {};
-	color_blend.sType = vk::StructureType::ePipelineColorBlendStateCreateInfo;
 	const std::array dynamic_states {vk::DynamicState::eViewport, vk::DynamicState::eScissor};
 	vk::PipelineDynamicStateCreateInfo dynamic {};
-	dynamic.sType             = vk::StructureType::ePipelineDynamicStateCreateInfo;
 	dynamic.dynamicStateCount = static_cast<uint32_t>(dynamic_states.size());
 	dynamic.pDynamicStates    = dynamic_states.data();
 
 	vk::PipelineRenderingCreateInfo rendering {};
-	rendering.sType                 = vk::StructureType::ePipelineRenderingCreateInfo;
 	rendering.depthAttachmentFormat = key.format;
 
 	vk::GraphicsPipelineCreateInfo create {};
-	create.sType               = vk::StructureType::eGraphicsPipelineCreateInfo;
 	create.pNext               = &rendering;
 	create.stageCount          = static_cast<uint32_t>(stages.size());
 	create.pStages             = stages.data();
@@ -186,7 +169,6 @@ void BlitHelper::ReinterpretColorAsMsDepth(Image& source, Image& destination) {
 	                    command);
 
 	vk::RenderingAttachmentInfo depth_attachment {};
-	depth_attachment.sType                   = vk::StructureType::eRenderingAttachmentInfo;
 	depth_attachment.imageView               = destination_view;
 	depth_attachment.imageLayout             = ColorToMsDepthLayout;
 	depth_attachment.loadOp                  = vk::AttachmentLoadOp::eClear;
@@ -194,7 +176,6 @@ void BlitHelper::ReinterpretColorAsMsDepth(Image& source, Image& destination) {
 	depth_attachment.clearValue.depthStencil = {0.0f, 0};
 
 	vk::RenderingInfo rendering {};
-	rendering.sType             = vk::StructureType::eRenderingInfo;
 	rendering.renderArea.extent = {destination_info.extent.width, destination_info.extent.height};
 	rendering.layerCount        = 1;
 	rendering.pDepthAttachment  = &depth_attachment;
@@ -204,7 +185,6 @@ void BlitHelper::ReinterpretColorAsMsDepth(Image& source, Image& destination) {
 	descriptor_image.imageView   = source_view;
 	descriptor_image.imageLayout = vk::ImageLayout::eShaderReadOnlyOptimal;
 	vk::WriteDescriptorSet descriptor_write {};
-	descriptor_write.sType           = vk::StructureType::eWriteDescriptorSet;
 	descriptor_write.dstBinding      = 0;
 	descriptor_write.descriptorCount = 1;
 	descriptor_write.descriptorType  = vk::DescriptorType::eSampledImage;

src/graphics/host_gpu/renderer/image/image.cpp

@@ -54,29 +54,22 @@ namespace {
 	}
 	if (DepthAspectTransferFormat(info.pixel_format) != vk::Format::eUndefined) {
 		usage |= vk::ImageUsageFlagBits::eDepthStencilAttachment;
+		if (graphics.attachment_feedback_loop_enabled && (usage & vk::ImageUsageFlagBits::eSampled)) {
+			usage |= vk::ImageUsageFlagBits::eAttachmentFeedbackLoopEXT;
+		}
 		return usage;
 	}
 	if (HasFormatFeature(properties, vk::FormatFeatureFlagBits::eColorAttachment)) {
 		usage |= vk::ImageUsageFlagBits::eColorAttachment;
 	}
-	if (info.samples == 1 &&
-	    HasFormatFeature(properties, vk::FormatFeatureFlagBits::eStorageImage)) {
+	if (info.samples == 1) {
 		usage |= vk::ImageUsageFlagBits::eStorage;
-	} else if (info.samples == 1) {
-		const auto compatible = SrgbStorageViewFormat(info.pixel_format);
-		if (compatible != vk::Format::eUndefined &&
-		    HasFormatFeature(graphics.GetFormatProperties(compatible),
-		                     vk::FormatFeatureFlagBits::eStorageImage)) {
-			usage |= vk::ImageUsageFlagBits::eStorage;
-		}
 	}
 	return usage;
 }
 
-void ValidateRange(GuestRange range, const char* name) {
-	if ((range.address == 0) != (range.size == 0) ||
-	    (range.address != 0 && (range.address >= TRACKER_ADDRESS_SIZE ||
-	                            range.size > TRACKER_ADDRESS_SIZE - range.address))) {
+void ValidateOptionalRange(GuestRange range, const char* name) {
+	if (!range.ValidOrEmpty()) {
 		EXIT("invalid %s image range: address=0x%016llx size=0x%016llx\n", name,
 		     static_cast<unsigned long long>(range.address),
 		     static_cast<unsigned long long>(range.size));
@@ -212,9 +205,7 @@ void Image::Transit(vk::ImageLayout destination_layout, vk::AccessFlags2 destina
 	if (barriers.empty()) {
 		return;
 	}
-	if (m_scheduler != nullptr) {
-		m_scheduler->EndRendering();
-	}
+	m_scheduler.EndRendering();
 	vk::DependencyInfo dependency {};
 	dependency.imageMemoryBarrierCount = static_cast<uint32_t>(barriers.size());
 	dependency.pImageMemoryBarriers    = barriers.data();
@@ -223,8 +214,8 @@ void Image::Transit(vk::ImageLayout destination_layout, vk::AccessFlags2 destina
 
 void Image::Upload(std::span<const vk::BufferImageCopy> copies, vk::Buffer buffer, uint64_t offset,
                    uint64_t size) {
-	EXIT_IF(m_scheduler == nullptr || copies.empty() || buffer == nullptr || size == 0);
-	m_scheduler->EndRendering();
+	EXIT_IF(copies.empty() || buffer == nullptr || size == 0);
+	m_scheduler.EndRendering();
 	vk::BufferMemoryBarrier2 buffer_barrier {};
 	buffer_barrier.srcStageMask        = vk::PipelineStageFlagBits2::eAllCommands;
 	buffer_barrier.srcAccessMask       = vk::AccessFlagBits2::eMemoryWrite;
@@ -244,7 +235,7 @@ void Image::Upload(std::span<const vk::BufferImageCopy> copies, vk::Buffer buffe
 	dependency.pBufferMemoryBarriers    = &buffer_barrier;
 	dependency.imageMemoryBarrierCount  = static_cast<uint32_t>(image_barriers.size());
 	dependency.pImageMemoryBarriers     = image_barriers.data();
-	auto command                        = m_scheduler->Current().Handle();
+	auto command                        = m_scheduler.Current().Handle();
 	command.pipelineBarrier2(dependency);
 	command.copyBufferToImage(buffer, backing.image, vk::ImageLayout::eTransferDstOptimal,
 	                          static_cast<uint32_t>(copies.size()), copies.data());
@@ -262,8 +253,8 @@ void Image::Upload(std::span<const vk::BufferImageCopy> copies, vk::Buffer buffe
 
 void Image::Download(std::span<const vk::BufferImageCopy> copies, vk::Buffer buffer,
                      uint64_t offset, uint64_t size) {
-	EXIT_IF(m_scheduler == nullptr || copies.empty() || buffer == nullptr || size == 0);
-	m_scheduler->EndRendering();
+	EXIT_IF(copies.empty() || buffer == nullptr || size == 0);
+	m_scheduler.EndRendering();
 	vk::BufferMemoryBarrier2 buffer_barrier {};
 	buffer_barrier.srcStageMask = vk::PipelineStageFlagBits2::eAllCommands;
 	buffer_barrier.srcAccessMask =
@@ -284,7 +275,7 @@ void Image::Download(std::span<const vk::BufferImageCopy> copies, vk::Buffer buf
 	dependency.pBufferMemoryBarriers    = &buffer_barrier;
 	dependency.imageMemoryBarrierCount  = static_cast<uint32_t>(image_barriers.size());
 	dependency.pImageMemoryBarriers     = image_barriers.data();
-	auto command                        = m_scheduler->Current().Handle();
+	auto command                        = m_scheduler.Current().Handle();
 	command.pipelineBarrier2(dependency);
 	command.copyImageToBuffer(backing.image, vk::ImageLayout::eTransferSrcOptimal, buffer,
 	                          static_cast<uint32_t>(copies.size()), copies.data());
@@ -321,8 +312,8 @@ std::pair<uint32_t, uint32_t> Image::SanitizeCopyLayers(const Image& source,
 }
 
 void Image::CopyImage(Image& source) {
-	EXIT_IF(m_scheduler == nullptr || source.backing.samples != backing.samples);
-	m_scheduler->EndRendering();
+	EXIT_IF(source.backing.samples != backing.samples);
+	m_scheduler.EndRendering();
 	const uint32_t levels     = std::min(source.backing.mip_levels, backing.mip_levels);
 	const uint32_t base_depth = backing.image_type == vk::ImageType::e3D
 	                                ? backing.extent.depth
@@ -361,7 +352,7 @@ void Image::CopyImage(Image& source) {
 	if (copies.empty()) {
 		return;
 	}
-	auto command = m_scheduler->Current().Handle();
+	auto command = m_scheduler.Current().Handle();
 	source.Transit(vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead, {},
 	               command);
 	Transit(vk::ImageLayout::eTransferDstOptimal, vk::AccessFlagBits2::eTransferWrite, {}, command);
@@ -374,8 +365,7 @@ void Image::CopyImage(Image& source) {
 
 void Image::Resolve(Image& source, const ImageSubresourceRange& source_range,
                     const ImageSubresourceRange& destination_range) {
-	EXIT_IF(m_scheduler == nullptr || backing.samples != 1 ||
-	        source.backing.image_type != vk::ImageType::e2D ||
+	EXIT_IF(backing.samples != 1 || source.backing.image_type != vk::ImageType::e2D ||
 	        backing.image_type != vk::ImageType::e2D || source_range.level_count != 1 ||
 	        destination_range.level_count != 1 ||
 	        source_range.base_level >= source.backing.mip_levels ||
@@ -403,8 +393,8 @@ void Image::Resolve(Image& source, const ImageSubresourceRange& source_range,
 	resolved_destination_range.layer_count = layers;
 	const vk::Extent3D resolve_extent {info.extent.width, info.extent.height, 1};
 
-	m_scheduler->EndRendering();
-	auto command = m_scheduler->Current().Handle();
+	m_scheduler.EndRendering();
+	auto command = m_scheduler.Current().Handle();
 	source.Transit(vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead,
 	               resolved_source_range, command);
 	Transit(vk::ImageLayout::eTransferDstOptimal, vk::AccessFlagBits2::eTransferWrite,
@@ -440,9 +430,8 @@ uint32_t Image::CopyRows(uint64_t row_size, uint32_t rows, uint64_t capacity) no
 }
 
 void Image::CopyImageWithBuffer(Image& source, Buffer& buffer) {
-	EXIT_IF(m_scheduler == nullptr || buffer.Handle() == nullptr || source.backing.samples != 1 ||
-	        backing.samples != 1);
-	m_scheduler->EndRendering();
+	EXIT_IF(buffer.Handle() == nullptr || source.backing.samples != 1 || backing.samples != 1);
+	m_scheduler.EndRendering();
 	const uint32_t levels = std::min(source.backing.mip_levels, backing.mip_levels);
 	const auto     source_aspect =
 	    FullAspectMask(source.backing.format) & ~vk::ImageAspectFlagBits::eStencil;
@@ -472,7 +461,7 @@ void Image::CopyImageWithBuffer(Image& source, Buffer& buffer) {
 	dependency.dependencyFlags          = vk::DependencyFlagBits::eByRegion;
 	dependency.bufferMemoryBarrierCount = 1;
 	dependency.pBufferMemoryBarriers    = &barrier;
-	auto command                        = m_scheduler->Current().Handle();
+	auto command                        = m_scheduler.Current().Handle();
 	source.Transit(vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead, {},
 	               command);
 	Transit(vk::ImageLayout::eTransferDstOptimal, vk::AccessFlagBits2::eTransferWrite, {}, command);
@@ -532,9 +521,9 @@ void Image::CopyImageWithBuffer(Image& source, Buffer& buffer) {
 }
 
 void Image::CopyMip(Image& source, uint32_t mip, uint32_t layer) {
-	EXIT_IF(m_scheduler == nullptr || source.backing.samples != backing.samples ||
-	        mip >= backing.mip_levels || layer >= backing.layers);
-	m_scheduler->EndRendering();
+	EXIT_IF(source.backing.samples != backing.samples || mip >= backing.mip_levels ||
+	        layer >= backing.layers);
+	m_scheduler.EndRendering();
 	const auto width  = std::max(backing.extent.width >> mip, 1u);
 	const auto height = std::max(backing.extent.height >> mip, 1u);
 	const auto depth  = std::max(backing.extent.depth >> mip, 1u);
@@ -554,7 +543,7 @@ void Image::CopyMip(Image& source, uint32_t mip, uint32_t layer) {
 		copy.dstSubresource = {aspect, mip, layer, destination_layers};
 		copy.extent         = {width, height, depth};
 	}
-	auto command = m_scheduler->Current().Handle();
+	auto command = m_scheduler.Current().Handle();
 	Transit(vk::ImageLayout::eTransferDstOptimal, vk::AccessFlagBits2::eTransferWrite, {}, command);
 	source.Transit(vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead, {},
 	               command);
@@ -567,13 +556,13 @@ void Image::CopyMip(Image& source, uint32_t mip, uint32_t layer) {
 namespace ImageOps {
 
 void Validate(const ImageInfo& info) {
-	ValidateRange(info.data, "data");
-	ValidateRange(info.stencil, "stencil");
+	ValidateOptionalRange(info.data, "data");
+	ValidateOptionalRange(info.stencil, "stencil");
 
 	if (info.pixel_format == vk::Format::eUndefined) {
 		const bool metadata_empty =
-		    info.metadata.range.address == 0 && info.metadata.range.size == 0 &&
-		    info.metadata.kind == ImageMetadataKind::None && info.metadata.control == 0 &&
+		    info.metadata.range.Empty() && info.metadata.kind == ImageMetadataKind::None &&
+		    info.metadata.control == 0 &&
 		    info.metadata.compression == VideoOutCompression::Uncompressed &&
 		    !info.metadata.stencil_compressed;
 		if (info.data.Empty() || info.HasStencil() || !metadata_empty || info.extent.width == 0 ||
@@ -620,16 +609,14 @@ void Validate(const ImageInfo& info) {
 	}
 	switch (info.metadata.kind) {
 		case ImageMetadataKind::None:
-			if (info.metadata.range.address != 0 || info.metadata.range.size != 0 ||
-			    info.metadata.control != 0 ||
+			if (!info.metadata.range.Empty() || info.metadata.control != 0 ||
 			    info.metadata.compression != VideoOutCompression::Uncompressed ||
 			    info.metadata.stencil_compressed) {
 				EXIT("metadata-free image has metadata state\n");
 			}
 			break;
 		case ImageMetadataKind::Htile:
-			ValidateRange(info.metadata.range, "HTILE");
-			if (info.metadata.range.Empty() ||
+			if (!info.metadata.range.Valid() ||
 			    info.metadata.compression != VideoOutCompression::Uncompressed) {
 				EXIT("invalid HTILE metadata\n");
 			}
@@ -660,7 +647,7 @@ Prospero::BufferFormat RenderTargetTransferFormat(uint32_t bytes_per_element) {
 } // namespace ImageOps
 
 Image::Image(GraphicContext& graphics, CommandScheduler& scheduler, const ImageInfo& image_info)
-    : info(image_info), m_graphics(&graphics), m_scheduler(&scheduler) {
+    : info(image_info), m_graphics(graphics), m_scheduler(scheduler) {
 	KYTY_PROFILER_FUNCTION();
 	ImageOps::Validate(info);
 	m_cpu_dirty =
@@ -680,7 +667,6 @@ Image::Image(GraphicContext& graphics, CommandScheduler& scheduler, const ImageI
 	backing.usage       = ImageUsageFlags(graphics, info);
 
 	vk::ImageCreateInfo create {};
-	create.sType         = vk::StructureType::eImageCreateInfo;
 	create.flags         = backing.flags;
 	create.imageType     = backing.image_type;
 	create.extent        = backing.extent;
@@ -690,7 +676,6 @@ Image::Image(GraphicContext& graphics, CommandScheduler& scheduler, const ImageI
 	create.tiling        = vk::ImageTiling::eOptimal;
 	create.initialLayout = backing.state.layout;
 	create.usage         = backing.usage;
-	create.sharingMode   = vk::SharingMode::eExclusive;
 	create.samples       = vulkan_sample_count(backing.samples);
 
 	vk::ImageFormatProperties properties {};
@@ -733,21 +718,13 @@ uint64_t Image::HashGuestEdges() const {
 
 Image::~Image() {
 	KYTY_PROFILER_FUNCTION();
-	if (m_graphics == nullptr) {
-		return;
-	}
-	{
-		std::lock_guard lock(views.mutex);
-		for (auto& cached: views.views) {
-			if (cached.view != nullptr) {
-				m_graphics->device.destroyImageView(cached.view, nullptr);
-				cached.view = nullptr;
-			}
+	for (const auto& cached: views) {
+		if (cached.view != nullptr) {
+			m_graphics.device.destroyImageView(cached.view, nullptr);
 		}
-		views.views.clear();
 	}
 	if (backing.image != nullptr) {
-		m_graphics->DeleteImage(backing);
+		m_graphics.DeleteImage(backing);
 	}
 }
 

src/graphics/host_gpu/renderer/image/image.h

@@ -2,12 +2,12 @@
 #define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_IMAGE_H_
 
 #include "common/assert.h"
+#include "common/slotVector.h"
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/host_gpu/renderer/image/imageInfo.h"
 
 #include <compare>
 #include <limits>
-#include <mutex>
 #include <optional>
 #include <span>
 #include <utility>
@@ -19,29 +19,13 @@ class Buffer;
 class CommandScheduler;
 struct ImageTestAccess;
 
-struct ImageId {
-	uint32_t index      = std::numeric_limits<uint32_t>::max();
-	uint32_t generation = 0;
-
-	[[nodiscard]] explicit operator bool() const noexcept {
-		return index != std::numeric_limits<uint32_t>::max();
-	}
-	auto operator<=>(const ImageId&) const = default;
-};
+using ImageId = Common::SlotId;
 
 struct CachedImageView {
 	ImageViewInfo info;
 	vk::ImageView view = nullptr;
 };
 
-struct ImageViewCache {
-	ImageViewCache() = default;
-	KYTY_CLASS_NO_COPY(ImageViewCache);
-
-	std::mutex                   mutex;
-	std::vector<CachedImageView> views;
-};
-
 struct ImageUsage {
 	bool texture       = false;
 	bool storage       = false;
@@ -51,11 +35,13 @@ struct ImageUsage {
 };
 
 struct ImageBinding {
-	bool is_bound      = false;
-	bool is_target     = false;
-	bool needs_rebind  = false;
-	bool force_general = false;
-	bool shader_write  = false;
+	vk::ImageLayout  attachment_layout = vk::ImageLayout::eUndefined;
+	vk::AccessFlags2 attachment_access;
+	bool             is_bound      = false;
+	bool             is_target     = false;
+	bool             needs_rebind  = false;
+	bool             force_general = false;
+	bool             shader_write  = false;
 };
 
 class Image final {
@@ -65,7 +51,6 @@ public:
 	KYTY_CLASS_NO_COPY(Image);
 
 	[[nodiscard]] vk::ImageView FindView(const ImageViewInfo& view_info);
-	void                        AssociateDepth(ImageId image_id) { depth_id = image_id; }
 	using Barriers = std::vector<vk::ImageMemoryBarrier2>;
 	[[nodiscard]] Barriers GetBarriers(vk::ImageLayout                      destination_layout,
 	                                   vk::AccessFlags2                     destination_access,
@@ -143,9 +128,6 @@ public:
 		return pages ? ImagePageRangesOverlap(info.data.address, info.data.size, address, size)
 		             : ImageRangeOverlaps(info.data.address, info.data.size, address, size);
 	}
-	[[nodiscard]] bool GpuOverlaps(uint64_t address, uint64_t size) const noexcept {
-		return IsGpuModified() && Overlaps(address, size);
-	}
 	[[nodiscard]] bool SafeToDownload() const noexcept {
 		return IsGpuModified() && !IsBufferModified() && !IsCpuDirty();
 	}
@@ -157,7 +139,7 @@ public:
 
 	ImageInfo        info;
 	VulkanImage      backing;
-	ImageViewCache   views;
+	std::vector<CachedImageView> views;
 	ImageUsage       usage;
 	ImageBinding     binding;
 	bool             registered     = false;
@@ -177,8 +159,8 @@ private:
 	[[nodiscard]] static std::pair<uint32_t, uint32_t>
 	SanitizeCopyLayers(const Image& source, const Image& destination, uint32_t depth);
 
-	GraphicContext*   m_graphics         = nullptr;
-	CommandScheduler* m_scheduler        = nullptr;
+	GraphicContext&   m_graphics;
+	CommandScheduler& m_scheduler;
 	uint64_t          m_maybe_cpu_hash   = 0;
 	bool              m_cpu_dirty        = false;
 	bool              m_maybe_cpu_dirty  = false;

src/graphics/host_gpu/renderer/image/imageInfo.h

@@ -15,18 +15,6 @@
 
 namespace Libs::Graphics {
 
-struct GuestRange {
-	uint64_t address = 0;
-	uint64_t size    = 0;
-
-	[[nodiscard]] constexpr bool Empty() const noexcept { return address == 0 || size == 0; }
-	[[nodiscard]] constexpr bool Valid() const noexcept {
-		return !Empty() && address < TRACKER_ADDRESS_SIZE && size <= TRACKER_ADDRESS_SIZE - address;
-	}
-	[[nodiscard]] constexpr uint64_t End() const noexcept { return address + size; }
-	auto                             operator<=>(const GuestRange&) const = default;
-};
-
 enum class VideoOutCompression : uint8_t { Uncompressed, Dcc256_256_0, Dcc256_64_64, Unsupported };
 
 enum class ImageMetadataKind : uint8_t { None, Htile, Dcc };
@@ -35,8 +23,11 @@ struct ImageMetadataInfo {
 	GuestRange          range;
 	ImageMetadataKind   kind               = ImageMetadataKind::None;
 	uint32_t            control            = 0;
+	uint32_t            dcc_clear_word           = 0;
 	VideoOutCompression compression        = VideoOutCompression::Uncompressed;
 	bool                stencil_compressed = false;
+	bool                dcc_clear_register_valid = false;
+	bool                dcc_alpha_msb            = true;
 };
 
 struct ImageSubresources {
@@ -56,6 +47,7 @@ struct ImageSubresourceRange {
 struct ImageMipInfo {
 	uint64_t offset                                 = 0;
 	uint64_t size                                   = 0;
+	// Padded dimensions in storage elements (compressed blocks for BC formats).
 	uint32_t pitch                                  = 0;
 	uint32_t height                                 = 0;
 	auto     operator<=>(const ImageMipInfo&) const = default;
@@ -443,6 +435,10 @@ IsSupportedDisplayRenderTargetTileMode(Prospero::TileMode tile_mode) noexcept {
 		return encoded <= 0.04045f ? encoded / 12.92f : std::pow((encoded + 0.055f) / 1.055f, 2.4f);
 	};
 	switch (format) {
+		// A single-plane float target carries its clear as raw float bits, the same encoding the
+		// depth decoder below uses. Without this the clear is discarded and the target keeps stale
+		// contents.
+		case vk::Format::eR32Sfloat: next.float32[0] = std::bit_cast<float>(packed); break;
 		case vk::Format::eR32Uint: next.uint32[0] = packed; break;
 		case vk::Format::eR32Sint: next.int32[0] = static_cast<int32_t>(packed); break;
 		case vk::Format::eR8G8B8A8Srgb:

src/graphics/host_gpu/renderer/image/imageView.cpp

@@ -4,8 +4,6 @@
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/host_gpu/renderer/image/image.h"
 
-#include <mutex>
-
 namespace Libs::Graphics {
 
 namespace {
@@ -23,10 +21,6 @@ namespace {
 	}
 }
 
-[[nodiscard]] bool IsCompatibleViewFormat(vk::Format image_format, vk::Format view_format) {
-	return ImageViewOps::FormatsCompatible(image_format, view_format);
-}
-
 [[nodiscard]] bool IsStencilViewFormat(vk::Format format) {
 	switch (format) {
 		case vk::Format::eS8Uint:
@@ -329,7 +323,7 @@ vk::ImageView Image::FindView(const ImageViewInfo& view_info) {
 	}
 	normalized.usage = is_storage ? vk::ImageUsageFlagBits::eStorage : vk::ImageUsageFlags {};
 	const bool format_compatible = normalized.format != vk::Format::eUndefined &&
-	                               IsCompatibleViewFormat(image.format, normalized.format);
+	                               ImageViewOps::FormatsCompatible(image.format, normalized.format);
 	const bool slice_view =
 	    image.image_type == vk::ImageType::e3D && (normalized.type == vk::ImageViewType::e2D ||
 	                                               normalized.type == vk::ImageViewType::e2DArray);
@@ -357,21 +351,18 @@ vk::ImageView Image::FindView(const ImageViewInfo& view_info) {
 		     image.layers);
 	}
 
-	std::lock_guard lock(views.mutex);
-	for (const auto& cached: views.views) {
+	for (const auto& cached: views) {
 		if (cached.info == normalized) {
 			return cached.view;
 		}
 	}
 
 	vk::ImageViewUsageCreateInfo usage {};
-	usage.sType = vk::StructureType::eImageViewUsageCreateInfo;
 	usage.usage = image.usage;
 	if (!is_storage) {
 		usage.usage &= ~vk::ImageUsageFlagBits::eStorage;
 	}
 	vk::ImageViewCreateInfo create {};
-	create.sType                           = vk::StructureType::eImageViewCreateInfo;
 	create.pNext                           = &usage;
 	create.image                           = image.image;
 	create.viewType                        = normalized.type;
@@ -384,7 +375,7 @@ vk::ImageView Image::FindView(const ImageViewInfo& view_info) {
 	create.subresourceRange.layerCount     = normalized.layer_count;
 
 	vk::ImageView view   = nullptr;
-	const auto    result = m_graphics->device.createImageView(&create, nullptr, &view);
+	const auto    result = m_graphics.device.createImageView(&create, nullptr, &view);
 	if (result != vk::Result::eSuccess || view == nullptr) {
 		EXIT("failed to create image view: result=%d image_format=%d view_format=%d type=%d "
 		     "aspect=0x%x mip=%u+%u layer=%u+%u usage=0x%x\n",
@@ -394,7 +385,7 @@ vk::ImageView Image::FindView(const ImageViewInfo& view_info) {
 		     view_info.level_count, view_info.base_layer, view_info.layer_count,
 		     static_cast<vk::ImageUsageFlags::MaskType>(view_info.usage));
 	}
-	views.views.push_back({normalized, view});
+	views.push_back({normalized, view});
 	return view;
 }
 

src/graphics/host_gpu/renderer/image/imageView.h

@@ -92,29 +92,18 @@ SelectSampledColorView(vk::Format image_format, vk::Format view_format, uint32_t
 	}
 }
 
-[[nodiscard]] inline uint32_t
-SelectSampledDepthView(vk::Format image_format, vk::Format view_format, uint32_t swizzle) noexcept {
-	if (IsSupportedSampledDepthView(image_format, view_format, swizzle)) {
-		return swizzle;
-	}
-	EXIT("unsupported sampled depth image view: image_format=%d view_format=%d swizzle=0x%03x\n",
-	     static_cast<int>(image_format), static_cast<int>(view_format), swizzle);
-}
-
 [[nodiscard]] inline bool
 IsSupportedSampledDepthResource(const ShaderRecompiler::IR::ImageResource& resource) noexcept {
-	if (resource.kind != ShaderRecompiler::IR::ResourceKind::Image &&
-	    resource.kind != ShaderRecompiler::IR::ResourceKind::ImageUint) {
+	if (resource.resource_class != ShaderRecompiler::IR::ImageResourceClass::Sampled) {
 		return false;
 	}
-	if (resource.kind == ShaderRecompiler::IR::ResourceKind::ImageUint && resource.depth_compare) {
+	if (resource.numeric_class == Prospero::TextureNumericClass::Unsupported) {
 		return false;
 	}
-	return (resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2D ||
-	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DArray ||
-	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaa ||
-	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaaArray) &&
-	       resource.mip_mode == ShaderRecompiler::IR::ImageMipMode::None && resource.read &&
+	if (resource.numeric_class != Prospero::TextureNumericClass::Float && resource.depth_compare) {
+		return false;
+	}
+	return resource.mip_mode == ShaderRecompiler::IR::ImageMipMode::None && resource.read &&
 	       !resource.written && !resource.atomic;
 }
 
@@ -128,8 +117,9 @@ inline void ValidateStorageColorView(vk::Format image_format, vk::Format view_fo
 
 [[nodiscard]] inline bool
 IsSupportedStorageImageResource(const ShaderRecompiler::IR::ImageResource& resource) noexcept {
-	return (resource.kind == ShaderRecompiler::IR::ResourceKind::StorageImage ||
-	        resource.kind == ShaderRecompiler::IR::ResourceKind::StorageImageUint) &&
+	return resource.resource_class == ShaderRecompiler::IR::ImageResourceClass::Storage &&
+	       (resource.numeric_class == Prospero::TextureNumericClass::Float ||
+	        resource.numeric_class == Prospero::TextureNumericClass::Uint) &&
 	       (resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim1D ||
 	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim1DArray ||
 	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2D ||
@@ -141,19 +131,19 @@ IsSupportedStorageImageResource(const ShaderRecompiler::IR::ImageResource& resou
 	         resource.mip_count != 0u)) &&
 	       resource.written &&
 	       (!resource.atomic ||
-	        (resource.kind == ShaderRecompiler::IR::ResourceKind::StorageImageUint &&
-	         resource.read)) &&
+	        (resource.numeric_class == Prospero::TextureNumericClass::Uint && resource.read)) &&
 	       !resource.depth_compare;
 }
 
 inline void
 ValidateStorageImageResource(const ShaderRecompiler::IR::ImageResource& resource) noexcept {
 	if (!IsSupportedStorageImageResource(resource)) {
-		EXIT("unsupported storage color image resource: kind=%u dimension=%u mip=%u "
+		EXIT("unsupported storage color image resource: class=%u numeric=%u dimension=%u mip=%u "
 		     "read=%d written=%d atomic=%d depth_compare=%d\n",
-		     static_cast<uint32_t>(resource.kind), static_cast<uint32_t>(resource.dimension),
-		     static_cast<uint32_t>(resource.mip_mode), resource.read, resource.written,
-		     resource.atomic, resource.depth_compare);
+		     static_cast<uint32_t>(resource.resource_class),
+		     static_cast<uint32_t>(resource.numeric_class),
+		     static_cast<uint32_t>(resource.dimension), static_cast<uint32_t>(resource.mip_mode),
+		     resource.read, resource.written, resource.atomic, resource.depth_compare);
 	}
 }
 

src/graphics/host_gpu/renderer/image/textureCommon.cpp

@@ -119,13 +119,16 @@ vk::ComponentMapping TextureGetComponentMapping(uint32_t swizzle) {
 	return components;
 }
 
-vk::Format TextureGetFormat(Prospero::BufferFormat format) {
-	const auto vk_format = VulkanFormat(format);
+SurfaceFormatInfo TextureGetSurfaceFormatInfo(Prospero::BufferFormat format) {
+	const auto backing_format    = Prospero::RemapTextureFormat(format);
+	const auto vk_format         = VulkanFormat(backing_format);
+	const auto conversion_format =
+	    backing_format != format ? format : Prospero::BufferFormat::kInvalid;
 	if (vk_format != vk::Format::eUndefined) {
-		return vk_format;
+		return SurfaceFormatInfo(vk_format, conversion_format);
 	}
 	EXIT("unknown format: fmt = %u\n", static_cast<uint32_t>(format));
-	return vk::Format::eUndefined;
+	return SurfaceFormatInfo(vk::Format::eUndefined, Prospero::BufferFormat::kInvalid);
 }
 
 namespace {

src/graphics/host_gpu/renderer/image/textureCommon.h

@@ -3,7 +3,7 @@
 
 #include "common/abi.h"
 #include "common/common.h"
-#include "graphics/guest_gpu/gpu_defs.h"
+#include "graphics/guest_gpu/gpu_format.h"
 #include "graphics/guest_gpu/tile.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 
@@ -19,6 +19,15 @@ struct RenderTargetFormatInfo {
 	Prospero::ColorComponentMapping export_mapping;
 };
 
+struct SurfaceFormatInfo {
+	explicit constexpr SurfaceFormatInfo(vk::Format format,
+	                                     Prospero::BufferFormat conversion_format)
+	    : vk_format(format), conversion_format(conversion_format) {}
+
+	vk::Format             vk_format;
+	Prospero::BufferFormat conversion_format;
+};
+
 struct TextureUploadMipLayout {
 	uint64_t offset       = 0;
 	uint64_t size         = 0;
@@ -35,7 +44,7 @@ struct TextureUploadLayout {
 };
 
 vk::ComponentMapping   TextureGetComponentMapping(uint32_t swizzle);
-vk::Format             TextureGetFormat(Prospero::BufferFormat format);
+SurfaceFormatInfo      TextureGetSurfaceFormatInfo(Prospero::BufferFormat format);
 RenderTargetFormatInfo TextureGetRenderTargetFormat(Prospero::ChannelLayout layout,
                                                     Prospero::ChannelType   type,
                                                     Prospero::ChannelOrder  order);

src/graphics/host_gpu/renderer/image/tiler.cpp

@@ -40,7 +40,6 @@ TileManager::TileManager(GraphicContext& graphics, CommandScheduler& scheduler,
 	               nullptr};
 
 	vk::DescriptorSetLayoutCreateInfo descriptor_info {};
-	descriptor_info.sType        = vk::StructureType::eDescriptorSetLayoutCreateInfo;
 	descriptor_info.flags        = vk::DescriptorSetLayoutCreateFlagBits::ePushDescriptorKHR;
 	descriptor_info.bindingCount = static_cast<uint32_t>(bindings.size());
 	descriptor_info.pBindings    = bindings.data();
@@ -50,7 +49,6 @@ TileManager::TileManager(GraphicContext& graphics, CommandScheduler& scheduler,
 
 	const vk::PushConstantRange  push_range {vk::ShaderStageFlagBits::eCompute, 0, sizeof(Push)};
 	vk::PipelineLayoutCreateInfo layout_info {};
-	layout_info.sType                  = vk::StructureType::ePipelineLayoutCreateInfo;
 	layout_info.setLayoutCount         = 1;
 	layout_info.pSetLayouts            = &m_descriptor_layout;
 	layout_info.pushConstantRangeCount = 1;
@@ -92,11 +90,9 @@ TileManager::~TileManager() {
 TileManager::Scratch TileManager::AllocateScratch(uint64_t size) {
 	EXIT_IF(size == 0);
 	vk::BufferCreateInfo create {};
-	create.sType = vk::StructureType::eBufferCreateInfo;
 	create.size  = size;
 	create.usage = vk::BufferUsageFlagBits::eStorageBuffer | vk::BufferUsageFlagBits::eTransferSrc |
 	               vk::BufferUsageFlagBits::eTransferDst;
-	create.sharingMode = vk::SharingMode::eExclusive;
 
 	VmaAllocationCreateInfo allocate {};
 	allocate.usage       = VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE;
@@ -264,20 +260,17 @@ vk::Pipeline TileManager::GetPipeline(uint32_t slot) {
 	const vk::SpecializationMapEntry entries[] {{0, 0, 4}, {1, 4, 4}};
 	const vk::SpecializationInfo     specialization {2, entries, sizeof(values), values};
 	vk::ShaderModuleCreateInfo       module_info {};
-	module_info.sType       = vk::StructureType::eShaderModuleCreateInfo;
 	module_info.codeSize    = shaders[family_index].words * sizeof(uint32_t);
 	module_info.pCode       = shaders[family_index].code;
 	vk::ShaderModule module = nullptr;
 	RequireVulkanSuccess(m_graphics.device.createShaderModule(&module_info, nullptr, &module),
 	                     "create TileManager shader module");
 	vk::PipelineShaderStageCreateInfo stage {};
-	stage.sType               = vk::StructureType::ePipelineShaderStageCreateInfo;
 	stage.stage               = vk::ShaderStageFlagBits::eCompute;
 	stage.module              = module;
 	stage.pName               = "main";
 	stage.pSpecializationInfo = &specialization;
 	vk::ComputePipelineCreateInfo create {};
-	create.sType  = vk::StructureType::eComputePipelineCreateInfo;
 	create.stage  = stage;
 	create.layout = m_pipeline_layout;
 	const auto result =
@@ -287,7 +280,7 @@ vk::Pipeline TileManager::GetPipeline(uint32_t slot) {
 	return m_pipelines[slot];
 }
 
-void TileManager::Record(bool tile, vk::Buffer source, uint64_t source_offset,
+void TileManager::Record(vk::Buffer source, uint64_t source_offset,
                          uint64_t source_capacity, vk::Buffer target, uint64_t target_offset,
                          uint64_t target_capacity, std::span<Dispatch> dispatches,
                          bool clear_target) {
@@ -307,7 +300,6 @@ void TileManager::Record(bool tile, vk::Buffer source, uint64_t source_offset,
 	m_scheduler.EndRendering();
 	auto                    command = m_scheduler.Current().Handle();
 	vk::BufferMemoryBarrier barriers[3] {};
-	barriers[0].sType         = vk::StructureType::eBufferMemoryBarrier;
 	barriers[0].srcAccessMask = vk::AccessFlagBits::eMemoryWrite | vk::AccessFlagBits::eHostWrite;
 	barriers[0].dstAccessMask = vk::AccessFlagBits::eShaderRead;
 	barriers[0].srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
@@ -316,7 +308,6 @@ void TileManager::Record(bool tile, vk::Buffer source, uint64_t source_offset,
 	barriers[0].offset              = source_offset;
 	barriers[0].size                = source_capacity;
 	barriers[1]                     = barriers[0];
-	barriers[1].srcAccessMask = vk::AccessFlagBits::eMemoryWrite | vk::AccessFlagBits::eHostWrite;
 	barriers[1].dstAccessMask =
 	    clear_target ? vk::AccessFlagBits::eTransferWrite
 	                 : vk::AccessFlagBits::eShaderRead | vk::AccessFlagBits::eShaderWrite;
@@ -352,7 +343,6 @@ void TileManager::Record(bool tile, vk::Buffer source, uint64_t source_offset,
 		const vk::DescriptorBufferInfo        infos[] {source_info, target_info, params_info};
 		std::array<vk::WriteDescriptorSet, 3> writes {};
 		for (uint32_t index = 0; index < writes.size(); index++) {
-			writes[index].sType           = vk::StructureType::eWriteDescriptorSet;
 			writes[index].dstBinding      = index;
 			writes[index].descriptorCount = 1;
 			writes[index].descriptorType  = index == 2 ? vk::DescriptorType::eUniformBuffer
@@ -384,7 +374,7 @@ TileManager::Result TileManager::Detile(vk::Buffer tiled, uint64_t tiled_offset,
 	Prepare(false, tiled_capacity, linear_capacity, infos, source_base, 0, dispatches);
 	auto scratch = AllocateScratch((linear_capacity + 3u) & ~uint64_t {3});
 	DeferDestroy(scratch);
-	Record(false, tiled, tiled_offset, tiled_capacity, scratch.buffer, 0, scratch.size, dispatches,
+	Record(tiled, tiled_offset, tiled_capacity, scratch.buffer, 0, scratch.size, dispatches,
 	       true);
 	return {scratch.buffer, 0, linear_capacity};
 }
@@ -399,7 +389,7 @@ void TileManager::Tile(vk::Buffer linear, uint64_t linear_offset, uint64_t linea
 	const uint64_t        target_base = tiled_offset & (descriptor_alignment - 1);
 	std::vector<Dispatch> dispatches;
 	Prepare(true, tiled_capacity, linear_capacity, infos, source_base, target_base, dispatches);
-	Record(true, linear, linear_offset, linear_capacity, tiled, tiled_offset, tiled_capacity,
+	Record(linear, linear_offset, linear_capacity, tiled, tiled_offset, tiled_capacity,
 	       dispatches, false);
 }
 
@@ -423,7 +413,7 @@ void TileManager::TileImage(Image& image, std::span<const vk::BufferImageCopy> r
 	if (transform == ColorTransform::SwapBgra16) {
 		source = SwapBgra16(source);
 	}
-	Record(true, source.buffer, source.offset, linear_capacity, tiled, tiled_offset, tiled_capacity,
+	Record(source.buffer, source.offset, linear_capacity, tiled, tiled_offset, tiled_capacity,
 	       dispatches, false);
 }
 
@@ -482,20 +472,17 @@ void TileManager::ConvertD16(Result source, Result target, D16Direction directio
 			words = std::size(GPU_TILER_DEMOTE_D16_SPV);
 		}
 		vk::ShaderModuleCreateInfo module_info {};
-		module_info.sType       = vk::StructureType::eShaderModuleCreateInfo;
 		module_info.codeSize    = words * sizeof(uint32_t);
 		module_info.pCode       = code;
 		vk::ShaderModule module = nullptr;
 		RequireVulkanSuccess(m_graphics.device.createShaderModule(&module_info, nullptr, &module),
 		                     "create D16 conversion shader module");
 		vk::PipelineShaderStageCreateInfo stage {};
-		stage.sType               = vk::StructureType::ePipelineShaderStageCreateInfo;
 		stage.stage               = vk::ShaderStageFlagBits::eCompute;
 		stage.module              = module;
 		stage.pName               = "main";
 		stage.pSpecializationInfo = &specialization;
 		vk::ComputePipelineCreateInfo create {};
-		create.sType  = vk::StructureType::eComputePipelineCreateInfo;
 		create.stage  = stage;
 		create.layout = m_pipeline_layout;
 		const auto result =
@@ -532,7 +519,6 @@ void TileManager::ConvertD16(Result source, Result target, D16Direction directio
 	m_scheduler.EndRendering();
 	auto                    command = m_scheduler.Current().Handle();
 	vk::BufferMemoryBarrier barriers[2] {};
-	barriers[0].sType               = vk::StructureType::eBufferMemoryBarrier;
 	barriers[0].srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
 	barriers[0].dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
 	barriers[0].buffer              = source.buffer;
@@ -542,7 +528,6 @@ void TileManager::ConvertD16(Result source, Result target, D16Direction directio
 	                            vk::AccessFlagBits::eTransferWrite |
 	                            vk::AccessFlagBits::eShaderWrite;
 	barriers[0].dstAccessMask = vk::AccessFlagBits::eShaderRead;
-	barriers[1].sType         = vk::StructureType::eBufferMemoryBarrier;
 	barriers[1].srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
 	barriers[1].dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
 	barriers[1].buffer              = target.buffer;
@@ -596,7 +581,6 @@ void TileManager::ConvertD16(Result source, Result target, D16Direction directio
 			};
 			std::array<vk::WriteDescriptorSet, 2> writes {};
 			for (uint32_t index = 0; index < writes.size(); index++) {
-				writes[index].sType           = vk::StructureType::eWriteDescriptorSet;
 				writes[index].dstBinding      = index;
 				writes[index].descriptorCount = 1;
 				writes[index].descriptorType  = vk::DescriptorType::eStorageBuffer;
@@ -627,19 +611,16 @@ void TileManager::ConvertD16(Result source, Result target, D16Direction directio
 void TileManager::SwapBgra16(Result input, Result output, uint32_t pixels) {
 	if (m_swap_bgra16 == nullptr) {
 		vk::ShaderModuleCreateInfo module_info {};
-		module_info.sType       = vk::StructureType::eShaderModuleCreateInfo;
 		module_info.codeSize    = std::size(GPU_TILER_SWAP_BGRA16_SPV) * sizeof(uint32_t);
 		module_info.pCode       = GPU_TILER_SWAP_BGRA16_SPV;
 		vk::ShaderModule module = nullptr;
 		RequireVulkanSuccess(m_graphics.device.createShaderModule(&module_info, nullptr, &module),
 		                     "create BGRA16 swap shader module");
 		vk::PipelineShaderStageCreateInfo stage {};
-		stage.sType  = vk::StructureType::ePipelineShaderStageCreateInfo;
 		stage.stage  = vk::ShaderStageFlagBits::eCompute;
 		stage.module = module;
 		stage.pName  = "main";
 		vk::ComputePipelineCreateInfo create {};
-		create.sType  = vk::StructureType::eComputePipelineCreateInfo;
 		create.stage  = stage;
 		create.layout = m_pipeline_layout;
 		const auto result =
@@ -658,7 +639,6 @@ void TileManager::SwapBgra16(Result input, Result output, uint32_t pixels) {
 	};
 	std::array<vk::WriteDescriptorSet, 2> writes {};
 	for (uint32_t index = 0; index < writes.size(); index++) {
-		writes[index].sType           = vk::StructureType::eWriteDescriptorSet;
 		writes[index].dstBinding      = index;
 		writes[index].descriptorCount = 1;
 		writes[index].descriptorType  = vk::DescriptorType::eStorageBuffer;
@@ -666,7 +646,6 @@ void TileManager::SwapBgra16(Result input, Result output, uint32_t pixels) {
 	}
 	vk::BufferMemoryBarrier barriers[2] {};
 	for (uint32_t index = 0; index < 2; index++) {
-		barriers[index].sType               = vk::StructureType::eBufferMemoryBarrier;
 		barriers[index].srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
 		barriers[index].dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
 		barriers[index].buffer              = infos[index].buffer;

src/graphics/host_gpu/renderer/image/tiler.h

@@ -125,7 +125,7 @@ private:
 	void Prepare(bool tile, uint64_t tiled_capacity, uint64_t linear_capacity,
 	             std::span<const GpuTileInfo> infos, uint64_t source_base, uint64_t target_base,
 	             std::vector<Dispatch>& dispatches);
-	void Record(bool tile, vk::Buffer source, uint64_t source_offset, uint64_t source_capacity,
+	void Record(vk::Buffer source, uint64_t source_offset, uint64_t source_capacity,
 	            vk::Buffer target, uint64_t target_offset, uint64_t target_capacity,
 	            std::span<Dispatch> dispatches, bool clear_target);
 	[[nodiscard]] vk::Pipeline GetPipeline(uint32_t slot);

src/graphics/host_gpu/renderer/masterSemaphore.cpp

@@ -7,12 +7,10 @@ namespace Libs::Graphics {
 
 MasterSemaphore::MasterSemaphore(GraphicContext& graphics): m_graphics(graphics) {
 	vk::SemaphoreTypeCreateInfo type_info {};
-	type_info.sType         = vk::StructureType::eSemaphoreTypeCreateInfo;
 	type_info.semaphoreType = vk::SemaphoreType::eTimeline;
 	type_info.initialValue  = 0;
 
 	vk::SemaphoreCreateInfo create_info {};
-	create_info.sType = vk::StructureType::eSemaphoreCreateInfo;
 	create_info.pNext = &type_info;
 
 	const auto result = m_graphics.device.createSemaphore(&create_info, nullptr, &m_semaphore);
@@ -47,7 +45,6 @@ void MasterSemaphore::Wait(uint64_t tick) {
 	}
 
 	vk::SemaphoreWaitInfo wait_info {};
-	wait_info.sType          = vk::StructureType::eSemaphoreWaitInfo;
 	wait_info.semaphoreCount = 1;
 	wait_info.pSemaphores    = &m_semaphore;
 	wait_info.pValues        = &tick;

src/graphics/host_gpu/renderer/pipeline/descriptorCache.cpp

@@ -1,313 +0,0 @@
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
-
-#include "common/assert.h"
-#include "common/profiler.h"
-#include "graphics/shader/recompiler/ir/ShaderIR.h"
-
-#include <array>
-
-namespace Libs::Graphics {
-namespace {
-
-using BindingKind = ShaderRecompiler::IR::DescriptorBindingKind;
-
-vk::ShaderStageFlags StageFlags(DescriptorCache::Stage stage) {
-	switch (stage) {
-		case DescriptorCache::Stage::Vertex: return vk::ShaderStageFlagBits::eVertex;
-		case DescriptorCache::Stage::Pixel: return vk::ShaderStageFlagBits::eFragment;
-		case DescriptorCache::Stage::Compute: return vk::ShaderStageFlagBits::eCompute;
-		default: EXIT("unknown descriptor stage\n");
-	}
-}
-
-bool IsSampledImage(BindingKind kind) {
-	switch (kind) {
-		case BindingKind::Sampled1D:
-		case BindingKind::Sampled1DArray:
-		case BindingKind::Sampled2D:
-		case BindingKind::Sampled2DArray:
-		case BindingKind::Sampled2DMsaa:
-		case BindingKind::Sampled2DMsaaArray:
-		case BindingKind::Sampled3D:
-		case BindingKind::SampledUint1D:
-		case BindingKind::SampledUint1DArray:
-		case BindingKind::SampledUint2D:
-		case BindingKind::SampledUint2DArray:
-		case BindingKind::SampledUint2DMsaa:
-		case BindingKind::SampledUint2DMsaaArray:
-		case BindingKind::SampledUint3D: return true;
-		default: return false;
-	}
-}
-
-bool IsStorageImage(BindingKind kind) {
-	switch (kind) {
-		case BindingKind::Storage1D:
-		case BindingKind::Storage1DArray:
-		case BindingKind::Storage2D:
-		case BindingKind::Storage2DArray:
-		case BindingKind::Storage3D:
-		case BindingKind::StorageUint1D:
-		case BindingKind::StorageUint1DArray:
-		case BindingKind::StorageUint2D:
-		case BindingKind::StorageUint2DArray:
-		case BindingKind::StorageUint3D: return true;
-		default: return false;
-	}
-}
-
-vk::DescriptorType DescriptorType(BindingKind kind) {
-	if (kind == BindingKind::Samplers) {
-		return vk::DescriptorType::eSampler;
-	}
-	if (IsSampledImage(kind)) {
-		return vk::DescriptorType::eSampledImage;
-	}
-	if (IsStorageImage(kind)) {
-		return vk::DescriptorType::eStorageImage;
-	}
-	return vk::DescriptorType::eStorageBuffer;
-}
-
-uint32_t DescriptorCount(const ShaderRecompiler::IR::DescriptorBinding& binding) {
-	return binding.resources.empty() ? 1u : static_cast<uint32_t>(binding.resources.size());
-}
-
-std::vector<uint32_t> LayoutKey(DescriptorCache::Stage               stage,
-                                const ShaderRecompiler::IR::Program& program) {
-	std::vector<uint32_t> key;
-	key.reserve(1u + program.bindings.descriptors.size() * 4u);
-	key.push_back(static_cast<uint32_t>(stage));
-	for (const auto& binding: program.bindings.descriptors) {
-		key.push_back(static_cast<uint32_t>(binding.kind));
-		key.push_back(binding.binding);
-		key.push_back(DescriptorCount(binding));
-		key.push_back(static_cast<uint32_t>(DescriptorType(binding.kind)));
-	}
-	return key;
-}
-
-vk::DescriptorBufferInfo BufferInfo(const BufferView& view) {
-	EXIT_IF(view.buffer == nullptr);
-	return {view.buffer, view.offset, view.range};
-}
-
-} // namespace
-
-vk::DescriptorImageInfo DescriptorCache::MakeImageInfo(const TextureBinding& texture,
-                                                       uint32_t              element) {
-	const auto view =
-	    texture.mip_views.empty()
-	        ? (element == 0u ? texture.image_view : vk::ImageView {})
-	        : (element < texture.mip_views.size() ? texture.mip_views[element] : vk::ImageView {});
-	EXIT_IF(!texture.image_id || view == nullptr || texture.layout == vk::ImageLayout::eUndefined);
-	return {nullptr, view, texture.layout};
-}
-
-DescriptorCache::~DescriptorCache() {
-	for (auto& [layout, sets]: m_free_sets_by_layout) {
-		(void)layout;
-		for (auto* set: sets) {
-			delete set;
-		}
-	}
-	for (const auto& [key, layout]: m_descriptor_set_layouts) {
-		(void)key;
-		m_graphics.device.destroyDescriptorSetLayout(layout, nullptr);
-	}
-	for (const auto& pool: m_pools) {
-		m_graphics.device.destroyDescriptorPool(pool.pool, nullptr);
-	}
-}
-
-vk::DescriptorSetLayout
-DescriptorCache::GetDescriptorSetLayoutInternal(Stage                                stage,
-                                                const ShaderRecompiler::IR::Program& program) {
-	const auto key = LayoutKey(stage, program);
-	if (const auto found = m_descriptor_set_layouts.find(key);
-	    found != m_descriptor_set_layouts.end()) {
-		return found->second;
-	}
-
-	std::vector<vk::DescriptorSetLayoutBinding> bindings;
-	bindings.reserve(program.bindings.descriptors.size());
-	for (const auto& descriptor: program.bindings.descriptors) {
-		bindings.push_back({descriptor.binding, DescriptorType(descriptor.kind),
-		                    DescriptorCount(descriptor), StageFlags(stage), nullptr});
-	}
-	if (bindings.empty()) {
-		return nullptr;
-	}
-
-	vk::DescriptorSetLayoutCreateInfo info {};
-	info.sType                     = vk::StructureType::eDescriptorSetLayoutCreateInfo;
-	info.bindingCount              = static_cast<uint32_t>(bindings.size());
-	info.pBindings                 = bindings.data();
-	vk::DescriptorSetLayout layout = nullptr;
-	const auto result = m_graphics.device.createDescriptorSetLayout(&info, nullptr, &layout);
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess || layout == nullptr);
-	m_descriptor_set_layouts.emplace(key, layout);
-	return layout;
-}
-
-void DescriptorCache::CreatePool() {
-	KYTY_PROFILER_FUNCTION();
-	constexpr uint32_t           MaxSets = 512;
-	const vk::DescriptorPoolSize sizes[] = {
-	    {vk::DescriptorType::eStorageBuffer,
-	     MaxSets * (ShaderRecompiler::IR::ShaderInfo::MaxBuffers +
-	                ShaderRecompiler::IR::ShaderInfo::MaxAddresses + 3u)},
-	    {vk::DescriptorType::eSampledImage, MaxSets * ShaderRecompiler::IR::ShaderInfo::MaxImages},
-	    {vk::DescriptorType::eStorageImage, MaxSets * ShaderRecompiler::IR::ShaderInfo::MaxImages},
-	    {vk::DescriptorType::eSampler, MaxSets * ShaderRecompiler::IR::ShaderInfo::MaxSamplers},
-	};
-	vk::DescriptorPoolCreateInfo info {};
-	info.sType         = vk::StructureType::eDescriptorPoolCreateInfo;
-	info.poolSizeCount = static_cast<uint32_t>(std::size(sizes));
-	info.pPoolSizes    = sizes;
-	info.maxSets       = MaxSets;
-	const auto pool_id = static_cast<int>(m_pools.size());
-	auto&      pool    = m_pools.emplace_back();
-	const auto result  = m_graphics.device.createDescriptorPool(&info, nullptr, &pool.pool);
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess || pool.pool == nullptr);
-	pool.next_free_pool = m_first_free_pool;
-	m_first_free_pool   = pool_id;
-}
-
-VulkanDescriptorSet* DescriptorCache::Allocate(Stage                                stage,
-                                               const ShaderRecompiler::IR::Program& program) {
-	KYTY_PROFILER_FUNCTION();
-	Common::LockGuard lock(m_mutex);
-	const auto        layout = GetDescriptorSetLayoutInternal(stage, program);
-	EXIT_IF(layout == nullptr);
-	auto& free_sets = m_free_sets_by_layout[layout];
-	if (!free_sets.empty()) {
-		auto* result = free_sets.back();
-		free_sets.pop_back();
-		return result;
-	}
-
-	constexpr uint32_t Batch = 32;
-	for (int attempt = 0; attempt < 2; attempt++) {
-		for (int pool_id = m_first_free_pool; pool_id != -1;
-		     pool_id     = m_pools[pool_id].next_free_pool) {
-			auto&                                      pool = m_pools[pool_id];
-			std::array<vk::DescriptorSetLayout, Batch> layouts;
-			layouts.fill(layout);
-			std::array<vk::DescriptorSet, Batch> sets {};
-			vk::DescriptorSetAllocateInfo        info {};
-			info.sType              = vk::StructureType::eDescriptorSetAllocateInfo;
-			info.descriptorPool     = pool.pool;
-			info.descriptorSetCount = Batch;
-			info.pSetLayouts        = layouts.data();
-			if (m_graphics.device.allocateDescriptorSets(&info, sets.data()) ==
-			    vk::Result::eSuccess) {
-				free_sets.reserve(free_sets.size() + Batch - 1u);
-				for (uint32_t i = 0; i + 1u < Batch; i++) {
-					free_sets.push_back(new VulkanDescriptorSet {sets[i], layout, pool_id});
-				}
-				return new VulkanDescriptorSet {sets.back(), layout, pool_id};
-			}
-			m_first_free_pool = pool.next_free_pool;
-			break;
-		}
-		CreatePool();
-	}
-	return nullptr;
-}
-
-void DescriptorCache::Recycle(VulkanDescriptorSet& set) {
-	EXIT_IF(set.set == nullptr || set.layout == nullptr);
-	Common::LockGuard lock(m_mutex);
-	m_free_sets_by_layout[set.layout].push_back(&set);
-}
-
-VulkanDescriptorSet& DescriptorCache::GetDescriptor(Stage                                stage,
-                                                    const ShaderRecompiler::IR::Program& program,
-                                                    const NativeDescriptors&             data) {
-	KYTY_PROFILER_FUNCTION();
-	EXIT_IF(data.buffers.size() != program.info.buffers.size() ||
-	        data.images.size() != program.info.images.size() ||
-	        data.samplers.size() != program.info.samplers.size() ||
-	        data.addresses.size() != program.info.addresses.size());
-	auto* set = Allocate(stage, program);
-	EXIT_NOT_IMPLEMENTED(set == nullptr);
-
-	size_t descriptor_count = 0;
-	for (const auto& binding: program.bindings.descriptors) {
-		descriptor_count += DescriptorCount(binding);
-	}
-	std::vector<vk::DescriptorBufferInfo> buffer_infos;
-	std::vector<vk::DescriptorImageInfo>  image_infos;
-	std::vector<vk::WriteDescriptorSet>   writes;
-	buffer_infos.reserve(descriptor_count);
-	image_infos.reserve(descriptor_count);
-	writes.reserve(program.bindings.descriptors.size());
-	std::vector<uint32_t> image_occurrences(data.images.size());
-
-	for (const auto& binding: program.bindings.descriptors) {
-		vk::WriteDescriptorSet write {};
-		write.sType             = vk::StructureType::eWriteDescriptorSet;
-		write.dstSet            = set->set;
-		write.dstBinding        = binding.binding;
-		write.descriptorType    = DescriptorType(binding.kind);
-		write.descriptorCount   = DescriptorCount(binding);
-		const auto buffer_start = buffer_infos.size();
-		const auto image_start  = image_infos.size();
-		switch (binding.kind) {
-			case BindingKind::Buffers:
-				for (const auto resource: binding.resources) {
-					buffer_infos.push_back(BufferInfo(data.buffers.at(resource)));
-				}
-				break;
-			case BindingKind::AddressMemory:
-				for (const auto resource: binding.resources) {
-					buffer_infos.push_back(BufferInfo(data.addresses.at(resource)));
-				}
-				break;
-			case BindingKind::FlattenedSrt:
-				buffer_infos.push_back(BufferInfo(data.flattened_srt));
-				break;
-			case BindingKind::UserData: buffer_infos.push_back(BufferInfo(data.user_data)); break;
-			case BindingKind::Gds: buffer_infos.push_back(BufferInfo(data.gds)); break;
-			case BindingKind::Samplers:
-				for (const auto resource: binding.resources) {
-					const auto sampler = data.samplers.at(resource);
-					EXIT_IF(sampler == nullptr);
-					image_infos.push_back({sampler, nullptr, vk::ImageLayout::eUndefined});
-				}
-				break;
-			default: {
-				for (const auto resource: binding.resources) {
-					const auto& texture = data.images.at(resource);
-					image_infos.push_back(MakeImageInfo(texture, image_occurrences.at(resource)++));
-				}
-				break;
-			}
-		}
-		if (buffer_infos.size() != buffer_start) {
-			write.pBufferInfo = buffer_infos.data() + buffer_start;
-		}
-		if (image_infos.size() != image_start) {
-			write.pImageInfo = image_infos.data() + image_start;
-		}
-		writes.push_back(write);
-	}
-	for (uint32_t i = 0; i < data.images.size(); i++) {
-		const auto expected = data.images[i].mip_views.empty()
-		                          ? 1u
-		                          : static_cast<uint32_t>(data.images[i].mip_views.size());
-		EXIT_IF(image_occurrences[i] != expected);
-	}
-	m_graphics.device.updateDescriptorSets(static_cast<uint32_t>(writes.size()), writes.data(), 0,
-	                                       nullptr);
-	return *set;
-}
-
-vk::DescriptorSetLayout
-DescriptorCache::GetDescriptorSetLayout(Stage stage, const ShaderRecompiler::IR::Program& program) {
-	Common::LockGuard lock(m_mutex);
-	return GetDescriptorSetLayoutInternal(stage, program);
-}
-
-} // namespace Libs::Graphics

src/graphics/host_gpu/renderer/pipeline/descriptorCache.h

@@ -1,116 +0,0 @@
-#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DESCRIPTORCACHE_H_
-#define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DESCRIPTORCACHE_H_
-
-#include "common/abi.h"
-#include "common/assert.h"
-#include "common/common.h"
-#include "common/threads.h"
-#include "graphics/host_gpu/graphicContext.h"
-#include "graphics/host_gpu/renderer/cache/textureCache.h"
-#include "graphics/host_gpu/vulkanCommon.h"
-#include "graphics/shader/shaderBindings.h"
-
-#include <map>
-#include <memory>
-#include <span>
-#include <unordered_map>
-#include <vector>
-
-namespace Libs::Graphics {
-
-namespace ShaderRecompiler::IR {
-struct DescriptorValue;
-struct ImageResource;
-struct Program;
-struct ResourceSnapshot;
-} // namespace ShaderRecompiler::IR
-
-class CommandBuffer;
-struct DescriptorCacheTestAccess;
-struct ShaderStageRuntime;
-
-struct VulkanDescriptorSet {
-	vk::DescriptorSet       set     = nullptr;
-	vk::DescriptorSetLayout layout  = nullptr;
-	int                     pool_id = -1;
-};
-
-struct BufferView {
-	std::shared_ptr<void> owner;
-	vk::Buffer            buffer = nullptr;
-	vk::DeviceSize        offset = 0;
-	vk::DeviceSize        range  = VK_WHOLE_SIZE;
-};
-
-class DescriptorCache {
-public:
-	enum class Stage { Unknown, Vertex, Pixel, Compute };
-
-	struct TextureBinding {
-		ImageId                    image_id;
-		vk::ImageView              image_view = nullptr;
-		TextureCache::ImageDesc    desc;
-		vk::ImageLayout            layout = vk::ImageLayout::eUndefined;
-		std::vector<vk::ImageView> mip_views;
-	};
-
-	struct NativeDescriptors {
-		std::vector<BufferView>     buffers;
-		std::vector<TextureBinding> images;
-		std::vector<vk::Sampler>    samplers;
-		std::vector<BufferView>     addresses;
-		BufferView                  gds;
-		BufferView                  flattened_srt;
-		BufferView                  user_data;
-	};
-
-	struct PreparedBindings {
-		std::shared_ptr<const ShaderRecompiler::IR::Program>          program;
-		std::shared_ptr<const ShaderRecompiler::IR::ResourceSnapshot> snapshot;
-		NativeDescriptors                                             resources;
-		std::vector<uint32_t>                                         flattened_srt;
-		std::vector<uint32_t>                                         user_data;
-		vk::ShaderStageFlags                                          shader_stage;
-		Stage                                                         stage     = Stage::Unknown;
-		bool                                                          committed = false;
-	};
-
-	explicit DescriptorCache(GraphicContext& graphics): m_graphics(graphics) {
-		EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
-	}
-	~DescriptorCache();
-	KYTY_CLASS_NO_COPY(DescriptorCache);
-
-	vk::DescriptorSetLayout GetDescriptorSetLayout(Stage                                stage,
-	                                               const ShaderRecompiler::IR::Program& program);
-	void                    Recycle(VulkanDescriptorSet& set);
-	VulkanDescriptorSet&    GetDescriptor(Stage stage, const ShaderRecompiler::IR::Program& program,
-	                                      const NativeDescriptors& descriptors);
-
-private:
-	friend struct DescriptorCacheTestAccess;
-
-	struct Pool {
-		vk::DescriptorPool pool           = nullptr;
-		int                next_free_pool = -1;
-	};
-
-	static vk::DescriptorImageInfo MakeImageInfo(const TextureBinding& texture,
-	                                             uint32_t              element = 0);
-	void                           CreatePool();
-	VulkanDescriptorSet* Allocate(Stage stage, const ShaderRecompiler::IR::Program& program);
-	vk::DescriptorSetLayout
-	GetDescriptorSetLayoutInternal(Stage stage, const ShaderRecompiler::IR::Program& program);
-
-	GraphicContext&   m_graphics;
-	Common::Mutex     m_mutex;
-	std::vector<Pool> m_pools;
-	int               m_first_free_pool = -1;
-	std::unordered_map<vk::DescriptorSetLayout, std::vector<VulkanDescriptorSet*>>
-	                                                         m_free_sets_by_layout;
-	std::map<std::vector<uint32_t>, vk::DescriptorSetLayout> m_descriptor_set_layouts;
-};
-
-} // namespace Libs::Graphics
-
-#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DESCRIPTORCACHE_H_

src/graphics/host_gpu/renderer/pipeline/descriptorHeap.cpp

@@ -0,0 +1,94 @@
+#include "graphics/host_gpu/renderer/pipeline/descriptorHeap.h"
+
+#include "common/assert.h"
+#include "common/profiler.h"
+#include "graphics/host_gpu/graphicContext.h"
+#include "graphics/host_gpu/renderer/masterSemaphore.h"
+
+namespace Libs::Graphics {
+namespace {
+
+constexpr uint32_t   DescriptorHeapCount = 1024;
+constexpr std::array DescriptorPoolSizes = {
+    vk::DescriptorPoolSize {vk::DescriptorType::eStorageBuffer, 8192},
+    vk::DescriptorPoolSize {vk::DescriptorType::eSampledImage, 8192},
+    vk::DescriptorPoolSize {vk::DescriptorType::eStorageImage, 1024},
+    vk::DescriptorPoolSize {vk::DescriptorType::eSampler, 1024},
+};
+
+} // namespace
+
+DescriptorHeap::DescriptorHeap(GraphicContext& graphics, MasterSemaphore& master_semaphore)
+    : m_graphics(graphics), m_master_semaphore(master_semaphore) {
+	CreatePool();
+}
+
+DescriptorHeap::~DescriptorHeap() {
+	m_graphics.device.destroyDescriptorPool(m_current_pool, nullptr);
+	for (const auto& [pool, tick]: m_pending_pools) {
+		m_master_semaphore.Wait(tick);
+		m_graphics.device.destroyDescriptorPool(pool, nullptr);
+	}
+}
+
+vk::DescriptorSet DescriptorHeap::Commit(vk::DescriptorSetLayout layout) {
+	KYTY_PROFILER_FUNCTION();
+	EXIT_IF(layout == nullptr);
+
+	auto& batch = m_sets[layout];
+	if (batch.size != 0) {
+		return batch.sets[--batch.size];
+	}
+	if (Allocate(layout, batch)) {
+		return batch.sets[--batch.size];
+	}
+
+	m_pending_pools.emplace_back(m_current_pool, m_master_semaphore.CurrentTick());
+	if (const auto& [pool, tick] = m_pending_pools.front(); m_master_semaphore.IsFree(tick)) {
+		m_current_pool = pool;
+		m_pending_pools.pop_front();
+		EXIT_IF(m_graphics.device.resetDescriptorPool(m_current_pool, {}) != vk::Result::eSuccess);
+	} else {
+		CreatePool();
+	}
+
+	m_sets.clear();
+	auto& fresh_batch = m_sets[layout];
+	EXIT_IF(!Allocate(layout, fresh_batch));
+	return fresh_batch.sets[--fresh_batch.size];
+}
+
+bool DescriptorHeap::Allocate(vk::DescriptorSetLayout layout, Batch& batch) {
+	std::array<vk::DescriptorSetLayout, DescriptorSetBatch> layouts;
+	layouts.fill(layout);
+
+	vk::DescriptorSetAllocateInfo allocate {};
+	allocate.descriptorPool = m_current_pool;
+	allocate.pSetLayouts    = layouts.data();
+
+	for (;;) {
+		allocate.descriptorSetCount = batch.allocation;
+		const auto result = m_graphics.device.allocateDescriptorSets(&allocate, batch.sets.data());
+		if (result == vk::Result::eSuccess) {
+			batch.size = batch.allocation;
+			return true;
+		}
+		EXIT_IF(result != vk::Result::eErrorOutOfPoolMemory &&
+		        result != vk::Result::eErrorFragmentedPool);
+		if (batch.allocation == 1) {
+			return false;
+		}
+		batch.allocation /= 2;
+	}
+}
+
+void DescriptorHeap::CreatePool() {
+	vk::DescriptorPoolCreateInfo create {};
+	create.maxSets       = DescriptorHeapCount;
+	create.poolSizeCount = static_cast<uint32_t>(DescriptorPoolSizes.size());
+	create.pPoolSizes    = DescriptorPoolSizes.data();
+	EXIT_IF(m_graphics.device.createDescriptorPool(&create, nullptr, &m_current_pool) !=
+	        vk::Result::eSuccess);
+}
+
+} // namespace Libs::Graphics

src/graphics/host_gpu/renderer/pipeline/descriptorHeap.h

@@ -0,0 +1,45 @@
+#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DESCRIPTORHEAP_H_
+#define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DESCRIPTORHEAP_H_
+
+#include "common/common.h"
+#include "graphics/host_gpu/vulkanCommon.h"
+
+#include <array>
+#include <deque>
+#include <unordered_map>
+
+namespace Libs::Graphics {
+
+struct GraphicContext;
+class MasterSemaphore;
+
+class DescriptorHeap {
+public:
+	DescriptorHeap(GraphicContext& graphics, MasterSemaphore& master_semaphore);
+	~DescriptorHeap();
+	KYTY_CLASS_NO_COPY(DescriptorHeap);
+
+	[[nodiscard]] vk::DescriptorSet Commit(vk::DescriptorSetLayout layout);
+
+private:
+	static constexpr uint32_t DescriptorSetBatch = 32;
+
+	struct Batch {
+		std::array<vk::DescriptorSet, DescriptorSetBatch> sets {};
+		uint32_t                                          size       = 0;
+		uint32_t                                          allocation = DescriptorSetBatch;
+	};
+
+	[[nodiscard]] bool Allocate(vk::DescriptorSetLayout layout, Batch& batch);
+	void               CreatePool();
+
+	GraphicContext&                                     m_graphics;
+	MasterSemaphore&                                    m_master_semaphore;
+	vk::DescriptorPool                                  m_current_pool = nullptr;
+	std::deque<std::pair<vk::DescriptorPool, uint64_t>> m_pending_pools;
+	std::unordered_map<vk::DescriptorSetLayout, Batch>  m_sets;
+};
+
+} // namespace Libs::Graphics
+
+#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DESCRIPTORHEAP_H_

src/graphics/host_gpu/renderer/pipeline/descriptors.h

@@ -2,6 +2,8 @@
 #define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DESCRIPTORS_H_
 
 #include "common/assert.h"
+#include "graphics/host_gpu/renderer/cache/bufferCache.h"
+#include "graphics/host_gpu/renderer/cache/textureCache.h"
 #include "graphics/host_gpu/renderer/image/image.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 #include "graphics/shader/recompiler/ir/ShaderIR.h"
@@ -10,9 +12,50 @@
 #include <cstdint>
 #include <cstring>
 #include <type_traits>
+#include <vector>
 
 namespace Libs::Graphics {
 
+namespace ShaderRecompiler::IR {
+struct ResourceSnapshot;
+}
+
+struct TextureBinding {
+	ImageId                    image_id;
+	vk::ImageView              image_view = nullptr;
+	TextureCache::ImageDesc    desc;
+	vk::ImageLayout            layout = vk::ImageLayout::eUndefined;
+	std::vector<vk::ImageView> mip_views;
+};
+
+struct PreparedBindings {
+	struct BufferSource {
+		uint64_t address = 0;
+		uint64_t size    = 0;
+		BufferId id;
+	};
+
+	const ShaderRecompiler::IR::CompiledShaderInfo* program  = nullptr;
+	const ShaderRecompiler::IR::ResourceSnapshot* snapshot = nullptr;
+	// Keep the resolved guest range through cache preparation; only the host buffer ID may
+	// become stale and need resolving again when bindings are rebound.
+	std::vector<BufferSource>             buffer_sources;
+	std::vector<vk::DescriptorBufferInfo> buffers;
+	std::vector<TextureBinding>           images;
+	std::vector<vk::Sampler>              samplers;
+	vk::DescriptorBufferInfo              gds {nullptr, 0, VK_WHOLE_SIZE};
+	vk::DescriptorBufferInfo              flattened_srt;
+	vk::DescriptorBufferInfo              shader_data_buffer;
+	std::vector<uint32_t>                 shader_data;
+};
+
+[[nodiscard]] vk::DescriptorType
+NativeDescriptorType(ShaderRecompiler::IR::DescriptorBindingKind kind);
+[[nodiscard]] uint32_t
+NativeDescriptorCount(const ShaderRecompiler::IR::DescriptorBinding& binding);
+[[nodiscard]] vk::DescriptorImageInfo MakeImageInfo(const TextureBinding& texture,
+                                                    uint32_t              element = 0);
+
 template <typename T>
 [[nodiscard]] T DecodeNativeDescriptor(const ShaderRecompiler::IR::DescriptorValue& value) {
 	static_assert(std::is_trivially_copyable_v<T>);
@@ -23,20 +66,8 @@ template <typename T>
 	return result;
 }
 
-struct TargetTextureViewInfo {
-	vk::ImageViewType type        = static_cast<vk::ImageViewType>(VK_IMAGE_VIEW_TYPE_MAX_ENUM);
-	uint32_t          base_layer  = 0;
-	uint32_t          layer_count = 0;
-};
-
-[[nodiscard]] TargetTextureViewInfo
-ResolveTargetTextureView(const ShaderRecompiler::IR::ImageResource& resource,
-                         Prospero::ImageType type, uint32_t base_layer, uint32_t image_layers);
-
-[[nodiscard]] bool IsSupportedDepthTargetDescriptor(const ShaderTextureResource& descriptor,
-                                                    const Image&                 image);
 [[nodiscard]] bool IsSupportedDepthTextureEncoding(const ShaderTextureResource& descriptor,
-                                                   const Image&                 image);
+                                                   bool r128 = false);
 [[nodiscard]] bool
 IsSupportedSampledVideoOutView(const ShaderRecompiler::IR::ImageResource& resource,
                                const ShaderTextureResource& descriptor, const Image& image);

src/graphics/host_gpu/renderer/pipeline/pipelineCache.cpp

@@ -1,6 +1,8 @@
 #include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
 
 #include "common/assert.h"
+#include "common/emulatorConfig.h"
+#include "common/file.h"
 #include "common/logging/log.h"
 #include "common/profiler.h"
 #include "graphics/guest_gpu/hardwareContext.h"
@@ -10,115 +12,680 @@
 #include "graphics/host_gpu/renderer/image/imageView.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
+#include "graphics/shader/recompiler/ShaderRecompiler.h"
+#include "graphics/shader/shaderCompiler.h"
+#include "kernel/memory.h"
+#include "kytyGitVersion.h"
+#include "loader/systemContent.h"
 
+#include <algorithm>
+#include <array>
 #include <atomic>
+#include <cctype>
+#include <cstdio>
 #include <cstring>
+#include <fmt/format.h>
+#include <limits>
 #include <span>
+#include <spirv-tools/libspirv.hpp>
+#include <string_view>
+#include <tuple>
 #include <utility>
+#include <vector>
+#include <xxhash.h>
 
 namespace Libs::Graphics {
 
 namespace {
 
-void NormalizeStaticParamsForDynamicState(PipelineStaticParameters& static_params) {
-	static_params.viewport_scale[0]  = 0.5f;
-	static_params.viewport_scale[1]  = 0.5f;
-	static_params.viewport_scale[2]  = 1.0f;
-	static_params.viewport_offset[0] = 0.5f;
-	static_params.viewport_offset[1] = 0.5f;
-	static_params.viewport_offset[2] = 0.0f;
+vk::PolygonMode ResolvePolygonMode(const HW::ModeControl& mode, bool cull_front, bool cull_back) {
+	// CxPrimitiveSetup::PolygonMode disables both per-face modes when it is zero.
+	if (mode.poly_mode == 0) {
+		return vk::PolygonMode::eFill;
+	}
+	EXIT_NOT_IMPLEMENTED(mode.poly_mode != 1);
+	if (cull_front && cull_back) {
+		return vk::PolygonMode::eFill;
+	}
+	if (!cull_front && !cull_back && mode.polymode_front_ptype != mode.polymode_back_ptype) {
+		EXIT("Pipeline: different polygon modes for two visible faces are unsupported\n");
+	}
+	// Vulkan has one polygon mode. A culled face does not constrain that mode.
+	const auto polygon_mode = cull_front ? mode.polymode_back_ptype : mode.polymode_front_ptype;
+	switch (polygon_mode) {
+		case 0: return vk::PolygonMode::ePoint;
+		case 1: return vk::PolygonMode::eLine;
+		case 2: return vk::PolygonMode::eFill;
+		default: EXIT("Pipeline: invalid polygon mode %u\n", polygon_mode);
+	}
+}
 
-	static_params.scissor_ltrb[0] = 0;
-	static_params.scissor_ltrb[1] = 0;
-	static_params.scissor_ltrb[2] = 1;
-	static_params.scissor_ltrb[3] = 1;
+std::string DriverCacheSignature(const vk::PhysicalDeviceProperties& properties) {
+	constexpr char hex[] = "0123456789abcdef";
+	std::string    uuid(VK_UUID_SIZE * 2, '0');
+	for (size_t i = 0; i < VK_UUID_SIZE; i++) {
+		uuid[i * 2]     = hex[properties.pipelineCacheUUID[i] >> 4u];
+		uuid[i * 2 + 1] = hex[properties.pipelineCacheUUID[i] & 0xfu];
+	}
+	return fmt::format("KytyPC1:{}:{:08x}:{:08x}:{:08x}:{}\n", KYTY_GIT_REVISION,
+	                   properties.vendorID, properties.deviceID, properties.driverVersion, uuid);
+}
+
+std::string PipelineCacheTitleId() {
+	std::string title_id;
+	if ((!Loader::SystemContentParamSfoGetString("TITLE_ID", &title_id) || title_id.empty()) &&
+	    (!Loader::SystemContentParamSfoGetString("CONTENT_ID", &title_id) || title_id.empty())) {
+		return {};
+	}
+	if (!std::ranges::all_of(title_id, [](unsigned char c) {
+		    return std::isalnum(c) != 0 || c == '-' || c == '_';
+	    })) {
+		return {};
+	}
+	return title_id;
+}
+
+template <typename... Args>
+void PipelineCacheLog(fmt::format_string<Args...> format, Args&&... args) {
+	auto message = fmt::format(format, std::forward<Args>(args)...);
+	message += '\n';
+	if (Log::GetDirection() != Log::Direction::Console) {
+		std::fwrite(message.data(), 1, message.size(), stdout);
+		std::fflush(stdout);
+	}
+	Log::Write(message);
+	Log::Flush();
+}
+
+bool ReadShaderGuestMemory(void*, uint64_t address, uint32_t* value) {
+	return value != nullptr &&
+	       Libs::LibKernel::Memory::TryReadGpuCleanBacking(address, value, sizeof(*value));
+}
+
+void DumpShaderSpirv(const char* stage_name, uint64_t shader_hash,
+                     const std::vector<uint32_t>& spirv) {
+	if (!Config::GraphicsDebugDumpEnabled()) {
+		return;
+	}
+	static std::atomic_int id = 0;
+	const auto path = Config::GetShaderLogFolder() / fmt::format("{:04d}_new_shader_{}_{:016x}.spv",
+	                                                             id++, stage_name, shader_hash);
+	Common::File::CreateDirectories(path.parent_path());
+	Common::File file(path);
+	if (file.IsInvalid()) {
+		const auto path_text = Common::PathToString(path);
+		LOGF_COLOR(Log::Color::BrightRed, "Can't create file: %s\n", path_text.c_str());
+		return;
+	}
+	file.Write(spirv.data(), spirv.size() * sizeof(uint32_t));
+}
+
+void DumpShaderOriginal(const char* stage_name, uint64_t shader_hash,
+                        std::span<const uint32_t> code, const std::string& decoded_dump) {
+	if (!Config::GraphicsDebugDumpEnabled()) {
+		return;
+	}
+	EXIT_IF(code.empty());
+	static std::atomic_int id = 0;
+	const auto base = Config::GetShaderLogFolder() / "original" /
+	                  fmt::format("{:04d}_new_shader_{}_{:016x}", id++, stage_name, shader_hash);
+	Common::File::CreateDirectories(base.parent_path());
+	for (const auto& [suffix, data, size]: {
+	         std::tuple {".bin", static_cast<const void*>(code.data()), code.size_bytes()},
+	         std::tuple {".rdna2", static_cast<const void*>(decoded_dump.data()),
+	                     decoded_dump.size()},
+	     }) {
+		if (size == 0) {
+			continue;
+		}
+		auto path = base;
+		path += suffix;
+		Common::File file(path);
+		if (file.IsInvalid()) {
+			const auto path_text = Common::PathToString(path);
+			LOGF_COLOR(Log::Color::BrightRed, "Can't create file: %s\n", path_text.c_str());
+		} else {
+			file.Write(data, size);
+		}
+	}
+}
+
+bool ValidateShaderSpirv(const char* label, uint64_t shader_hash,
+                         const std::vector<uint32_t>& spirv) {
+	if (!Config::ShaderValidationEnabled()) {
+		return true;
+	}
+	spvtools::SpirvTools tools(SPV_ENV_VULKAN_1_3);
+	std::string          messages;
+	tools.SetMessageConsumer([&messages](spv_message_level_t, const char*,
+	                                     const spv_position_t& position, const char* message) {
+		messages += fmt::format("{}: {} ({}) {}\n", static_cast<int>(position.line),
+		                        static_cast<int>(position.column), static_cast<int>(position.index),
+		                        message);
+	});
+	if (tools.Validate(spirv)) {
+		return true;
+	}
+	spvtools::SpirvTools disassembler(SPV_ENV_VULKAN_1_2);
+	std::string          text;
+	disassembler.Disassemble(spirv, &text,
+	                         static_cast<uint32_t>(SPV_BINARY_TO_TEXT_OPTION_NO_HEADER) |
+	                             static_cast<uint32_t>(SPV_BINARY_TO_TEXT_OPTION_FRIENDLY_NAMES) |
+	                             static_cast<uint32_t>(SPV_BINARY_TO_TEXT_OPTION_COMMENT) |
+	                             static_cast<uint32_t>(SPV_BINARY_TO_TEXT_OPTION_INDENT) |
+	                             static_cast<uint32_t>(SPV_BINARY_TO_TEXT_OPTION_COLOR));
+	LOGF_COLOR(Log::Color::BrightRed, "%s SPIR-V validation failed hash=0x%016" PRIx64 ":\n%s",
+	           label, shader_hash, messages.c_str());
+	LOGF("%s\n", text.c_str());
+	return false;
 }
 
 } // namespace
 
+struct PipelineCache::ProgramCache {
+	struct ProgramKey {
+		ShaderType            stage           = ShaderType::Unknown;
+		uint64_t              hash            = 0;
+		uint32_t              user_data_count = 0;
+		uint32_t              code_size       = 0;
+		std::vector<uint32_t> static_state;
+
+		bool operator==(const ProgramKey&) const = default;
+	};
+
+	struct Permutation {
+		ShaderRecompiler::IR::ResourceSpecialization specialization;
+		ShaderRecompiler::IR::CompiledShaderInfo     program;
+		ShaderProgram                                handle;
+	};
+
+	struct SourceEntry {
+		explicit SourceEntry(ShaderRecompiler::IR::ResourcePlan plan)
+		    : resource_plan(std::move(plan)) {
+			permutations.reserve(8);
+		}
+
+		ShaderRecompiler::IR::ResourcePlan resource_plan;
+		std::vector<Permutation>           permutations;
+	};
+
+	struct ProgramKeyHash {
+		std::size_t operator()(const ProgramKey& key) const {
+			std::size_t hash = static_cast<std::size_t>(key.stage);
+			PipelineKeyHash::Mix(hash, static_cast<std::size_t>(key.hash));
+			if constexpr (sizeof(std::size_t) < sizeof(uint64_t)) {
+				PipelineKeyHash::Mix(hash, static_cast<std::size_t>(key.hash >> 32u));
+			}
+			PipelineKeyHash::Mix(hash, key.user_data_count);
+			PipelineKeyHash::Mix(hash, key.code_size);
+			PipelineKeyHash::Mix(hash, key.static_state.size());
+			// Bucket same-shape static variants by source. ProgramKey equality performs the one
+			// exact state comparison needed on a stable hit without hashing up to 429 words first.
+			return hash;
+		}
+	};
+
+	static constexpr std::size_t MaxStaticKeyWords = 13 + ShaderVertexInputInfo::RES_MAX * 13;
+
+	Permutation CompilePermutation(const ShaderParams&                          params,
+	                               const ShaderRecompiler::CompileOptions&      options,
+	                               ShaderRecompiler::TranslateResult            translated,
+	                               ShaderRecompiler::IR::ResourceSpecialization specialization,
+	                               uint32_t push_data_start_dword) {
+		const char* stage_name = nullptr;
+		switch (options.stage) {
+			case ShaderType::Vertex: stage_name = "vs"; break;
+			case ShaderType::Mesh: stage_name = "ms"; break;
+			case ShaderType::Pixel: stage_name = "ps"; break;
+			case ShaderType::Compute: stage_name = "cs"; break;
+			default: EXIT("invalid pipeline shader stage\n");
+		}
+		auto result = ShaderRecompiler::CompileProgram(std::move(translated), options,
+		                                               specialization, push_data_start_dword);
+		DumpShaderOriginal(stage_name, options.shader_hash, params.code, result.decoded_dump);
+		if (!ValidateShaderSpirv(options.dump_label, options.shader_hash, result.spirv)) {
+			DumpShaderSpirv(stage_name, options.shader_hash, result.spirv);
+			EXIT("%s failed hash=0x%016" PRIx64 ": SPIR-V validation failed\n", options.dump_label,
+			     options.shader_hash);
+		}
+		DumpShaderSpirv(stage_name, options.shader_hash, result.spirv);
+
+		vk::ShaderModuleCreateInfo create_info {};
+		create_info.codeSize    = result.spirv.size() * sizeof(uint32_t);
+		create_info.pCode       = result.spirv.data();
+		vk::ShaderModule module = nullptr;
+		RequireVulkanSuccess(device.createShaderModule(&create_info, nullptr, &module),
+		                     "create recompiled shader module");
+		EXIT_IF(module == nullptr);
+		if (options.dump_ir) {
+			if (!options.early_dump) {
+				LOGF("%s decoded RDNA2:\n%s", options.dump_label, result.decoded_dump.c_str());
+				LOGF("%s IR:\n%s", options.dump_label, result.ir_dump.c_str());
+			}
+			LOGF("%s SPIR-V words=%" PRIu64 " wave_size=%u\n", options.dump_label,
+			     static_cast<uint64_t>(result.spirv.size()), options.wave_size);
+		}
+		return {
+		    .specialization = std::move(specialization),
+		    .program        = std::move(result.program).TakeCompiledInfo(),
+		    .handle         = {.id = ++next_shader_id, .module = module},
+		};
+	}
+
+	template <typename InputInfo>
+	ShaderProgram Get(const ShaderParams& params, InputInfo& input_info,
+	                  uint32_t& push_data_cursor) {
+		ShaderType stage;
+		if constexpr (std::is_same_v<InputInfo, ShaderVertexInputInfo>) {
+			stage = input_info.mesh.threads_num[0] != 0 ? ShaderType::Mesh : ShaderType::Vertex;
+		} else if constexpr (std::is_same_v<InputInfo, ShaderPixelInputInfo>) {
+			stage = ShaderType::Pixel;
+		} else {
+			static_assert(std::is_same_v<InputInfo, ShaderComputeInputInfo>);
+			stage = ShaderType::Compute;
+		}
+
+		lookup_key.stage           = stage;
+		lookup_key.hash            = params.hash;
+		lookup_key.user_data_count = static_cast<uint32_t>(params.user_data.size());
+		lookup_key.code_size       = static_cast<uint32_t>(params.code.size());
+		BuildStageStaticKey(input_info, lookup_key.static_state);
+		auto                                         entry = programs.find(lookup_key);
+		ShaderRecompiler::IR::ResourceSnapshot       resources;
+		ShaderRecompiler::IR::ResourceSpecialization specialization;
+		const ShaderRecompiler::IR::SrtRuntime       runtime {
+		    .user_data                  = params.user_data,
+		    .shader_base                = params.Base(),
+		    .read_specialization_memory = ReadShaderGuestMemory,
+		};
+		if (entry != programs.end()) {
+			EXIT_IF(!ShaderRecompiler::IR::MaterializeResources(
+			    entry->second.resource_plan, runtime, resources, specialization));
+			if (const auto permutation = std::ranges::find_if(
+			        entry->second.permutations, [&](const Permutation& candidate) {
+				        const auto& layout = candidate.program.bindings;
+				        return layout.push_data_start_dword ==
+				                   ShaderRecompiler::IR::PushData::StartFor(
+				                       push_data_cursor, layout.ShaderDataDwords()) &&
+				               candidate.specialization == specialization;
+			        });
+			    permutation != entry->second.permutations.end()) {
+				input_info.stage = {.program   = &permutation->program,
+				                    .resources = std::move(resources)};
+				permutation->program.bindings.AdvancePushData(push_data_cursor);
+				return permutation->handle;
+			}
+		}
+
+		ShaderStageInputInfo stage_input {};
+		if constexpr (std::is_same_v<InputInfo, ShaderVertexInputInfo>) {
+			stage_input.vertex = &input_info;
+		} else if constexpr (std::is_same_v<InputInfo, ShaderPixelInputInfo>) {
+			stage_input.pixel = &input_info;
+		} else {
+			stage_input.compute = &input_info;
+		}
+		const char* label = nullptr;
+		switch (stage) {
+			case ShaderType::Vertex: label = "ShaderRecompiler VS"; break;
+			case ShaderType::Mesh: label = "ShaderRecompiler MS"; break;
+			case ShaderType::Pixel: label = "ShaderRecompiler PS"; break;
+			case ShaderType::Compute: label = "ShaderRecompiler CS"; break;
+			default: EXIT("invalid pipeline shader stage\n");
+		}
+		ShaderRecompiler::CompileOptions options;
+		options.stage       = stage;
+		options.shader_hash = params.hash;
+		options.user_data   = params.user_data;
+		options.back_code      = params.back_code;
+		options.dump_ir     = Config::GetShaderLogDirection() != Config::ShaderLogDirection::Silent;
+		options.early_dump  = options.dump_ir;
+		options.dump_label  = label;
+		options.input_info  = stage_input;
+		options.scratch_dwords = input_info.scratch_size_dwords;
+		if constexpr (std::is_same_v<InputInfo, ShaderVertexInputInfo>) {
+			options.user_data_base = 8;
+			if (stage == ShaderType::Mesh) {
+				options.user_data_base = 0;
+				options.wave_size      = input_info.mesh.wave_size;
+				options.scratch_dwords = input_info.mesh.scratch_size_dwords;
+			}
+		} else if constexpr (std::is_same_v<InputInfo, ShaderComputeInputInfo>) {
+			options.wave_size = input_info.wave_size;
+		}
+		auto translated = ShaderRecompiler::TranslateProgram(params.code, options);
+		if (entry == programs.end()) {
+			auto resource_plan = ShaderRecompiler::IR::ExtractResourcePlan(translated.program);
+			EXIT_IF(!ShaderRecompiler::IR::MaterializeResources(resource_plan, runtime, resources,
+			                                                    specialization));
+			entry = programs.try_emplace(lookup_key, std::move(resource_plan)).first;
+		}
+		entry->second.permutations.push_back(CompilePermutation(
+		    params, options, std::move(translated), std::move(specialization), push_data_cursor));
+		const auto& permutation = entry->second.permutations.back();
+		input_info.stage = {.program = &permutation.program, .resources = std::move(resources)};
+		permutation.program.bindings.AdvancePushData(push_data_cursor);
+
+		std::array<size_t, static_cast<size_t>(ShaderType::Mesh) + 1> counts {};
+		for (const auto& [key, source]: programs) {
+			counts[static_cast<size_t>(key.stage)] += source.permutations.size();
+		}
+		// Guest geometry shaders are compiled through the host mesh stage.
+		std::printf("Shaders: VS %zu | PS %zu | CS %zu | GS %zu\n",
+		            counts[static_cast<size_t>(ShaderType::Vertex)],
+		            counts[static_cast<size_t>(ShaderType::Pixel)],
+		            counts[static_cast<size_t>(ShaderType::Compute)],
+		            counts[static_cast<size_t>(ShaderType::Mesh)]);
+		return permutation.handle;
+	}
+
+	explicit ProgramCache(vk::Device device): device(device) {
+		lookup_key.static_state.reserve(MaxStaticKeyWords);
+	}
+	~ProgramCache() {
+		for (const auto& [key, entry]: programs) {
+			(void)key;
+			for (const auto& permutation: entry.permutations) {
+				device.destroyShaderModule(permutation.handle.module, nullptr);
+			}
+		}
+	}
+
+	std::unordered_map<ProgramKey, SourceEntry, ProgramKeyHash> programs;
+	ProgramKey                                                  lookup_key;
+	vk::Device                                                  device;
+	uint64_t                                                    next_shader_id = 0;
+};
+
+PipelineCache::PipelineCache(GraphicContext& graphics)
+    : m_graphics(graphics), m_program_cache(std::make_unique<ProgramCache>(graphics.device)) {
+	EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
+	InitializeDriverCache();
+}
+
 PipelineCache::~PipelineCache() {
+	Save();
 	auto destroy = [this](const auto& pipelines) {
 		for (const auto& [key, pipeline]: pipelines) {
 			(void)key;
 			m_graphics.device.destroyPipeline(pipeline->pipeline, nullptr);
 			m_graphics.device.destroyPipelineLayout(pipeline->pipeline_layout, nullptr);
+			m_graphics.device.destroyDescriptorSetLayout(pipeline->descriptor_set_layout, nullptr);
 		}
 	};
 	destroy(m_graphics_pipelines);
 	destroy(m_compute_pipelines);
+	if (m_driver_cache != nullptr) {
+		m_graphics.device.destroyPipelineCache(m_driver_cache, nullptr);
+	}
+}
+
+void PipelineCache::InitializeDriverCache() {
+	const auto title_id = PipelineCacheTitleId();
+	if (title_id.empty()) {
+		return;
+	}
+	if (KYTY_BUILD != KYTY_BUILD_RELEASE) {
+		PipelineCacheLog("Vulkan pipeline cache: disabled (non-Release build)");
+		return;
+	}
+	const std::string_view git_hash     = KYTY_GIT_HASH;
+	const std::string_view git_revision = KYTY_GIT_REVISION;
+	if (git_hash == "unknown" || git_revision == "unknown") {
+		PipelineCacheLog("Vulkan pipeline cache: disabled (unknown git revision)");
+		return;
+	}
+	if (git_hash.ends_with("-dirty")) {
+		PipelineCacheLog("Vulkan pipeline cache: disabled (dirty build)");
+		return;
+	}
+
+	m_driver_cache_path     = std::filesystem::path("_PipelineCache") / (title_id + ".bin");
+	const auto path         = Common::PathToString(m_driver_cache_path);
+	const bool cache_exists = Common::File::IsFileExisting(m_driver_cache_path);
+	if (cache_exists) {
+		PipelineCacheLog("Vulkan pipeline cache: loading {}", path);
+	} else {
+		PipelineCacheLog("Vulkan pipeline cache: initializing {}", path);
+	}
+	std::vector<uint8_t> initial_data;
+	if (cache_exists) {
+		Common::File file(m_driver_cache_path, Common::File::Mode::Read);
+		const auto   file_size = file.IsInvalid() ? 0 : file.Size();
+		const auto   signature = DriverCacheSignature(m_graphics.GetPhysicalDeviceProperties());
+		if (file_size >= signature.size() + sizeof(uint64_t) &&
+		    file_size <= std::numeric_limits<uint32_t>::max()) {
+			std::string cached_signature(signature.size(), '\0');
+			uint64_t    payload_hash = 0;
+			initial_data.resize(file_size - signature.size() - sizeof(payload_hash));
+			uint32_t signature_read = 0;
+			uint32_t hash_read      = 0;
+			uint32_t payload_read   = 0;
+			file.Read(cached_signature.data(), static_cast<uint32_t>(cached_signature.size()),
+			          &signature_read);
+			file.Read(&payload_hash, sizeof(payload_hash), &hash_read);
+			file.Read(initial_data.data(), static_cast<uint32_t>(initial_data.size()),
+			          &payload_read);
+			file.Close();
+			if (signature_read != cached_signature.size() || hash_read != sizeof(payload_hash) ||
+			    payload_read != initial_data.size() || cached_signature != signature ||
+			    XXH3_64bits(initial_data.data(), initial_data.size()) != payload_hash) {
+				initial_data.clear();
+				PipelineCacheLog(
+				    "Vulkan pipeline cache: invalidating {} (driver, emulator, or data mismatch)",
+				    path);
+			}
+		} else {
+			file.Close();
+			PipelineCacheLog("Vulkan pipeline cache: invalidating {} (invalid file size)", path);
+		}
+	}
+
+	vk::PipelineCacheCreateInfo create {};
+	create.initialDataSize = initial_data.size();
+	create.pInitialData    = initial_data.empty() ? nullptr : initial_data.data();
+	auto result = m_graphics.device.createPipelineCache(&create, nullptr, &m_driver_cache);
+	if (result != vk::Result::eSuccess && !initial_data.empty()) {
+		PipelineCacheLog("Vulkan pipeline cache: driver rejected {} ({}); starting empty", path,
+		                 vk::to_string(result));
+		initial_data.clear();
+		create.initialDataSize = 0;
+		create.pInitialData    = nullptr;
+		result = m_graphics.device.createPipelineCache(&create, nullptr, &m_driver_cache);
+	}
+	if (result != vk::Result::eSuccess) {
+		PipelineCacheLog("Vulkan pipeline cache: disabled ({})", vk::to_string(result));
+		m_driver_cache = nullptr;
+		return;
+	}
+	if (!initial_data.empty()) {
+		PipelineCacheLog("Vulkan pipeline cache: loaded {} bytes from {}", initial_data.size(),
+		                 path);
+	} else {
+		PipelineCacheLog("Vulkan pipeline cache: initialized empty");
+	}
+}
+
+void PipelineCache::Save() {
+	Common::LockGuard lock(m_mutex);
+	if (m_driver_cache == nullptr) {
+		return;
+	}
+
+	size_t               size = 0;
+	vk::Result           result;
+	std::vector<uint8_t> payload;
+	for (uint32_t attempt = 0; attempt < 3; attempt++) {
+		size   = 0;
+		result = m_graphics.device.getPipelineCacheData(m_driver_cache, &size, nullptr);
+		if (result != vk::Result::eSuccess || size == 0 ||
+		    size > std::numeric_limits<uint32_t>::max()) {
+			break;
+		}
+		payload.resize(size);
+		result = m_graphics.device.getPipelineCacheData(m_driver_cache, &size, payload.data());
+		if (result != vk::Result::eIncomplete) {
+			break;
+		}
+	}
+	if (result != vk::Result::eSuccess || size == 0 ||
+	    size > std::numeric_limits<uint32_t>::max()) {
+		PipelineCacheLog("Vulkan pipeline cache: save failed ({}, {} bytes)",
+		                 vk::to_string(result), size);
+		return;
+	}
+	payload.resize(size);
+	auto       prefix       = DriverCacheSignature(m_graphics.GetPhysicalDeviceProperties());
+	const auto payload_hash = XXH3_64bits(payload.data(), payload.size());
+	prefix.append(reinterpret_cast<const char*>(&payload_hash), sizeof(payload_hash));
+	if (!Common::File::CreateDirectories(m_driver_cache_path.parent_path())) {
+		PipelineCacheLog("Vulkan pipeline cache: failed to create cache directory");
+		return;
+	}
+	auto temp_path = m_driver_cache_path;
+	temp_path += ".tmp";
+	Common::File file;
+	uint32_t     prefix_written  = 0;
+	uint32_t     payload_written = 0;
+	if (file.Create(temp_path)) {
+		file.Write(prefix.data(), static_cast<uint32_t>(prefix.size()), &prefix_written);
+		file.Write(payload.data(), static_cast<uint32_t>(payload.size()), &payload_written);
+	}
+	const bool flushed = !file.IsInvalid() && file.Flush();
+	file.Close();
+	if (prefix_written != prefix.size() || payload_written != payload.size() || !flushed ||
+	    !Common::File::RenameFile(temp_path, m_driver_cache_path)) {
+		PipelineCacheLog("Vulkan pipeline cache: failed to write {}",
+		                 Common::PathToString(m_driver_cache_path));
+		return;
+	}
+	PipelineCacheLog("Vulkan pipeline cache: saved {} bytes to {}", payload.size(),
+	                 Common::PathToString(m_driver_cache_path));
+	m_graphics.device.destroyPipelineCache(m_driver_cache, nullptr);
+	m_driver_cache = nullptr;
+}
+
+PipelineCache::GraphicsPrograms PipelineCache::GetGraphicsPrograms(
+    const HW::VertexShaderInfo& vertex_regs, const HW::PixelShaderInfo& pixel_regs,
+    const HW::ShaderRegisters& sh, const HW::Context& context, const HW::UserConfig& user_config,
+    std::span<const Prospero::ColorComponentMapping, 8> target_export_mapping, bool pixel_active,
+    ShaderVertexInputInfo& vertex_info, ShaderPixelInputInfo& pixel_info) {
+	const auto vertex_params = PrepareProgram(vertex_regs, context, user_config, vertex_info);
+	const bool mesh_active   = vertex_info.mesh.threads_num[0] != 0;
+	if (mesh_active) {
+		EXIT_NOT_IMPLEMENTED(!m_graphics.mesh_shader_enabled);
+		auto& mesh              = vertex_info.mesh;
+		mesh.host_subgroup_size = m_graphics.subgroup_size;
+		const auto& limits      = m_graphics.mesh_shader_properties;
+		const auto  logical_threads =
+		    mesh.threads_num[0] * mesh.threads_num[1] * mesh.threads_num[2];
+		const auto host_threads = ((logical_threads + mesh.wave_size - 1u) / mesh.wave_size) *
+		                          std::min(mesh.host_subgroup_size, mesh.wave_size);
+		if (host_threads > limits.maxMeshWorkGroupInvocations ||
+		    host_threads > limits.maxMeshWorkGroupSize[0] ||
+		    mesh.max_vertices > limits.maxMeshOutputVertices ||
+		    mesh.max_primitives > limits.maxMeshOutputPrimitives ||
+		    mesh.lds_size_dwords * sizeof(uint32_t) > limits.maxMeshSharedMemorySize) {
+			EXIT("mesh shader exceeds host limits: threads=%u vertices=%u primitives=%u LDS=%u\n",
+			     host_threads, mesh.max_vertices, mesh.max_primitives, mesh.lds_size_dwords);
+		}
+	}
+	ShaderParams pixel_params;
+	if (pixel_active) {
+		pixel_params = PrepareProgram(pixel_regs, sh, target_export_mapping, pixel_info);
+	}
+	if (context.GetClipControl().clip_disable) {
+		const auto& viewport = context.GetScreenViewport().viewports[0];
+		const auto& limits   = m_graphics.GetPhysicalDeviceProperties().limits;
+		auto&       clip     = vertex_info.clip_space;
+		clip.scale[0]        = viewport.xscale;
+		clip.scale[1]        = viewport.yscale;
+		clip.offset[0]       = viewport.xoffset;
+		clip.offset[1]       = viewport.yoffset;
+		clip.half_extent[0] =
+		    static_cast<float>(std::min(limits.maxViewportDimensions[0], 16384u)) * 0.5f;
+		clip.half_extent[1] =
+		    static_cast<float>(std::min(limits.maxViewportDimensions[1], 16384u)) * 0.5f;
+		clip.enabled = true;
+	}
+	Common::LockGuard lock(m_mutex);
+	uint32_t          push_data_cursor =
+	    mesh_active ? ShaderRecompiler::IR::PushData::MeshDrawDwordCount : 0;
+	GraphicsPrograms  result;
+	if (pixel_active) {
+		result.pixel = m_program_cache->Get(pixel_params, pixel_info, push_data_cursor);
+	}
+	result.vertex = m_program_cache->Get(vertex_params, vertex_info, push_data_cursor);
+	return result;
+}
+
+ShaderProgram PipelineCache::GetComputeProgram(const HW::ComputeShaderInfo& regs,
+                                               const HW::ShaderRegisters&   sh,
+                                               ShaderComputeInputInfo&      input_info) {
+	input_info.host_subgroup_size = m_graphics.SupportsComputeWave64() ? 64u : 32u;
+	const auto        params      = PrepareProgram(regs, sh, input_info);
+	Common::LockGuard lock(m_mutex);
+	uint32_t          push_data_cursor = 0;
+	return m_program_cache->Get(params, input_info, push_data_cursor);
 }
 
 bool PipelineStaticParameters::operator==(const PipelineStaticParameters& other) const noexcept {
 	return std::memcmp(this, &other, sizeof(*this)) == 0;
 }
 
-PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
-    RenderColorInfo* colors, uint32_t color_count, RenderDepthInfo& depth,
-    ShaderVertexInputInfo& vs_input_info, RenderCommandBuffer& command,
-    ShaderPixelInputInfo* ps_input_info, vk::PrimitiveTopology topology,
-    bool primitive_restart_enable, bool ps_active, std::span<const uint32_t> vs_spirv,
-    std::span<const uint32_t> ps_spirv) {
+PipelineCache::Pipeline& PipelineCache::CreateGraphicsPipeline(
+    std::span<const RenderColorInfo> colors, const RenderDepthInfo& depth,
+    const ShaderVertexInputInfo& vs_input_info, CommandBuffer& command,
+    const ShaderPixelInputInfo* ps_input_info, vk::PrimitiveTopology topology,
+    bool primitive_restart_enable, const ShaderProgram& vertex_program,
+    const ShaderProgram& pixel_program) {
 	KYTY_PROFILER_BLOCK("PipelineCache::CreatePipeline(Gfx)", profiler::colors::DeepOrangeA200);
 
-	EXIT_IF(colors == nullptr);
-	EXIT_IF(color_count > RENDER_COLOR_ATTACHMENTS_MAX);
-	EXIT_IF(vs_spirv.empty());
-	EXIT_IF(ps_active && ps_spirv.empty());
+	EXIT_IF(colors.size() > RENDER_COLOR_ATTACHMENTS_MAX);
+	EXIT_IF(!vertex_program);
+	const bool ps_active = ps_input_info != nullptr;
+	EXIT_IF(ps_active && !pixel_program);
+	const auto color_count = static_cast<uint32_t>(colors.size());
 
 	Common::LockGuard lock(m_mutex);
-	auto&             ctx    = command.GetRegisters();
-	auto&             sh_ctx = command.GetShaders();
+	auto&             ctx = command.GetRegisters();
 
-	const auto&           vertex_info                              = sh_ctx.GetVs();
-	const auto&           ps_regs                                  = sh_ctx.GetPs();
-	const HW::BlendColor& bclr                                     = ctx.GetBlendColor();
-	uint32_t              color_mask[RENDER_COLOR_ATTACHMENTS_MAX] = {};
-	for (uint32_t i = 0; i < color_count; i++) {
-		color_mask[i] =
-		    (colors[i].image_id ? colors[i].export_mapping.ApplyMask(render_target_mask_slot(
-		                              ctx.GetRenderTargetMask(), colors[i].target_slot))
-		                        : 0);
-	}
 	const HW::ModeControl& mc = ctx.GetModeControl();
 
-	auto     vs_id = ShaderGetIdVS(vertex_info, vs_input_info, true);
-	ShaderId ps_id {};
-	if (ps_active) {
-		ps_id = ShaderGetIdPS(ps_regs, *ps_input_info, true);
-	}
+	const auto vs_id = vertex_program.id;
+	const auto ps_id = ps_active ? pixel_program.id : 0;
 
 	PipelineStaticParameters static_params {};
-	GraphicsPipeline         p {};
-	p.ps_shader_id = ps_id;
-	p.vs_shader_id = vs_id;
-
-	static_params.color_count = color_count;
-	PipelineRenderingState rendering {};
+	PipelineRenderingState   rendering {};
 	rendering.color_count       = color_count;
 	uint32_t attachment_samples = 0;
 	for (uint32_t i = 0; i < color_count; i++) {
-		EXIT_IF(!colors[i].image_id || colors[i].format == vk::Format::eUndefined);
-		rendering.color_formats[i] = colors[i].format;
+		EXIT_IF(!colors[i].image_id || colors[i].desc.view_info.format == vk::Format::eUndefined);
+		static_params.color_mask[i] = colors[i].export_mapping.ApplyMask(
+		    render_target_mask_slot(ctx.GetRenderTargetMask(), colors[i].target_slot));
+		rendering.color_formats[i] = colors[i].desc.view_info.format;
 		if (attachment_samples == 0) {
-			attachment_samples = colors[i].samples;
-		} else if (attachment_samples != colors[i].samples) {
+			attachment_samples = colors[i].desc.info.samples;
+		} else if (attachment_samples != colors[i].desc.info.samples) {
 			EXIT("mixed color attachment sample counts are unsupported: %u and %u\n",
-			     attachment_samples, colors[i].samples);
+			     attachment_samples, colors[i].desc.info.samples);
 		}
 	}
 	const bool with_depth =
-	    depth.format != vk::Format::eUndefined && static_cast<bool>(depth.image_id);
+	    depth.desc.view_info.format != vk::Format::eUndefined && static_cast<bool>(depth.image_id);
 	if (with_depth) {
-		const auto aspects = ImageViewOps::DepthAspectMask(depth.format);
-		rendering.depth_format =
-		    aspects & vk::ImageAspectFlagBits::eDepth ? depth.format : vk::Format::eUndefined;
-		rendering.stencil_format =
-		    aspects & vk::ImageAspectFlagBits::eStencil ? depth.format : vk::Format::eUndefined;
+		const auto aspects       = ImageViewOps::DepthAspectMask(depth.desc.view_info.format);
+		rendering.depth_format   = aspects & vk::ImageAspectFlagBits::eDepth
+		                               ? depth.desc.view_info.format
+		                               : vk::Format::eUndefined;
+		rendering.stencil_format = aspects & vk::ImageAspectFlagBits::eStencil
+		                               ? depth.desc.view_info.format
+		                               : vk::Format::eUndefined;
 		if (attachment_samples == 0) {
-			attachment_samples = depth.samples;
-		} else if (attachment_samples != depth.samples) {
+			attachment_samples = depth.desc.info.samples;
+		} else if (attachment_samples != depth.desc.info.samples) {
 			EXIT("mixed color/depth sample counts are unsupported: %u and %u\n", attachment_samples,
-			     depth.samples);
+			     depth.desc.info.samples);
 		}
 	}
 	if (color_count == 0 && !with_depth) {
@@ -148,23 +715,18 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 	if (static_params.sample_shading_enable && !m_graphics.sample_rate_shading_enabled) {
 		EXIT("Pipeline: sample-rate shading is required but unsupported by the host\n");
 	}
-	static_params.with_depth         = with_depth;
-	static_params.depth_test_enable  = depth.depth_test_enable;
-	static_params.depth_write_enable = (depth.depth_write_enable && !depth.depth_clear_enable);
-	static_params.depth_compare_op   = depth.depth_compare_op;
 	static_params.depth_bounds_test_enable = depth.depth_bounds_test_enable;
 	static_params.depth_min_bounds         = depth.depth_min_bounds;
 	static_params.depth_max_bounds         = depth.depth_max_bounds;
 	static_params.stencil_test_enable      = depth.stencil_test_enable;
 	static_params.stencil_front            = depth.stencil_static_front;
 	static_params.stencil_back             = depth.stencil_static_back;
-	for (uint32_t i = 0; i < RENDER_COLOR_ATTACHMENTS_MAX; i++) {
-		static_params.color_mask[i] = color_mask[i];
-	}
 	const bool rect_list     = topology == vk::PrimitiveTopology::ePatchList;
 	static_params.cull_back  = !rect_list && mc.cull_back;
 	static_params.cull_front = !rect_list && mc.cull_front;
 	static_params.face       = mc.face;
+	static_params.polygon_mode =
+	    ResolvePolygonMode(mc, static_params.cull_front, static_params.cull_back);
 
 	for (uint32_t i = 0; i < color_count; i++) {
 		const auto& rt                        = ctx.GetRenderTarget(colors[i].target_slot);
@@ -179,18 +741,37 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 		static_params.blend_enable[i]         = bc.enable;
 		static_params.blend_bypass[i]         = rt.info.blend_bypass;
 	}
-	static_params.blend_color_red   = bclr.red;
-	static_params.blend_color_green = bclr.green;
-	static_params.blend_color_blue  = bclr.blue;
-	static_params.blend_color_alpha = bclr.alpha;
-
-	NormalizeStaticParamsForDynamicState(static_params);
-
 	GraphicsPipelineKey key {};
 	key.rendering     = rendering;
-	key.vs_shader_id  = p.vs_shader_id;
-	key.ps_shader_id  = p.ps_shader_id;
+	key.vs_shader_id  = vs_id;
+	key.ps_shader_id  = ps_id;
 	key.static_params = static_params;
+	if (vs_input_info.stage.program->stage != ShaderType::Mesh) {
+		EXIT_IF(vs_input_info.buffers_num < 0 ||
+		        vs_input_info.buffers_num > ShaderVertexInputInfo::RES_MAX ||
+		        vs_input_info.resources_num < 0 ||
+		        vs_input_info.resources_num > ShaderVertexInputInfo::RES_MAX);
+		key.vertex_input.binding_count   = static_cast<uint8_t>(vs_input_info.buffers_num);
+		key.vertex_input.attribute_count = static_cast<uint8_t>(vs_input_info.resources_num);
+		uint32_t attributes_num          = 0;
+		for (int binding = 0; binding < vs_input_info.buffers_num; binding++) {
+			const auto& buffer = vs_input_info.buffers[binding];
+			EXIT_IF(buffer.attr_num < 0 || buffer.attr_num > ShaderVertexInputBuffer::ATTR_MAX);
+			attributes_num += static_cast<uint32_t>(buffer.attr_num);
+			EXIT_IF(attributes_num > static_cast<uint32_t>(vs_input_info.resources_num));
+			key.vertex_input.bindings[binding] = {.stride   = buffer.stride,
+			                                      .instance = buffer.fetch_index != 0};
+			for (int attribute = 0; attribute < buffer.attr_num; attribute++) {
+				const auto index = buffer.attr_indices[attribute];
+				EXIT_IF(index < 0 || index >= vs_input_info.resources_num);
+				key.vertex_input.attributes[index] = {
+				    .offset  = buffer.attr_offsets[attribute],
+				    .binding = static_cast<uint8_t>(binding),
+				};
+			}
+		}
+		EXIT_IF(attributes_num != static_cast<uint32_t>(vs_input_info.resources_num));
+	}
 
 	if (auto iter = m_graphics_pipelines.find(key); iter != m_graphics_pipelines.end()) {
 		return *iter->second;
@@ -201,20 +782,17 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 		if (ps_active) {
 			ShaderDbgDumpInputInfo(*ps_input_info);
 		}
-		LOGF("PipelineTrace: shader binaries VS=0x%08" PRIx32 "/0x%08" PRIx32 " words=%" PRIu64
-		     " PS=0x%08" PRIx32 "/0x%08" PRIx32 " words=%" PRIu64 "\n",
-		     vs_id.hash0, vs_id.crc32, static_cast<uint64_t>(vs_spirv.size()), ps_id.hash0,
-		     ps_id.crc32, static_cast<uint64_t>(ps_spirv.size()));
+		LOGF("PipelineTrace: shader modules VS=%" PRIu64 " module=%p PS=%" PRIu64 " module=%p\n",
+		     vs_id, static_cast<void*>(vertex_program.module), ps_id,
+		     static_cast<void*>(pixel_program.module));
 	}
 
-	auto cached = std::make_unique<GraphicsPipeline>(p);
-	LogPipelineTrace("CreatePipelineInternal begin", vs_id.hash0, vs_id.crc32, ps_id.hash0,
-	                 ps_id.crc32);
-	CreatePipelineInternal(m_graphics, m_descriptor_cache, *cached, rendering, vs_input_info,
-	                       vs_spirv, ps_input_info, ps_spirv, static_params, vs_id.hash0,
-	                       vs_id.crc32, ps_id.hash0, ps_id.crc32, ps_active);
-	LogPipelineTrace("CreatePipelineInternal done", vs_id.hash0, vs_id.crc32, ps_id.hash0,
-	                 ps_id.crc32);
+	auto cached = std::make_unique<Pipeline>();
+	LogPipelineTrace("CreatePipelineInternal begin", vs_id, ps_id);
+	CreatePipelineInternal(m_graphics, *cached, rendering, key.vertex_input, vs_input_info,
+	                       vertex_program, ps_input_info, pixel_program, static_params,
+	                       m_driver_cache);
+	LogPipelineTrace("CreatePipelineInternal done", vs_id, ps_id);
 
 	EXIT_NOT_IMPLEMENTED(cached->pipeline == nullptr);
 	EXIT_NOT_IMPLEMENTED(cached->pipeline_layout == nullptr);
@@ -225,25 +803,17 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 	return *iter->second;
 }
 
-PipelineCache::ComputePipeline&
-PipelineCache::CreateComputePipeline(ShaderComputeInputInfo&      input_info,
-                                     const HW::ComputeShaderInfo& cs_regs,
-                                     std::span<const uint32_t>    cs_spirv) {
+PipelineCache::Pipeline&
+PipelineCache::CreateComputePipeline(const ShaderComputeInputInfo& input_info,
+                                     const ShaderProgram&          compute_program) {
 	KYTY_PROFILER_BLOCK("PipelineCache::CreatePipeline(Compute)", profiler::colors::RedA100);
 
-	EXIT_IF(cs_spirv.empty());
+	EXIT_IF(!compute_program);
 
 	Common::LockGuard lock(m_mutex);
 
-	auto cs_id = ShaderGetIdCS(cs_regs, input_info, true);
-
-	ComputePipeline p {};
-	p.cs_shader_id = cs_id;
-
-	ComputePipelineKey key {};
-	key.cs_shader_id = p.cs_shader_id;
-
-	if (auto iter = m_compute_pipelines.find(key); iter != m_compute_pipelines.end()) {
+	if (auto iter = m_compute_pipelines.find(compute_program.id);
+	    iter != m_compute_pipelines.end()) {
 		return *iter->second;
 	}
 
@@ -251,13 +821,13 @@ PipelineCache::CreateComputePipeline(ShaderComputeInputInfo&      input_info,
 		ShaderDbgDumpInputInfo(input_info);
 	}
 
-	auto cached = std::make_unique<ComputePipeline>(p);
-	CreatePipelineInternal(m_graphics, m_descriptor_cache, *cached, input_info, cs_spirv);
+	auto cached = std::make_unique<Pipeline>();
+	CreatePipelineInternal(m_graphics, *cached, input_info, compute_program.module, m_driver_cache);
 
 	EXIT_NOT_IMPLEMENTED(cached->pipeline == nullptr);
 	EXIT_NOT_IMPLEMENTED(cached->pipeline_layout == nullptr);
 
-	auto [iter, inserted] = m_compute_pipelines.emplace(std::move(key), std::move(cached));
+	auto [iter, inserted] = m_compute_pipelines.emplace(compute_program.id, std::move(cached));
 	EXIT_IF(!inserted);
 
 	return *iter->second;

src/graphics/host_gpu/renderer/pipeline/pipelineCache.h

@@ -10,6 +10,7 @@
 #include "graphics/shader/shader.h"
 
 #include <cstddef>
+#include <filesystem>
 #include <memory>
 #include <span>
 #include <type_traits>
@@ -20,42 +21,35 @@ namespace Libs::Graphics {
 struct GraphicContext;
 struct RenderColorInfo;
 struct RenderDepthInfo;
-class RenderCommandBuffer;
-class DescriptorCache;
+class CommandBuffer;
 
 namespace HW {
 class Context;
 class Shader;
+class UserConfig;
 struct ComputeShaderInfo;
 } // namespace HW
 
 #pragma pack(push, 1)
 
 struct PipelineStaticParameters {
-	float                      viewport_scale[3]        = {};
-	float                      viewport_offset[3]       = {};
 	bool                       negative_one_to_one      = false;
 	bool                       depth_clip_enable        = true;
-	int                        scissor_ltrb[4]          = {0};
 	vk::PrimitiveTopology      topology                 = vk::PrimitiveTopology::ePointList;
 	bool                       primitive_restart_enable = false;
 	uint32_t                   samples                  = 1;
 	bool                       sample_shading_enable    = false;
-	bool                       with_depth               = false;
-	bool                       depth_test_enable        = false;
-	bool                       depth_write_enable       = false;
-	vk::CompareOp              depth_compare_op         = vk::CompareOp::eNever;
 	bool                       depth_bounds_test_enable = false;
 	float                      depth_min_bounds         = 0.0f;
 	float                      depth_max_bounds         = 0.0f;
 	bool                       stencil_test_enable      = false;
 	PipelineStencilStaticState stencil_front;
 	PipelineStencilStaticState stencil_back;
-	uint32_t                   color_count                                        = 1;
 	uint32_t                   color_mask[RENDER_COLOR_ATTACHMENTS_MAX]           = {};
 	bool                       cull_front                                         = false;
 	bool                       cull_back                                          = false;
 	bool                       face                                               = false;
+	vk::PolygonMode            polygon_mode                                       = vk::PolygonMode::eFill;
 	uint8_t                    color_srcblend[RENDER_COLOR_ATTACHMENTS_MAX]       = {};
 	uint8_t                    color_comb_fcn[RENDER_COLOR_ATTACHMENTS_MAX]       = {};
 	uint8_t                    color_destblend[RENDER_COLOR_ATTACHMENTS_MAX]      = {};
@@ -65,10 +59,6 @@ struct PipelineStaticParameters {
 	bool                       separate_alpha_blend[RENDER_COLOR_ATTACHMENTS_MAX] = {};
 	bool                       blend_enable[RENDER_COLOR_ATTACHMENTS_MAX]         = {};
 	bool                       blend_bypass[RENDER_COLOR_ATTACHMENTS_MAX]         = {};
-	float                      blend_color_red                                    = 0.0f;
-	float                      blend_color_green                                  = 0.0f;
-	float                      blend_color_blue                                   = 0.0f;
-	float                      blend_color_alpha                                  = 0.0f;
 
 	bool operator==(const PipelineStaticParameters& other) const noexcept;
 };
@@ -78,14 +68,7 @@ struct PipelineStaticParameters {
 static_assert(std::is_trivially_copyable_v<PipelineStaticParameters>);
 static_assert(std::is_standard_layout_v<PipelineStaticParameters>);
 static_assert(alignof(PipelineStaticParameters) == 1);
-static_assert(sizeof(PipelineStaticParameters) ==
-              sizeof(float[3]) + sizeof(float[3]) + sizeof(bool) * 2 + sizeof(int[4]) +
-                  sizeof(vk::PrimitiveTopology) + sizeof(bool) + sizeof(uint32_t) +
-                  sizeof(bool) * 4 + sizeof(vk::CompareOp) + sizeof(bool) + sizeof(float) * 2 +
-                  sizeof(bool) + sizeof(PipelineStencilStaticState) * 2 + sizeof(uint32_t) +
-                  sizeof(uint32_t[RENDER_COLOR_ATTACHMENTS_MAX]) + sizeof(bool) * 3 +
-                  sizeof(uint8_t[RENDER_COLOR_ATTACHMENTS_MAX]) * 6 +
-                  sizeof(bool[RENDER_COLOR_ATTACHMENTS_MAX]) * 3 + sizeof(float) * 4);
+static_assert(sizeof(PipelineStaticParameters) == 165);
 
 struct PipelineRenderingState {
 	std::array<vk::Format, RENDER_COLOR_ATTACHMENTS_MAX> color_formats {};
@@ -96,57 +79,86 @@ struct PipelineRenderingState {
 	bool operator==(const PipelineRenderingState&) const = default;
 };
 
+struct PipelineVertexInputState {
+	struct Binding {
+		uint32_t stride                           = 0;
+		bool     instance                         = false;
+		bool     operator==(const Binding&) const = default;
+	};
+	struct Attribute {
+		uint32_t offset                             = 0;
+		uint8_t  binding                            = 0;
+		bool     operator==(const Attribute&) const = default;
+	};
+
+	std::array<Binding, ShaderVertexInputInfo::RES_MAX>   bindings {};
+	std::array<Attribute, ShaderVertexInputInfo::RES_MAX> attributes {};
+	uint8_t                                               binding_count   = 0;
+	uint8_t                                               attribute_count = 0;
+
+	bool operator==(const PipelineVertexInputState&) const = default;
+};
+
+struct ShaderProgram {
+	uint64_t         id     = 0;
+	vk::ShaderModule module = nullptr;
+
+	explicit operator bool() const { return id != 0 && module != nullptr; }
+};
+
 class PipelineCache {
 public:
-	PipelineCache(GraphicContext& graphics, DescriptorCache& descriptor_cache)
-	    : m_graphics(graphics), m_descriptor_cache(descriptor_cache) {
-		EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
-	}
+	explicit PipelineCache(GraphicContext& graphics);
 	~PipelineCache();
 	KYTY_CLASS_NO_COPY(PipelineCache);
+	void Save();
 
 	struct Pipeline {
-		vk::PipelineLayout pipeline_layout = nullptr;
-		vk::Pipeline       pipeline        = nullptr;
+		vk::PipelineLayout      pipeline_layout       = nullptr;
+		vk::Pipeline            pipeline              = nullptr;
+		vk::DescriptorSetLayout descriptor_set_layout = nullptr;
+		bool                    uses_push_descriptors = false;
 	};
 
-	struct GraphicsPipeline: Pipeline {
-		ShaderId vs_shader_id;
-		ShaderId ps_shader_id;
+	struct GraphicsPrograms {
+		ShaderProgram vertex;
+		ShaderProgram pixel;
 	};
 
-	struct ComputePipeline: Pipeline {
-		ShaderId cs_shader_id;
-	};
+	GraphicsPrograms
+	GetGraphicsPrograms(const HW::VertexShaderInfo& vertex_regs,
+	                    const HW::PixelShaderInfo& pixel_regs, const HW::ShaderRegisters& sh,
+	                    const HW::Context& context, const HW::UserConfig& user_config,
+	                    std::span<const Prospero::ColorComponentMapping, 8> target_export_mapping,
+	                    bool pixel_active, ShaderVertexInputInfo& vertex_info,
+	                    ShaderPixelInputInfo& pixel_info);
+	ShaderProgram GetComputeProgram(const HW::ComputeShaderInfo& regs,
+	                                const HW::ShaderRegisters&   sh,
+	                                ShaderComputeInputInfo&      input_info);
 
-	GraphicsPipeline&
-	CreateGraphicsPipeline(RenderColorInfo* colors, uint32_t color_count, RenderDepthInfo& depth,
-	                       ShaderVertexInputInfo& vs_input_info, RenderCommandBuffer& command,
-	                       ShaderPixelInputInfo* ps_input_info, vk::PrimitiveTopology topology,
-	                       bool primitive_restart_enable, bool ps_active,
-	                       std::span<const uint32_t> vs_spirv, std::span<const uint32_t> ps_spirv);
-	ComputePipeline& CreateComputePipeline(ShaderComputeInputInfo&      input_info,
-	                                       const HW::ComputeShaderInfo& cs_regs,
-	                                       std::span<const uint32_t>    cs_spirv);
+	Pipeline&
+	CreateGraphicsPipeline(std::span<const RenderColorInfo> colors, const RenderDepthInfo& depth,
+	                       const ShaderVertexInputInfo& vs_input_info, CommandBuffer& command,
+	                       const ShaderPixelInputInfo* ps_input_info,
+	                       vk::PrimitiveTopology topology, bool primitive_restart_enable,
+	                       const ShaderProgram& vertex_program, const ShaderProgram& pixel_program);
+	Pipeline& CreateComputePipeline(const ShaderComputeInputInfo& input_info,
+	                                const ShaderProgram&          compute_program);
 
 private:
+	struct ProgramCache;
+
 	struct GraphicsPipelineKey {
 		PipelineRenderingState   rendering;
-		ShaderId                 vs_shader_id;
-		ShaderId                 ps_shader_id;
+		uint64_t                 vs_shader_id = 0;
+		uint64_t                 ps_shader_id = 0;
+		PipelineVertexInputState vertex_input;
 		PipelineStaticParameters static_params;
 
 		bool operator==(const GraphicsPipelineKey& other) const {
 			return rendering == other.rendering && vs_shader_id == other.vs_shader_id &&
-			       ps_shader_id == other.ps_shader_id && static_params == other.static_params;
-		}
-	};
-
-	struct ComputePipelineKey {
-		ShaderId cs_shader_id;
-
-		bool operator==(const ComputePipelineKey& other) const {
-			return cs_shader_id == other.cs_shader_id;
+			       ps_shader_id == other.ps_shader_id && vertex_input == other.vertex_input &&
+			       static_params == other.static_params;
 		}
 	};
 
@@ -156,15 +168,6 @@ private:
 			        (hash >> 2u);
 		}
 
-		static void MixShaderId(std::size_t& hash, const ShaderId& id) {
-			Mix(hash, id.hash0);
-			Mix(hash, id.crc32);
-			Mix(hash, id.ids.size());
-			for (auto value: id.ids) {
-				Mix(hash, value);
-			}
-		}
-
 		static void MixStaticParams(std::size_t& hash, const PipelineStaticParameters& params) {
 			const auto* bytes = reinterpret_cast<const uint8_t*>(&params);
 			for (std::size_t i = 0; i < sizeof(params); i++) {
@@ -186,44 +189,45 @@ private:
 		std::size_t operator()(const GraphicsPipelineKey& key) const {
 			std::size_t hash = 0;
 			PipelineKeyHash::MixRendering(hash, key.rendering);
-			PipelineKeyHash::MixShaderId(hash, key.vs_shader_id);
-			PipelineKeyHash::MixShaderId(hash, key.ps_shader_id);
+			PipelineKeyHash::Mix(hash, key.vs_shader_id);
+			PipelineKeyHash::Mix(hash, key.ps_shader_id);
+			PipelineKeyHash::Mix(hash, key.vertex_input.binding_count);
+			for (uint32_t i = 0; i < key.vertex_input.binding_count; i++) {
+				PipelineKeyHash::Mix(hash, key.vertex_input.bindings[i].stride);
+				PipelineKeyHash::Mix(hash, key.vertex_input.bindings[i].instance);
+			}
+			PipelineKeyHash::Mix(hash, key.vertex_input.attribute_count);
+			for (uint32_t i = 0; i < key.vertex_input.attribute_count; i++) {
+				PipelineKeyHash::Mix(hash, key.vertex_input.attributes[i].offset);
+				PipelineKeyHash::Mix(hash, key.vertex_input.attributes[i].binding);
+			}
 			PipelineKeyHash::MixStaticParams(hash, key.static_params);
 			return hash;
 		}
 	};
 
-	struct ComputePipelineKeyHash {
-		std::size_t operator()(const ComputePipelineKey& key) const {
-			std::size_t hash = 0;
-			PipelineKeyHash::MixShaderId(hash, key.cs_shader_id);
-			return hash;
-		}
-	};
-
-	GraphicContext&  m_graphics;
-	DescriptorCache& m_descriptor_cache;
-	std::unordered_map<GraphicsPipelineKey, std::unique_ptr<GraphicsPipeline>,
-	                   GraphicsPipelineKeyHash>
-	    m_graphics_pipelines;
-	std::unordered_map<ComputePipelineKey, std::unique_ptr<ComputePipeline>, ComputePipelineKeyHash>
-	              m_compute_pipelines;
+	GraphicContext&               m_graphics;
+	std::unique_ptr<ProgramCache> m_program_cache;
+	vk::PipelineCache             m_driver_cache = nullptr;
+	std::filesystem::path         m_driver_cache_path;
+	std::unordered_map<GraphicsPipelineKey, std::unique_ptr<Pipeline>, GraphicsPipelineKeyHash>
+	                                                        m_graphics_pipelines;
+	std::unordered_map<uint64_t, std::unique_ptr<Pipeline>> m_compute_pipelines;
 	Common::Mutex m_mutex;
+
+	void InitializeDriverCache();
 };
 
-void LogPipelineTrace(const char* phase, uint32_t vs_hash0, uint32_t vs_crc32, uint32_t ps_hash0,
-                      uint32_t ps_crc32);
+void LogPipelineTrace(const char* phase, uint64_t vertex_program_id, uint64_t pixel_program_id);
 void CreatePipelineInternal(
-    GraphicContext& graphics, DescriptorCache& descriptor_cache,
-    PipelineCache::GraphicsPipeline& pipeline, const PipelineRenderingState& rendering,
-    const ShaderVertexInputInfo& vs_input_info, std::span<const uint32_t> vs_shader,
-    const ShaderPixelInputInfo* ps_input_info, std::span<const uint32_t> ps_shader,
-    const PipelineStaticParameters& static_params, uint32_t vs_hash0, uint32_t vs_crc32,
-    uint32_t ps_hash0, uint32_t ps_crc32, bool ps_active);
-void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descriptor_cache,
-                            PipelineCache::ComputePipeline& pipeline,
-                            const ShaderComputeInputInfo&   input_info,
-                            std::span<const uint32_t>       cs_shader);
+    GraphicContext& graphics, PipelineCache::Pipeline& pipeline,
+    const PipelineRenderingState& rendering, const PipelineVertexInputState& vertex_input,
+    const ShaderVertexInputInfo& vs_input_info, const ShaderProgram& vertex_program,
+    const ShaderPixelInputInfo* ps_input_info, const ShaderProgram& pixel_program,
+    const PipelineStaticParameters& static_params, vk::PipelineCache driver_cache);
+void CreatePipelineInternal(GraphicContext& graphics, PipelineCache::Pipeline& pipeline,
+                            const ShaderComputeInputInfo& input_info,
+                            vk::ShaderModule compute_module, vk::PipelineCache driver_cache);
 
 } // namespace Libs::Graphics
 

src/graphics/host_gpu/renderer/pipeline/shaderResourceBarrier.cpp

@@ -13,6 +13,9 @@ vk::PipelineStageFlags ShaderPipelineStages(vk::ShaderStageFlags stages) {
 	if (stages & vk::ShaderStageFlagBits::eVertex) {
 		result |= vk::PipelineStageFlagBits::eVertexShader;
 	}
+	if (stages & vk::ShaderStageFlagBits::eMeshEXT) {
+		result |= vk::PipelineStageFlagBits::eMeshShaderEXT;
+	}
 	if (stages & vk::ShaderStageFlagBits::eFragment) {
 		result |= vk::PipelineStageFlagBits::eFragmentShader;
 	}
@@ -25,7 +28,6 @@ vk::PipelineStageFlags ShaderPipelineStages(vk::ShaderStageFlags stages) {
 
 VulkanMemoryBarrier MakeShaderWriteDependency() {
 	VulkanMemoryBarrier barrier {};
-	barrier.sType         = vk::StructureType::eMemoryBarrier;
 	barrier.srcAccessMask = vk::AccessFlagBits::eShaderWrite;
 	barrier.dstAccessMask = vk::AccessFlagBits::eShaderRead | vk::AccessFlagBits::eShaderWrite |
 	                        vk::AccessFlagBits::eVertexAttributeRead |
@@ -38,7 +40,6 @@ VulkanMemoryBarrier MakeShaderWriteDependency() {
 
 VulkanMemoryBarrier MakeShaderAccessDependency() {
 	VulkanMemoryBarrier barrier {};
-	barrier.sType         = vk::StructureType::eMemoryBarrier;
 	barrier.srcAccessMask = vk::AccessFlagBits::eShaderRead | vk::AccessFlagBits::eShaderWrite;
 	barrier.dstAccessMask = vk::AccessFlagBits::eMemoryRead | vk::AccessFlagBits::eMemoryWrite;
 	return barrier;
@@ -46,7 +47,6 @@ VulkanMemoryBarrier MakeShaderAccessDependency() {
 
 VulkanMemoryBarrier MakeShaderWriteHazardDependency() {
 	VulkanMemoryBarrier barrier {};
-	barrier.sType         = vk::StructureType::eMemoryBarrier;
 	barrier.srcAccessMask = vk::AccessFlagBits::eMemoryRead | vk::AccessFlagBits::eMemoryWrite;
 	barrier.dstAccessMask = vk::AccessFlagBits::eShaderRead | vk::AccessFlagBits::eShaderWrite;
 	return barrier;
@@ -56,7 +56,6 @@ vk::BufferMemoryBarrier MakeGdsDependency(vk::Buffer buffer) {
 	EXIT_IF(buffer == nullptr);
 
 	vk::BufferMemoryBarrier barrier {};
-	barrier.sType         = vk::StructureType::eBufferMemoryBarrier;
 	barrier.srcAccessMask = vk::AccessFlagBits::eHostWrite | vk::AccessFlagBits::eTransferWrite |
 	                        vk::AccessFlagBits::eShaderWrite;
 	barrier.dstAccessMask = vk::AccessFlagBits::eShaderRead | vk::AccessFlagBits::eShaderWrite;
@@ -68,11 +67,12 @@ vk::BufferMemoryBarrier MakeGdsDependency(vk::Buffer buffer) {
 	return barrier;
 }
 
-std::vector<ShaderBufferWriteRange>
-CollectShaderBufferWrites(const ShaderRecompiler::IR::Program&          program,
-                          const ShaderRecompiler::IR::ResourceSnapshot& resources) {
+bool HasShaderBufferWrites(const ShaderStageRuntime& runtime) {
+	EXIT_IF(!runtime);
+	const auto& program   = *runtime.program;
+	const auto& resources = runtime.resources;
 	EXIT_IF(resources.buffers.size() != program.info.buffers.size());
-	std::vector<ShaderBufferWriteRange> writes;
+	bool has_writes = false;
 	for (uint32_t i = 0; i < program.info.buffers.size(); i++) {
 		if (!program.info.buffers[i].written) {
 			continue;
@@ -81,23 +81,11 @@ CollectShaderBufferWrites(const ShaderRecompiler::IR::Program&          program,
 		EXIT_IF(value.dword_count < 4);
 		ShaderBufferResource descriptor;
 		std::memcpy(descriptor.fields, value.dwords.data(), sizeof(descriptor.fields));
-		const auto address = descriptor.Base48();
-		const auto records = static_cast<uint64_t>(descriptor.NumRecords());
-		const auto stride  = static_cast<uint64_t>(descriptor.Stride());
-		if (stride != 0 && records > UINT64_MAX / stride) {
-			EXIT("shader resource barrier buffer footprint overflow\n");
-		}
-		const auto size = stride == 0 ? records : stride * records;
-		if (address != 0 && size != 0) {
-			writes.push_back({address, size});
-		}
+		// A zero stride means byte addressing. For either addressing mode a nonzero record
+		// count is exactly the condition for a nonempty descriptor range.
+		has_writes |= descriptor.Base48() != 0 && descriptor.NumRecords() != 0;
 	}
-	return writes;
-}
-
-bool HasShaderBufferWrites(const ShaderStageRuntime& runtime) {
-	EXIT_IF(!runtime);
-	return !CollectShaderBufferWrites(*runtime.program, *runtime.resources).empty();
+	return has_writes;
 }
 
 void ShaderAccessBarrier(vk::CommandBuffer vk_buffer, vk::PipelineStageFlags source_stages) {
@@ -118,13 +106,11 @@ void ShaderWriteHazardBarrier(vk::CommandBuffer      vk_buffer,
 void ShaderWriteBarrier(vk::CommandBuffer vk_buffer, vk::PipelineStageFlags source_stages) {
 	EXIT_IF(vk_buffer == nullptr || !source_stages);
 	const auto barrier = MakeShaderWriteDependency();
-	vk_buffer.pipelineBarrier(
-	    source_stages,
-	    vk::PipelineStageFlagBits::eComputeShader | vk::PipelineStageFlagBits::eVertexInput |
-	        vk::PipelineStageFlagBits::eVertexShader | vk::PipelineStageFlagBits::eFragmentShader |
-	        vk::PipelineStageFlagBits::eTransfer |
-	        vk::PipelineStageFlagBits::eColorAttachmentOutput,
-	    vk::DependencyFlags {}, 1, &barrier, 0, nullptr, 0, nullptr);
+	vk_buffer.pipelineBarrier(source_stages,
+	                          vk::PipelineStageFlagBits::eComputeShader |
+	                              vk::PipelineStageFlagBits::eAllGraphics |
+	                              vk::PipelineStageFlagBits::eTransfer,
+	                          vk::DependencyFlags {}, 1, &barrier, 0, nullptr, 0, nullptr);
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/pipeline/shaderResourceBarrier.h

@@ -4,27 +4,15 @@
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/shader/recompiler/ir/passes/ResourceMaterialization.h"
 
-#include <vector>
-
 namespace Libs::Graphics {
 
 struct ShaderStageRuntime;
 
-struct ShaderBufferWriteRange {
-	uint64_t address = 0;
-	uint64_t size    = 0;
-
-	bool operator==(const ShaderBufferWriteRange& other) const = default;
-};
-
 vk::PipelineStageFlags  ShaderPipelineStages(vk::ShaderStageFlags stages);
 VulkanMemoryBarrier     MakeShaderAccessDependency();
 VulkanMemoryBarrier     MakeShaderWriteHazardDependency();
 VulkanMemoryBarrier     MakeShaderWriteDependency();
 vk::BufferMemoryBarrier MakeGdsDependency(vk::Buffer buffer);
-std::vector<ShaderBufferWriteRange>
-CollectShaderBufferWrites(const ShaderRecompiler::IR::Program&          program,
-                          const ShaderRecompiler::IR::ResourceSnapshot& resources);
 bool HasShaderBufferWrites(const ShaderStageRuntime& runtime);
 void ShaderAccessBarrier(vk::CommandBuffer vk_buffer, vk::PipelineStageFlags source_stages);
 void ShaderWriteHazardBarrier(vk::CommandBuffer      vk_buffer,

src/graphics/host_gpu/renderer/pipeline/shaderSubgroup.cpp

@@ -1,60 +0,0 @@
-#include "graphics/host_gpu/renderer/pipeline/shaderSubgroup.h"
-
-#include "graphics/shader/recompiler/backend/spirv/SpirvEmitter.h"
-
-namespace Libs::Graphics {
-
-ShaderSubgroupCapabilities::ShaderSubgroupCapabilities(const GraphicContext& graphics)
-    : subgroup_size(graphics.subgroup_size), min_subgroup_size(graphics.min_subgroup_size),
-      max_subgroup_size(graphics.max_subgroup_size),
-      required_subgroup_size_stages(graphics.required_subgroup_size_stages),
-      subgroup_size_control_enabled(graphics.subgroup_size_control_enabled) {}
-
-ShaderLaneMaskMode SelectGraphicsLaneMaskMode(uint32_t guest_wave_size) {
-	// A 64-wide guest wave can't assume the host GPU's lane layout matches
-	// GCN's. AMD RDNA also reports a 64-wide subgroup but uses a different
-	// lane/pixel layout, which corrupted rendering under "native wave" mode.
-	// Per-invocation emulation works regardless of the host's lane layout.
-	if (guest_wave_size == 64u) {
-		return ShaderLaneMaskMode::PerInvocation;
-	}
-	return ShaderLaneMaskMode::NativeWave;
-}
-
-ShaderSubgroupConfiguration ConfigureShaderSubgroup(const ShaderSubgroupCapabilities& capabilities,
-                                                    vk::ShaderStageFlagBits           stage,
-                                                    const ShaderRecompiler::IR::Program& program) {
-	const auto guest_wave_size = program.wave_size;
-	if (guest_wave_size != 32u && guest_wave_size != 64u) {
-		return {};
-	}
-	if (stage != vk::ShaderStageFlagBits::eCompute) {
-		const auto expected = SelectGraphicsLaneMaskMode(guest_wave_size);
-		if (program.lane_mask_mode != expected || (capabilities.subgroup_size != guest_wave_size &&
-		                                           expected != ShaderLaneMaskMode::PerInvocation)) {
-			return {};
-		}
-		return {expected == ShaderLaneMaskMode::NativeWave
-		            ? ShaderSubgroupMode::Natural
-		            : ShaderSubgroupMode::PerInvocationGraphics,
-		        0};
-	}
-	if (program.lane_mask_mode != ShaderLaneMaskMode::NativeWave) {
-		return {};
-	}
-	if (capabilities.subgroup_size == guest_wave_size) {
-		return {ShaderSubgroupMode::Natural, 0};
-	}
-	if (capabilities.subgroup_size_control_enabled &&
-	    (capabilities.required_subgroup_size_stages & stage) &&
-	    guest_wave_size >= capabilities.min_subgroup_size &&
-	    guest_wave_size <= capabilities.max_subgroup_size) {
-		return {ShaderSubgroupMode::Controlled, guest_wave_size};
-	}
-	if (!ShaderRecompiler::Spirv::ProgramRequiresExactSubgroupSize(program)) {
-		return {ShaderSubgroupMode::FlattenedMasks, 0};
-	}
-	return {};
-}
-
-} // namespace Libs::Graphics

src/graphics/host_gpu/renderer/pipeline/shaderSubgroup.h

@@ -1,40 +0,0 @@
-#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_SHADERSUBGROUP_H_
-#define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_SHADERSUBGROUP_H_
-
-#include "graphics/host_gpu/graphicContext.h"
-#include "graphics/shader/recompiler/ir/ShaderIR.h"
-
-namespace Libs::Graphics {
-
-enum class ShaderSubgroupMode {
-	Natural,
-	Controlled,
-	PerInvocationGraphics,
-	FlattenedMasks,
-	Unsupported
-};
-
-struct ShaderSubgroupConfiguration {
-	ShaderSubgroupMode mode          = ShaderSubgroupMode::Unsupported;
-	uint32_t           required_size = 0;
-};
-
-struct ShaderSubgroupCapabilities {
-	explicit ShaderSubgroupCapabilities(const GraphicContext& graphics);
-
-	uint32_t             subgroup_size     = 0;
-	uint32_t             min_subgroup_size = 0;
-	uint32_t             max_subgroup_size = 0;
-	vk::ShaderStageFlags required_subgroup_size_stages;
-	bool                 subgroup_size_control_enabled = false;
-};
-
-ShaderLaneMaskMode SelectGraphicsLaneMaskMode(uint32_t guest_wave_size);
-
-ShaderSubgroupConfiguration ConfigureShaderSubgroup(const ShaderSubgroupCapabilities& capabilities,
-                                                    vk::ShaderStageFlagBits           stage,
-                                                    const ShaderRecompiler::IR::Program& program);
-
-} // namespace Libs::Graphics
-
-#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_SHADERSUBGROUP_H_

src/graphics/host_gpu/renderer/pipeline/shaders.cpp

@@ -6,9 +6,8 @@
 #include "graphics/guest_gpu/gpu_defs.h"
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/host_gpu/renderer/debug.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
+#include "graphics/host_gpu/renderer/pipeline/descriptors.h"
 #include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
-#include "graphics/host_gpu/renderer/pipeline/shaderSubgroup.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/renderer/renderTarget.h"
@@ -20,6 +19,7 @@
 #include <algorithm>
 #include <limits>
 #include <span>
+#include <vector>
 
 namespace Libs::Graphics {
 
@@ -103,7 +103,7 @@ static void GetInputFormat(const ShaderBufferResource& res, vk::Format& format,
 		size   = 4;
 		if (NarrowInputFormat(format, size, used_components)) {
 			LOGF("InputFormat: narrowing fmt=%u to %s for used_components=%u\n", raw_format,
-			     VulkanToString(format).c_str(), used_components);
+			     vk::to_string(format).c_str(), used_components);
 		}
 		return;
 	}
@@ -117,7 +117,7 @@ static void GetInputFormat(const ShaderBufferResource& res, vk::Format& format,
 		size   = 2;
 		if (NarrowInputFormat(format, size, used_components)) {
 			LOGF("InputFormat: narrowing fmt=%u to %s for used_components=%u\n", raw_format,
-			     VulkanToString(format).c_str(), used_components);
+			     vk::to_string(format).c_str(), used_components);
 		}
 		return;
 	}
@@ -389,116 +389,58 @@ static vk::BlendOp GetBlendOp(uint32_t op) {
 	return vk::BlendOp::eAdd;
 }
 
-static void CreateLayout(DescriptorCache&                   descriptor_cache,
-                         std::span<vk::DescriptorSetLayout> set_layouts, uint32_t& set_layouts_num,
-                         std::span<vk::PushConstantRange>     push_constant_info,
-                         uint32_t&                            push_constant_info_num,
-                         const ShaderRecompiler::IR::Program& program,
-                         vk::ShaderStageFlags vk_stage, DescriptorCache::Stage stage) {
-	const auto& bindings        = program.bindings;
-	const bool  need_descriptor = !bindings.descriptors.empty();
-
-	if (bindings.push_constant_size != 0) {
-		auto index = push_constant_info_num;
-
-		push_constant_info[index].stageFlags = vk_stage;
-		push_constant_info[index].offset     = bindings.push_constant_offset;
-		push_constant_info[index].size       = bindings.push_constant_size;
-		push_constant_info_num++;
-	}
-
-	if (need_descriptor) {
-		EXIT_IF(bindings.descriptor_set != set_layouts_num);
-
-		set_layouts[set_layouts_num] = descriptor_cache.GetDescriptorSetLayout(stage, program);
-		set_layouts_num++;
+static void AddLayoutBindings(std::vector<vk::DescriptorSetLayoutBinding>& descriptor_bindings,
+                              const ShaderRecompiler::IR::CompiledShaderInfo& program,
+                              vk::ShaderStageFlagBits              stage) {
+	for (const auto& binding: program.bindings.descriptors) {
+		descriptor_bindings.push_back(
+		    {ShaderRecompiler::IR::NativeBinding(program.stage, binding.kind),
+		     NativeDescriptorType(binding.kind), NativeDescriptorCount(binding), stage, nullptr});
 	}
 }
 
-static void ConfigureSubgroupSize(const GraphicContext& graphics, vk::ShaderStageFlagBits vk_stage,
-                                  const ShaderRecompiler::IR::Program&                   program,
-                                  vk::PipelineShaderStageRequiredSubgroupSizeCreateInfo& required,
-                                  vk::PipelineShaderStageCreateInfo&                     stage) {
-	const auto config =
-	    ConfigureShaderSubgroup(ShaderSubgroupCapabilities {graphics}, vk_stage, program);
-	switch (config.mode) {
-		case ShaderSubgroupMode::Natural: return;
-		case ShaderSubgroupMode::PerInvocationGraphics: {
-			static std::atomic<uint32_t> log_count {0};
-			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {
-				LOGF("Vulkan running %u-lane graphics shader stage 0x%08x with per-invocation "
-				     "EXEC/VCC on the host-native subgroup\n",
-				     program.wave_size, static_cast<uint32_t>(vk_stage));
-			}
-			return;
-		}
-		case ShaderSubgroupMode::FlattenedMasks: {
-			static std::atomic<uint32_t> log_count {0};
-			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {
-				LOGF("Vulkan running %u-lane shader stage 0x%08x with flattened per-invocation "
-				     "EXEC/VCC\n",
-				     program.wave_size, static_cast<uint32_t>(vk_stage));
-			}
-			return;
-		}
-		case ShaderSubgroupMode::Controlled: break;
-		case ShaderSubgroupMode::Unsupported:
-		default:
-			EXIT("Vulkan cannot run %u-lane shader stage 0x%08x: default=%u min=%u max=%u "
-			     "controlled_stages=0x%08x\n",
-			     program.wave_size, static_cast<uint32_t>(vk_stage), graphics.subgroup_size,
-			     graphics.min_subgroup_size, graphics.max_subgroup_size,
-			     static_cast<vk::ShaderStageFlags::MaskType>(
-			         graphics.required_subgroup_size_stages));
+static void CreateDescriptorLayout(GraphicContext& graphics, PipelineCache::Pipeline& pipeline,
+                                   std::span<const vk::DescriptorSetLayoutBinding> bindings) {
+	uint32_t descriptor_count = 0;
+	for (const auto& binding: bindings) {
+		descriptor_count += binding.descriptorCount;
 	}
+	pipeline.uses_push_descriptors = descriptor_count <= graphics.max_push_descriptors;
 
-	required.sType = vk::StructureType::ePipelineShaderStageRequiredSubgroupSizeCreateInfo;
-	required.requiredSubgroupSize = config.required_size;
-	stage.pNext                   = &required;
+	vk::DescriptorSetLayoutCreateInfo create {};
+	create.flags        = pipeline.uses_push_descriptors
+	                          ? vk::DescriptorSetLayoutCreateFlagBits::ePushDescriptorKHR
+	                          : vk::DescriptorSetLayoutCreateFlags {};
+	create.bindingCount = static_cast<uint32_t>(bindings.size());
+	create.pBindings    = bindings.data();
+	EXIT_IF(graphics.device.createDescriptorSetLayout(
+	            &create, nullptr, &pipeline.descriptor_set_layout) != vk::Result::eSuccess);
 }
 
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
 void CreatePipelineInternal(
-    GraphicContext& graphics, DescriptorCache& descriptor_cache,
-    PipelineCache::GraphicsPipeline& pipeline, const PipelineRenderingState& rendering,
-    const ShaderVertexInputInfo& vs_input_info, std::span<const uint32_t> vs_shader,
-    const ShaderPixelInputInfo* ps_input_info, std::span<const uint32_t> ps_shader,
-    const PipelineStaticParameters& static_params, uint32_t vs_hash0, uint32_t vs_crc32,
-    uint32_t ps_hash0, uint32_t ps_crc32, bool ps_active) {
-	EXIT_IF(ps_active && ps_input_info == nullptr);
+    GraphicContext& graphics, PipelineCache::Pipeline& pipeline,
+    const PipelineRenderingState& rendering, const PipelineVertexInputState& vertex_input,
+    const ShaderVertexInputInfo& vs_input_info, const ShaderProgram& vertex_program,
+    const ShaderPixelInputInfo* ps_input_info, const ShaderProgram& pixel_program,
+    const PipelineStaticParameters& static_params, vk::PipelineCache driver_cache) {
+	const bool ps_active = ps_input_info != nullptr;
+	EXIT_IF(!vertex_program || (ps_active && !pixel_program));
+	const bool with_depth = rendering.depth_format != vk::Format::eUndefined ||
+	                        rendering.stencil_format != vk::Format::eUndefined;
+	EXIT_IF(!vs_input_info.stage);
+	const bool mesh = vs_input_info.stage.program->stage == ShaderType::Mesh;
+	EXIT_NOT_IMPLEMENTED(mesh && !graphics.mesh_shader_enabled);
+	const auto vertex_stage =
+	    mesh ? vk::ShaderStageFlagBits::eMeshEXT : vk::ShaderStageFlagBits::eVertex;
 
-	const bool rect_list = static_params.topology == vk::PrimitiveTopology::ePatchList;
+	const bool rect_list = !mesh && static_params.topology == vk::PrimitiveTopology::ePatchList;
 
-	vk::ShaderModule vert_shader_module         = nullptr;
 	vk::ShaderModule tess_control_shader_module = nullptr;
 	vk::ShaderModule tess_eval_shader_module    = nullptr;
-	vk::ShaderModule frag_shader_module         = nullptr;
 
 	vk::ShaderModuleCreateInfo create_info {};
-
-	create_info.sType = vk::StructureType::eShaderModuleCreateInfo;
-	create_info.pNext = nullptr;
-	create_info.flags = {};
-
-	create_info.codeSize = vs_shader.size() * 4;
-	create_info.pCode    = vs_shader.data();
-	auto result = graphics.device.createShaderModule(&create_info, nullptr, &vert_shader_module);
-	if (graphics_debug_dump_enabled()) {
-		LOGF("PipelineTrace: vkCreateShaderModule VS done result=%s module=%p\n",
-		     VulkanToString(result).c_str(), static_cast<void*>(vert_shader_module));
-	}
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
-
-	if (ps_active) {
-		create_info.codeSize = ps_shader.size() * 4;
-		create_info.pCode    = ps_shader.data();
-		result = graphics.device.createShaderModule(&create_info, nullptr, &frag_shader_module);
-		if (graphics_debug_dump_enabled()) {
-			LOGF("PipelineTrace: vkCreateShaderModule PS done result=%s module=%p\n",
-			     VulkanToString(result).c_str(), static_cast<void*>(frag_shader_module));
-		}
-		EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
-	}
+	vk::Result result {};
 	if (rect_list) {
 		const auto shaders =
 		    BuildRectListShaders(vs_input_info, ps_active ? ps_input_info : nullptr);
@@ -508,7 +450,7 @@ void CreatePipelineInternal(
 		    graphics.device.createShaderModule(&create_info, nullptr, &tess_control_shader_module);
 		if (graphics_debug_dump_enabled()) {
 			LOGF("PipelineTrace: vkCreateShaderModule RectList TCS done result=%s module=%p\n",
-			     VulkanToString(result).c_str(), static_cast<void*>(tess_control_shader_module));
+			     vk::to_string(result).c_str(), static_cast<void*>(tess_control_shader_module));
 		}
 		EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
 
@@ -518,54 +460,30 @@ void CreatePipelineInternal(
 		    graphics.device.createShaderModule(&create_info, nullptr, &tess_eval_shader_module);
 		if (graphics_debug_dump_enabled()) {
 			LOGF("PipelineTrace: vkCreateShaderModule RectList TES done result=%s module=%p\n",
-			     VulkanToString(result).c_str(), static_cast<void*>(tess_eval_shader_module));
+			     vk::to_string(result).c_str(), static_cast<void*>(tess_eval_shader_module));
 		}
 		EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
 	}
 
-	EXIT_NOT_IMPLEMENTED(vert_shader_module == nullptr);
 	EXIT_NOT_IMPLEMENTED(
 	    rect_list && (tess_control_shader_module == nullptr || tess_eval_shader_module == nullptr));
-	EXIT_NOT_IMPLEMENTED(ps_active && frag_shader_module == nullptr);
 
-	vk::PipelineShaderStageCreateInfo                     vert_shader_stage_info {};
-	vk::PipelineShaderStageRequiredSubgroupSizeCreateInfo vert_subgroup_size {};
+	vk::PipelineShaderStageCreateInfo vert_shader_stage_info {};
+	vert_shader_stage_info.stage  = vertex_stage;
+	vert_shader_stage_info.module = vertex_program.module;
+	vert_shader_stage_info.pName  = "main";
 
-	vert_shader_stage_info.sType               = vk::StructureType::ePipelineShaderStageCreateInfo;
-	vert_shader_stage_info.pNext               = nullptr;
-	vert_shader_stage_info.flags               = {};
-	vert_shader_stage_info.stage               = vk::ShaderStageFlagBits::eVertex;
-	vert_shader_stage_info.module              = vert_shader_module;
-	vert_shader_stage_info.pName               = "main";
-	vert_shader_stage_info.pSpecializationInfo = nullptr;
-	EXIT_IF(!vs_input_info.stage);
-	ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eVertex, *vs_input_info.stage.program,
-	                      vert_subgroup_size, vert_shader_stage_info);
-
-	vk::PipelineShaderStageCreateInfo                     frag_shader_stage_info {};
-	vk::PipelineShaderStageRequiredSubgroupSizeCreateInfo frag_subgroup_size {};
-	frag_shader_stage_info.sType               = vk::StructureType::ePipelineShaderStageCreateInfo;
-	frag_shader_stage_info.pNext               = nullptr;
-	frag_shader_stage_info.flags               = {};
-	frag_shader_stage_info.stage               = vk::ShaderStageFlagBits::eFragment;
-	frag_shader_stage_info.module              = frag_shader_module;
-	frag_shader_stage_info.pName               = "main";
-	frag_shader_stage_info.pSpecializationInfo = nullptr;
-	if (ps_active) {
-		EXIT_IF(!ps_input_info->stage);
-		ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eFragment,
-		                      *ps_input_info->stage.program, frag_subgroup_size,
-		                      frag_shader_stage_info);
-	}
+	vk::PipelineShaderStageCreateInfo frag_shader_stage_info {};
+	frag_shader_stage_info.stage  = vk::ShaderStageFlagBits::eFragment;
+	frag_shader_stage_info.module = pixel_program.module;
+	frag_shader_stage_info.pName  = "main";
 
 	vk::PipelineShaderStageCreateInfo tess_control_shader_stage_info {};
-	tess_control_shader_stage_info.sType  = vk::StructureType::ePipelineShaderStageCreateInfo;
 	tess_control_shader_stage_info.stage  = vk::ShaderStageFlagBits::eTessellationControl;
 	tess_control_shader_stage_info.module = tess_control_shader_module;
 	tess_control_shader_stage_info.pName  = "main";
 
 	vk::PipelineShaderStageCreateInfo tess_eval_shader_stage_info {};
-	tess_eval_shader_stage_info.sType  = vk::StructureType::ePipelineShaderStageCreateInfo;
 	tess_eval_shader_stage_info.stage  = vk::ShaderStageFlagBits::eTessellationEvaluation;
 	tess_eval_shader_stage_info.module = tess_eval_shader_module;
 	tess_eval_shader_stage_info.pName  = "main";
@@ -581,181 +499,81 @@ void CreatePipelineInternal(
 		shader_stages[shader_stage_count++] = frag_shader_stage_info;
 	}
 
-	vk::VertexInputAttributeDescription input_attr[ShaderVertexInputInfo::RES_MAX];
-	vk::VertexInputBindingDescription   input_desc[ShaderVertexInputInfo::RES_MAX];
+	vk::VertexInputAttributeDescription input_attr[ShaderVertexInputInfo::RES_MAX] {};
+	vk::VertexInputBindingDescription   input_desc[ShaderVertexInputInfo::RES_MAX] {};
 
-	for (int bi = 0; bi < vs_input_info.buffers_num; bi++) {
-		const auto& b          = vs_input_info.buffers[bi];
-		input_desc[bi].binding = bi;
-		input_desc[bi].stride  = b.stride;
-		input_desc[bi].inputRate =
-		    (b.fetch_index == 0 ? vk::VertexInputRate::eVertex : vk::VertexInputRate::eInstance);
+	for (uint32_t binding = 0; binding < vertex_input.binding_count; binding++) {
+		input_desc[binding].binding   = binding;
+		input_desc[binding].stride    = vertex_input.bindings[binding].stride;
+		input_desc[binding].inputRate = vertex_input.bindings[binding].instance
+		                                    ? vk::VertexInputRate::eInstance
+		                                    : vk::VertexInputRate::eVertex;
+	}
+	for (uint32_t index = 0; index < vertex_input.attribute_count; index++) {
+		input_attr[index].binding  = vertex_input.attributes[index].binding;
+		input_attr[index].location = index;
+		input_attr[index].offset   = vertex_input.attributes[index].offset;
 
-		for (int ai = 0; ai < b.attr_num; ai++) {
-			auto index                 = b.attr_indices[ai];
-			input_attr[index].binding  = bi;
-			input_attr[index].location = index;
-			input_attr[index].offset   = b.attr_offsets[ai];
+		uint32_t   attr_size     = 4;
+		const auto registers_num = vs_input_info.resources_dst[index].registers_num;
+		const auto compiled_components =
+		    vs_input_info.stage.program->info.vertex_fetch_components[index];
+		const auto used_components =
+		    compiled_components > 0 ? static_cast<int>(compiled_components) : registers_num;
+		GetInputFormat(vs_input_info.resources[index], input_attr[index].format, attr_size,
+		               static_cast<uint32_t>(used_components));
 
-			uint32_t attr_size       = 4;
-			auto     registers_num   = vs_input_info.resources_dst[index].registers_num;
-			auto     used_components = (vs_input_info.resource_fetch_components[index] > 0
-			                                ? vs_input_info.resource_fetch_components[index]
-			                                : registers_num);
-			GetInputFormat(vs_input_info.resources[index], input_attr[index].format, attr_size,
-			               static_cast<uint32_t>(used_components));
-
-			if (graphics_debug_dump_enabled()) {
-				static std::atomic_uint log_count = 0;
-				const auto              log_id = log_count.fetch_add(1, std::memory_order_relaxed);
-				if (log_id < 128) {
-					LOGF("VertexInputState[%u]: attr=%u binding=%u offset=%u stride=%u fmt=%d "
-					     "src_fmt=%u dst=v%u regs=%u"
-					     " fetched_components=%u attr_size=%u swizzle=%u,%u,%u,%u\n",
-					     log_id, index, input_attr[index].binding, input_attr[index].offset,
-					     input_desc[bi].stride, static_cast<int>(input_attr[index].format),
-					     static_cast<uint32_t>(vs_input_info.resources[index].Format()),
-					     static_cast<uint32_t>(vs_input_info.resources_dst[index].register_start),
-					     static_cast<uint32_t>(registers_num),
-					     static_cast<uint32_t>(used_components), attr_size,
-					     static_cast<uint32_t>(vs_input_info.resources[index].DstSelX()),
-					     static_cast<uint32_t>(vs_input_info.resources[index].DstSelY()),
-					     static_cast<uint32_t>(vs_input_info.resources[index].DstSelZ()),
-					     static_cast<uint32_t>(vs_input_info.resources[index].DstSelW()));
-				}
-			}
-
-			if (vs_input_info.resources[index].OutOfBounds() != 0) {
-				static bool logged = false;
-				if (!logged) {
-					LOGF("VertexInput: temporary: accepting PS5 out-of-bounds behavior %" PRIu8
-					     "\n",
-					     vs_input_info.resources[index].OutOfBounds());
-					logged = true;
-				}
-			}
-
-			EXIT_NOT_IMPLEMENTED(vs_input_info.resources[index].AddTid());
-			EXIT_NOT_IMPLEMENTED(vs_input_info.resources[index].SwizzleEnabled());
-
-			auto log_unsupported_vertex_swizzle = [index, attr_size, registers_num](
-			                                          uint32_t swizzle, uint32_t expected) {
-				static bool logged = false;
-				if (!logged) {
-					LOGF(
-					    "VertexInput: temporary: accepting unsupported dst swizzle at attr %" PRIu32
-					    " (attr_size=%" PRIu32 ", regs=%" PRIu32 ", swizzle=0x%03" PRIx32
-					    ", expected=0x%03" PRIx32 ")\n",
-					    static_cast<uint32_t>(index), attr_size, registers_num, swizzle, expected);
-					logged = true;
-				}
-			};
-
-			switch (registers_num) {
-				case 1: {
-					auto swizzle = vs_input_info.resources[index].DstSelX();
-					if (swizzle != DstSel(4)) {
-						log_unsupported_vertex_swizzle(swizzle, DstSel(4));
-					}
-					break;
-				}
-				case 2: {
-					auto swizzle  = vs_input_info.resources[index].DstSelXY();
-					auto expected = (attr_size == 1 ? DstSel(4, 0) : DstSel(4, 5));
-					if (swizzle != expected) {
-						log_unsupported_vertex_swizzle(swizzle, expected);
-					}
-					break;
-				}
-				case 3: {
-					auto swizzle  = vs_input_info.resources[index].DstSelXYZ();
-					auto expected = DstSel(4, 5, 6);
-					switch (attr_size) {
-						case 1: expected = DstSel(4, 0, 0); break;
-						case 2: expected = DstSel(4, 5, 0); break;
-						default: break;
-					}
-					if (swizzle != expected) {
-						log_unsupported_vertex_swizzle(swizzle, expected);
-					}
-					break;
-				}
-				case 4: {
-					auto swizzle   = vs_input_info.resources[index].DstSelXYZW();
-					auto expected  = DstSel(4, 5, 6, 7);
-					bool supported = false;
-					switch (attr_size) {
-						case 1:
-							expected  = DstSel(4, 0, 0, 1);
-							supported = (swizzle == expected);
-							break;
-						case 2:
-							expected  = DstSel(4, 5, 0, 1);
-							supported = (swizzle == expected);
-							break;
-						case 3:
-							expected  = DstSel(4, 5, 6, 1);
-							supported = (swizzle == expected || swizzle == DstSel(4, 5, 6, 0));
-							break;
-						default:
-							supported = (swizzle == expected || swizzle == DstSel(4, 5, 6, 1) ||
-							             swizzle == DstSel(4, 5, 6, 0));
-							break;
-					}
-					if (!supported) {
-						log_unsupported_vertex_swizzle(swizzle, expected);
-					}
-					break;
-				}
-				default: EXIT("invalid registers_num");
+		if (graphics_debug_dump_enabled()) {
+			static std::atomic_uint log_count = 0;
+			const auto              log_id    = log_count.fetch_add(1, std::memory_order_relaxed);
+			if (log_id < 128) {
+				LOGF("VertexInputState[%u]: attr=%u binding=%u offset=%u stride=%u fmt=%d "
+				     "src_fmt=%u dst=v%u regs=%u"
+				     " fetched_components=%u attr_size=%u swizzle=%u,%u,%u,%u\n",
+				     log_id, index, input_attr[index].binding, input_attr[index].offset,
+				     input_desc[input_attr[index].binding].stride,
+				     static_cast<int>(input_attr[index].format),
+				     static_cast<uint32_t>(vs_input_info.resources[index].Format()),
+				     static_cast<uint32_t>(vs_input_info.resources_dst[index].register_start),
+				     static_cast<uint32_t>(registers_num), static_cast<uint32_t>(used_components),
+				     attr_size, static_cast<uint32_t>(vs_input_info.resources[index].DstSelX()),
+				     static_cast<uint32_t>(vs_input_info.resources[index].DstSelY()),
+				     static_cast<uint32_t>(vs_input_info.resources[index].DstSelZ()),
+				     static_cast<uint32_t>(vs_input_info.resources[index].DstSelW()));
 			}
 		}
+
+		if (vs_input_info.resources[index].OutOfBounds() != 0) {
+			static bool logged = false;
+			if (!logged) {
+				LOGF("VertexInput: temporary: accepting PS5 out-of-bounds behavior %" PRIu8 "\n",
+				     vs_input_info.resources[index].OutOfBounds());
+				logged = true;
+			}
+		}
+
+		EXIT_NOT_IMPLEMENTED(vs_input_info.resources[index].AddTid());
+		EXIT_NOT_IMPLEMENTED(vs_input_info.resources[index].SwizzleEnabled());
+
+		EXIT_IF(registers_num < 1 || registers_num > 4);
 	}
 
 	vk::PipelineVertexInputStateCreateInfo vertex_input_info {};
-	vertex_input_info.sType = vk::StructureType::ePipelineVertexInputStateCreateInfo;
-	vertex_input_info.pNext = nullptr;
-	vertex_input_info.flags = {};
-	vertex_input_info.vertexBindingDescriptionCount   = vs_input_info.buffers_num;
+	vertex_input_info.vertexBindingDescriptionCount   = vertex_input.binding_count;
 	vertex_input_info.pVertexBindingDescriptions      = input_desc;
-	vertex_input_info.vertexAttributeDescriptionCount = vs_input_info.resources_num;
+	vertex_input_info.vertexAttributeDescriptionCount = vertex_input.attribute_count;
 	vertex_input_info.pVertexAttributeDescriptions    = input_attr;
 
 	vk::PipelineInputAssemblyStateCreateInfo input_assembly {};
-	input_assembly.sType    = vk::StructureType::ePipelineInputAssemblyStateCreateInfo;
-	input_assembly.pNext    = nullptr;
-	input_assembly.flags    = {};
 	input_assembly.topology = static_params.topology;
 	input_assembly.primitiveRestartEnable =
 	    static_params.primitive_restart_enable ? VK_TRUE : VK_FALSE;
 
-	vk::Viewport viewport {};
-	viewport.x        = static_params.viewport_offset[0] - static_params.viewport_scale[0];
-	viewport.y        = static_params.viewport_offset[1] - static_params.viewport_scale[1];
-	viewport.width    = static_params.viewport_scale[0] * 2.0f;
-	viewport.height   = static_params.viewport_scale[1] * 2.0f;
-	viewport.minDepth = static_params.viewport_offset[2];
-	viewport.maxDepth = static_params.viewport_scale[2] + static_params.viewport_offset[2];
-
-	vk::Rect2D scissor {};
-	scissor.offset = {static_params.scissor_ltrb[0], static_params.scissor_ltrb[1]};
-	scissor.extent = {
-	    static_cast<uint32_t>(static_params.scissor_ltrb[2] - static_params.scissor_ltrb[0]),
-	    static_cast<uint32_t>(static_params.scissor_ltrb[3] - static_params.scissor_ltrb[1])};
-
 	vk::PipelineViewportDepthClipControlCreateInfoEXT depth_clip_control {};
-	depth_clip_control.sType = vk::StructureType::ePipelineViewportDepthClipControlCreateInfoEXT;
-	depth_clip_control.pNext = nullptr;
 	depth_clip_control.negativeOneToOne = (static_params.negative_one_to_one ? VK_TRUE : VK_FALSE);
 
 	vk::PipelineViewportStateCreateInfo viewport_state {};
-	viewport_state.sType         = vk::StructureType::ePipelineViewportStateCreateInfo;
-	viewport_state.pNext         = &depth_clip_control;
-	viewport_state.flags         = {};
-	viewport_state.viewportCount = 1;
-	viewport_state.pViewports    = &viewport;
-	viewport_state.scissorCount  = 1;
-	viewport_state.pScissors     = &scissor;
+	viewport_state.pNext = &depth_clip_control;
 
 	vk::CullModeFlags cull_mode = vk::CullModeFlagBits::eNone;
 	if (static_params.cull_back) {
@@ -769,65 +587,29 @@ void CreatePipelineInternal(
 	    (static_params.face ? vk::FrontFace::eClockwise : vk::FrontFace::eCounterClockwise);
 
 	vk::PipelineRasterizationDepthClipStateCreateInfoEXT clip_ext {};
-	clip_ext.sType           = vk::StructureType::ePipelineRasterizationDepthClipStateCreateInfoEXT;
-	clip_ext.pNext           = nullptr;
-	clip_ext.flags           = {};
 	clip_ext.depthClipEnable = static_params.depth_clip_enable ? VK_TRUE : VK_FALSE;
 
 	vk::PipelineRasterizationStateCreateInfo rasterizer {};
-	rasterizer.sType = vk::StructureType::ePipelineRasterizationStateCreateInfo;
 	// MoltenVK lacks VK_EXT_depth_clip_enable; omit the depth-clip struct on macOS and accept
 	// Vulkan's default depth clipping (enabled) instead of the PS5's clamp behavior.
-#if defined(__APPLE__)
-	rasterizer.pNext = nullptr;
-#else
+#if !defined(__APPLE__)
 	rasterizer.pNext = &clip_ext;
 #endif
-	rasterizer.flags                   = {};
-	rasterizer.depthClampEnable        = VK_FALSE;
-	rasterizer.rasterizerDiscardEnable = VK_FALSE;
-	rasterizer.polygonMode             = vk::PolygonMode::eFill;
-	rasterizer.cullMode                = cull_mode;
-	rasterizer.frontFace               = front_face;
-	rasterizer.depthBiasEnable         = VK_FALSE;
-	rasterizer.depthBiasConstantFactor = 0.0f;
-	rasterizer.depthBiasClamp          = 0.0f;
-	rasterizer.depthBiasSlopeFactor    = 0.0f;
-	rasterizer.lineWidth               = 1.0f;
+	rasterizer.cullMode  = cull_mode;
+	rasterizer.frontFace = front_face;
+	rasterizer.polygonMode = static_params.polygon_mode;
+	rasterizer.lineWidth = 1.0f;
 
 	vk::PipelineMultisampleStateCreateInfo multisampling {};
-	multisampling.sType                 = vk::StructureType::ePipelineMultisampleStateCreateInfo;
-	multisampling.pNext                 = nullptr;
-	multisampling.flags                 = {};
-	multisampling.sampleShadingEnable   = static_params.sample_shading_enable ? VK_TRUE : VK_FALSE;
-	multisampling.rasterizationSamples  = vulkan_sample_count(static_params.samples);
-	multisampling.minSampleShading      = 1.0f;
-	multisampling.pSampleMask           = nullptr;
-	multisampling.alphaToCoverageEnable = VK_FALSE;
-	multisampling.alphaToOneEnable      = VK_FALSE;
+	multisampling.sampleShadingEnable  = static_params.sample_shading_enable ? VK_TRUE : VK_FALSE;
+	multisampling.rasterizationSamples = vulkan_sample_count(static_params.samples);
+	multisampling.minSampleShading     = 1.0f;
 
 	vk::PipelineColorBlendAttachmentState color_blend_attachment[RENDER_COLOR_ATTACHMENTS_MAX] = {};
-	for (uint32_t i = 0; i < static_params.color_count; i++) {
-		vk::ColorComponentFlags color_write_mask = {};
+	for (uint32_t i = 0; i < rendering.color_count; i++) {
 		EXIT_NOT_IMPLEMENTED((static_params.color_mask[i] & ~0x0fu) != 0);
-		if ((static_params.color_mask[i] & 0x1u) != 0) {
-			color_write_mask |=
-			    static_cast<vk::ColorComponentFlags>(vk::ColorComponentFlagBits::eR);
-		}
-		if ((static_params.color_mask[i] & 0x2u) != 0) {
-			color_write_mask |=
-			    static_cast<vk::ColorComponentFlags>(vk::ColorComponentFlagBits::eG);
-		}
-		if ((static_params.color_mask[i] & 0x4u) != 0) {
-			color_write_mask |=
-			    static_cast<vk::ColorComponentFlags>(vk::ColorComponentFlagBits::eB);
-		}
-		if ((static_params.color_mask[i] & 0x8u) != 0) {
-			color_write_mask |=
-			    static_cast<vk::ColorComponentFlags>(vk::ColorComponentFlagBits::eA);
-		}
-
-		color_blend_attachment[i].colorWriteMask = color_write_mask;
+		color_blend_attachment[i].colorWriteMask =
+		    vk::ColorComponentFlags {static_params.color_mask[i]};
 		color_blend_attachment[i].blendEnable =
 		    (static_params.blend_enable[i] && !static_params.blend_bypass[i]) ? VK_TRUE : VK_FALSE;
 		color_blend_attachment[i].srcColorBlendFactor =
@@ -848,87 +630,60 @@ void CreatePipelineInternal(
 	}
 
 	vk::Bool32 color_write_enable[RENDER_COLOR_ATTACHMENTS_MAX] = {};
-	for (uint32_t i = 0; i < static_params.color_count; i++) {
+	for (uint32_t i = 0; i < rendering.color_count; i++) {
 		color_write_enable[i] = VK_TRUE;
 	}
 
 	vk::PipelineColorWriteCreateInfoEXT color_write {};
-	color_write.sType              = vk::StructureType::ePipelineColorWriteCreateInfoEXT;
-	color_write.pNext              = nullptr;
-	color_write.attachmentCount    = static_params.color_count;
+	color_write.attachmentCount    = rendering.color_count;
 	color_write.pColorWriteEnables = color_write_enable;
 
 	vk::PipelineColorBlendStateCreateInfo color_blending {};
-	color_blending.sType = vk::StructureType::ePipelineColorBlendStateCreateInfo;
 	// MoltenVK lacks VK_EXT_color_write_enable; drop the dynamic color-write struct on macOS
 	// and rely on each attachment's static colorWriteMask (all channels enabled by default).
-#if defined(__APPLE__)
-	color_blending.pNext = nullptr;
-#else
+#if !defined(__APPLE__)
 	color_blending.pNext = &color_write;
 #endif
-	color_blending.flags             = {};
-	color_blending.logicOpEnable     = VK_FALSE;
-	color_blending.logicOp           = vk::LogicOp::eCopy;
-	color_blending.attachmentCount   = static_params.color_count;
-	color_blending.pAttachments      = color_blend_attachment;
-	color_blending.blendConstants[0] = static_params.blend_color_red;
-	color_blending.blendConstants[1] = static_params.blend_color_green;
-	color_blending.blendConstants[2] = static_params.blend_color_blue;
-	color_blending.blendConstants[3] = static_params.blend_color_alpha;
+	color_blending.logicOp         = vk::LogicOp::eCopy;
+	color_blending.attachmentCount = rendering.color_count;
+	color_blending.pAttachments    = color_blend_attachment;
 
-	vk::DescriptorSetLayout set_layouts[2]  = {};
-	uint32_t                set_layouts_num = 0;
-
-	vk::PushConstantRange push_constant_info[2];
-	uint32_t              push_constant_info_num = 0;
-
-	EXIT_IF(!vs_input_info.stage);
-	CreateLayout(descriptor_cache, set_layouts, set_layouts_num, push_constant_info,
-	             push_constant_info_num, *vs_input_info.stage.program,
-	             vk::ShaderStageFlagBits::eVertex, DescriptorCache::Stage::Vertex);
+	std::vector<vk::DescriptorSetLayoutBinding> descriptor_bindings;
+	AddLayoutBindings(descriptor_bindings, *vs_input_info.stage.program, vertex_stage);
 	if (ps_active) {
 		EXIT_IF(!ps_input_info->stage);
-		CreateLayout(descriptor_cache, set_layouts, set_layouts_num, push_constant_info,
-		             push_constant_info_num, *ps_input_info->stage.program,
-		             vk::ShaderStageFlagBits::eFragment, DescriptorCache::Stage::Pixel);
+		AddLayoutBindings(descriptor_bindings, *ps_input_info->stage.program,
+		                  vk::ShaderStageFlagBits::eFragment);
 	}
+	CreateDescriptorLayout(graphics, pipeline, descriptor_bindings);
+	const auto                  GraphicsStages = vertex_stage | vk::ShaderStageFlagBits::eFragment;
+	const vk::PushConstantRange push_constants {GraphicsStages, 0,
+	                                            ShaderRecompiler::IR::NativePushConstantSize};
 
 	vk::PipelineLayoutCreateInfo pipeline_layout_info {};
-	pipeline_layout_info.sType                  = vk::StructureType::ePipelineLayoutCreateInfo;
-	pipeline_layout_info.pNext                  = nullptr;
-	pipeline_layout_info.flags                  = {};
-	pipeline_layout_info.setLayoutCount         = set_layouts_num;
-	pipeline_layout_info.pSetLayouts            = (set_layouts_num > 0 ? set_layouts : nullptr);
-	pipeline_layout_info.pushConstantRangeCount = push_constant_info_num;
-	pipeline_layout_info.pPushConstantRanges =
-	    push_constant_info_num > 0 ? push_constant_info : nullptr;
+	pipeline_layout_info.setLayoutCount         = 1;
+	pipeline_layout_info.pSetLayouts            = &pipeline.descriptor_set_layout;
+	pipeline_layout_info.pushConstantRangeCount = 1;
+	pipeline_layout_info.pPushConstantRanges    = &push_constants;
 
 	EXIT_IF(pipeline.pipeline_layout != nullptr);
 
 	if (graphics_debug_dump_enabled()) {
-		LOGF("PipelineTrace: vkCreatePipelineLayout begin VS=0x%08" PRIx32 "/0x%08" PRIx32
-		     " PS=0x%08" PRIx32 "/0x%08" PRIx32 " set_layouts=%" PRIu32 " push_constants=%" PRIu32
-		     "\n",
-		     vs_hash0, vs_crc32, ps_hash0, ps_crc32, set_layouts_num, push_constant_info_num);
+		LOGF("PipelineTrace: vkCreatePipelineLayout begin VS=%" PRIu64 " PS=%" PRIu64
+		     " set_layouts=1 push_constants=%" PRIu32 "\n",
+		     vertex_program.id, ps_active ? pixel_program.id : 0, 1u);
 	}
 	result = graphics.device.createPipelineLayout(&pipeline_layout_info, nullptr,
 	                                              &pipeline.pipeline_layout);
 	if (graphics_debug_dump_enabled()) {
 		LOGF("PipelineTrace: vkCreatePipelineLayout done result=%s layout=%p\n",
-		     VulkanToString(result).c_str(), static_cast<void*>(pipeline.pipeline_layout));
+		     vk::to_string(result).c_str(), static_cast<void*>(pipeline.pipeline_layout));
 	}
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
 
 	EXIT_NOT_IMPLEMENTED(pipeline.pipeline_layout == nullptr);
 
 	vk::PipelineDepthStencilStateCreateInfo depth_stencil_info {};
-	depth_stencil_info.sType            = vk::StructureType::ePipelineDepthStencilStateCreateInfo;
-	depth_stencil_info.pNext            = nullptr;
-	depth_stencil_info.flags            = {};
-	depth_stencil_info.depthTestEnable  = (static_params.depth_test_enable ? VK_TRUE : VK_FALSE);
-	depth_stencil_info.depthWriteEnable = (static_params.depth_write_enable ? VK_TRUE : VK_FALSE);
-	depth_stencil_info.depthCompareOp   = static_params.depth_compare_op;
 	depth_stencil_info.depthBoundsTestEnable =
 #if defined(__APPLE__)
 	    VK_FALSE; // MoltenVK lacks the depthBounds feature; depth-bounds testing is disabled
@@ -947,190 +702,145 @@ void CreatePipelineInternal(
 	depth_stencil_info.minDepthBounds    = static_params.depth_min_bounds;
 	depth_stencil_info.maxDepthBounds    = static_params.depth_max_bounds;
 
-	const vk::DynamicState dynamic_states[] = {
-	    vk::DynamicState::eViewport,
-	    vk::DynamicState::eScissor,
+	std::vector<vk::DynamicState> dynamic_states {
+	    vk::DynamicState::eViewportWithCount,
+	    vk::DynamicState::eScissorWithCount,
 	    vk::DynamicState::eLineWidth,
+	    vk::DynamicState::eDepthTestEnable,
+	    vk::DynamicState::eDepthWriteEnable,
+	    vk::DynamicState::eDepthCompareOp,
 	    vk::DynamicState::eDepthBiasEnable,
 	    vk::DynamicState::eDepthBias,
 	    vk::DynamicState::eStencilCompareMask,
 	    vk::DynamicState::eStencilReference,
 	    vk::DynamicState::eStencilWriteMask,
-#if !defined(__APPLE__)
-	    // Keep last so depth-only pipelines can omit this dynamic state.
-	    vk::DynamicState::eColorWriteEnableEXT, // unsupported by MoltenVK; static mask instead
-#endif
+	    vk::DynamicState::eBlendConstants,
 	};
-	auto dynamic_states_count =
-	    static_cast<uint32_t>(sizeof(dynamic_states) / sizeof(dynamic_states[0]));
 #if !defined(__APPLE__)
-	if (static_params.color_count == 0) {
-		dynamic_states_count--;
+	if (rendering.color_count != 0) {
+		dynamic_states.push_back(vk::DynamicState::eColorWriteEnableEXT);
 	}
 #endif
+	if (graphics.attachment_feedback_loop_enabled) {
+		dynamic_states.push_back(vk::DynamicState::eAttachmentFeedbackLoopEnableEXT);
+	}
 
 	vk::PipelineDynamicStateCreateInfo dynamic_state {};
-	dynamic_state.sType             = vk::StructureType::ePipelineDynamicStateCreateInfo;
-	dynamic_state.pNext             = nullptr;
-	dynamic_state.flags             = {};
-	dynamic_state.dynamicStateCount = dynamic_states_count;
-	dynamic_state.pDynamicStates    = dynamic_states;
+	dynamic_state.dynamicStateCount = static_cast<uint32_t>(dynamic_states.size());
+	dynamic_state.pDynamicStates    = dynamic_states.data();
 
 	vk::GraphicsPipelineCreateInfo  pipeline_info {};
 	vk::PipelineRenderingCreateInfo rendering_info {};
-	rendering_info.sType                   = vk::StructureType::ePipelineRenderingCreateInfo;
 	rendering_info.colorAttachmentCount    = rendering.color_count;
 	rendering_info.pColorAttachmentFormats = rendering.color_formats.data();
 	rendering_info.depthAttachmentFormat   = rendering.depth_format;
 	rendering_info.stencilAttachmentFormat = rendering.stencil_format;
-	pipeline_info.sType                    = vk::StructureType::eGraphicsPipelineCreateInfo;
 	pipeline_info.pNext                    = &rendering_info;
-	pipeline_info.flags                    = {};
 	pipeline_info.stageCount               = shader_stage_count;
 	pipeline_info.pStages                  = shader_stages;
-	pipeline_info.pVertexInputState        = &vertex_input_info;
-	pipeline_info.pInputAssemblyState      = &input_assembly;
+	pipeline_info.pVertexInputState        = mesh ? nullptr : &vertex_input_info;
+	pipeline_info.pInputAssemblyState      = mesh ? nullptr : &input_assembly;
 	vk::PipelineTessellationStateCreateInfo tessellation_state {};
-	tessellation_state.sType              = vk::StructureType::ePipelineTessellationStateCreateInfo;
 	tessellation_state.patchControlPoints = 3;
 	pipeline_info.pTessellationState      = (rect_list ? &tessellation_state : nullptr);
 	pipeline_info.pViewportState          = &viewport_state;
 	pipeline_info.pRasterizationState     = &rasterizer;
 	pipeline_info.pMultisampleState       = &multisampling;
-	pipeline_info.pDepthStencilState = (static_params.with_depth ? &depth_stencil_info : nullptr);
-	pipeline_info.pColorBlendState   = &color_blending;
-	pipeline_info.pDynamicState      = &dynamic_state;
-	pipeline_info.layout             = pipeline.pipeline_layout;
-	pipeline_info.renderPass         = nullptr;
-	pipeline_info.subpass            = 0;
-	pipeline_info.basePipelineHandle = nullptr;
-	pipeline_info.basePipelineIndex  = -1;
+	pipeline_info.pDepthStencilState      = (with_depth ? &depth_stencil_info : nullptr);
+	pipeline_info.pColorBlendState        = &color_blending;
+	pipeline_info.pDynamicState           = &dynamic_state;
+	pipeline_info.layout                  = pipeline.pipeline_layout;
+	pipeline_info.basePipelineIndex       = -1;
 
 	EXIT_IF(pipeline.pipeline != nullptr);
 
 	if (graphics_debug_dump_enabled()) {
-		LOGF("PipelineTrace: vkCreateGraphicsPipelines begin VS=0x%08" PRIx32 "/0x%08" PRIx32
-		     " PS=0x%08" PRIx32 "/0x%08" PRIx32 " topology=%" PRIu32 " color_mask=0x%08" PRIx32
-		     " depth=%s blend=%s dyn_states=%" PRIu32 " viewport=%f,%f %fx%f"
-		     " scissor=%d,%d %ux%u\n",
-		     vs_hash0, vs_crc32, ps_hash0, ps_crc32, static_cast<uint32_t>(static_params.topology),
-		     static_params.color_mask[0], (static_params.with_depth ? "true" : "false"),
-		     (static_params.blend_enable[0] ? "true" : "false"), dynamic_states_count, viewport.x,
-		     viewport.y, viewport.width, viewport.height, scissor.offset.x, scissor.offset.y,
-		     scissor.extent.width, scissor.extent.height);
+		LOGF("PipelineTrace: vkCreateGraphicsPipelines begin VS=%" PRIu64 " PS=%" PRIu64
+		     " topology=%" PRIu32 " color_mask=0x%08" PRIx32
+		     " depth=%s blend=%s dyn_states=%" PRIu32 "\n",
+		     vertex_program.id, ps_active ? pixel_program.id : 0,
+		     static_cast<uint32_t>(static_params.topology), static_params.color_mask[0],
+		     (with_depth ? "true" : "false"), (static_params.blend_enable[0] ? "true" : "false"),
+		     dynamic_state.dynamicStateCount);
 	}
-	result = graphics.device.createGraphicsPipelines(nullptr, 1, &pipeline_info, nullptr,
+	result = graphics.device.createGraphicsPipelines(driver_cache, 1, &pipeline_info, nullptr,
 	                                                 &pipeline.pipeline);
 	if (graphics_debug_dump_enabled()) {
 		LOGF("PipelineTrace: vkCreateGraphicsPipelines done result=%s pipeline=%p\n",
-		     VulkanToString(result).c_str(), static_cast<void*>(pipeline.pipeline));
+		     vk::to_string(result).c_str(), static_cast<void*>(pipeline.pipeline));
 	}
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
 
 	EXIT_NOT_IMPLEMENTED(pipeline.pipeline == nullptr);
 
-	if (frag_shader_module != nullptr) {
-		graphics.device.destroyShaderModule(frag_shader_module, nullptr);
-	}
 	if (tess_control_shader_module != nullptr) {
 		graphics.device.destroyShaderModule(tess_control_shader_module, nullptr);
 	}
 	if (tess_eval_shader_module != nullptr) {
 		graphics.device.destroyShaderModule(tess_eval_shader_module, nullptr);
 	}
-	graphics.device.destroyShaderModule(vert_shader_module, nullptr);
 }
 
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
-void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descriptor_cache,
-                            PipelineCache::ComputePipeline& pipeline,
-                            const ShaderComputeInputInfo&   input_info,
-                            std::span<const uint32_t>       cs_shader) {
-	vk::ShaderModule comp_shader_module = nullptr;
-
-	vk::ShaderModuleCreateInfo create_info {};
-
-	create_info.sType    = vk::StructureType::eShaderModuleCreateInfo;
-	create_info.pNext    = nullptr;
-	create_info.flags    = {};
-	create_info.codeSize = cs_shader.size() * 4;
-	create_info.pCode    = cs_shader.data();
-	LOGF("PipelineTrace: vkCreateShaderModule CS begin words=%" PRIu64 "\n",
-	     static_cast<uint64_t>(cs_shader.size()));
-	auto result = graphics.device.createShaderModule(&create_info, nullptr, &comp_shader_module);
-	LOGF("PipelineTrace: vkCreateShaderModule CS done result=%s module=%p\n",
-	     VulkanToString(result).c_str(), static_cast<void*>(comp_shader_module));
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
-
-	EXIT_NOT_IMPLEMENTED(comp_shader_module == nullptr);
+void CreatePipelineInternal(GraphicContext& graphics, PipelineCache::Pipeline& pipeline,
+                            const ShaderComputeInputInfo& input_info,
+                            vk::ShaderModule compute_module, vk::PipelineCache driver_cache) {
+	EXIT_IF(compute_module == nullptr);
 
 	vk::PipelineShaderStageCreateInfo                     comp_shader_stage_info {};
 	vk::PipelineShaderStageRequiredSubgroupSizeCreateInfo comp_subgroup_size {};
-	comp_shader_stage_info.sType               = vk::StructureType::ePipelineShaderStageCreateInfo;
-	comp_shader_stage_info.pNext               = nullptr;
-	comp_shader_stage_info.flags               = {};
-	comp_shader_stage_info.stage               = vk::ShaderStageFlagBits::eCompute;
-	comp_shader_stage_info.module              = comp_shader_module;
-	comp_shader_stage_info.pName               = "main";
-	comp_shader_stage_info.pSpecializationInfo = nullptr;
+	comp_shader_stage_info.stage  = vk::ShaderStageFlagBits::eCompute;
+	comp_shader_stage_info.module = compute_module;
+	comp_shader_stage_info.pName  = "main";
 	EXIT_IF(!input_info.stage);
-	ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eCompute, *input_info.stage.program,
-	                      comp_subgroup_size, comp_shader_stage_info);
+	const auto wave_size = input_info.stage.program->wave_size;
+	if (graphics.compute_subgroup_size_control_enabled &&
+	    wave_size >= graphics.min_subgroup_size && wave_size <= graphics.max_subgroup_size) {
+		comp_subgroup_size.requiredSubgroupSize = wave_size;
+		comp_shader_stage_info.pNext            = &comp_subgroup_size;
+	}
 
-	vk::DescriptorSetLayout set_layouts[1]  = {};
-	uint32_t                set_layouts_num = 0;
-
-	vk::PushConstantRange push_constant_info[1];
-	uint32_t              push_constant_info_num = 0;
-
-	EXIT_IF(!input_info.stage);
-	CreateLayout(descriptor_cache, set_layouts, set_layouts_num, push_constant_info,
-	             push_constant_info_num, *input_info.stage.program,
-	             vk::ShaderStageFlagBits::eCompute, DescriptorCache::Stage::Compute);
+	std::vector<vk::DescriptorSetLayoutBinding> descriptor_bindings;
+	AddLayoutBindings(descriptor_bindings, *input_info.stage.program,
+	                  vk::ShaderStageFlagBits::eCompute);
+	CreateDescriptorLayout(graphics, pipeline, descriptor_bindings);
+	const vk::PushConstantRange push_constants {vk::ShaderStageFlagBits::eCompute, 0,
+	                                            ShaderRecompiler::IR::NativePushConstantSize};
 
 	vk::PipelineLayoutCreateInfo pipeline_layout_info {};
-	pipeline_layout_info.sType                  = vk::StructureType::ePipelineLayoutCreateInfo;
-	pipeline_layout_info.pNext                  = nullptr;
-	pipeline_layout_info.flags                  = {};
-	pipeline_layout_info.setLayoutCount         = set_layouts_num;
-	pipeline_layout_info.pSetLayouts            = (set_layouts_num > 0 ? set_layouts : nullptr);
-	pipeline_layout_info.pushConstantRangeCount = push_constant_info_num;
-	pipeline_layout_info.pPushConstantRanges =
-	    push_constant_info_num > 0 ? push_constant_info : nullptr;
+	pipeline_layout_info.setLayoutCount         = 1;
+	pipeline_layout_info.pSetLayouts            = &pipeline.descriptor_set_layout;
+	pipeline_layout_info.pushConstantRangeCount = 1;
+	pipeline_layout_info.pPushConstantRanges    = &push_constants;
 
 	EXIT_IF(pipeline.pipeline_layout != nullptr);
 
-	LOGF("PipelineTrace: vkCreatePipelineLayout CS begin set_layouts=%u push_constants=%u\n",
-	     set_layouts_num, push_constant_info_num);
-	result = graphics.device.createPipelineLayout(&pipeline_layout_info, nullptr,
-	                                              &pipeline.pipeline_layout);
+	LOGF("PipelineTrace: vkCreatePipelineLayout CS begin set_layouts=1 push_constants=%u\n",
+	     1u);
+	auto result = graphics.device.createPipelineLayout(&pipeline_layout_info, nullptr,
+	                                                  &pipeline.pipeline_layout);
 	LOGF("PipelineTrace: vkCreatePipelineLayout CS done result=%s layout=%p\n",
-	     VulkanToString(result).c_str(), static_cast<void*>(pipeline.pipeline_layout));
+	     vk::to_string(result).c_str(), static_cast<void*>(pipeline.pipeline_layout));
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
 
 	EXIT_NOT_IMPLEMENTED(pipeline.pipeline_layout == nullptr);
 
 	vk::ComputePipelineCreateInfo info {};
-	info.sType              = vk::StructureType::eComputePipelineCreateInfo;
-	info.pNext              = nullptr;
-	info.flags              = {};
-	info.stage              = comp_shader_stage_info;
-	info.layout             = pipeline.pipeline_layout;
-	info.basePipelineHandle = nullptr;
-	info.basePipelineIndex  = -1;
+	info.stage             = comp_shader_stage_info;
+	info.layout            = pipeline.pipeline_layout;
+	info.basePipelineIndex = -1;
 
 	EXIT_IF(pipeline.pipeline != nullptr);
 
 	LOGF("PipelineTrace: vkCreateComputePipelines begin layout=%p\n",
 	     static_cast<void*>(pipeline.pipeline_layout));
-	result = graphics.device.createComputePipelines(nullptr, 1, &info, nullptr, &pipeline.pipeline);
+	result = graphics.device.createComputePipelines(driver_cache, 1, &info, nullptr,
+	                                                &pipeline.pipeline);
 	LOGF("PipelineTrace: vkCreateComputePipelines done result=%s pipeline=%p\n",
-	     VulkanToString(result).c_str(), static_cast<void*>(pipeline.pipeline));
+	     vk::to_string(result).c_str(), static_cast<void*>(pipeline.pipeline));
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
 
 	EXIT_NOT_IMPLEMENTED(pipeline.pipeline == nullptr);
-
-	graphics.device.destroyShaderModule(comp_shader_module, nullptr);
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/render.h

@@ -4,12 +4,14 @@
 #include "common/abi.h"
 #include "common/assert.h"
 #include "common/common.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
+#include "graphics/host_gpu/renderer/pipeline/descriptors.h"
+#include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
 #include "graphics/host_gpu/renderer/renderTarget.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 
 #include <array>
-#include <memory>
+#include <optional>
+#include <span>
 #include <vector>
 
 namespace Libs::Graphics {
@@ -23,9 +25,6 @@ class Shader;
 struct GraphicContext;
 struct ShaderBufferResource;
 struct ShaderComputeInputInfo;
-struct CommandSlot;
-struct VulkanBuffer;
-struct VulkanDescriptorSet;
 struct RenderDepthInfo;
 struct RenderColorInfo;
 struct DrawCallInfo;
@@ -49,18 +48,29 @@ enum class CommandBufferDebugOp : uint32_t {
 	Unknown,
 };
 
-class FenceResourceRetainer {
-public:
-	FenceResourceRetainer() = default;
-	~FenceResourceRetainer();
-	KYTY_CLASS_NO_COPY(FenceResourceRetainer);
+enum class DrawOffsetSource : uint8_t {
+	DrawState,
+	IndirectArgs,
+};
 
-	void               Retain(std::shared_ptr<void> resource);
-	void               ReleaseAfterFence() noexcept;
-	[[nodiscard]] bool Empty() const noexcept { return m_resources.empty(); }
+struct DrawIndexArgs {
+	uint32_t         index_count                = 0;
+	const void*      index_addr                 = nullptr;
+	uint32_t         instance_count             = 0;
+	uint32_t         index_type_and_size        = 0;
+	int32_t          base_vertex                = 0;
+	uint32_t         first_instance             = 0;
+	DrawOffsetSource offset_source              = DrawOffsetSource::DrawState;
+	uint32_t         render_target_slice_offset = 0;
+};
 
-private:
-	std::vector<std::shared_ptr<void>> m_resources;
+struct DrawAutoArgs {
+	uint32_t         vertex_count               = 0;
+	uint32_t         instance_count             = 0;
+	uint32_t         first_vertex               = 0;
+	uint32_t         first_instance             = 0;
+	DrawOffsetSource offset_source              = DrawOffsetSource::DrawState;
+	uint32_t         render_target_slice_offset = 0;
 };
 
 struct SubmitInfo {
@@ -91,78 +101,52 @@ struct SubmitInfo {
 
 class CommandBuffer {
 public:
-	explicit CommandBuffer(CommandScheduler& scheduler);
-	~CommandBuffer();
+	~CommandBuffer() = default;
 
 	KYTY_CLASS_NO_COPY(CommandBuffer);
 
 	[[nodiscard]] bool IsInvalid() const;
 
-	void Begin() const;
-	void End() const;
-	void Execute(const SubmitInfo& submit = {});
 	void SetDebugInfo(uint32_t op, uint64_t submit_id, uint32_t arg0 = 0, uint32_t arg1 = 0,
 	                  uint32_t arg2 = 0, uint32_t arg3 = 0, uint64_t arg4 = 0);
 	void BeginRendering(const RenderState& state) const;
 	void EndRendering() const;
-	void WaitForFenceOnly();
-	void WaitForFence();
-	void WaitForFenceAndReset();
-	void RetireBufferAfterFence(std::unique_ptr<VulkanBuffer> buffer);
-	void RetainResourceUntilFence(std::shared_ptr<void> resource);
-	void RecycleDescriptorAfterFence(VulkanDescriptorSet& set);
 
 	[[nodiscard]] vk::CommandBuffer Handle() const;
 	[[nodiscard]] GraphicContext&   GetGraphics() const noexcept { return m_graphics; }
 	[[nodiscard]] RenderContext&    GetContext() const noexcept { return m_context; }
-	[[nodiscard]] bool              IsExecute() const { return m_execute; }
+	[[nodiscard]] HW::Context&      GetRegisters() const noexcept { return *m_registers; }
+	[[nodiscard]] HW::UserConfig&   GetUserConfig() const noexcept { return *m_user_config; }
+	[[nodiscard]] HW::Shader&       GetShaders() const noexcept { return *m_shaders; }
 
 private:
-	void Release();
-	void FinalizeFence(bool reset_recording);
-	void ReleaseResourcesAfterFence();
-	void DeleteBuffersAfterFence();
-	void RecycleDescriptorsAfterFence();
-
-	RenderContext&                             m_context;
-	CommandScheduler&                          m_scheduler;
-	GraphicContext&                            m_graphics;
-	CommandSlot*                               m_slot            = nullptr;
-	bool                                       m_execute         = false;
-	bool                                       m_fence_waited    = false;
-	uint64_t                                   m_submit_seq      = 0;
-	uint32_t                                   m_debug_op        = 0;
-	uint64_t                                   m_debug_submit_id = 0;
-	uint32_t                                   m_debug_arg0      = 0;
-	uint32_t                                   m_debug_arg1      = 0;
-	uint32_t                                   m_debug_arg2      = 0;
-	uint32_t                                   m_debug_arg3      = 0;
-	uint64_t                                   m_debug_arg4      = 0;
-	std::vector<std::unique_ptr<VulkanBuffer>> m_retired_buffers;
-	FenceResourceRetainer                      m_fence_resources;
-	std::vector<VulkanDescriptorSet*>          m_descriptor_sets_after_fence;
-	mutable RenderState                        m_render_state;
-	mutable bool                               m_rendering = false;
-};
-
-class RenderCommandBuffer final: public CommandBuffer {
-public:
-	explicit RenderCommandBuffer(CommandScheduler& scheduler): CommandBuffer(scheduler) {}
-
+	explicit CommandBuffer(CommandScheduler& scheduler);
 	void Bind(HW::Context& registers, HW::UserConfig& user_config, HW::Shader& shaders) noexcept {
 		m_registers   = &registers;
 		m_user_config = &user_config;
 		m_shaders     = &shaders;
 	}
 
-	[[nodiscard]] HW::Context&    GetRegisters() const noexcept { return *m_registers; }
-	[[nodiscard]] HW::UserConfig& GetUserConfig() const noexcept { return *m_user_config; }
-	[[nodiscard]] HW::Shader&     GetShaders() const noexcept { return *m_shaders; }
+	void Begin();
+	void End() const;
 
-private:
-	HW::Context*    m_registers   = nullptr;
-	HW::UserConfig* m_user_config = nullptr;
-	HW::Shader*     m_shaders     = nullptr;
+	RenderContext&      m_context;
+	GraphicContext&     m_graphics;
+	vk::CommandBuffer   m_buffer          = nullptr;
+	uint32_t            m_debug_op        = 0;
+	uint64_t            m_debug_submit_id = 0;
+	uint32_t            m_debug_arg0      = 0;
+	uint32_t            m_debug_arg1      = 0;
+	uint32_t            m_debug_arg2      = 0;
+	uint32_t            m_debug_arg3      = 0;
+	uint64_t            m_debug_arg4      = 0;
+	mutable RenderState m_render_state;
+	mutable bool        m_rendering   = false;
+	HW::Context*        m_registers   = nullptr;
+	HW::UserConfig*     m_user_config = nullptr;
+	HW::Shader*         m_shaders     = nullptr;
+
+	friend class CommandScheduler;
 };
 
 class RenderExecutor {
@@ -170,72 +154,73 @@ public:
 	explicit RenderExecutor(RenderContext& context): m_context(context) {}
 	KYTY_CLASS_NO_COPY(RenderExecutor);
 
-	void DrawIndex(uint64_t submit_id, RenderCommandBuffer& buffer, uint32_t index_type_and_size,
-	               uint32_t index_count, const void* index_addr, uint32_t flags, uint32_t type,
-	               uint32_t instance_count = 1, uint32_t render_target_slice_offset = 0,
-	               int32_t vertex_offset_add = 0, uint32_t first_instance = 0);
-	void DrawAuto(uint64_t submit_id, RenderCommandBuffer& buffer, uint32_t index_count,
-	              uint32_t flags, uint32_t render_target_slice_offset = 0,
-	              uint32_t instance_count = 1, uint32_t first_vertex = 0,
-	              uint32_t first_instance = 0);
-	void DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buffer, uint32_t thread_group_x,
+	void DispatchDirect(uint64_t submit_id, CommandBuffer& buffer, uint32_t thread_group_x,
 	                    uint32_t thread_group_y, uint32_t thread_group_z, uint32_t mode);
 
-	[[nodiscard]] DescriptorCache::PreparedBindings
-	     PrepareBindings(CommandBuffer& buffer, const ShaderStageRuntime& runtime,
-	                     vk::ShaderStageFlags shader_stage, DescriptorCache::Stage stage);
-	void RebindBuffers(CommandBuffer& buffer, DescriptorCache::PreparedBindings& bindings);
-	void RebindImages(CommandBuffer& buffer, DescriptorCache::PreparedBindings& bindings);
+	[[nodiscard]] PreparedBindings PrepareBindings(const ShaderStageRuntime& runtime);
+	void                           FindBuffers(PreparedBindings& bindings);
+	void                           RebindBuffers(PreparedBindings& bindings);
+	void                           RebindImages(PreparedBindings& bindings);
 	void CommitBindings(CommandBuffer& buffer, vk::PipelineBindPoint pipeline_bind_point,
-	                    vk::PipelineLayout layout, DescriptorCache::PreparedBindings& bindings);
+	                    const PipelineCache::Pipeline&     pipeline,
+	                    std::span<PreparedBindings* const> bindings);
 
 private:
+	void DrawIndex(uint64_t submit_id, CommandBuffer& buffer, const DrawIndexArgs& args);
+	void DrawAuto(uint64_t submit_id, CommandBuffer& buffer, const DrawAutoArgs& args);
+
 	struct GraphicsBindings {
-		DescriptorCache::PreparedBindings                vertex;
-		std::optional<DescriptorCache::PreparedBindings> pixel;
+		PreparedBindings                vertex;
+		std::optional<PreparedBindings> pixel;
 	};
 
-	[[nodiscard]] DescriptorCache::TextureBinding
-	ResolveTexture(const ShaderRecompiler::IR::ImageResource&   resource,
-	               const ShaderRecompiler::IR::DescriptorValue& value);
-	[[nodiscard]] GraphicsBindings PrepareGraphicsBindings(CommandBuffer&            buffer,
-	                                                       const ShaderStageRuntime& vertex,
+	[[nodiscard]] TextureBinding ResolveTexture(const ShaderRecompiler::IR::ImageResource& resource,
+	                                            const ShaderRecompiler::IR::DescriptorValue& value);
+	[[nodiscard]] GraphicsBindings PrepareGraphicsBindings(const ShaderStageRuntime& vertex,
 	                                                       const ShaderStageRuntime& pixel,
 	                                                       bool                      pixel_active);
-	void ResolveRenderColorTarget(uint64_t submit_id, RenderCommandBuffer& buffer,
+	void ResolveRenderColorTarget(uint64_t submit_id, CommandBuffer& buffer,
 	                              RenderColorInfo& target, uint32_t render_target_slice_offset = 0,
 	                              uint32_t render_target_slot = UINT32_MAX,
 	                              bool ignore_target_mask = false, bool exact_format = false);
-	void ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandBuffer& buffer,
+	void ResolveRenderDepthTarget(uint64_t submit_id, CommandBuffer& buffer,
 	                              RenderDepthInfo& target);
-	[[nodiscard]] bool PrepareDrawRenderState(uint64_t submit_id, RenderCommandBuffer& buffer,
+	[[nodiscard]] bool PrepareDrawRenderState(uint64_t submit_id, CommandBuffer& buffer,
 	                                          const DrawCallInfo& draw,
 	                                          uint32_t            render_target_slice_offset,
 	                                          bool log_setup_phases, DrawRenderState& state);
-	void               ExecutePreparedDraw(uint64_t submit_id, RenderCommandBuffer& buffer,
-	                                       const DrawCallInfo& draw, DrawRenderState& state,
-	                                       vk::PrimitiveTopology topology, const DrawEmitInfo& emit,
-	                                       const DrawIndexBufferSource& index_source,
-	                                       bool primitive_restart_enable, bool log_pipeline_phase,
-	                                       bool set_bind_debug, bool set_auto_debug);
+	void ExecutePreparedDraw(uint64_t submit_id, CommandBuffer& buffer, const DrawCallInfo& draw,
+	                         DrawRenderState& state, vk::PrimitiveTopology topology,
+	                         const DrawEmitInfo& emit, const DrawIndexBufferSource& index_source,
+	                         bool primitive_restart_enable, bool log_pipeline_phase,
+	                         bool set_bind_debug, bool set_auto_debug);
 	[[nodiscard]] RenderState AcquireRenderTargets(CommandBuffer& buffer, RenderColorInfo* colors,
-	                                               uint32_t color_count, RenderDepthInfo& depth);
-	[[nodiscard]] bool        ResolveColorTargets(uint64_t submit_id, RenderCommandBuffer& buffer,
+	                                               uint32_t color_count, RenderDepthInfo& depth,
+	                                               const std::optional<PreparedBindings>& pixel = std::nullopt);
+	[[nodiscard]] bool        ResolveColorTargets(uint64_t submit_id, CommandBuffer& buffer,
 	                                              uint32_t render_target_slice_offset);
 	void                      BindImage(ImageId id, bool storage);
 	void                      BindRenderTarget(ImageId id);
 	void                      TrackImageBinding(ImageId id);
 	void                      ResetBindings();
 	[[nodiscard]] bool        TryConsumeComputeMetaClear(const ShaderComputeInputInfo& input,
-	                                                     const RenderCommandBuffer&    buffer);
+	                                                     const CommandBuffer&          buffer);
+	[[nodiscard]] bool TryConsumeComputeImageClear(const ShaderComputeInputInfo& input,
+	                                              CommandBuffer& command, uint32_t group_x,
+	                                              uint32_t group_y, uint32_t group_z, uint32_t mode);
 
-	RenderContext&                      m_context;
-	std::vector<std::shared_ptr<Image>> m_bound_images;
+	RenderContext&                        m_context;
+	std::vector<ImageId>                  m_bound_images;
+	std::vector<vk::DescriptorBufferInfo> m_descriptor_buffers;
+	std::vector<vk::DescriptorImageInfo>  m_descriptor_images;
+	std::vector<vk::WriteDescriptorSet>   m_descriptor_writes;
+	std::vector<uint32_t>                 m_image_occurrences;
 
+	friend class CommandProcessor;
 	friend struct RenderExecutorTestAccess;
 };
 
-[[nodiscard]] bool ResolveComputeImageClear(const ShaderComputeInputInfo& input, uint32_t group_x,
+[[nodiscard]] bool ResolveComputeBufferFill(const ShaderComputeInputInfo& input, uint32_t group_x,
                                             uint32_t group_y, uint32_t group_z, uint32_t mode,
                                             ShaderBufferResource& descriptor,
                                             uint32_t& packed_clear, uint64_t& size);

src/graphics/host_gpu/renderer/renderCompute.cpp

@@ -11,11 +11,9 @@
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/host_gpu/renderer/image/imageInfo.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/pipeline/descriptors.h"
 #include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
 #include "graphics/host_gpu/renderer/pipeline/shaderResourceBarrier.h"
-#include "graphics/host_gpu/renderer/pipeline/shaderSubgroup.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vulkanCommon.h"
@@ -27,10 +25,12 @@
 #include "libs/errno.h"
 
 #include <algorithm>
+#include <array>
 #include <atomic>
 #include <cmath>
 #include <cstring>
 #include <limits>
+#include <mutex>
 #include <span>
 #include <unordered_map>
 #include <vector>
@@ -39,16 +39,25 @@ namespace Libs::Graphics {
 static uint64_t BufferDescriptorSize(const ShaderBufferResource& descriptor) {
 	const uint64_t records = descriptor.NumRecords();
 	const uint64_t stride  = descriptor.Stride();
-	if (stride != 0 && records > UINT64_MAX / stride) {
-		EXIT("compute buffer descriptor footprint overflow\n");
-	}
 	return stride == 0 ? records : records * stride;
 }
 
+static bool FillSourcesDisjoint(std::span<const ShaderRecompiler::IR::DescriptorValue> sources,
+                                 GuestRange destination, uint32_t output_buffer = UINT32_MAX) {
+	for (uint32_t i = 0; i < sources.size(); ++i) {
+		if (i == output_buffer) continue;
+		const auto source = DecodeNativeDescriptor<ShaderBufferResource>(sources[i]);
+		const auto bytes  = BufferDescriptorSize(source);
+		if (source.Base48() < destination.End() && destination.address < source.Base48() + bytes)
+			return false;
+	}
+	return true;
+}
+
 bool RenderExecutor::TryConsumeComputeMetaClear(const ShaderComputeInputInfo& input,
-                                                const RenderCommandBuffer&    buffer) {
+                                                const CommandBuffer&          buffer) {
 	const auto& program   = *input.stage.program;
-	const auto& resources = *input.stage.resources;
+	const auto& resources = input.stage.resources;
 	if (resources.buffers.size() != program.info.buffers.size()) {
 		EXIT("compute runtime buffer count does not match shader metadata\n");
 	}
@@ -56,7 +65,9 @@ bool RenderExecutor::TryConsumeComputeMetaClear(const ShaderComputeInputInfo& in
 	for (uint32_t i = 0; i < program.info.buffers.size(); i++) {
 		const auto& resource   = program.info.buffers[i];
 		const auto  descriptor = DecodeNativeDescriptor<ShaderBufferResource>(resources.buffers[i]);
-		if (!resource.written && cache.IsMeta(descriptor.Base48())) {
+		// A metadata resource that is also read is not proven to be a full overwrite. Execute it
+		// conservatively instead of replacing the dispatch with a coarse full-surface clear.
+		if (cache.IsMeta(descriptor.Base48()) && (!resource.written || resource.read)) {
 			return false;
 		}
 	}
@@ -76,69 +87,116 @@ bool RenderExecutor::TryConsumeComputeMetaClear(const ShaderComputeInputInfo& in
 	return false;
 }
 
-bool ResolveComputeImageClear(const ShaderComputeInputInfo& input, uint32_t group_x,
+bool ResolveComputeBufferFill(const ShaderComputeInputInfo& input, uint32_t group_x,
                               uint32_t group_y, uint32_t group_z, uint32_t mode,
                               ShaderBufferResource& resolved_descriptor, uint32_t& resolved_clear,
                               uint64_t& resolved_size) {
-	const auto& program   = *input.stage.program;
-	const auto& resources = *input.stage.resources;
-	if (program.info.buffers.size() != 1 || resources.buffers.size() != 1 ||
-	    !program.info.images.empty() || !program.info.samplers.empty() ||
-	    !program.info.addresses.empty() || !resources.images.empty() ||
-	    !resources.samplers.empty() || !resources.addresses.empty()) {
+	const auto& resources = input.stage.resources;
+	const auto& fill      = resources.uniform_fill;
+	if (fill.kind != ShaderRecompiler::IR::UniformFillKind::Buffer) {
 		return false;
 	}
-	const auto& resource   = program.info.buffers.front();
-	const auto& raw        = resources.buffers.front();
-	const auto  descriptor = DecodeNativeDescriptor<ShaderBufferResource>(raw);
-	if (!resource.formatted || !resource.written || resource.read || resource.atomic ||
-	    resource.scalar || resource.max_byte_extent != 16 || descriptor.Stride() != 16 ||
-	    descriptor.Format() != Prospero::BufferFormat::k32_32_32_32UInt ||
+	const auto element_size = fill.words * sizeof(uint32_t);
+	const auto descriptor =
+	    DecodeNativeDescriptor<ShaderBufferResource>(resources.buffers[fill.resource]);
+	constexpr std::array formats {
+	    Prospero::BufferFormat::k32UInt, Prospero::BufferFormat::k32_32UInt,
+	    Prospero::BufferFormat::k32_32_32UInt, Prospero::BufferFormat::k32_32_32_32UInt};
+	if (descriptor.Stride() != element_size || descriptor.Format() != formats[fill.words - 1] ||
 	    descriptor.SwizzleEnabled() || descriptor.IndexStride() != 0 || descriptor.AddTid() ||
-	    resource.packed_stride != descriptor.PackedStride() || raw.dword_count != 4 ||
-	    program.user_data_base != 0 || resources.user_data.size() != 8) {
+	    descriptor.Base48() == 0) {
 		return false;
 	}
-	for (uint32_t i = 0; i < raw.dword_count; i++) {
-		if (raw.dwords[i] != resources.user_data[i]) {
-			return false;
-		}
-	}
-	const uint32_t clear = resources.user_data[4];
-	if (resources.user_data[5] != clear || resources.user_data[6] != clear ||
-	    resources.user_data[7] != clear) {
+	if (input.threads_num[0] == 0 || input.threads_num[0] != fill.group_stride[0] ||
+	    input.threads_num[1] != 1 || input.threads_num[2] != 1 || group_x == 0 || group_y != 1 ||
+	    group_z != 1 || mode != (input.dispatch_thread_dimensions ? 0x61u : 0x41u)) {
 		return false;
 	}
-	const bool full_dispatch =
-	    input.dispatch_thread_dimensions && input.threads_num[0] == 64 &&
-	    input.threads_num[1] == 1 && input.threads_num[2] == 1 && group_x != 0 && group_y == 1 &&
-	    group_z == 1 && input.dispatch_threads_num[0] == group_x &&
-	    input.dispatch_threads_num[1] == 1 && input.dispatch_threads_num[2] == 1 &&
-	    input.group_id[0] && !input.group_id[1] && !input.group_id[2] &&
-	    input.thread_ids_num == 1 && input.wave_size == 32 && !input.tg_size_en && mode == 0x61u &&
-	    group_x % input.threads_num[0] == 0 && descriptor.NumRecords() == group_x;
-	const auto size = BufferDescriptorSize(descriptor);
-	if (!full_dispatch || size == 0) {
+	const uint64_t invocations = input.dispatch_thread_dimensions
+	                                 ? group_x
+	                                 : static_cast<uint64_t>(group_x) * input.threads_num[0];
+	const auto     size        = BufferDescriptorSize(descriptor);
+	if (invocations != descriptor.NumRecords() || size == 0 || size > UINT32_MAX ||
+	    (input.dispatch_thread_dimensions &&
+	     (group_x % input.threads_num[0] != 0 || input.dispatch_threads_num[0] != group_x ||
+	      input.dispatch_threads_num[1] != 1 || input.dispatch_threads_num[2] != 1))) {
 		return false;
 	}
+	if (!FillSourcesDisjoint(resources.buffers, {descriptor.Base48(), size}, fill.resource))
+		return false;
 	resolved_descriptor = descriptor;
-	resolved_clear      = clear;
+	resolved_clear      = fill.value;
 	resolved_size       = size;
 	return true;
 }
 
-static bool TryConsumeComputeImageClear(const ShaderComputeInputInfo& input, CommandBuffer& command,
-                                        uint32_t group_x, uint32_t group_y, uint32_t group_z,
-                                        uint32_t mode) {
+bool RenderExecutor::TryConsumeComputeImageClear(const ShaderComputeInputInfo& input,
+                                                CommandBuffer& command, uint32_t group_x,
+                                                uint32_t group_y, uint32_t group_z, uint32_t mode) {
+	const auto& program   = *input.stage.program;
+	const auto& resources = input.stage.resources;
+	const auto& fill      = resources.uniform_fill;
+	auto&       cache     = command.GetContext().GetTextureCache();
+	if (fill.kind == ShaderRecompiler::IR::UniformFillKind::Image) {
+		if (mode != 0x41u || input.dispatch_thread_dimensions || fill.value > 255 ||
+		    input.threads_num[2] != 1)
+			return false;
+		const auto  descriptor = DecodeNativeDescriptor<ShaderTextureResource>(resources.images[0]);
+		const auto& resource   = program.info.images[0];
+		if (descriptor.IsNull() || descriptor.Format() != Prospero::BufferFormat::k8UInt ||
+		    descriptor.Type() != Prospero::ImageType::kColor2DArray || descriptor.MetaCompress() ||
+		    descriptor.WriteCompress() || descriptor.BaseLevel() > descriptor.LastLevel() ||
+		    descriptor.BaseLevel() > descriptor.MaxMip() ||
+		    descriptor.BaseArray5() > descriptor.Depth() || descriptor.DstSelX() != 4)
+			return false;
+		const std::array extents {
+		    std::max(1u, (descriptor.Width5() + 1u) >> descriptor.BaseLevel()),
+		    std::max(1u, (descriptor.Height5() + 1u) >> descriptor.BaseLevel()),
+		    descriptor.Depth() - descriptor.BaseArray5() + 1u};
+		const std::array groups {group_x, group_y, group_z};
+		for (uint32_t axis = 0; axis < 3; ++axis) {
+			const uint64_t threads = input.threads_num[axis];
+			// Guest image writes outside the descriptor dimensions are discarded. Only the
+			// final workgroup may extend beyond the selected image view.
+			if (threads == 0 || threads != fill.group_stride[axis] ||
+			    groups[axis] != (extents[axis] + threads - 1) / threads ||
+			    groups[axis] * threads > UINT32_MAX) return false;
+		}
+		const auto  binding     = ResolveTexture(resource, resources.images[0]);
+		const auto& destination = binding.desc.info.data;
+		if (!FillSourcesDisjoint(resources.buffers, destination)) return false;
+		std::scoped_lock lock {cache.m_lock};
+		const auto&      image = cache.GetImage(binding.image_id);
+		const auto&      view  = binding.desc.view_info;
+		if (image.backing.format != vk::Format::eD32SfloatS8Uint || image.info.samples != 1 ||
+		    image.info.stencil != destination || view.base_level >= image.backing.mip_levels ||
+		    view.base_layer >= image.backing.layers || view.layer_count != extents[2] ||
+		    view.layer_count > image.backing.layers - view.base_layer ||
+		    std::max(1u, image.info.extent.width >> view.base_level) != extents[0] ||
+		    std::max(1u, image.info.extent.height >> view.base_level) != extents[1]) return false;
+		const vk::ImageSubresourceRange range {vk::ImageAspectFlagBits::eStencil, view.base_level,
+		                                       1, view.base_layer, view.layer_count};
+		vk::ClearValue clear {};
+		clear.depthStencil = vk::ClearDepthStencilValue {0.0f, fill.value};
+		cache.ClearImage(command, binding.image_id, range, clear);
+		return true;
+	}
 	ShaderBufferResource descriptor;
 	uint32_t             packed_clear = 0;
 	uint64_t             size         = 0;
-	if (!ResolveComputeImageClear(input, group_x, group_y, group_z, mode, descriptor, packed_clear,
+	if (!ResolveComputeBufferFill(input, group_x, group_y, group_z, mode, descriptor, packed_clear,
 	                              size)) {
 		return false;
 	}
-	auto& cache = command.GetContext().GetTextureCache();
 	if (!cache.ClearImageFromBuffer(command, descriptor.Base48(), size, packed_clear)) {
+		// Track deferred DCC state while the original dispatch writes the metadata allocation.
+		cache.TrackDccFill(descriptor.Base48(), size, packed_clear);
+		static std::atomic<uint32_t> logged_metadata_clears {0};
+		if (logged_metadata_clears.fetch_add(1, std::memory_order_relaxed) < 32) {
+			LOGF("GraphicsRenderDispatchDirect: metadata fill shader=0x%016" PRIx64
+			     " addr=0x%016" PRIx64 " size=0x%016" PRIx64 " value=0x%08" PRIx32 "\n",
+			     input.stage.program->shader_hash, descriptor.Base48(), size, packed_clear);
+		}
 		return false;
 	}
 	static std::atomic<uint32_t> logged_clears {0};
@@ -150,10 +208,11 @@ static bool TryConsumeComputeImageClear(const ShaderComputeInputInfo& input, Com
 	return true;
 }
 
-void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buffer,
+void RenderExecutor::DispatchDirect(uint64_t submit_id, CommandBuffer& buffer,
                                     uint32_t thread_group_x, uint32_t thread_group_y,
                                     uint32_t thread_group_z, uint32_t mode) {
 	EXIT_IF(buffer.IsInvalid());
+	m_context.GetCommandScheduler().PopPendingOperations();
 	auto& ctx    = buffer.GetRegisters();
 	auto& sh_ctx = buffer.GetShaders();
 
@@ -194,16 +253,12 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 	const auto& cs_regs = sh_ctx.GetCs();
 	const auto& sh_regs = ctx.GetShaderRegisters();
 
-	ShaderComputeInputInfo    input_info {};
-	std::span<const uint32_t> cs_shader;
-	if (!ShaderCompileInfoCS(cs_regs, sh_regs, input_info, cs_shader)) {
-		EXIT("ShaderCompileInfoCS failed for dispatch with CS shader 0x%016" PRIx64 "\n",
-		     cs_regs.cs_regs.data_addr);
-	}
-
+	ShaderComputeInputInfo input_info {};
 	const bool use_thread_dimensions = (mode & DISPATCH_INITIATOR_USE_THREAD_DIMENSIONS) != 0;
+	input_info.dispatch_thread_dimensions = use_thread_dimensions;
+	const auto compute_program =
+	    m_context.GetPipelineCache().GetComputeProgram(cs_regs, sh_regs, input_info);
 	if (use_thread_dimensions) {
-		input_info.dispatch_thread_dimensions = true;
 		input_info.dispatch_threads_num[0]    = thread_group_x;
 		input_info.dispatch_threads_num[1]    = thread_group_y;
 		input_info.dispatch_threads_num[2]    = thread_group_z;
@@ -213,7 +268,7 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 	const bool     large_workgroup =
 	    (input_info.threads_num[0] * input_info.threads_num[1] * input_info.threads_num[2] >= 512);
 	const auto& program   = *input_info.stage.program;
-	const auto& resources = *input_info.stage.resources;
+	const auto& resources = input_info.stage.resources;
 	if (TryConsumeComputeMetaClear(input_info, buffer)) {
 		ResetBindings();
 		return;
@@ -225,8 +280,7 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 	}
 	const auto sampled_images = std::count_if(
 	    program.info.images.begin(), program.info.images.end(), [](const auto& image) {
-		    return image.kind == ShaderRecompiler::IR::ResourceKind::Image ||
-		           image.kind == ShaderRecompiler::IR::ResourceKind::ImageUint;
+		    return image.resource_class == ShaderRecompiler::IR::ImageResourceClass::Sampled;
 	    });
 	const bool                   has_sampler = !program.info.samplers.empty();
 	static std::atomic<uint32_t> dispatch_log_count {0};
@@ -239,7 +293,10 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 		     thread_group_z, mode, input_info.threads_num[0], input_info.threads_num[1],
 		     input_info.threads_num[2], program.info.buffers.size(), program.info.images.size(),
 		     sampled_images, program.info.images.size() - sampled_images,
-		     program.info.samplers.size(), program.bindings.push_constant_size);
+		     program.info.samplers.size(),
+		     program.bindings.UsesPushData()
+		         ? static_cast<uint32_t>(sizeof(ShaderRecompiler::IR::PushData))
+		         : 0u);
 		for (uint32_t i = 0; i < program.info.buffers.size(); i++) {
 			const auto& buffer = program.info.buffers[i];
 			const auto  r      = DecodeNativeDescriptor<ShaderBufferResource>(resources.buffers[i]);
@@ -254,8 +311,7 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 			LOGF("  CS texture[%u]: source=%u usage=%s sampled=%s addr=0x%010" PRIx64
 			     " type=%u fmt=%u extent=%ux%u depth=%u levels=%u tile=%u\n",
 			     i, image.source, image.written ? "read-write" : "read-only",
-			     (image.kind == ShaderRecompiler::IR::ResourceKind::Image ||
-			      image.kind == ShaderRecompiler::IR::ResourceKind::ImageUint)
+			     image.resource_class == ShaderRecompiler::IR::ImageResourceClass::Sampled
 			         ? "true"
 			         : "false",
 			     r.Base40(), static_cast<uint32_t>(r.Type()), static_cast<uint32_t>(r.Format()),
@@ -264,7 +320,7 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 			     r.Type() == Prospero::ImageType::kColor2DMsaa ||
 			             r.Type() == Prospero::ImageType::kColor2DMsaaArray
 			         ? 1u
-			         : static_cast<uint32_t>(r.MaxMip()) + 1u,
+			         : static_cast<uint32_t>(image.r128 ? r.LastLevel() : r.MaxMip()) + 1u,
 			     static_cast<uint32_t>(r.TileMode()));
 		}
 		for (uint32_t i = 0; i < program.info.samplers.size(); i++) {
@@ -318,21 +374,26 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 
 	buffer.EndRendering();
 	auto& pipeline =
-	    m_context.GetPipelineCache().CreateComputePipeline(input_info, sh_ctx.GetCs(), cs_shader);
-	auto bindings = PrepareBindings(buffer, input_info.stage, vk::ShaderStageFlagBits::eCompute,
-	                                DescriptorCache::Stage::Compute);
-	RebindBuffers(buffer, bindings);
-	RebindImages(buffer, bindings);
+	    m_context.GetPipelineCache().CreateComputePipeline(input_info, compute_program);
+	auto bindings = PrepareBindings(input_info.stage);
+	FindBuffers(bindings);
+	if (program.info.uses_dma) {
+		m_context.GetGpuResources().PrepareBda();
+	}
+	RebindBuffers(bindings);
+	RebindImages(bindings);
 
-	auto vk_buffer = buffer.Handle();
-	CommitBindings(buffer, vk::PipelineBindPoint::eCompute, pipeline.pipeline_layout, bindings);
+	auto              vk_buffer        = buffer.Handle();
+	PreparedBindings* descriptor_stage = &bindings;
+	CommitBindings(buffer, vk::PipelineBindPoint::eCompute, pipeline,
+	               std::span {&descriptor_stage, 1u});
 	bool has_storage_writes = HasShaderBufferWrites(input_info.stage);
 	has_storage_writes =
 	    std::any_of(program.info.images.begin(), program.info.images.end(),
 	                [](const auto& image) {
 		                return image.written &&
-		                       (image.kind == ShaderRecompiler::IR::ResourceKind::StorageImage ||
-		                        image.kind == ShaderRecompiler::IR::ResourceKind::StorageImageUint);
+		                       image.resource_class ==
+		                           ShaderRecompiler::IR::ImageResourceClass::Storage;
 	                }) ||
 	    has_storage_writes;
 	if (has_storage_writes) {

src/graphics/host_gpu/renderer/renderContext.cpp

@@ -4,7 +4,6 @@
 #include "common/logging/log.h"
 #include "graphics/guest_gpu/graphicsRun.h"
 #include "graphics/presentation/videoOut.h"
-#include "kernel/pthread.h"
 #include "libs/errno.h"
 
 #include <algorithm>
@@ -13,8 +12,9 @@ namespace Libs::Graphics {
 
 RenderContext::RenderContext(GraphicContext& graphics)
     : m_graphics(graphics), m_render_executor(*this), m_command_scheduler(*this, graphics),
-      m_descriptor_cache(graphics), m_pipeline_cache(graphics, m_descriptor_cache),
-      m_sampler_cache(graphics), m_gpu_resources(graphics, m_command_scheduler) {
+      m_descriptor_heap(graphics, m_command_scheduler.GetMasterSemaphore()),
+      m_pipeline_cache(graphics), m_sampler_cache(graphics),
+      m_gpu_resources(graphics, m_command_scheduler) {
 	EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
 }
 
@@ -38,7 +38,7 @@ void RenderContext::ShutdownGpu() {
 	}
 	if (m_video_out != nullptr) {
 		if (m_command_scheduler.Active()) {
-			m_command_scheduler.FinishCurrent();
+			m_command_scheduler.Finish();
 		}
 		m_command_scheduler.DrainPriorityOperations();
 		m_video_out = nullptr;
@@ -55,60 +55,54 @@ VideoOut::VideoOutDriver& RenderContext::GetVideoOut() const {
 	return *m_video_out;
 }
 
-void RenderContext::AddEopEq(LibKernel::EventQueue::KernelEqueue eq, int id) {
-	auto queue = LibKernel::EventQueue::KernelPinEqueue(eq);
-	if (!queue) {
+void RenderContext::AddInterruptEq(LibKernel::EventQueue::KernelEqueue eq, int event_id) {
+	Common::LockGuard lock(m_interrupt_mutex);
+
+	auto it = std::find_if(
+	    m_interrupt_eqs.begin(), m_interrupt_eqs.end(),
+	    [eq, event_id](const auto& entry) { return entry.eq == eq && entry.event_id == event_id; });
+	if (it != m_interrupt_eqs.end()) {
 		return;
 	}
 
-	Common::LockGuard lock(m_eop_mutex);
-
-	auto it = std::find_if(m_eop_eqs.begin(), m_eop_eqs.end(), [eq, id](const auto& entry) {
-		return entry.eq == eq && entry.id == id;
-	});
-	if (it != m_eop_eqs.end()) {
-		return;
-	}
-
-	m_eop_eqs.push_back({eq, std::move(queue), id});
+	m_interrupt_eqs.push_back({eq, event_id});
 }
 
-void RenderContext::DeleteEopEq(LibKernel::EventQueue::KernelEqueue eq, int id) {
-	Common::LockGuard lock(m_eop_mutex);
+void RenderContext::DeleteInterruptEq(LibKernel::EventQueue::KernelEqueue eq, int event_id) {
+	Common::LockGuard lock(m_interrupt_mutex);
 
-	auto it = std::find_if(m_eop_eqs.begin(), m_eop_eqs.end(), [eq, id](const auto& entry) {
-		return entry.eq == eq && entry.id == id;
-	});
-	if (it == m_eop_eqs.end()) {
+	auto it = std::find_if(
+	    m_interrupt_eqs.begin(), m_interrupt_eqs.end(),
+	    [eq, event_id](const auto& entry) { return entry.eq == eq && entry.event_id == event_id; });
+	if (it == m_interrupt_eqs.end()) {
 		return;
 	}
 
-	m_eop_eqs.erase(it);
+	m_interrupt_eqs.erase(it);
 }
 
-void RenderContext::TriggerEopEvent(uint32_t context_id) {
-	std::vector<EopEqRegistration> registrations;
+void RenderContext::TriggerInterrupt(int event_id, uint32_t context_id) {
+	std::vector<InterruptEqRegistration> registrations;
 	{
-		Common::LockGuard lock(m_eop_mutex);
-		registrations = m_eop_eqs;
+		Common::LockGuard lock(m_interrupt_mutex);
+		for (const auto& registration: m_interrupt_eqs) {
+			if (registration.event_id == event_id) {
+				registrations.push_back(registration);
+			}
+		}
 	}
 
 	for (const auto& registration: registrations) {
 		const auto result = LibKernel::EventQueue::KernelTriggerEvent(
-		    registration.eq, static_cast<uintptr_t>(registration.id),
+		    registration.eq, static_cast<uintptr_t>(registration.event_id),
 		    LibKernel::EventQueue::KERNEL_EVFILT_GRAPHICS,
 		    reinterpret_cast<void*>(static_cast<uintptr_t>(context_id)));
 		if (result == LibKernel::KERNEL_ERROR_EBADF || result == LibKernel::KERNEL_ERROR_ENOENT) {
-			DeleteEopEq(registration.eq, registration.id);
+			DeleteInterruptEq(registration.eq, registration.event_id);
 			continue;
 		}
 		EXIT_NOT_IMPLEMENTED(result != OK);
 	}
-
-	auto tsc    = LibKernel::KernelReadTsc();
-	auto result = LibKernel::EventQueue::KernelTriggerUserEventForAll(AGC_USER_INTERRUPT_EVENT,
-	                                                                  reinterpret_cast<void*>(tsc));
-	EXIT_NOT_IMPLEMENTED(result != OK && result != LibKernel::KERNEL_ERROR_ENOENT);
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/renderContext.h

@@ -10,7 +10,7 @@
 #include "graphics/host_gpu/renderer/cache/samplerCache.h"
 #include "graphics/host_gpu/renderer/cache/textureCache.h"
 #include "graphics/host_gpu/renderer/commandScheduler.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
+#include "graphics/host_gpu/renderer/pipeline/descriptorHeap.h"
 #include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
 #include "kernel/eventQueue.h"
 
@@ -23,7 +23,6 @@ class VideoOutDriver;
 
 namespace Libs::Graphics {
 
-constexpr int AGC_USER_INTERRUPT_EVENT = 0x1800;
 class GuestGpu;
 
 class RenderContext {
@@ -41,37 +40,36 @@ public:
 	Common::Mutex&      GetMutex() { return m_mutex; }
 	CommandScheduler&   GetCommandScheduler() { return m_command_scheduler; }
 	PipelineCache&      GetPipelineCache() { return m_pipeline_cache; }
-	DescriptorCache&    GetDescriptorCache() { return m_descriptor_cache; }
+	DescriptorHeap&     GetDescriptorHeap() { return m_descriptor_heap; }
 	SamplerCache&       GetSamplerCache() { return m_sampler_cache; }
 	GpuResourceManager& GetGpuResources() { return m_gpu_resources; }
 	BufferCache&        GetBufferCache() { return m_gpu_resources.GetBufferCache(); }
 	TextureCache&       GetTextureCache() { return m_gpu_resources.GetTextureCache(); }
 	RenderExecutor&     GetRenderExecutor() { return m_render_executor; }
 
-	void AddEopEq(LibKernel::EventQueue::KernelEqueue eq, int id);
-	void DeleteEopEq(LibKernel::EventQueue::KernelEqueue eq, int id);
-	void TriggerEopEvent(uint32_t context_id);
+	void AddInterruptEq(LibKernel::EventQueue::KernelEqueue eq, int event_id);
+	void DeleteInterruptEq(LibKernel::EventQueue::KernelEqueue eq, int event_id);
+	void TriggerInterrupt(int event_id, uint32_t context_id);
 
 private:
-	struct EopEqRegistration {
-		LibKernel::EventQueue::KernelEqueue    eq = LibKernel::EventQueue::KERNEL_EQUEUE_INVALID;
-		LibKernel::EventQueue::KernelEqueueRef queue;
-		int                                    id = 0;
+	struct InterruptEqRegistration {
+		LibKernel::EventQueue::KernelEqueue eq       = LibKernel::EventQueue::KERNEL_EQUEUE_INVALID;
+		int                                 event_id = 0;
 	};
 
 	GraphicContext&           m_graphics;
 	Common::Mutex             m_mutex;
 	RenderExecutor            m_render_executor;
 	CommandScheduler          m_command_scheduler;
-	DescriptorCache           m_descriptor_cache;
+	DescriptorHeap            m_descriptor_heap;
 	PipelineCache             m_pipeline_cache;
 	SamplerCache              m_sampler_cache;
 	GpuResourceManager        m_gpu_resources;
 	std::unique_ptr<GuestGpu> m_gpu;
 	VideoOut::VideoOutDriver* m_video_out = nullptr;
 
-	Common::Mutex                  m_eop_mutex;
-	std::vector<EopEqRegistration> m_eop_eqs;
+	Common::Mutex                        m_interrupt_mutex;
+	std::vector<InterruptEqRegistration> m_interrupt_eqs;
 };
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/renderDraw.h

@@ -1,45 +1,15 @@
 #ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_RENDERDRAW_H_
 #define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_RENDERDRAW_H_
 
-#include "graphics/host_gpu/renderer/renderTarget.h"
-
 #include <cstdint>
-#include <type_traits>
 
 namespace Libs::Graphics {
 
-struct RenderColorInfo;
 struct ShaderVertexInputInfo;
 
-#pragma pack(push, 1)
-
-struct PipelineDynamicParameters {
-	bool stencil_test_enable = false;
-	bool depth_bias_enable   = false;
-
-	float viewport_scale[3]  = {};
-	float viewport_offset[3] = {};
-	int   scissor_ltrb[4]    = {0};
-
-	float    line_width                                       = 1.0f;
-	float    depth_bias_constant_factor                       = 0.0f;
-	float    depth_bias_clamp                                 = 0.0f;
-	float    depth_bias_slope_factor                          = 0.0f;
-	uint32_t color_write_count                                = 1;
-	bool     color_write_enable[RENDER_COLOR_ATTACHMENTS_MAX] = {true, true};
-
-	PipelineStencilDynamicState stencil_front;
-	PipelineStencilDynamicState stencil_back;
-};
-
-#pragma pack(pop)
-
-static_assert(std::is_trivially_copyable_v<PipelineDynamicParameters>);
-static_assert(std::is_standard_layout_v<PipelineDynamicParameters>);
-static_assert(alignof(PipelineDynamicParameters) == 1);
-
-[[nodiscard]] int32_t ResolveVertexOffset(uint32_t                     index_offset,
-                                          const ShaderVertexInputInfo& vs_input_info);
+[[nodiscard]] int32_t  ResolveVertexOffset(uint32_t                     index_offset,
+                                           const ShaderVertexInputInfo& vs_input_info);
+[[nodiscard]] uint32_t ResolveInstanceOffset(const ShaderVertexInputInfo& vs_input_info);
 
 } // namespace Libs::Graphics
 

src/graphics/host_gpu/renderer/sync.cpp

@@ -20,9 +20,7 @@
 
 namespace Libs::Graphics::Sync {
 
-constexpr int      GRAPHICS_EVENT_QUEUED_GRAPHICS_INTERRUPT = 0x00;
-constexpr int      GRAPHICS_EVENT_EOP                       = 0x40;
-constexpr uint64_t GRAPHICS_REFERENCE_CLOCK_FREQUENCY       = 100000000;
+constexpr uint64_t GRAPHICS_REFERENCE_CLOCK_FREQUENCY = 100000000;
 
 bool ScaleReferenceClock(uint64_t host_ticks, uint64_t host_frequency, uint64_t& value) {
 	if (host_frequency == 0) {
@@ -66,8 +64,9 @@ struct EndOfPipeSignal {
 	std::array<uint32_t, 4> debug_args      = {};
 	uint64_t                debug_data      = 0;
 	std::optional<uint64_t> destination;
-	EndOfPipeCompletion     completion      = EndOfPipeCompletion::None;
-	uint64_t                completion_data = 0;
+	EndOfPipeCompletion     completion         = EndOfPipeCompletion::None;
+	uint64_t                completion_data    = 0;
+	int                     interrupt_event_id = 0;
 };
 
 enum class EndOfPipeWriteSize : uint32_t { Dword = 4, Qword = 8 };
@@ -96,8 +95,10 @@ static void RecordEndOfPipeSignal(const EndOfPipeSignal& signal) {
 		case EndOfPipeCompletion::None: return;
 		case EndOfPipeCompletion::Interrupt: {
 			const auto context_id = static_cast<uint32_t>(signal.completion_data);
-			scheduler.DeferPriorityOperation(
-			    [&renderer, context_id] { renderer.TriggerEopEvent(context_id); });
+			const auto event_id   = signal.interrupt_event_id;
+			scheduler.DeferPriorityOperation([&renderer, event_id, context_id] {
+				renderer.TriggerInterrupt(event_id, context_id);
+			});
 			return;
 		}
 		case EndOfPipeCompletion::Flip: {
@@ -108,9 +109,10 @@ static void RecordEndOfPipeSignal(const EndOfPipeSignal& signal) {
 		}
 		case EndOfPipeCompletion::FlipAndInterrupt: {
 			const auto request_id = signal.completion_data;
-			scheduler.DeferPriorityOperation([&renderer, request_id] {
+			const auto event_id   = signal.interrupt_event_id;
+			scheduler.DeferPriorityOperation([&renderer, event_id, request_id] {
 				renderer.GetVideoOut().CompleteFlip(request_id);
-				renderer.TriggerEopEvent(0);
+				renderer.TriggerInterrupt(event_id, 0);
 			});
 			return;
 		}
@@ -135,7 +137,8 @@ static bool TriggersInterrupt(EndOfPipeWriteAction action) {
 
 static void RecordEndOfPipeWrite(uint64_t submit_id, CommandBuffer& buffer, uint64_t destination,
                                  uint64_t value, EndOfPipeWriteSize size,
-                                 EndOfPipeWriteAction action, uint32_t context_id = 0) {
+                                 EndOfPipeWriteAction action, int interrupt_event_id = 0,
+                                 uint32_t context_id = 0) {
 	const auto width      = static_cast<uint32_t>(size);
 	const auto value_low  = static_cast<uint32_t>(value);
 	const auto value_high = static_cast<uint32_t>(value >> 32u);
@@ -150,18 +153,12 @@ static void RecordEndOfPipeWrite(uint64_t submit_id, CommandBuffer& buffer, uint
 	    .debug_data      = destination,
 	    .destination     = destination,
 	    .completion      = interrupt ? EndOfPipeCompletion::Interrupt : EndOfPipeCompletion::None,
-	    .completion_data = context_id != 0 ? context_id : value,
+	    .completion_data = interrupt ? context_id : value,
+	    .interrupt_event_id = interrupt_event_id,
 	};
 	RecordEndOfPipeSignal(signal);
 }
 
-void TriggerAgcUserInterrupt() {
-	auto tsc    = LibKernel::KernelReadTsc();
-	auto result = LibKernel::EventQueue::KernelTriggerUserEventForAll(AGC_USER_INTERRUPT_EVENT,
-	                                                                  reinterpret_cast<void*>(tsc));
-	EXIT_NOT_IMPLEMENTED(result != OK && result != LibKernel::KERNEL_ERROR_ENOENT);
-}
-
 void WriteAtEndOfPipe32(uint64_t submit_id, CommandBuffer& buffer, uint32_t* dst_gpu_addr,
                         uint32_t value) {
 	RecordEndOfPipeWrite(submit_id, buffer, reinterpret_cast<uint64_t>(dst_gpu_addr), value,
@@ -220,31 +217,35 @@ void WriteAtEndOfPipeWithWriteBack32(uint64_t submit_id, CommandBuffer& buffer,
 }
 
 void WriteAtEndOfPipeWithInterruptWriteBack64(uint64_t submit_id, CommandBuffer& buffer,
-                                              uint64_t* dst_gpu_addr, uint64_t value,
+                                              uint64_t* dst_gpu_addr, uint64_t value, int event_id,
                                               uint32_t context_id) {
 	RecordEndOfPipeWrite(submit_id, buffer, reinterpret_cast<uint64_t>(dst_gpu_addr), value,
 	                     EndOfPipeWriteSize::Qword, EndOfPipeWriteAction::InterruptWriteBack,
-	                     context_id);
+	                     event_id, context_id);
 }
 
 void WriteAtEndOfPipeWithInterruptWriteBack32(uint64_t submit_id, CommandBuffer& buffer,
-                                              uint32_t* dst_gpu_addr, uint32_t value,
+                                              uint32_t* dst_gpu_addr, uint32_t value, int event_id,
                                               uint32_t context_id) {
 	RecordEndOfPipeWrite(submit_id, buffer, reinterpret_cast<uint64_t>(dst_gpu_addr), value,
 	                     EndOfPipeWriteSize::Dword, EndOfPipeWriteAction::InterruptWriteBack,
-	                     context_id);
+	                     event_id, context_id);
 }
 
 void WriteAtEndOfPipeWithInterrupt64(uint64_t submit_id, CommandBuffer& buffer,
-                                     uint64_t* dst_gpu_addr, uint64_t value, uint32_t context_id) {
+                                     uint64_t* dst_gpu_addr, uint64_t value, int event_id,
+                                     uint32_t context_id) {
 	RecordEndOfPipeWrite(submit_id, buffer, reinterpret_cast<uint64_t>(dst_gpu_addr), value,
-	                     EndOfPipeWriteSize::Qword, EndOfPipeWriteAction::Interrupt, context_id);
+	                     EndOfPipeWriteSize::Qword, EndOfPipeWriteAction::Interrupt, event_id,
+	                     context_id);
 }
 
 void WriteAtEndOfPipeWithInterrupt32(uint64_t submit_id, CommandBuffer& buffer,
-                                     uint32_t* dst_gpu_addr, uint32_t value, uint32_t context_id) {
+                                     uint32_t* dst_gpu_addr, uint32_t value, int event_id,
+                                     uint32_t context_id) {
 	RecordEndOfPipeWrite(submit_id, buffer, reinterpret_cast<uint64_t>(dst_gpu_addr), value,
-	                     EndOfPipeWriteSize::Dword, EndOfPipeWriteAction::Interrupt, context_id);
+	                     EndOfPipeWriteSize::Dword, EndOfPipeWriteAction::Interrupt, event_id,
+	                     context_id);
 }
 
 uint64_t PrepareVideoOutFlip(CommandBuffer& buffer, int handle, int index, int flip_mode,
@@ -270,18 +271,20 @@ uint64_t PrepareVideoOutFlip(CommandBuffer& buffer, int handle, int index, int f
 void WriteAtEndOfPipeWithInterruptWriteBackFlip32(uint64_t submit_id, CommandBuffer& buffer,
                                                   uint32_t* dst_gpu_addr, uint32_t value,
                                                   int handle, int index, int flip_mode,
-                                                  int64_t flip_arg, uint64_t request_id) {
+                                                  int64_t flip_arg, uint64_t request_id,
+                                                  int event_id) {
 	const auto destination = reinterpret_cast<uint64_t>(dst_gpu_addr);
 	RecordEndOfPipeSignal({
-	    .buffer          = &buffer,
-	    .submit_id       = submit_id,
-	    .debug_operation = CommandBufferDebugOp::EopWriteBackFlip,
-	    .debug_args      = {static_cast<uint32_t>(handle), static_cast<uint32_t>(index),
-	                        static_cast<uint32_t>(flip_mode), value},
-	    .debug_data      = static_cast<uint64_t>(flip_arg),
-	    .destination     = destination,
-	    .completion      = EndOfPipeCompletion::FlipAndInterrupt,
-	    .completion_data = request_id,
+	    .buffer             = &buffer,
+	    .submit_id          = submit_id,
+	    .debug_operation    = CommandBufferDebugOp::EopWriteBackFlip,
+	    .debug_args         = {static_cast<uint32_t>(handle), static_cast<uint32_t>(index),
+	                           static_cast<uint32_t>(flip_mode), value},
+	    .debug_data         = static_cast<uint64_t>(flip_arg),
+	    .destination        = destination,
+	    .completion         = EndOfPipeCompletion::FlipAndInterrupt,
+	    .completion_data    = request_id,
+	    .interrupt_event_id = event_id,
 	});
 }
 
@@ -316,36 +319,24 @@ void WriteAtEndOfPipeOnlyFlip(uint64_t submit_id, CommandBuffer& buffer, int han
 	});
 }
 
-void TriggerEopEventAtEndOfPipe(CommandBuffer& buffer, uint32_t context_id) {
+void TriggerEopEventAtEndOfPipe(CommandBuffer& buffer, int event_id, uint32_t context_id) {
 	ValidateEndOfPipeSignal({.buffer = &buffer});
 	auto& renderer  = buffer.GetContext();
 	auto& scheduler = renderer.GetCommandScheduler();
 	EXIT_IF(!scheduler.Active() || &buffer != &scheduler.Current());
 	scheduler.DeferPriorityOperation(
-	    [&renderer, context_id] { renderer.TriggerEopEvent(context_id); });
+	    [&renderer, event_id, context_id] { renderer.TriggerInterrupt(event_id, context_id); });
 }
 
-static void EopEventResetFunc(LibKernel::EventQueue::KernelEqueueEvent* event) {
+static void InterruptEventResetFunc(LibKernel::EventQueue::KernelEqueueEvent* event) {
 	EXIT_IF(event == nullptr);
 	event->triggered    = false;
 	event->event.fflags = 0;
 	event->event.data   = 0;
 }
 
-static void EopEventDeleteFunc(LibKernel::EventQueue::KernelEqueue       eq,
-                               LibKernel::EventQueue::KernelEqueueEvent* event) {
-	EXIT_IF(event == nullptr);
-	EXIT_NOT_IMPLEMENTED(event->event.filter != LibKernel::EventQueue::KERNEL_EVFILT_GRAPHICS);
-	if (event->event.ident == GRAPHICS_EVENT_QUEUED_GRAPHICS_INTERRUPT ||
-	    event->event.ident == GRAPHICS_EVENT_EOP) {
-		auto* renderer = static_cast<RenderContext*>(event->filter.data);
-		EXIT_IF(renderer == nullptr);
-		renderer->DeleteEopEq(eq, static_cast<int>(event->event.ident));
-	}
-}
-
-static void EopEventTriggerFunc(LibKernel::EventQueue::KernelEqueueEvent* event,
-                                void*                                     trigger_data) {
+static void InterruptEventTriggerFunc(LibKernel::EventQueue::KernelEqueueEvent* event,
+                                      void*                                     trigger_data) {
 	EXIT_IF(event == nullptr);
 
 	auto triggered_event = event->event;
@@ -362,35 +353,35 @@ static void EopEventTriggerFunc(LibKernel::EventQueue::KernelEqueueEvent* event,
 int AddEqEvent(RenderContext& renderer, LibKernel::EventQueue::KernelEqueue eq, int id,
                void* udata) {
 	LibKernel::EventQueue::KernelEqueueEvent event;
-	event.triggered                = false;
-	event.event.ident              = static_cast<uintptr_t>(id);
-	event.event.filter             = LibKernel::EventQueue::KERNEL_EVFILT_GRAPHICS;
-	event.event.udata              = udata;
-	event.event.fflags             = 0;
-	event.event.data               = id;
-	event.filter.delete_event_func = EopEventDeleteFunc;
-	event.filter.reset_func        = EopEventResetFunc;
-	event.filter.trigger_func      = EopEventTriggerFunc;
-	event.filter.data              = &renderer;
+	event.triggered           = false;
+	event.event.ident         = static_cast<uintptr_t>(id);
+	event.event.filter        = LibKernel::EventQueue::KERNEL_EVFILT_GRAPHICS;
+	event.event.udata         = udata;
+	event.event.fflags        = 0;
+	event.event.data          = id;
+	event.filter.reset_func   = InterruptEventResetFunc;
+	event.filter.trigger_func = InterruptEventTriggerFunc;
 
 	int result = LibKernel::EventQueue::KernelAddEvent(eq, event);
 
-	if (result == 0 &&
-	    (id == GRAPHICS_EVENT_QUEUED_GRAPHICS_INTERRUPT || id == GRAPHICS_EVENT_EOP)) {
-		renderer.AddEopEq(eq, id);
+	if (result == 0) {
+		renderer.AddInterruptEq(eq, id);
 	}
 
 	return result;
 }
 
-int DeleteEqEvent(LibKernel::EventQueue::KernelEqueue eq, int id) {
+int DeleteEqEvent(RenderContext& renderer, LibKernel::EventQueue::KernelEqueue eq, int id) {
 	int result = LibKernel::EventQueue::KernelDeleteEvent(
 	    eq, static_cast<uintptr_t>(id), LibKernel::EventQueue::KERNEL_EVFILT_GRAPHICS);
+	if (result == OK || result == LibKernel::KERNEL_ERROR_ENOENT) {
+		renderer.DeleteInterruptEq(eq, id);
+	}
 
 	return result;
 }
 
-void ReadGds(Buffer& gds, uint32_t* dst, uint32_t dw_offset, uint32_t dw_size) {
+void ReadGds(const Buffer& gds, uint32_t* dst, uint32_t dw_offset, uint32_t dw_size) {
 	const auto offset = uint64_t {dw_offset} * sizeof(uint32_t);
 	const auto size   = uint64_t {dw_size} * sizeof(uint32_t);
 	EXIT_IF(dst == nullptr || offset > gds.Size() || size > gds.Size() - offset ||

src/graphics/host_gpu/renderer/sync.h

@@ -17,8 +17,7 @@ namespace Sync {
                                            uint64_t& value);
 [[nodiscard]] uint64_t ReadReferenceClock();
 
-void TriggerAgcUserInterrupt();
-void TriggerEopEventAtEndOfPipe(CommandBuffer& buffer, uint32_t context_id);
+void TriggerEopEventAtEndOfPipe(CommandBuffer& buffer, int event_id, uint32_t context_id);
 
 void WriteAtEndOfPipe32(uint64_t submit_id, CommandBuffer& buffer, uint32_t* dst_gpu_addr,
                         uint32_t value);
@@ -35,16 +34,16 @@ void WriteAtEndOfPipeWithWriteBack32(uint64_t submit_id, CommandBuffer& buffer,
 void WriteAtEndOfPipeWithWriteBack64(uint64_t submit_id, CommandBuffer& buffer,
                                      uint64_t* dst_gpu_addr, uint64_t value);
 void WriteAtEndOfPipeWithInterrupt32(uint64_t submit_id, CommandBuffer& buffer,
-                                     uint32_t* dst_gpu_addr, uint32_t value,
+                                     uint32_t* dst_gpu_addr, uint32_t value, int event_id,
                                      uint32_t context_id = 0);
 void WriteAtEndOfPipeWithInterrupt64(uint64_t submit_id, CommandBuffer& buffer,
-                                     uint64_t* dst_gpu_addr, uint64_t value,
+                                     uint64_t* dst_gpu_addr, uint64_t value, int event_id,
                                      uint32_t context_id = 0);
 void WriteAtEndOfPipeWithInterruptWriteBack32(uint64_t submit_id, CommandBuffer& buffer,
-                                              uint32_t* dst_gpu_addr, uint32_t value,
+                                              uint32_t* dst_gpu_addr, uint32_t value, int event_id,
                                               uint32_t context_id = 0);
 void WriteAtEndOfPipeWithInterruptWriteBack64(uint64_t submit_id, CommandBuffer& buffer,
-                                              uint64_t* dst_gpu_addr, uint64_t value,
+                                              uint64_t* dst_gpu_addr, uint64_t value, int event_id,
                                               uint32_t context_id = 0);
 
 [[nodiscard]] uint64_t PrepareVideoOutFlip(CommandBuffer& buffer, int handle, int index,
@@ -57,12 +56,13 @@ void WriteAtEndOfPipeWithFlip32(uint64_t submit_id, CommandBuffer& buffer, uint3
 void WriteAtEndOfPipeWithInterruptWriteBackFlip32(uint64_t submit_id, CommandBuffer& buffer,
                                                   uint32_t* dst_gpu_addr, uint32_t value,
                                                   int handle, int index, int flip_mode,
-                                                  int64_t flip_arg, uint64_t request_id);
+                                                  int64_t flip_arg, uint64_t request_id,
+                                                  int event_id);
 
 int  AddEqEvent(RenderContext& renderer, LibKernel::EventQueue::KernelEqueue eq, int id,
                 void* udata);
-int  DeleteEqEvent(LibKernel::EventQueue::KernelEqueue eq, int id);
-void ReadGds(Buffer& gds, uint32_t* dst, uint32_t dw_offset, uint32_t dw_size);
+int  DeleteEqEvent(RenderContext& renderer, LibKernel::EventQueue::KernelEqueue eq, int id);
+void ReadGds(const Buffer& gds, uint32_t* dst, uint32_t dw_offset, uint32_t dw_size);
 
 } // namespace Sync
 } // namespace Libs::Graphics

src/graphics/host_gpu/shaders/fault_buffer_process.comp

@@ -0,0 +1,35 @@
+// SPDX-FileCopyrightText: Copyright 2024 shadPS4 Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#version 450
+#extension GL_ARB_gpu_shader_int64 : enable
+
+layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in;
+
+layout(std430, binding = 0) buffer input_buf {
+    uint fault_buffer[];
+};
+
+layout(std430, binding = 1) buffer output_buf {
+    uint64_t download_buffer[];
+};
+layout(std430, binding = 1) buffer output_buf32 {
+    uint download_buffer32[];
+};
+
+void main() {
+    const uint id = gl_GlobalInvocationID.x;
+    uint word = fault_buffer[id];
+    fault_buffer[id] = 0u;
+    const uint base_bit = id * 32u;
+    while (word != 0u) {
+        const uint store_index = atomicAdd(download_buffer32[0], 1u) + 1u;
+        if (store_index >= MAX_PAGE_FAULTS) {
+            return;
+        }
+        const uint bit = findLSB(word);
+        word &= word - 1;
+        const uint page = base_bit + bit;
+        download_buffer[store_index] = uint64_t(page) << CACHING_PAGEBITS;
+    }
+}

src/graphics/host_gpu/vma.cpp

@@ -35,28 +35,20 @@ struct MemoryStats {
 
 MemoryStats g_memory_stats;
 
-void TrackAllocationImpl(const VulkanMemory& memory) {
+} // namespace
+
+void VulkanTrackAllocation(const VulkanMemory& memory) {
 	g_memory_stats.allocated[memory.type] += memory.requirements.size;
 	g_memory_stats.count[memory.type]++;
 }
 
-void UntrackAllocationImpl(const VulkanMemory& memory) {
+void VulkanUntrackAllocation(const VulkanMemory& memory) {
 	EXIT_IF(g_memory_stats.allocated[memory.type] < memory.requirements.size);
 	EXIT_IF(g_memory_stats.count[memory.type] == 0);
 	g_memory_stats.allocated[memory.type] -= memory.requirements.size;
 	g_memory_stats.count[memory.type]--;
 }
 
-} // namespace
-
-void VulkanTrackAllocation(const VulkanMemory& memory) {
-	TrackAllocationImpl(memory);
-}
-
-void VulkanUntrackAllocation(const VulkanMemory& memory) {
-	UntrackAllocationImpl(memory);
-}
-
 bool GraphicContext::CreateAllocator() {
 	KYTY_PROFILER_FUNCTION();
 	EXIT_IF(instance == nullptr || physical_device == nullptr || device == nullptr ||
@@ -72,11 +64,14 @@ bool GraphicContext::CreateAllocator() {
 	info.device           = device;
 	info.pVulkanFunctions = &functions;
 	info.vulkanApiVersion = VULKAN_TARGET_API_VERSION;
-	info.flags = memory_budget_ext_enabled ? VMA_ALLOCATOR_CREATE_EXT_MEMORY_BUDGET_BIT : 0;
+	info.flags = VMA_ALLOCATOR_CREATE_BUFFER_DEVICE_ADDRESS_BIT;
+	if (memory_budget_ext_enabled) {
+		info.flags |= VMA_ALLOCATOR_CREATE_EXT_MEMORY_BUDGET_BIT;
+	}
 
 	const auto result = static_cast<vk::Result>(vmaCreateAllocator(&info, &allocator));
 	if (result != vk::Result::eSuccess) {
-		LOGF("vmaCreateAllocator failed: %s\n", VulkanToString(result).c_str());
+		LOGF("vmaCreateAllocator failed: %s\n", vk::to_string(result).c_str());
 		return false;
 	}
 	return true;
@@ -90,11 +85,6 @@ void GraphicContext::DestroyAllocator() {
 	allocator = nullptr;
 }
 
-uint64_t VulkanNextMemoryUniqueId() {
-	static std::atomic_uint64_t sequence = 0;
-	return ++sequence;
-}
-
 void GraphicContext::LogMemoryBudget() const {
 	if (allocator == nullptr || physical_device == nullptr) {
 		return;
@@ -163,56 +153,6 @@ uint64_t GraphicContext::GetTotalMemoryBudget() const {
 	return std::max(local, available > system_reserve ? available - system_reserve : uint64_t {0});
 }
 
-void GraphicContext::CreateBuffer(uint64_t size, VulkanBuffer& buffer) {
-	KYTY_PROFILER_FUNCTION();
-	EXIT_IF(allocator == nullptr || buffer.buffer != nullptr ||
-	        buffer.memory.allocation != nullptr || size == 0);
-
-	vk::BufferCreateInfo buffer_info {};
-	buffer_info.sType       = vk::StructureType::eBufferCreateInfo;
-	buffer_info.size        = size;
-	buffer_info.usage       = buffer.usage;
-	buffer_info.sharingMode = vk::SharingMode::eExclusive;
-
-	VmaAllocationCreateInfo alloc_info {};
-	alloc_info.requiredFlags =
-	    static_cast<vk::MemoryPropertyFlags::MaskType>(buffer.memory.property);
-	alloc_info.preferredFlags =
-	    static_cast<vk::MemoryPropertyFlags::MaskType>(buffer.memory.preferred_property);
-
-	vk::Buffer::CType native_buffer = VK_NULL_HANDLE;
-	const auto        result        = static_cast<vk::Result>(vmaCreateBuffer(
-	    allocator, static_cast<const vk::BufferCreateInfo::NativeType*>(buffer_info), &alloc_info,
-	    &native_buffer, &buffer.memory.allocation, &buffer.memory.allocation_info));
-	buffer.buffer                   = native_buffer;
-	if (result != vk::Result::eSuccess) {
-		LogMemoryBudget();
-	}
-	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
-
-	device.getBufferMemoryRequirements(buffer.buffer, &buffer.memory.requirements);
-	buffer.memory.type      = buffer.memory.allocation_info.memoryType;
-	buffer.memory.memory    = buffer.memory.allocation_info.deviceMemory;
-	buffer.memory.offset    = buffer.memory.allocation_info.offset;
-	buffer.memory.unique_id = VulkanNextMemoryUniqueId();
-	buffer.buffer_size      = size;
-	VulkanTrackAllocation(buffer.memory);
-}
-
-void GraphicContext::DeleteBuffer(VulkanBuffer& buffer) {
-	KYTY_PROFILER_FUNCTION();
-	EXIT_IF(allocator == nullptr || buffer.buffer == nullptr ||
-	        buffer.memory.allocation == nullptr);
-
-	VulkanUntrackAllocation(buffer.memory);
-	vmaDestroyBuffer(allocator, buffer.buffer, buffer.memory.allocation);
-	buffer.buffer                 = nullptr;
-	buffer.memory.memory          = nullptr;
-	buffer.memory.allocation      = nullptr;
-	buffer.memory.allocation_info = {};
-	buffer.memory.offset          = 0;
-}
-
 bool GraphicContext::CreateImage(const vk::ImageCreateInfo& image_info, VulkanImage& image) {
 	KYTY_PROFILER_FUNCTION();
 	EXIT_IF(allocator == nullptr || image.image != nullptr || image.memory.allocation != nullptr);
@@ -223,10 +163,11 @@ bool GraphicContext::CreateImage(const vk::ImageCreateInfo& image_info, VulkanIm
 	alloc_info.preferredFlags =
 	    static_cast<vk::MemoryPropertyFlags::MaskType>(memory.preferred_property);
 
+	VmaAllocationInfo allocation_result {};
 	vk::Image::CType native_image = VK_NULL_HANDLE;
-	const auto       result       = static_cast<vk::Result>(
+	const auto        result       = static_cast<vk::Result>(
 	    vmaCreateImage(allocator, static_cast<const vk::ImageCreateInfo::NativeType*>(image_info),
-	                   &alloc_info, &native_image, &memory.allocation, &memory.allocation_info));
+	                   &alloc_info, &native_image, &memory.allocation, &allocation_result));
 	image.image = native_image;
 	if (result != vk::Result::eSuccess) {
 		LogMemoryBudget();
@@ -234,10 +175,7 @@ bool GraphicContext::CreateImage(const vk::ImageCreateInfo& image_info, VulkanIm
 	}
 
 	device.getImageMemoryRequirements(image.image, &memory.requirements);
-	memory.type      = memory.allocation_info.memoryType;
-	memory.memory    = memory.allocation_info.deviceMemory;
-	memory.offset    = memory.allocation_info.offset;
-	memory.unique_id = VulkanNextMemoryUniqueId();
+	memory.type = allocation_result.memoryType;
 	VulkanTrackAllocation(memory);
 	return true;
 }
@@ -250,23 +188,7 @@ void GraphicContext::DeleteImage(VulkanImage& image) {
 	VulkanUntrackAllocation(memory);
 	vmaDestroyImage(allocator, image.image, memory.allocation);
 	image.image            = nullptr;
-	memory.memory          = nullptr;
 	memory.allocation      = nullptr;
-	memory.allocation_info = {};
-	memory.offset          = 0;
-}
-
-void GraphicContext::MapMemory(VulkanMemory& memory, void*& data) {
-	KYTY_PROFILER_FUNCTION();
-	EXIT_IF(allocator == nullptr || memory.allocation == nullptr);
-	EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(vmaMapMemory(allocator, memory.allocation,
-	                                                          &data)) != vk::Result::eSuccess);
-}
-
-void GraphicContext::UnmapMemory(VulkanMemory& memory) {
-	KYTY_PROFILER_FUNCTION();
-	EXIT_IF(allocator == nullptr || memory.allocation == nullptr);
-	vmaUnmapMemory(allocator, memory.allocation);
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/vma.h

@@ -6,7 +6,6 @@
 
 namespace Libs::Graphics {
 
-uint64_t VulkanNextMemoryUniqueId();
 void     VulkanTrackAllocation(const VulkanMemory& memory);
 void     VulkanUntrackAllocation(const VulkanMemory& memory);
 

src/graphics/host_gpu/vulkanCommon.cpp

@@ -65,7 +65,6 @@ constexpr FormatMapping kFormatMappings[] = {
     {Prospero::BufferFormat::k5_6_5UNorm, vk::Format::eB5G6R5UnormPack16},
     {Prospero::BufferFormat::k5_5_5_1UNorm, vk::Format::eR5G5B5A1UnormPack16},
     {Prospero::BufferFormat::k4_4_4_4UNorm, vk::Format::eR4G4B4A4UnormPack16},
-    {Prospero::BufferFormat::kFmask8_S4_F4, vk::Format::eR32Sfloat},
     {Prospero::BufferFormat::kBc1UNorm, vk::Format::eBc1RgbaUnormBlock},
     {Prospero::BufferFormat::kBc1Srgb, vk::Format::eBc1RgbaSrgbBlock},
     {Prospero::BufferFormat::kBc2UNorm, vk::Format::eBc2UnormBlock},
@@ -96,22 +95,6 @@ constexpr auto kFormatLookup = MakeFormatLookup();
 
 } // namespace
 
-std::string VulkanToString(vk::Result value) {
-	return vk::to_string(value);
-}
-
-std::string VulkanToString(vk::Format value) {
-	return vk::to_string(value);
-}
-
-std::string VulkanToString(vk::ImageLayout value) {
-	return vk::to_string(value);
-}
-
-std::string VulkanToString(vk::QueueFlags value) {
-	return vk::to_string(value);
-}
-
 vk::Format VulkanFormat(Prospero::BufferFormat guest_format) {
 	const auto index = static_cast<size_t>(guest_format);
 	return index < kFormatLookup.size() ? kFormatLookup[index] : vk::Format::eUndefined;
@@ -119,7 +102,7 @@ vk::Format VulkanFormat(Prospero::BufferFormat guest_format) {
 
 void RequireVulkanSuccess(vk::Result result, const char* operation) {
 	if (result != vk::Result::eSuccess) {
-		EXIT("%s failed: %s (%d)\n", operation, VulkanToString(result).c_str(),
+		EXIT("%s failed: %s (%d)\n", operation, vk::to_string(result).c_str(),
 		     static_cast<int>(result));
 	}
 }

src/graphics/host_gpu/vulkanCommon.h

@@ -10,10 +10,13 @@
 #define VMA_STATIC_VULKAN_FUNCTIONS  0
 #define VMA_DYNAMIC_VULKAN_FUNCTIONS 1
 
+#include "common/emulatorConfig.h"
 #include "graphics/guest_gpu/gpu_defs.h"
 
 #include <cstdint>
+#include <fmt/format.h>
 #include <string>
+#include <utility>
 #include <vector>
 #include <vulkan/vulkan.hpp>
 #include <vulkan/vulkan_hash.hpp>
@@ -22,13 +25,27 @@ namespace Libs::Graphics {
 
 using VulkanMemoryBarrier = vk::MemoryBarrier;
 
-std::string VulkanToString(vk::Result value);
-std::string VulkanToString(vk::Format value);
-std::string VulkanToString(vk::ImageLayout value);
-std::string VulkanToString(vk::QueueFlags value);
 vk::Format  VulkanFormat(Prospero::BufferFormat guest_format);
 void        RequireVulkanSuccess(vk::Result result, const char* operation);
 
+template <typename Handle, typename... Args>
+void SetVulkanObjectNameF(vk::Device device, Handle handle, fmt::format_string<Args...> format,
+                          Args&&... args) {
+	if (!Config::GraphicsDebugDumpEnabled() || device == nullptr || handle == nullptr ||
+	    VULKAN_HPP_DEFAULT_DISPATCHER.vkSetDebugUtilsObjectNameEXT == nullptr) {
+		return;
+	}
+
+	const auto                      name = fmt::format(format, std::forward<Args>(args)...);
+	vk::DebugUtilsObjectNameInfoEXT info {};
+	info.sType        = vk::StructureType::eDebugUtilsObjectNameInfoEXT;
+	info.objectType   = Handle::objectType;
+	info.objectHandle = static_cast<uint64_t>(
+	    reinterpret_cast<uintptr_t>(static_cast<typename Handle::CType>(handle)));
+	info.pObjectName = name.c_str();
+	(void)device.setDebugUtilsObjectNameEXT(&info);
+}
+
 template <typename T, typename Enumerator>
 [[nodiscard]] std::vector<T> EnumerateVulkan(const char* operation, Enumerator&& enumerate) {
 	for (;;) {