fiber: save contexts without compiler frames

This reverts commit 7012634f32.

.github/ISSUE_TEMPLATE/kytyps5-game-emulation.yaml

@@ -1,6 +1,6 @@
-name: Game Emulation Bug Report
-description: Report KytyPS5 game boot, crash, freeze, rendering, audio, input, or gameplay behavior problems.
-title: "[GAME BUG]: "
+name: Game Emulation Status Report
+description: Submit a KytyPS5 game compatibility result, including boot, rendering, audio, input, or gameplay behavior.
+title: "[GAME STATUS]: "
 body:
   - type: markdown
     attributes:
@@ -10,6 +10,7 @@ body:
         Before submitting, please:
         - Search existing issues first.
         - Re-test with a clean game dump.
+        - Use the latest KytyPS5 build available to you.
         - Attach the complete log file instead of edited snippets.
 
   - type: markdown
@@ -44,6 +45,28 @@ body:
     validations:
       required: true
 
+  - type: dropdown
+    id: compatibility-status
+    attributes:
+      label: Compatibility status
+      description: Select the furthest point the game consistently reaches on this build.
+      options:
+        - "Doesn't boot"
+        - Logo
+        - Main menu
+        - In game
+    validations:
+      required: true
+
+  - type: markdown
+    attributes:
+      value: |
+        **Status guide**
+        - **Doesn't boot:** The game does not show its first logo or startup screen.
+        - **Logo:** The game shows a logo or startup screen but does not reach the main menu.
+        - **Main menu:** The game reaches its menus but does not enter gameplay.
+        - **In game:** The game reaches controllable gameplay, even if bugs prevent completion.
+
   - type: input
     id: regression-builds
     attributes:
@@ -54,22 +77,22 @@ body:
   - type: markdown
     attributes:
       value: |
-        ## Problem details
+        ## Test result
 
   - type: textarea
     id: what-happened
     attributes:
-      label: What happened?
-      description: Describe the failure or incorrect behavior as directly as possible.
-      placeholder: "Example: The game reaches the main menu, then freezes after selecting Continue."
+      label: Result details
+      description: Describe what worked, what failed, and any visible or audible problems.
+      placeholder: "Example: The game reaches the main menu, then freezes after selecting Continue. Audio plays correctly until the freeze."
     validations:
       required: true
 
   - type: textarea
     id: steps-to-reproduce
     attributes:
-      label: Steps to reproduce
-      description: List the shortest path that triggers the problem from a fresh KytyPS5 start.
+      label: Steps to reproduce the result
+      description: List the shortest path to the reported status or failure from a fresh KytyPS5 start.
       placeholder: |
         1. Open KytyPS5.
         2. Boot the game.
@@ -81,7 +104,7 @@ body:
     id: expected-behavior
     attributes:
       label: Expected behavior
-      description: Explain what should happen at the same point in the game.
+      description: Explain what should happen at the same point in the game. If there is no known problem, write "No known issue".
       placeholder: "Example: The save should load and gameplay should continue normally."
     validations:
       required: true

.github/workflows/build.yml

@@ -94,6 +94,7 @@ jobs:
           cmake -S . -B _Build/windows ^
             -G Ninja ^
             -DCMAKE_BUILD_TYPE=Release ^
+            -DCMAKE_INTERPROCEDURAL_OPTIMIZATION=ON ^
             -DKYTY_BUILD_ORIGIN=%KYTY_BUILD_ORIGIN% ^
             -DKYTY_BUILD_REPOSITORY=%KYTY_BUILD_REPOSITORY% ^
             -DKYTY_RELEASE_TAG=${{ needs.metadata.outputs.release_tag }} ^
@@ -104,12 +105,12 @@ jobs:
       - name: Build
         shell: cmd
         run: |
-          cmake --build _Build/windows --target launcher audio_out2_port_tests virtual_memory_allocation_tests --parallel
+          cmake --build _Build/windows --target launcher audio_out2_port_tests ime_dialog_tests virtual_memory_allocation_tests --parallel
 
       - name: Test
         shell: cmd
         run: |
-          ctest --test-dir _Build/windows --output-on-failure -R "^(audio_out2_port|virtual_memory_allocation)$"
+          ctest --test-dir _Build/windows --output-on-failure -R "^(audio_out2_port|ime_dialog|virtual_memory_allocation)$"
 
       - name: Install
         shell: cmd
@@ -172,6 +173,7 @@ jobs:
           cmake -S . -B _Build/macos \
             -G Ninja \
             -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_INTERPROCEDURAL_OPTIMIZATION=ON \
             -DKYTY_BUILD_ORIGIN="$KYTY_BUILD_ORIGIN" \
             -DKYTY_BUILD_REPOSITORY="$KYTY_BUILD_REPOSITORY" \
             -DKYTY_RELEASE_TAG="${{ needs.metadata.outputs.release_tag }}" \
@@ -184,14 +186,14 @@ jobs:
         shell: bash
         run: |
           cmake --build _Build/macos \
-            --target launcher audio_out2_port_tests virtual_memory_allocation_tests \
+            --target launcher audio_out2_port_tests ime_dialog_tests virtual_memory_allocation_tests \
             --parallel
 
       - name: Test
         shell: bash
         run: |
           ctest --test-dir _Build/macos --output-on-failure \
-            -R '^(audio_out2_port|virtual_memory_allocation)$'
+            -R '^(audio_out2_port|ime_dialog|virtual_memory_allocation)$'
 
       - name: Install
         shell: bash
@@ -299,6 +301,7 @@ jobs:
           cmake -S . -B _Build/linux \
             -G Ninja \
             -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_INTERPROCEDURAL_OPTIMIZATION=ON \
             -DKYTY_BUILD_ORIGIN="$KYTY_BUILD_ORIGIN" \
             -DKYTY_BUILD_REPOSITORY="$KYTY_BUILD_REPOSITORY" \
             -DKYTY_RELEASE_TAG="${{ needs.metadata.outputs.release_tag }}" \
@@ -326,14 +329,14 @@ jobs:
         run: |
           cmake --build _Build/linux \
             --target launcher page_manager_tests memory_tracker_tests \
-              audio_out2_port_tests virtual_memory_allocation_tests \
+              audio_out2_port_tests ime_dialog_tests virtual_memory_allocation_tests \
             --parallel
 
       - name: Test
         shell: bash
         run: |
           ctest --test-dir _Build/linux --output-on-failure \
-            -R '^(audio_out2_port|page_manager|memory_tracker|virtual_memory_allocation)$'
+            -R '^(audio_out2_port|ime_dialog|page_manager|memory_tracker|virtual_memory_allocation)$'
 
       - name: Install
         shell: bash
@@ -370,7 +373,13 @@ jobs:
 
   release:
     name: Release KytyPS5
-    if: github.event_name == 'push' && github.repository == 'KytyPS5/KytyPS5'
+    if: >-
+      always() &&
+      github.event_name == 'push' &&
+      github.repository == 'KytyPS5/KytyPS5' &&
+      (needs.windows.result == 'success' ||
+       needs.macos.result == 'success' ||
+       needs.linux.result == 'success')
     needs: [metadata, windows, macos, linux]
     runs-on: ubuntu-latest
     permissions:
@@ -379,34 +388,62 @@ jobs:
       RELEASE_NAME: ${{ needs.metadata.outputs.release_tag }}
 
     steps:
-      - name: Download builds
+      - name: Download Windows build
+        if: needs.windows.result == 'success'
         uses: actions/download-artifact@v4
         with:
-          path: artifacts
+          name: KytyPS5-Windows-x64
+          path: artifacts/KytyPS5-Windows-x64
 
-      - name: Package builds
+      - name: Download macOS build
+        if: needs.macos.result == 'success'
+        uses: actions/download-artifact@v4
+        with:
+          name: KytyPS5-macOS-x86_64
+          path: artifacts/KytyPS5-macOS-x86_64
+
+      - name: Download Linux build
+        if: needs.linux.result == 'success'
+        uses: actions/download-artifact@v4
+        with:
+          name: KytyPS5-Linux-x86_64
+          path: artifacts/KytyPS5-Linux-x86_64
+
+      - name: Package Windows build
+        if: needs.windows.result == 'success'
         shell: bash
         run: |
           windows_dir="artifacts/KytyPS5-Windows-x64"
-          macos_dir="artifacts/KytyPS5-macOS-x86_64"
-          linux_dir="artifacts/KytyPS5-Linux-x86_64"
           test -d "$windows_dir"
-          test -d "$macos_dir"
-          test -d "$linux_dir"
-          chmod a+x \
-            "$macos_dir/launcher" \
-            "$macos_dir/kyty_emulator" \
-            "$macos_dir/libMoltenVK.dylib" \
-            "$linux_dir/launcher" \
-            "$linux_dir/kyty_emulator"
           (
             cd "$windows_dir"
             zip -r "$GITHUB_WORKSPACE/$RELEASE_NAME-Windows-x64.zip" .
           )
+
+      - name: Package macOS build
+        if: needs.macos.result == 'success'
+        shell: bash
+        run: |
+          macos_dir="artifacts/KytyPS5-macOS-x86_64"
+          test -d "$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" .
           )
+
+      - name: Package Linux build
+        if: needs.linux.result == 'success'
+        shell: bash
+        run: |
+          linux_dir="artifacts/KytyPS5-Linux-x86_64"
+          test -d "$linux_dir"
+          chmod a+x \
+            "$linux_dir/launcher" \
+            "$linux_dir/kyty_emulator"
           tar -C "$linux_dir" -czf \
             "$GITHUB_WORKSPACE/$RELEASE_NAME-Linux-x86_64.tar.gz" .
 
@@ -415,11 +452,16 @@ jobs:
         env:
           GH_TOKEN: ${{ github.token }}
         run: |
-          assets=(
-            "$RELEASE_NAME-Windows-x64.zip"
-            "$RELEASE_NAME-macOS-x86_64.zip"
-            "$RELEASE_NAME-Linux-x86_64.tar.gz"
-          )
+          assets=()
+          for artifact in \
+            "$RELEASE_NAME-Windows-x64.zip" \
+            "$RELEASE_NAME-macOS-x86_64.zip" \
+            "$RELEASE_NAME-Linux-x86_64.tar.gz"; do
+            if [[ -f "$artifact" ]]; then
+              assets+=("$artifact")
+            fi
+          done
+          test "${#assets[@]}" -gt 0
           if gh release view "$RELEASE_NAME" --repo "$GITHUB_REPOSITORY" > /dev/null 2>&1; then
             gh release upload "$RELEASE_NAME" "${assets[@]}" \
               --clobber \

.gitmodules

@@ -39,3 +39,6 @@
 [submodule "3rdparty/nlohmann_json"]
 	path = 3rdparty/nlohmann_json
 	url = https://github.com/nlohmann/json.git
+[submodule "3rdparty/imgui"]
+	path = 3rdparty/imgui
+	url = https://github.com/ocornut/imgui.git

3rdparty/CMakeLists.txt

@@ -2,6 +2,32 @@ set(BUILD_SHARED_LIBS OFF CACHE BOOL "" FORCE)
 
 add_subdirectory(cpuinfo)
 
+# The Windows guest red-zone patcher needs a decoder/encoder and a small
+# trampoline assembler. Pin the same revisions used by shadPS4's implementation.
+include(FetchContent)
+
+set(XBYAK_BUILD_TESTS OFF CACHE BOOL "" FORCE)
+FetchContent_Declare(
+	xbyak
+	GIT_REPOSITORY https://github.com/herumi/xbyak.git
+	GIT_TAG 44a72f369268f7d552650891b296693e91db86bb
+	GIT_SHALLOW FALSE
+)
+
+set(ZYDIS_BUILD_EXAMPLES OFF CACHE BOOL "" FORCE)
+set(ZYDIS_BUILD_TOOLS OFF CACHE BOOL "" FORCE)
+set(ZYDIS_BUILD_TESTS OFF CACHE BOOL "" FORCE)
+set(ZYDIS_BUILD_DOXYGEN OFF CACHE BOOL "" FORCE)
+FetchContent_Declare(
+	zydis
+	GIT_REPOSITORY https://github.com/zyantific/zydis.git
+	GIT_TAG 120e0e705f8e3b507dc49377ac2879979f0d545c
+	GIT_SUBMODULES_RECURSE TRUE
+	GIT_SHALLOW FALSE
+)
+
+FetchContent_MakeAvailable(xbyak zydis)
+
 set(SDL_SHARED OFF CACHE BOOL "" FORCE)
 add_subdirectory(SDL2)
 
@@ -11,6 +37,25 @@ target_include_directories(Vulkan-Headers INTERFACE
 	"${CMAKE_CURRENT_SOURCE_DIR}/Vulkan-Headers/include"
 )
 
+add_library(imgui STATIC
+	imgui/imgui.cpp
+	imgui/imgui_draw.cpp
+	imgui/imgui_tables.cpp
+	imgui/imgui_widgets.cpp
+	imgui/backends/imgui_impl_vulkan.cpp
+)
+target_include_directories(imgui PUBLIC
+	"${CMAKE_CURRENT_SOURCE_DIR}/imgui"
+	"${CMAKE_CURRENT_SOURCE_DIR}/imgui/backends"
+)
+target_compile_definitions(imgui PRIVATE IMGUI_IMPL_VULKAN_NO_PROTOTYPES)
+target_link_libraries(imgui PUBLIC Vulkan::Headers)
+if(MSVC)
+	target_compile_options(imgui PRIVATE /W0)
+else()
+	target_compile_options(imgui PRIVATE -w)
+endif()
+
 set(SPIRV_SKIP_TESTS ON CACHE BOOL "" FORCE)
 set(SPIRV_SKIP_EXECUTABLES ON CACHE BOOL "" FORCE)
 set(SKIP_SPIRV_TOOLS_INSTALL ON CACHE BOOL "" FORCE)

3rdparty/renderdoc/renderdoc_app.h

@@ -0,0 +1,875 @@
+/******************************************************************************
+ * The MIT License (MIT)
+ *
+ * Copyright (c) 2015-2026 Baldur Karlsson
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ ******************************************************************************/
+
+#pragma once
+
+//////////////////////////////////////////////////////////////////////////////////////////////////
+//
+// Documentation for the API is available at https://renderdoc.org/docs/in_application_api.html
+//
+
+#if !defined(RENDERDOC_NO_STDINT)
+#include <stdint.h>
+#endif
+
+#if defined(WIN32) || defined(__WIN32__) || defined(_WIN32) || defined(_MSC_VER)
+#define RENDERDOC_CC __cdecl
+#elif defined(__linux__) || defined(__FreeBSD__) || defined(__sun__) || defined(__OpenBSD__)
+#define RENDERDOC_CC
+#elif defined(__APPLE__)
+#define RENDERDOC_CC
+#else
+#error "Unknown platform"
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+//////////////////////////////////////////////////////////////////////////////////////////////////
+// Constants not used directly in below API
+
+// This is a GUID/magic value used for when applications pass a path where shader debug
+// information can be found to match up with a stripped shader.
+// the define can be used like so: const GUID RENDERDOC_ShaderDebugMagicValue =
+// RENDERDOC_ShaderDebugMagicValue_value
+#define RENDERDOC_ShaderDebugMagicValue_struct                                \
+  {                                                                           \
+    0xeab25520, 0x6670, 0x4865, 0x84, 0x29, 0x6c, 0x8, 0x51, 0x54, 0x00, 0xff \
+  }
+
+// as an alternative when you want a byte array (assuming x86 endianness):
+#define RENDERDOC_ShaderDebugMagicValue_bytearray                                                 \
+  {                                                                                               \
+    0x20, 0x55, 0xb2, 0xea, 0x70, 0x66, 0x65, 0x48, 0x84, 0x29, 0x6c, 0x8, 0x51, 0x54, 0x00, 0xff \
+  }
+
+// truncated version when only a uint64_t is available (e.g. Vulkan tags):
+#define RENDERDOC_ShaderDebugMagicValue_truncated 0x48656670eab25520ULL
+
+// this is a magic value for vulkan user tags to indicate which dispatchable API objects are which
+// for object annotations
+#define RENDERDOC_APIObjectAnnotationHelper 0xfbb3b337b664d0adULL
+
+//////////////////////////////////////////////////////////////////////////////////////////////////
+// RenderDoc capture options
+//
+
+typedef enum RENDERDOC_CaptureOption
+{
+  // Allow the application to enable vsync
+  //
+  // Default - enabled
+  //
+  // 1 - The application can enable or disable vsync at will
+  // 0 - vsync is force disabled
+  eRENDERDOC_Option_AllowVSync = 0,
+
+  // Allow the application to enable fullscreen
+  //
+  // Default - enabled
+  //
+  // 1 - The application can enable or disable fullscreen at will
+  // 0 - fullscreen is force disabled
+  eRENDERDOC_Option_AllowFullscreen = 1,
+
+  // Record API debugging events and messages
+  //
+  // Default - disabled
+  //
+  // 1 - Enable built-in API debugging features and records the results into
+  //     the capture, which is matched up with events on replay
+  // 0 - no API debugging is forcibly enabled
+  eRENDERDOC_Option_APIValidation = 2,
+  eRENDERDOC_Option_DebugDeviceMode = 2,    // deprecated name of this enum
+
+  // Capture CPU callstacks for API events
+  //
+  // Default - disabled
+  //
+  // 1 - Enables capturing of callstacks
+  // 0 - no callstacks are captured
+  eRENDERDOC_Option_CaptureCallstacks = 3,
+
+  // When capturing CPU callstacks, only capture them from actions.
+  // This option does nothing without the above option being enabled
+  //
+  // Default - disabled
+  //
+  // 1 - Only captures callstacks for actions.
+  //     Ignored if CaptureCallstacks is disabled
+  // 0 - Callstacks, if enabled, are captured for every event.
+  eRENDERDOC_Option_CaptureCallstacksOnlyDraws = 4,
+  eRENDERDOC_Option_CaptureCallstacksOnlyActions = 4,
+
+  // Specify a delay in seconds to wait for a debugger to attach, after
+  // creating or injecting into a process, before continuing to allow it to run.
+  //
+  // 0 indicates no delay, and the process will run immediately after injection
+  //
+  // Default - 0 seconds
+  //
+  eRENDERDOC_Option_DelayForDebugger = 5,
+
+  // Verify buffer access. This includes checking the memory returned by a Map() call to
+  // detect any out-of-bounds modification, as well as initialising buffers with undefined contents
+  // to a marker value to catch use of uninitialised memory.
+  //
+  // NOTE: This option is only valid for OpenGL and D3D11. Explicit APIs such as D3D12 and Vulkan do
+  // not do the same kind of interception & checking and undefined contents are really undefined.
+  //
+  // Default - disabled
+  //
+  // 1 - Verify buffer access
+  // 0 - No verification is performed, and overwriting bounds may cause crashes or corruption in
+  //     RenderDoc.
+  eRENDERDOC_Option_VerifyBufferAccess = 6,
+
+  // The old name for eRENDERDOC_Option_VerifyBufferAccess was eRENDERDOC_Option_VerifyMapWrites.
+  // This option now controls the filling of uninitialised buffers with 0xdddddddd which was
+  // previously always enabled
+  eRENDERDOC_Option_VerifyMapWrites = eRENDERDOC_Option_VerifyBufferAccess,
+
+  // Hooks any system API calls that create child processes, and injects
+  // RenderDoc into them recursively with the same options.
+  //
+  // Default - disabled
+  //
+  // 1 - Hooks into spawned child processes
+  // 0 - Child processes are not hooked by RenderDoc
+  eRENDERDOC_Option_HookIntoChildren = 7,
+
+  // By default RenderDoc only includes resources in the final capture necessary
+  // for that frame, this allows you to override that behaviour.
+  //
+  // Default - disabled
+  //
+  // 1 - all live resources at the time of capture are included in the capture
+  //     and available for inspection
+  // 0 - only the resources referenced by the captured frame are included
+  eRENDERDOC_Option_RefAllResources = 8,
+
+  // **NOTE**: As of RenderDoc v1.1 this option has been deprecated. Setting or
+  // getting it will be ignored, to allow compatibility with older versions.
+  // In v1.1 the option acts as if it's always enabled.
+  //
+  // By default RenderDoc skips saving initial states for resources where the
+  // previous contents don't appear to be used, assuming that writes before
+  // reads indicate previous contents aren't used.
+  //
+  // Default - disabled
+  //
+  // 1 - initial contents at the start of each captured frame are saved, even if
+  //     they are later overwritten or cleared before being used.
+  // 0 - unless a read is detected, initial contents will not be saved and will
+  //     appear as black or empty data.
+  eRENDERDOC_Option_SaveAllInitials = 9,
+
+  // In APIs that allow for the recording of command lists to be replayed later,
+  // RenderDoc may choose to not capture command lists before a frame capture is
+  // triggered, to reduce overheads. This means any command lists recorded once
+  // and replayed many times will not be available and may cause a failure to
+  // capture.
+  //
+  // NOTE: This is only true for APIs where multithreading is difficult or
+  // discouraged. Newer APIs like Vulkan and D3D12 will ignore this option
+  // and always capture all command lists since the API is heavily oriented
+  // around it and the overheads have been reduced by API design.
+  //
+  // 1 - All command lists are captured from the start of the application
+  // 0 - Command lists are only captured if their recording begins during
+  //     the period when a frame capture is in progress.
+  eRENDERDOC_Option_CaptureAllCmdLists = 10,
+
+  // Mute API debugging output when the API validation mode option is enabled
+  //
+  // Default - enabled
+  //
+  // 1 - Mute any API debug messages from being displayed or passed through
+  // 0 - API debugging is displayed as normal
+  eRENDERDOC_Option_DebugOutputMute = 11,
+
+  // Option to allow vendor extensions to be used even when they may be
+  // incompatible with RenderDoc and cause corrupted replays or crashes.
+  //
+  // Default - inactive
+  //
+  // No values are documented, this option should only be used when absolutely
+  // necessary as directed by a RenderDoc developer.
+  eRENDERDOC_Option_AllowUnsupportedVendorExtensions = 12,
+
+  // Define a soft memory limit which some APIs may aim to keep overhead under where
+  // possible. Anything above this limit will where possible be saved directly to disk during
+  // capture.
+  // This will cause increased disk space use (which may cause a capture to fail if disk space is
+  // exhausted) as well as slower capture times.
+  //
+  // Not all memory allocations may be deferred like this so it is not a guarantee of a memory
+  // limit.
+  //
+  // Units are in MBs, suggested values would range from 200MB to 1000MB.
+  //
+  // Default - 0 Megabytes
+  eRENDERDOC_Option_SoftMemoryLimit = 13,
+} RENDERDOC_CaptureOption;
+
+// Sets an option that controls how RenderDoc behaves on capture.
+//
+// Returns 1 if the option and value are valid
+// Returns 0 if either is invalid and the option is unchanged
+typedef int(RENDERDOC_CC *pRENDERDOC_SetCaptureOptionU32)(RENDERDOC_CaptureOption opt, uint32_t val);
+typedef int(RENDERDOC_CC *pRENDERDOC_SetCaptureOptionF32)(RENDERDOC_CaptureOption opt, float val);
+
+// Gets the current value of an option as a uint32_t
+//
+// If the option is invalid, 0xffffffff is returned
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_GetCaptureOptionU32)(RENDERDOC_CaptureOption opt);
+
+// Gets the current value of an option as a float
+//
+// If the option is invalid, -FLT_MAX is returned
+typedef float(RENDERDOC_CC *pRENDERDOC_GetCaptureOptionF32)(RENDERDOC_CaptureOption opt);
+
+typedef enum RENDERDOC_InputButton
+{
+  // '0' - '9' matches ASCII values
+  eRENDERDOC_Key_0 = 0x30,
+  eRENDERDOC_Key_1 = 0x31,
+  eRENDERDOC_Key_2 = 0x32,
+  eRENDERDOC_Key_3 = 0x33,
+  eRENDERDOC_Key_4 = 0x34,
+  eRENDERDOC_Key_5 = 0x35,
+  eRENDERDOC_Key_6 = 0x36,
+  eRENDERDOC_Key_7 = 0x37,
+  eRENDERDOC_Key_8 = 0x38,
+  eRENDERDOC_Key_9 = 0x39,
+
+  // 'A' - 'Z' matches ASCII values
+  eRENDERDOC_Key_A = 0x41,
+  eRENDERDOC_Key_B = 0x42,
+  eRENDERDOC_Key_C = 0x43,
+  eRENDERDOC_Key_D = 0x44,
+  eRENDERDOC_Key_E = 0x45,
+  eRENDERDOC_Key_F = 0x46,
+  eRENDERDOC_Key_G = 0x47,
+  eRENDERDOC_Key_H = 0x48,
+  eRENDERDOC_Key_I = 0x49,
+  eRENDERDOC_Key_J = 0x4A,
+  eRENDERDOC_Key_K = 0x4B,
+  eRENDERDOC_Key_L = 0x4C,
+  eRENDERDOC_Key_M = 0x4D,
+  eRENDERDOC_Key_N = 0x4E,
+  eRENDERDOC_Key_O = 0x4F,
+  eRENDERDOC_Key_P = 0x50,
+  eRENDERDOC_Key_Q = 0x51,
+  eRENDERDOC_Key_R = 0x52,
+  eRENDERDOC_Key_S = 0x53,
+  eRENDERDOC_Key_T = 0x54,
+  eRENDERDOC_Key_U = 0x55,
+  eRENDERDOC_Key_V = 0x56,
+  eRENDERDOC_Key_W = 0x57,
+  eRENDERDOC_Key_X = 0x58,
+  eRENDERDOC_Key_Y = 0x59,
+  eRENDERDOC_Key_Z = 0x5A,
+
+  // leave the rest of the ASCII range free
+  // in case we want to use it later
+  eRENDERDOC_Key_NonPrintable = 0x100,
+
+  eRENDERDOC_Key_Divide,
+  eRENDERDOC_Key_Multiply,
+  eRENDERDOC_Key_Subtract,
+  eRENDERDOC_Key_Plus,
+
+  eRENDERDOC_Key_F1,
+  eRENDERDOC_Key_F2,
+  eRENDERDOC_Key_F3,
+  eRENDERDOC_Key_F4,
+  eRENDERDOC_Key_F5,
+  eRENDERDOC_Key_F6,
+  eRENDERDOC_Key_F7,
+  eRENDERDOC_Key_F8,
+  eRENDERDOC_Key_F9,
+  eRENDERDOC_Key_F10,
+  eRENDERDOC_Key_F11,
+  eRENDERDOC_Key_F12,
+
+  eRENDERDOC_Key_Home,
+  eRENDERDOC_Key_End,
+  eRENDERDOC_Key_Insert,
+  eRENDERDOC_Key_Delete,
+  eRENDERDOC_Key_PageUp,
+  eRENDERDOC_Key_PageDn,
+
+  eRENDERDOC_Key_Backspace,
+  eRENDERDOC_Key_Tab,
+  eRENDERDOC_Key_PrtScrn,
+  eRENDERDOC_Key_Pause,
+
+  eRENDERDOC_Key_Max,
+} RENDERDOC_InputButton;
+
+// Sets which key or keys can be used to toggle focus between multiple windows
+//
+// If keys is NULL or num is 0, toggle keys will be disabled
+typedef void(RENDERDOC_CC *pRENDERDOC_SetFocusToggleKeys)(RENDERDOC_InputButton *keys, int num);
+
+// Sets which key or keys can be used to capture the next frame
+//
+// If keys is NULL or num is 0, captures keys will be disabled
+typedef void(RENDERDOC_CC *pRENDERDOC_SetCaptureKeys)(RENDERDOC_InputButton *keys, int num);
+
+typedef enum RENDERDOC_OverlayBits
+{
+  // This single bit controls whether the overlay is enabled or disabled globally
+  eRENDERDOC_Overlay_Enabled = 0x1,
+
+  // Show the average framerate over several seconds as well as min/max
+  eRENDERDOC_Overlay_FrameRate = 0x2,
+
+  // Show the current frame number
+  eRENDERDOC_Overlay_FrameNumber = 0x4,
+
+  // Show a list of recent captures, and how many captures have been made
+  eRENDERDOC_Overlay_CaptureList = 0x8,
+
+  // Default values for the overlay mask
+  eRENDERDOC_Overlay_Default = (eRENDERDOC_Overlay_Enabled | eRENDERDOC_Overlay_FrameRate |
+                                eRENDERDOC_Overlay_FrameNumber | eRENDERDOC_Overlay_CaptureList),
+
+  // Enable all bits
+  eRENDERDOC_Overlay_All = 0x7ffffff,
+
+  // Disable all bits
+  eRENDERDOC_Overlay_None = 0,
+} RENDERDOC_OverlayBits;
+
+// returns the overlay bits that have been set
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_GetOverlayBits)(void);
+// sets the overlay bits with an and & or mask
+typedef void(RENDERDOC_CC *pRENDERDOC_MaskOverlayBits)(uint32_t And, uint32_t Or);
+
+// this function will attempt to remove RenderDoc's hooks in the application.
+//
+// Note: that this can only work correctly if done immediately after
+// the module is loaded, before any API work happens. RenderDoc will remove its
+// injected hooks and shut down. Behaviour is undefined if this is called
+// after any API functions have been called, and there is still no guarantee of
+// success.
+typedef void(RENDERDOC_CC *pRENDERDOC_RemoveHooks)(void);
+
+// DEPRECATED: compatibility for code compiled against pre-1.4.1 headers.
+typedef pRENDERDOC_RemoveHooks pRENDERDOC_Shutdown;
+
+// This function will unload RenderDoc's crash handler.
+//
+// If you use your own crash handler and don't want RenderDoc's handler to
+// intercede, you can call this function to unload it and any unhandled
+// exceptions will pass to the next handler.
+typedef void(RENDERDOC_CC *pRENDERDOC_UnloadCrashHandler)(void);
+
+// Sets the capture file path template
+//
+// pathtemplate is a UTF-8 string that gives a template for how captures will be named
+// and where they will be saved.
+//
+// Any extension is stripped off the path, and captures are saved in the directory
+// specified, and named with the filename and the frame number appended. If the
+// directory does not exist it will be created, including any parent directories.
+//
+// If pathtemplate is NULL, the template will remain unchanged
+//
+// Example:
+//
+// SetCaptureFilePathTemplate("my_captures/example");
+//
+// Capture #1 -> my_captures/example_frame123.rdc
+// Capture #2 -> my_captures/example_frame456.rdc
+typedef void(RENDERDOC_CC *pRENDERDOC_SetCaptureFilePathTemplate)(const char *pathtemplate);
+
+// returns the current capture path template, see SetCaptureFileTemplate above, as a UTF-8 string
+typedef const char *(RENDERDOC_CC *pRENDERDOC_GetCaptureFilePathTemplate)(void);
+
+// DEPRECATED: compatibility for code compiled against pre-1.1.2 headers.
+typedef pRENDERDOC_SetCaptureFilePathTemplate pRENDERDOC_SetLogFilePathTemplate;
+typedef pRENDERDOC_GetCaptureFilePathTemplate pRENDERDOC_GetLogFilePathTemplate;
+
+// returns the number of captures that have been made
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_GetNumCaptures)(void);
+
+// This function returns the details of a capture, by index. New captures are added
+// to the end of the list.
+//
+// filename will be filled with the absolute path to the capture file, as a UTF-8 string
+// pathlength will be written with the length in bytes of the filename string
+// timestamp will be written with the time of the capture, in seconds since the Unix epoch
+//
+// Any of the parameters can be NULL and they'll be skipped.
+//
+// The function will return 1 if the capture index is valid, or 0 if the index is invalid
+// If the index is invalid, the values will be unchanged
+//
+// Note: when captures are deleted in the UI they will remain in this list, so the
+// capture path may not exist anymore.
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_GetCapture)(uint32_t idx, char *filename,
+                                                      uint32_t *pathlength, uint64_t *timestamp);
+
+// Sets the comments associated with a capture file. These comments are displayed in the
+// UI program when opening.
+//
+// filePath should be a path to the capture file to add comments to. If set to NULL or ""
+// the most recent capture file created made will be used instead.
+// comments should be a NULL-terminated UTF-8 string to add as comments.
+//
+// Any existing comments will be overwritten.
+typedef void(RENDERDOC_CC *pRENDERDOC_SetCaptureFileComments)(const char *filePath,
+                                                              const char *comments);
+
+// returns 1 if the RenderDoc UI is connected to this application, 0 otherwise
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_IsTargetControlConnected)(void);
+
+// DEPRECATED: compatibility for code compiled against pre-1.1.1 headers.
+// This was renamed to IsTargetControlConnected in API 1.1.1, the old typedef is kept here for
+// backwards compatibility with old code, it is castable either way since it's ABI compatible
+// as the same function pointer type.
+typedef pRENDERDOC_IsTargetControlConnected pRENDERDOC_IsRemoteAccessConnected;
+
+// This function will launch the Replay UI associated with the RenderDoc library injected
+// into the running application.
+//
+// if connectTargetControl is 1, the Replay UI will be launched with a command line parameter
+// to connect to this application
+// cmdline is the rest of the command line, as a UTF-8 string. E.g. a captures to open
+// if cmdline is NULL, the command line will be empty.
+//
+// returns the PID of the replay UI if successful, 0 if not successful.
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_LaunchReplayUI)(uint32_t connectTargetControl,
+                                                          const char *cmdline);
+
+// RenderDoc can return a higher version than requested if it's backwards compatible,
+// this function returns the actual version returned. If a parameter is NULL, it will be
+// ignored and the others will be filled out.
+typedef void(RENDERDOC_CC *pRENDERDOC_GetAPIVersion)(int *major, int *minor, int *patch);
+
+// Requests that the replay UI show itself (if hidden or not the current top window). This can be
+// used in conjunction with IsTargetControlConnected and LaunchReplayUI to intelligently handle
+// showing the UI after making a capture.
+//
+// This will return 1 if the request was successfully passed on, though it's not guaranteed that
+// the UI will be on top in all cases depending on OS rules. It will return 0 if there is no current
+// target control connection to make such a request, or if there was another error
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_ShowReplayUI)(void);
+
+//////////////////////////////////////////////////////////////////////////
+// Capturing functions
+//
+
+// A device pointer is a pointer to the API's root handle.
+//
+// This would be an ID3D11Device, HGLRC/GLXContext, ID3D12Device, etc
+typedef void *RENDERDOC_DevicePointer;
+
+// A window handle is the OS's native window handle
+//
+// This would be an HWND, GLXDrawable, etc
+typedef void *RENDERDOC_WindowHandle;
+
+// A helper macro for Vulkan, where the device handle cannot be used directly.
+//
+// Passing the VkInstance to this macro will return the RENDERDOC_DevicePointer to use.
+//
+// Specifically, the value needed is the dispatch table pointer, which sits as the first
+// pointer-sized object in the memory pointed to by the VkInstance. Thus we cast to a void** and
+// indirect once.
+#define RENDERDOC_DEVICEPOINTER_FROM_VKINSTANCE(inst) (*((void **)(inst)))
+
+// This sets the RenderDoc in-app overlay in the API/window pair as 'active' and it will
+// respond to keypresses. Neither parameter can be NULL
+typedef void(RENDERDOC_CC *pRENDERDOC_SetActiveWindow)(RENDERDOC_DevicePointer device,
+                                                       RENDERDOC_WindowHandle wndHandle);
+
+// capture the next frame on whichever window and API is currently considered active
+typedef void(RENDERDOC_CC *pRENDERDOC_TriggerCapture)(void);
+
+// capture the next N frames on whichever window and API is currently considered active
+typedef void(RENDERDOC_CC *pRENDERDOC_TriggerMultiFrameCapture)(uint32_t numFrames);
+
+// When choosing either a device pointer or a window handle to capture, you can pass NULL.
+// Passing NULL specifies a 'wildcard' match against anything. This allows you to specify
+// any API rendering to a specific window, or a specific API instance rendering to any window,
+// or in the simplest case of one window and one API, you can just pass NULL for both.
+//
+// In either case, if there are two or more possible matching (device,window) pairs it
+// is undefined which one will be captured.
+//
+// Note: for headless rendering you can pass NULL for the window handle and either specify
+// a device pointer or leave it NULL as above.
+
+// Immediately starts capturing API calls on the specified device pointer and window handle.
+//
+// If there is no matching thing to capture (e.g. no supported API has been initialised),
+// this will do nothing.
+//
+// The results are undefined (including crashes) if two captures are started overlapping,
+// even on separate devices and/oror windows.
+typedef void(RENDERDOC_CC *pRENDERDOC_StartFrameCapture)(RENDERDOC_DevicePointer device,
+                                                         RENDERDOC_WindowHandle wndHandle);
+
+// Returns whether or not a frame capture is currently ongoing anywhere.
+//
+// This will return 1 if a capture is ongoing, and 0 if there is no capture running
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_IsFrameCapturing)(void);
+
+// Ends capturing immediately.
+//
+// This will return 1 if the capture succeeded, and 0 if there was an error capturing.
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_EndFrameCapture)(RENDERDOC_DevicePointer device,
+                                                           RENDERDOC_WindowHandle wndHandle);
+
+// Ends capturing immediately and discard any data stored without saving to disk.
+//
+// This will return 1 if the capture was discarded, and 0 if there was an error or no capture
+// was in progress
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_DiscardFrameCapture)(RENDERDOC_DevicePointer device,
+                                                               RENDERDOC_WindowHandle wndHandle);
+
+// Only valid to be called between a call to StartFrameCapture and EndFrameCapture. Gives a custom
+// title to the capture produced which will be displayed in the UI.
+//
+// If multiple captures are ongoing, this title will be applied to the first capture to end after
+// this call. The second capture to end will have no title, unless this function is called again.
+//
+// Calling this function has no effect if no capture is currently running, and if it is called
+// multiple times only the last title will be used.
+typedef void(RENDERDOC_CC *pRENDERDOC_SetCaptureTitle)(const char *title);
+
+// Annotations API:
+//
+// These functions allow you to specify annotations either on a per-command level, or a per-object
+// level.
+//
+// Basic types of annotations are supported, as well as vector versions and references to API objects.
+//
+// The annotations are stored as keys, with the key being a dot-separated path allowing arbitrary
+// nesting and user organisation. The keys are sorted in human order so `foo.2.bar` will be displayed
+// before `foo.10.bar` to allow creation of arrays if desired.
+//
+// Deleting an annotation can be done by assigning an empty value to it.
+
+// the type of an annotation value, or Empty to delete an annotation
+typedef enum RENDERDOC_AnnotationType
+{
+  eRENDERDOC_Empty,
+  eRENDERDOC_Bool,
+  eRENDERDOC_Int32,
+  eRENDERDOC_UInt32,
+  eRENDERDOC_Int64,
+  eRENDERDOC_UInt64,
+  eRENDERDOC_Float,
+  eRENDERDOC_Double,
+  eRENDERDOC_String,
+  eRENDERDOC_APIObject,
+  eRENDERDOC_AnnotationMax = 0x7FFFFFFF,
+} RENDERDOC_AnnotationType;
+
+// a union with vector annotation value data
+typedef union RENDERDOC_AnnotationVectorValue
+{
+  bool boolean[4];
+  int32_t int32[4];
+  int64_t int64[4];
+  uint32_t uint32[4];
+  uint64_t uint64[4];
+  float float32[4];
+  double float64[4];
+} RENDERDOC_AnnotationVectorValue;
+
+// a union with scalar annotation value data
+typedef union RENDERDOC_AnnotationValue
+{
+  bool boolean;
+  int32_t int32;
+  int64_t int64;
+  uint32_t uint32;
+  uint64_t uint64;
+  float float32;
+  double float64;
+
+  RENDERDOC_AnnotationVectorValue vector;
+
+  const char *string;
+  void *apiObject;
+} RENDERDOC_AnnotationValue;
+
+// a struct for specifying a GL object, as we don't have pointers we can use so instead we specify a
+// pointer to this struct giving both the type and the name
+typedef struct RENDERDOC_GLResourceReference
+{
+  // this is the same GLenum identifier as passed to glObjectLabel
+  uint32_t identifier;
+  uint32_t name;
+} GLResourceReference;
+
+// simple C++ helpers to avoid the need for a temporary objects for value passing and GL object specification
+#ifdef __cplusplus
+struct RDGLObjectHelper
+{
+  RENDERDOC_GLResourceReference gl;
+
+  RDGLObjectHelper(uint32_t identifier, uint32_t name)
+  {
+    gl.identifier = identifier;
+    gl.name = name;
+  }
+
+  operator RENDERDOC_GLResourceReference *() { return &gl; }
+};
+
+struct RDAnnotationHelper
+{
+  RENDERDOC_AnnotationValue val;
+
+  RDAnnotationHelper(bool b) { val.boolean = b; }
+  RDAnnotationHelper(int32_t i) { val.int32 = i; }
+  RDAnnotationHelper(int64_t i) { val.int64 = i; }
+  RDAnnotationHelper(uint32_t i) { val.uint32 = i; }
+  RDAnnotationHelper(uint64_t i) { val.uint64 = i; }
+  RDAnnotationHelper(float f) { val.float32 = f; }
+  RDAnnotationHelper(double d) { val.float64 = d; }
+  RDAnnotationHelper(const char *s) { val.string = s; }
+
+  operator RENDERDOC_AnnotationValue *() { return &val; }
+};
+#endif
+
+// The device is specified in the same way as other API calls that take a RENDERDOC_DevicePointer
+// to specify the device.
+//
+// The object or queue/commandbuffer will depend on the graphics API in question.
+//
+// Return value:
+// 0 - The annotation was applied successfully.
+// 1 - The device is unknown/invalid
+// 2 - The device is valid but the annotation is not supported for API-specific reasons, such as an
+//     unrecognised or invalid object or queue/commandbuffer
+// 3 - The call is ill-formed or invalid e.g. empty is specified with a value pointer, or non-empty
+//     is specified with a NULL value pointer
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_SetObjectAnnotation)(RENDERDOC_DevicePointer device,
+                                                               void *object, const char *key,
+                                                               RENDERDOC_AnnotationType valueType,
+                                                               uint32_t valueVectorWidth,
+                                                               const RENDERDOC_AnnotationValue *value);
+
+typedef uint32_t(RENDERDOC_CC *pRENDERDOC_SetCommandAnnotation)(
+    RENDERDOC_DevicePointer device, void *queueOrCommandBuffer, const char *key,
+    RENDERDOC_AnnotationType valueType, uint32_t valueVectorWidth,
+    const RENDERDOC_AnnotationValue *value);
+
+//////////////////////////////////////////////////////////////////////////////////////////////////
+// RenderDoc API versions
+//
+
+// RenderDoc uses semantic versioning (http://semver.org/).
+//
+// MAJOR version is incremented when incompatible API changes happen.
+// MINOR version is incremented when functionality is added in a backwards-compatible manner.
+// PATCH version is incremented when backwards-compatible bug fixes happen.
+//
+// Note that this means the API returned can be higher than the one you might have requested.
+// e.g. if you are running against a newer RenderDoc that supports 1.0.1, it will be returned
+// instead of 1.0.0. You can check this with the GetAPIVersion entry point
+typedef enum RENDERDOC_Version
+{
+  eRENDERDOC_API_Version_1_0_0 = 10000,    // RENDERDOC_API_1_0_0 = 1 00 00
+  eRENDERDOC_API_Version_1_0_1 = 10001,    // RENDERDOC_API_1_0_1 = 1 00 01
+  eRENDERDOC_API_Version_1_0_2 = 10002,    // RENDERDOC_API_1_0_2 = 1 00 02
+  eRENDERDOC_API_Version_1_1_0 = 10100,    // RENDERDOC_API_1_1_0 = 1 01 00
+  eRENDERDOC_API_Version_1_1_1 = 10101,    // RENDERDOC_API_1_1_1 = 1 01 01
+  eRENDERDOC_API_Version_1_1_2 = 10102,    // RENDERDOC_API_1_1_2 = 1 01 02
+  eRENDERDOC_API_Version_1_2_0 = 10200,    // RENDERDOC_API_1_2_0 = 1 02 00
+  eRENDERDOC_API_Version_1_3_0 = 10300,    // RENDERDOC_API_1_3_0 = 1 03 00
+  eRENDERDOC_API_Version_1_4_0 = 10400,    // RENDERDOC_API_1_4_0 = 1 04 00
+  eRENDERDOC_API_Version_1_4_1 = 10401,    // RENDERDOC_API_1_4_1 = 1 04 01
+  eRENDERDOC_API_Version_1_4_2 = 10402,    // RENDERDOC_API_1_4_2 = 1 04 02
+  eRENDERDOC_API_Version_1_5_0 = 10500,    // RENDERDOC_API_1_5_0 = 1 05 00
+  eRENDERDOC_API_Version_1_6_0 = 10600,    // RENDERDOC_API_1_6_0 = 1 06 00
+  eRENDERDOC_API_Version_1_7_0 = 10700,    // RENDERDOC_API_1_7_0 = 1 07 00
+} RENDERDOC_Version;
+
+// API version changelog:
+//
+// 1.0.0 - initial release
+// 1.0.1 - Bugfix: IsFrameCapturing() was returning false for captures that were triggered
+//         by keypress or TriggerCapture, instead of Start/EndFrameCapture.
+// 1.0.2 - Refactor: Renamed eRENDERDOC_Option_DebugDeviceMode to eRENDERDOC_Option_APIValidation
+// 1.1.0 - Add feature: TriggerMultiFrameCapture(). Backwards compatible with 1.0.x since the new
+//         function pointer is added to the end of the struct, the original layout is identical
+// 1.1.1 - Refactor: Renamed remote access to target control (to better disambiguate from remote
+//         replay/remote server concept in replay UI)
+// 1.1.2 - Refactor: Renamed "log file" in function names to just capture, to clarify that these
+//         are captures and not debug logging files. This is the first API version in the v1.0
+//         branch.
+// 1.2.0 - Added feature: SetCaptureFileComments() to add comments to a capture file that will be
+//         displayed in the UI program on load.
+// 1.3.0 - Added feature: New capture option eRENDERDOC_Option_AllowUnsupportedVendorExtensions
+//         which allows users to opt-in to allowing unsupported vendor extensions to function.
+//         Should be used at the user's own risk.
+//         Refactor: Renamed eRENDERDOC_Option_VerifyMapWrites to
+//         eRENDERDOC_Option_VerifyBufferAccess, which now also controls initialisation to
+//         0xdddddddd of uninitialised buffer contents.
+// 1.4.0 - Added feature: DiscardFrameCapture() to discard a frame capture in progress and stop
+//         capturing without saving anything to disk.
+// 1.4.1 - Refactor: Renamed Shutdown to RemoveHooks to better clarify what is happening
+// 1.4.2 - Refactor: Renamed 'draws' to 'actions' in callstack capture option.
+// 1.5.0 - Added feature: ShowReplayUI() to request that the replay UI show itself if connected
+// 1.6.0 - Added feature: SetCaptureTitle() which can be used to set a title for a
+//         capture made with StartFrameCapture() or EndFrameCapture()
+// 1.7.0 - Added feature: SetObjectAnnotation() / SetCommandAnnotation() for adding rich
+//         annotations to objects and command streams
+
+typedef struct RENDERDOC_API_1_7_0
+{
+  pRENDERDOC_GetAPIVersion GetAPIVersion;
+
+  pRENDERDOC_SetCaptureOptionU32 SetCaptureOptionU32;
+  pRENDERDOC_SetCaptureOptionF32 SetCaptureOptionF32;
+
+  pRENDERDOC_GetCaptureOptionU32 GetCaptureOptionU32;
+  pRENDERDOC_GetCaptureOptionF32 GetCaptureOptionF32;
+
+  pRENDERDOC_SetFocusToggleKeys SetFocusToggleKeys;
+  pRENDERDOC_SetCaptureKeys SetCaptureKeys;
+
+  pRENDERDOC_GetOverlayBits GetOverlayBits;
+  pRENDERDOC_MaskOverlayBits MaskOverlayBits;
+
+  // Shutdown was renamed to RemoveHooks in 1.4.1.
+  // These unions allow old code to continue compiling without changes
+  union
+  {
+    pRENDERDOC_Shutdown Shutdown;
+    pRENDERDOC_RemoveHooks RemoveHooks;
+  };
+  pRENDERDOC_UnloadCrashHandler UnloadCrashHandler;
+
+  // Get/SetLogFilePathTemplate was renamed to Get/SetCaptureFilePathTemplate in 1.1.2.
+  // These unions allow old code to continue compiling without changes
+  union
+  {
+    // deprecated name
+    pRENDERDOC_SetLogFilePathTemplate SetLogFilePathTemplate;
+    // current name
+    pRENDERDOC_SetCaptureFilePathTemplate SetCaptureFilePathTemplate;
+  };
+  union
+  {
+    // deprecated name
+    pRENDERDOC_GetLogFilePathTemplate GetLogFilePathTemplate;
+    // current name
+    pRENDERDOC_GetCaptureFilePathTemplate GetCaptureFilePathTemplate;
+  };
+
+  pRENDERDOC_GetNumCaptures GetNumCaptures;
+  pRENDERDOC_GetCapture GetCapture;
+
+  pRENDERDOC_TriggerCapture TriggerCapture;
+
+  // IsRemoteAccessConnected was renamed to IsTargetControlConnected in 1.1.1.
+  // This union allows old code to continue compiling without changes
+  union
+  {
+    // deprecated name
+    pRENDERDOC_IsRemoteAccessConnected IsRemoteAccessConnected;
+    // current name
+    pRENDERDOC_IsTargetControlConnected IsTargetControlConnected;
+  };
+  pRENDERDOC_LaunchReplayUI LaunchReplayUI;
+
+  pRENDERDOC_SetActiveWindow SetActiveWindow;
+
+  pRENDERDOC_StartFrameCapture StartFrameCapture;
+  pRENDERDOC_IsFrameCapturing IsFrameCapturing;
+  pRENDERDOC_EndFrameCapture EndFrameCapture;
+
+  // new function in 1.1.0
+  pRENDERDOC_TriggerMultiFrameCapture TriggerMultiFrameCapture;
+
+  // new function in 1.2.0
+  pRENDERDOC_SetCaptureFileComments SetCaptureFileComments;
+
+  // new function in 1.4.0
+  pRENDERDOC_DiscardFrameCapture DiscardFrameCapture;
+
+  // new function in 1.5.0
+  pRENDERDOC_ShowReplayUI ShowReplayUI;
+
+  // new function in 1.6.0
+  pRENDERDOC_SetCaptureTitle SetCaptureTitle;
+
+  // new functions in 1.7.0
+  pRENDERDOC_SetObjectAnnotation SetObjectAnnotation;
+  pRENDERDOC_SetCommandAnnotation SetCommandAnnotation;
+} RENDERDOC_API_1_7_0;
+
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_0_0;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_0_1;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_0_2;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_1_0;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_1_1;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_1_2;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_2_0;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_3_0;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_4_0;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_4_1;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_4_2;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_5_0;
+typedef RENDERDOC_API_1_7_0 RENDERDOC_API_1_6_0;
+
+//////////////////////////////////////////////////////////////////////////////////////////////////
+// RenderDoc API entry point
+//
+// This entry point can be obtained via GetProcAddress/dlsym if RenderDoc is available.
+//
+// The name is the same as the typedef - "RENDERDOC_GetAPI"
+//
+// This function is not thread safe, and should not be called on multiple threads at once.
+// Ideally, call this once as early as possible in your application's startup, before doing
+// any API work, since some configuration functionality etc has to be done also before
+// initialising any APIs.
+//
+// Parameters:
+//   version is a single value from the RENDERDOC_Version above.
+//
+//   outAPIPointers will be filled out with a pointer to the corresponding struct of function
+//   pointers.
+//
+// Returns:
+//   1 - if the outAPIPointers has been filled with a pointer to the API struct requested
+//   0 - if the requested version is not supported or the arguments are invalid.
+//
+typedef int(RENDERDOC_CC *pRENDERDOC_GetAPI)(RENDERDOC_Version version, void **outAPIPointers);
+
+#ifdef __cplusplus
+}    // extern "C"
+#endif

CMakeLists.txt

@@ -4,6 +4,12 @@ if(CMAKE_VERSION VERSION_GREATER_EQUAL 4.0)
 	set(CMAKE_POLICY_VERSION_MINIMUM 3.5 CACHE STRING "Minimum policy version for bundled dependencies")
 endif()
 
+get_property(KYTY_GENERATOR_IS_MULTI_CONFIG GLOBAL PROPERTY GENERATOR_IS_MULTI_CONFIG)
+if(NOT KYTY_GENERATOR_IS_MULTI_CONFIG AND NOT CMAKE_BUILD_TYPE)
+	set(CMAKE_BUILD_TYPE Release CACHE STRING "Choose the type of build" FORCE)
+	set_property(CACHE CMAKE_BUILD_TYPE PROPERTY STRINGS Debug Release RelWithDebInfo MinSizeRel)
+endif()
+
 project(Kyty)
 
 set(KYTY_SOURCE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/src")
@@ -197,14 +203,21 @@ file(GLOB kyty_emulator_src CONFIGURE_DEPENDS
 	"${KYTY_SOURCE_DIR}/graphics/shader/*.h"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/*.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/*.h"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/cfg/*.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/cfg/*.h"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/decompiler/*.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/decompiler/*.h"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/emitter/*.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/emitter/*.h"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/frontend/cfg/*.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/frontend/cfg/*.h"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/frontend/decode/*.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/frontend/decode/*.h"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/frontend/translate/*.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/frontend/translate/*.h"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/backend/spirv/*.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/backend/spirv/*.h"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/*.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/*.h"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/opcodes/*.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/opcodes/*.h"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/opcodes/*.inc"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/*.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/*.h"
 	"${KYTY_SOURCE_DIR}/graphics/presentation/*.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/presentation/*.h"
 	"${KYTY_SOURCE_DIR}/graphics/presentation/window/*.cpp"
@@ -241,7 +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 "${KYTY_SOURCE_DIR}/embed_spirv.cmake"
@@ -250,6 +263,22 @@ foreach(shader_name IN LISTS gpu_tiler_shader_names)
 	)
 	list(APPEND gpu_tiler_shader_headers "${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 "${PROJECT_BINARY_DIR}/gpu_blit_shaders")
@@ -265,7 +294,7 @@ foreach(shader_source IN LISTS gpu_blit_shader_sources)
 	add_custom_command(
 		OUTPUT "${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 "${KYTY_SOURCE_DIR}/embed_spirv.cmake"
@@ -286,7 +315,7 @@ if (KYTY_CLANG_CL)
 	set_source_files_properties(${kyty_emulator_src} PROPERTIES COMPILE_FLAGS "-Wno-pragma-pack -Wno-deprecated-declarations -D_TIMESPEC_DEFINED")
 endif()
 
-set(kyty_emulator_link_libraries common Vulkan::Headers spirv-tools-opt spirv-tools SDL2-static xxhash FFmpeg::ffmpeg fmt::fmt nlohmann_json::nlohmann_json LibAtrac9 kyty_git_version)
+set(kyty_emulator_link_libraries common Vulkan::Headers spirv-tools-opt spirv-tools SDL2-static imgui xxhash FFmpeg::ffmpeg fmt::fmt nlohmann_json::nlohmann_json LibAtrac9 kyty_git_version Zydis::Zydis xbyak::xbyak)
 
 # Linux system libraries required by the static FFmpeg archive.
 if(LINUX)
@@ -316,6 +345,7 @@ set(inc_headers
 	${KYTY_THIRD_PARTY_DIR}/fmt/include
 	${KYTY_THIRD_PARTY_DIR}/xxhash
 	${KYTY_THIRD_PARTY_DIR}/cpuinfo/include
+	${KYTY_THIRD_PARTY_DIR}/renderdoc
 	${KYTY_THIRD_PARTY_DIR}/stb
 )
 
@@ -361,9 +391,8 @@ add_kyty_full_emulator_test(shader_cfg_tests "${KYTY_TESTS_DIR}/shaderCfgTests.c
 add_executable(scalar_provenance_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/ScalarProvenanceTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/hostMemory.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ReadLaneElimination.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ScalarProvenance.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/SrtWalker.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ReadLaneElimination.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/SrtWalker.cpp"
 )
 target_link_libraries(scalar_provenance_tests fmt::fmt)
 target_include_directories(scalar_provenance_tests PRIVATE ${inc_headers})
@@ -372,6 +401,7 @@ add_executable(page_manager_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/PageManagerTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/pageManager.cpp"
 )
+target_link_libraries(page_manager_tests common)
 target_include_directories(page_manager_tests PRIVATE ${inc_headers})
 
 add_executable(bit_array_tests EXCLUDE_FROM_ALL
@@ -379,6 +409,13 @@ add_executable(bit_array_tests EXCLUDE_FROM_ALL
 )
 target_include_directories(bit_array_tests PRIVATE ${inc_headers})
 
+add_executable(ime_dialog_tests EXCLUDE_FROM_ALL
+	"${KYTY_TESTS_DIR}/ImeDialogTests.cpp"
+	"${KYTY_SOURCE_DIR}/libs/imeCommon.cpp"
+	"${KYTY_SOURCE_DIR}/libs/imeDialog.cpp"
+)
+target_include_directories(ime_dialog_tests PRIVATE ${inc_headers})
+
 add_executable(memory_tracker_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/MemoryTrackerTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/pageManager.cpp"
@@ -399,9 +436,8 @@ add_executable(shader_stage_runtime_tests EXCLUDE_FROM_ALL
 	"${KYTY_SOURCE_DIR}/graphics/guest_gpu/gpu_format.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/hostMemory.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/shader/shaderStageRuntime.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ResourceMaterialization.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ScalarProvenance.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/SrtWalker.cpp"
+	"${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 ${inc_headers})
@@ -410,24 +446,16 @@ add_executable(resource_tracking_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/ResourceTrackingTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/guest_gpu/gpu_format.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/hostMemory.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ScalarProvenance.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/SrtWalker.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/SrtPatcher.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ResourceTracking.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ResourceMaterialization.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/ShaderInfoCollection.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/BindingLayout.cpp"
+	"${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"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ResourceMaterialization.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ShaderInfoCollection.cpp"
+	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/BindingLayout.cpp"
 )
 target_link_libraries(resource_tracking_tests fmt::fmt)
 target_include_directories(resource_tracking_tests PRIVATE ${inc_headers})
 
-add_executable(resource_mutex_tests EXCLUDE_FROM_ALL
-	"${KYTY_TESTS_DIR}/ResourceMutexTests.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/host_gpu/renderer/cache/resourceMutex.cpp"
-)
-target_link_libraries(resource_mutex_tests common)
-target_include_directories(resource_mutex_tests PRIVATE ${inc_headers})
-
 add_executable(audio_out2_port_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/AudioOut2PortTests.cpp"
 	"${KYTY_SOURCE_DIR}/libs/libAudio2.cpp"
@@ -444,6 +472,16 @@ add_executable(event_queue_lifetime_tests EXCLUDE_FROM_ALL
 target_link_libraries(event_queue_lifetime_tests common fmt::fmt)
 target_include_directories(event_queue_lifetime_tests PRIVATE ${inc_headers})
 
+add_kyty_full_emulator_test(kernel_file_system_tests
+	"${KYTY_TESTS_DIR}/KernelFileSystemTests.cpp")
+
+add_executable(sync_on_address_tests EXCLUDE_FROM_ALL
+	"${KYTY_TESTS_DIR}/SyncOnAddressTests.cpp"
+	"${KYTY_SOURCE_DIR}/kernel/syncOnAddress.cpp"
+)
+target_link_libraries(sync_on_address_tests common)
+target_include_directories(sync_on_address_tests PRIVATE ${inc_headers})
+
 add_executable(image_page_table_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/ImagePageTableTests.cpp"
 )
@@ -462,7 +500,7 @@ 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
@@ -486,6 +524,7 @@ if(NOT KYTY_CLANG_CL)
 endif()
 
 if(BUILD_TESTING)
+	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>)
 	add_test(NAME image_page_table COMMAND $<TARGET_FILE:image_page_table_tests>)
@@ -495,18 +534,25 @@ if(BUILD_TESTING)
 	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_tracking COMMAND $<TARGET_FILE:resource_tracking_tests>)
-	add_test(NAME resource_mutex COMMAND $<TARGET_FILE:resource_mutex_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>)
+	add_test(NAME sync_on_address COMMAND $<TARGET_FILE:sync_on_address_tests>)
 	add_test(NAME audio_out2_port COMMAND $<TARGET_FILE:audio_out2_port_tests>)
 	add_test(NAME shader_recompiler_compute COMMAND $<TARGET_FILE:shader_recompiler_compute_tests>)
+	add_test(NAME shader_recompiler_alignbyte
+		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --alignbyte-only)
 	add_test(NAME virtual_memory_allocation
 		COMMAND $<TARGET_FILE:virtual_memory_allocation_tests>)
+	add_test(NAME guest_red_zone_patcher
+		COMMAND $<TARGET_FILE:virtual_memory_allocation_tests> --red-zone-patcher-only)
 	add_test(NAME command_scheduler_timeline
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --scheduler-only)
 	add_test(NAME stream_buffer_ring
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --stream-buffer-only)
 	add_test(NAME gpu_command_lane
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --gpu-command-lane-only)
+	add_test(NAME pm4_context_state
+		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --context-state-only)
 	add_test(NAME gpu_tiler
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --gpu-tiler-only)
 
@@ -532,6 +578,7 @@ if(BUILD_TESTING)
 	endif()
 
 	add_custom_target(kyty_tests DEPENDS
+		ime_dialog_tests
 		shader_cfg_tests
 		scalar_provenance_tests
 		image_page_table_tests
@@ -541,10 +588,12 @@ if(BUILD_TESTING)
 		shader_vertex_metadata_tests
 		shader_stage_runtime_tests
 		resource_tracking_tests
-		resource_mutex_tests
 		event_queue_lifetime_tests
+		kernel_file_system_tests
+		sync_on_address_tests
 		shader_recompiler_compute_tests
 		virtual_memory_allocation_tests
+		audio_out2_port_tests
 	)
 endif()
 

README.md

@@ -31,6 +31,9 @@ macOS support is experimental. The emulator is built for x86-64 and runs on Appl
 Rosetta 2, with Vulkan provided by MoltenVK. A small number of titles have been verified in-game
 on Apple Silicon hardware; see [Building on macOS](#building-on-macos).
 
+Community game test results are available in the
+[KytyPS5 Compatibility List](https://kytyps5.github.io/).
+
 ## Bugs and Issues
 
 The project is in an early stage, so please be mindful when opening new issues. Expect crashes,
@@ -52,7 +55,7 @@ graphical glitches, low compatibility, and poor performance.
   <tr>
     <td align="center">
       <strong>Neptunia ReVerse</strong><br>
-      <img src="docs/screenshots/ps5-04.png" width="300" alt="Minecraft Legends running in KytyPS5">
+      <img src="docs/screenshots/ps5-04.png" width="300" alt="Neptunia ReVerse running in KytyPS5">
     </td>
     <td align="center">
       <strong>SILENT HILL: The Short Message</strong><br>
@@ -62,11 +65,11 @@ graphical glitches, low compatibility, and poor performance.
   <tr>
     <td align="center">
       <strong>Hellboy</strong><br>
-      <img src="docs/screenshots/ps5-02.png" width="300" alt="Disgaea 6 running in KytyPS5">
+      <img src="docs/screenshots/ps5-02.png" width="300" alt="Hellboy running in KytyPS5">
     </td>
     <td align="center">
       <strong>Paleo Pines</strong><br>
-      <img src="docs/screenshots/ps5-06.png" width="300" alt="Dreaming Sarah running in KytyPS5">
+      <img src="docs/screenshots/ps5-06.png" width="300" alt="Paleo Pines running in KytyPS5">
     </td>
   </tr>
 </table>

src/common/CMakeLists.txt

@@ -24,6 +24,7 @@ target_link_libraries(common
 		Tracy::TracyClient
 	PRIVATE
 		cpuinfo
+		kyty_git_version
 		spdlog::spdlog
 )
 

src/common/assert.cpp

@@ -3,6 +3,7 @@
 #include "common/debug.h"
 #include "common/logging/log.h"
 #include "common/subsystems.h"
+#include "kytyGitVersion.h"
 
 #include <cstdio>
 #include <cstdlib>
@@ -23,7 +24,7 @@ static std::string BuildFatalReport(const char* title, std::string_view text, co
 	DebugStack stack;
 	DebugStack::Trace(&stack);
 
-	std::string report = "--- Stack Trace ---\n";
+	std::string report = "--- Build ---\n" KYTY_BUILD_LABEL "\n--- Stack Trace ---\n";
 	for (int i = PRINT_STACK_FROM; i < stack.depth; i++) {
 		report += fmt::format("[{}] {:016x}\n", i - PRINT_STACK_FROM,
 		                      static_cast<uint64_t>(stack.GetAddr(i)));
@@ -34,7 +35,7 @@ static std::string BuildFatalReport(const char* title, std::string_view text, co
 
 static int DbgReport(const char* title, std::string_view text, const char* file, int line) {
 	Log::WriteFatal(BuildFatalReport(title, text, file, line));
-	SubsystemsListSingleton::Instance()->ShutdownAll();
+	Subsystems::EmergencyShutdownActive();
 	return 1;
 }
 

src/common/commonSubsystem.cpp

@@ -1,21 +0,0 @@
-#include "common/commonSubsystem.h"
-
-#include "common/byteBuffer.h" // IWYU pragma: associated
-#include "common/common.h"     // IWYU pragma: associated
-#include "common/debug.h"      // IWYU pragma: associated
-#include "common/hash.h"       // IWYU pragma: associated
-#include "common/magicEnum.h"  // IWYU pragma: associated
-#include "common/singleton.h"  // IWYU pragma: associated
-#include "common/virtualMemory.h"
-
-namespace Common {
-
-KYTY_SUBSYSTEM_INIT(Common) {
-	VirtualMemory::Init();
-}
-
-KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Common) {}
-
-KYTY_SUBSYSTEM_DESTROY(Common) {}
-
-} // namespace Common

src/common/commonSubsystem.h

@@ -1,12 +0,0 @@
-#ifndef KYTY_COMMON_COMMON_SUBSYSTEM_H_
-#define KYTY_COMMON_COMMON_SUBSYSTEM_H_
-
-#include "common/subsystems.h"
-
-namespace Common {
-
-KYTY_SUBSYSTEM_DEFINE(Common);
-
-} // namespace Common
-
-#endif /* KYTY_COMMON_COMMON_SUBSYSTEM_H_ */

src/common/emulatorConfig.cpp

@@ -9,15 +9,15 @@ namespace Config {
 
 static std::unique_ptr<ConfigOptions> g_config;
 
-KYTY_SUBSYSTEM_INIT(Config) {
+void Initialize() {
 	EXIT_IF(g_config != nullptr);
 
 	g_config = std::make_unique<ConfigOptions>();
 }
 
-KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Config) {}
-
-KYTY_SUBSYSTEM_DESTROY(Config) {}
+void Shutdown() {
+	g_config.reset();
+}
 
 void Load(const ConfigOptions& cfg) {
 	EXIT_IF(g_config == nullptr);
@@ -33,6 +33,10 @@ uint32_t GetScreenHeight() {
 	return g_config->screen_height;
 }
 
+bool FullscreenEnabled() {
+	return g_config->fullscreen_enabled;
+}
+
 uint32_t GetVblankFrequency() {
 	return std::clamp(g_config->vblank_frequency, 30u, 360u);
 }
@@ -89,6 +93,10 @@ 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;
 }
@@ -97,4 +105,18 @@ bool ReadbackLinearImagesEnabled() {
 	return g_config->readback_linear_images;
 }
 
+bool PlayGoHackEnabled() {
+	return g_config->playgo_hack_enabled;
+}
+
+#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
+bool RedZoneProtectionEnabled() {
+	return g_config->red_zone_protection_enabled;
+}
+#endif
+
+const Keymap& GetKeymap() {
+	return g_config->keymap;
+}
+
 } // namespace Config

src/common/emulatorConfig.h

@@ -2,13 +2,21 @@
 #define KYTY_COMMON_EMULATOR_CONFIG_H_
 
 #include "common/common.h"
-#include "common/subsystems.h"
 
 #include <filesystem>
+#include <string>
+#include <vector>
 
 namespace Config {
 
-KYTY_SUBSYSTEM_DEFINE(Config);
+void Initialize();
+void Shutdown();
+
+struct Lifecycle {
+	static constexpr const char* name       = "Config";
+	static constexpr auto        initialize = Config::Initialize;
+	static constexpr auto        shutdown   = Config::Shutdown;
+};
 
 enum class ShaderOptimizationType { None, Size, Performance };
 
@@ -18,12 +26,15 @@ enum class ProfilerDirection { None, Network };
 
 enum class OutputDirection { Silent, Console, File };
 
+using Keymap = std::vector<std::string>;
+
 constexpr uint32_t DEFAULT_CONSOLE_LANGUAGE = 1;
 constexpr uint32_t MAX_CONSOLE_LANGUAGE     = 29;
 
 struct ConfigOptions {
 	uint32_t               screen_width                = 1280;
 	uint32_t               screen_height               = 720;
+	bool                   fullscreen_enabled          = false;
 	uint32_t               vblank_frequency            = 60;
 	uint32_t               console_language            = DEFAULT_CONSOLE_LANGUAGE;
 	bool                   vulkan_validation_enabled   = false;
@@ -38,14 +49,21 @@ struct ConfigOptions {
 	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;
+#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
+	bool red_zone_protection_enabled = false;
+#endif
+	Keymap keymap;
 };
 
 void Load(const ConfigOptions& cfg);
 
 uint32_t GetScreenWidth();
 uint32_t GetScreenHeight();
+bool     FullscreenEnabled();
 uint32_t GetVblankFrequency();
 uint32_t GetConsoleLanguage();
 bool     VulkanValidationEnabled();
@@ -67,8 +85,16 @@ ProfilerDirection GetProfilerDirection();
 
 bool SpirvDebugPrintfEnabled();
 
+bool GpuAssistedValidationEnabled();
+
 bool RenderDocEnabled();
 bool ReadbackLinearImagesEnabled();
+bool PlayGoHackEnabled();
+#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
+bool RedZoneProtectionEnabled();
+#endif
+
+const Keymap& GetKeymap();
 
 } // namespace Config
 

src/common/hostException.cpp

@@ -2,7 +2,6 @@
 
 #include <atomic>
 #include <cstdio>
-#include <cstdlib>
 
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
 #include <windows.h> // IWYU pragma: keep
@@ -28,54 +27,14 @@ namespace Common::HostException {
 
 static std::atomic<Handler> g_handler {nullptr};
 static std::atomic_uint32_t g_install_state {0};
-static thread_local bool    g_in_exception_filter = false;
 
 static_assert(decltype(g_handler)::is_always_lock_free);
 static_assert(decltype(g_install_state)::is_always_lock_free);
-
-[[noreturn]] static void FailFast(const char* reason) noexcept {
-	std::fputs("HostException fail-fast: ", stderr);
-	std::fputs(reason != nullptr ? reason : "unspecified", stderr);
-	std::fputc('\n', stderr);
-	std::fflush(stderr);
-#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
-	TerminateProcess(GetCurrentProcess(), static_cast<UINT>(EXCEPTION_NONCONTINUABLE_EXCEPTION));
-#endif
-	std::_Exit(321);
-}
-
-class FilterScope final {
-public:
-	FilterScope() noexcept {
-		if (g_in_exception_filter) {
-			FailFast("nested exception while resolving a host fault");
-		}
-		g_in_exception_filter = true;
-	}
-
-	~FilterScope() { g_in_exception_filter = false; }
-
-	KYTY_CLASS_NO_COPY(FilterScope);
-};
-
-static Handler LoadInstalledHandler() noexcept {
-	if (g_install_state.load(std::memory_order_acquire) == 0) {
-		FailFast("host exception handler is not installed");
-	}
-
-	const auto handler = g_handler.load(std::memory_order_acquire);
-	if (handler == nullptr) {
-		FailFast("host exception callback is null");
-	}
-	return handler;
-}
 #endif
 
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
 
-static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) {
-	FilterScope filter_scope;
-
+static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) noexcept {
 	auto* exception_record = exception->ExceptionRecord;
 
 	if (exception_record->ExceptionCode == DBG_PRINTEXCEPTION_C ||
@@ -105,12 +64,6 @@ static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) {
 	} else if (exception_record->ExceptionCode == EXCEPTION_ILLEGAL_INSTRUCTION) {
 		info.type = ExceptionType::IllegalInstruction;
 	} else {
-		printf("Unhandled win exception: code=0x%08" PRIx32 ", addr=0x%016" PRIx64
-		       ", rip=0x%016" PRIx64 ", rsp=0x%016" PRIx64 ", rbp=0x%016" PRIx64 "\n",
-		       static_cast<uint32_t>(exception_record->ExceptionCode),
-		       reinterpret_cast<uint64_t>(exception_record->ExceptionAddress),
-		       exception->ContextRecord->Rip, exception->ContextRecord->Rsp,
-		       exception->ContextRecord->Rbp);
 		return EXCEPTION_CONTINUE_SEARCH;
 	}
 
@@ -131,28 +84,21 @@ static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) {
 	info.r14 = exception->ContextRecord->R14;
 	info.r15 = exception->ContextRecord->R15;
 
-	const auto handler = LoadInstalledHandler();
-
-	return handler(info) ? EXCEPTION_CONTINUE_EXECUTION : EXCEPTION_CONTINUE_SEARCH;
+	const auto handler = g_handler.load(std::memory_order_acquire);
+	if (handler != nullptr && handler(info)) {
+		return EXCEPTION_CONTINUE_EXECUTION;
+	}
+	return EXCEPTION_CONTINUE_SEARCH;
 }
 
 #elif defined(__APPLE__)
 
 static std::atomic<Handler> g_handler {nullptr};
 static std::atomic_uint32_t g_install_state {0};
-static thread_local bool    g_in_exception_filter = false;
 
 static_assert(decltype(g_handler)::is_always_lock_free);
 static_assert(decltype(g_install_state)::is_always_lock_free);
 
-[[noreturn]] static void FailFast(const char* reason) noexcept {
-	std::fputs("HostException fail-fast: ", stderr);
-	std::fputs(reason != nullptr ? reason : "unspecified", stderr);
-	std::fputc('\n', stderr);
-	std::fflush(stderr);
-	std::_Exit(321);
-}
-
 // Translate the x86-64 page-fault error code (mcontext __es.__err) into an access type.
 // bit 1 (0x2) = write, bit 4 (0x10) = instruction fetch, otherwise a read.
 static AccessViolationType DecodeAccess(uint64_t err) {
@@ -170,11 +116,6 @@ static AccessViolationType DecodeAccess(uint64_t err) {
 // re-executing the faulting instruction against the now-fixed protection. An unresolved
 // fault restores the default disposition so the retry terminates the process.
 static void SignalHandler(int sig, siginfo_t* si, void* uctx) {
-	if (g_in_exception_filter) {
-		FailFast("nested exception while resolving a host fault");
-	}
-	g_in_exception_filter = true;
-
 	auto*       uc = static_cast<ucontext_t*>(uctx);
 	const auto* mc = uc->uc_mcontext;
 	const auto& ss = mc->__ss;
@@ -210,14 +151,7 @@ static void SignalHandler(int sig, siginfo_t* si, void* uctx) {
 	info.r15 = ss.__r15;
 
 	const auto handler = g_handler.load(std::memory_order_acquire);
-	if (handler == nullptr) {
-		FailFast("host exception callback is null");
-	}
-
-	const bool resolved   = handler(info);
-	g_in_exception_filter = false;
-
-	if (resolved) {
+	if (handler != nullptr && handler(info)) {
 		return; // retry the faulting instruction against the fixed mapping
 	}
 
@@ -244,8 +178,6 @@ static void ChainToDefault(int signal_number) noexcept {
 }
 
 static void SignalHandler(int signal_number, siginfo_t* signal_info, void* native_context) {
-	FilterScope filter_scope;
-
 	auto* context = static_cast<ucontext_t*>(native_context);
 	auto* gregs   = context->uc_mcontext.gregs;
 
@@ -289,9 +221,8 @@ static void SignalHandler(int signal_number, siginfo_t* signal_info, void* nativ
 	info.r14 = static_cast<uint64_t>(gregs[REG_R14]);
 	info.r15 = static_cast<uint64_t>(gregs[REG_R15]);
 
-	const auto handler = LoadInstalledHandler();
-
-	if (handler(info)) {
+	const auto handler = g_handler.load(std::memory_order_acquire);
+	if (handler != nullptr && handler(info)) {
 		return;
 	}
 
@@ -313,7 +244,7 @@ bool InstallHandler(Handler handler) {
 	g_handler.store(handler, std::memory_order_release);
 
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
-	if (AddVectoredExceptionHandler(1, ExceptionFilter) == nullptr) {
+	if (AddVectoredExceptionHandler(0, ExceptionFilter) == nullptr) {
 		g_handler.store(nullptr, std::memory_order_release);
 		g_install_state.store(0, std::memory_order_release);
 		printf("AddVectoredExceptionHandler() failed\n");

src/common/logging/log.cpp

@@ -149,7 +149,7 @@ void WriteFatal(fmt::text_style style, std::string_view text) {
 	Flush();
 }
 
-KYTY_SUBSYSTEM_INIT(Log) {
+void Initialize() {
 	g_initialized = true;
 	switch (Config::GetPrintfDirection()) {
 		case Config::OutputDirection::Silent: g_direction = Direction::Silent; break;
@@ -161,13 +161,7 @@ KYTY_SUBSYSTEM_INIT(Log) {
 	SetupLogger();
 }
 
-KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Log) {
-	Flush();
-	std::lock_guard lock(g_logger_mutex);
-	g_logger.reset();
-}
-
-KYTY_SUBSYSTEM_DESTROY(Log) {
+void Shutdown() {
 	Flush();
 	std::lock_guard lock(g_logger_mutex);
 	g_logger.reset();

src/common/logging/log.h

@@ -2,7 +2,6 @@
 #define KYTY_COMMON_LOGGING_LOG_H_
 
 #include "common/common.h"
-#include "common/subsystems.h"
 
 #include <fmt/color.h>
 #include <fmt/printf.h>
@@ -10,7 +9,15 @@
 
 namespace Log {
 
-KYTY_SUBSYSTEM_DEFINE(Log);
+void Initialize();
+void Shutdown();
+
+struct Lifecycle {
+	static constexpr const char* name               = "Log";
+	static constexpr auto        initialize         = Log::Initialize;
+	static constexpr auto        shutdown           = Log::Shutdown;
+	static constexpr auto        emergency_shutdown = Log::Shutdown;
+};
 
 enum class Direction { Silent, Console, File };
 

src/common/platform/sysWindowsDbg.cpp

@@ -15,56 +15,8 @@
 #include "common/assert.h"
 #include "common/platform/sysDbg.h"
 
-struct FrameS {
-	struct FrameS* next;
-	void*          ret_addr;
-};
-
-int WalkStack(int z_stack_depth, void** z_stack_trace) {
-	CONTEXT context;
-	//	KNONVOLATILE_CONTEXT_POINTERS NvContext;
-	PRUNTIME_FUNCTION runtime_function  = nullptr;
-	PVOID             handler_data      = nullptr;
-	ULONG64           establisher_frame = 0;
-	ULONG64           image_base        = 0;
-
-	RtlCaptureContext(&context);
-
-	int frame = 0;
-
-	while (true) {
-		if (frame >= z_stack_depth) {
-			break;
-		}
-
-		z_stack_trace[frame] = reinterpret_cast<void*>(context.Rip);
-
-		frame++;
-
-		runtime_function = RtlLookupFunctionEntry(context.Rip, &image_base, nullptr);
-
-		if (runtime_function == nullptr) {
-			break;
-		}
-
-		// RtlZeroMemory(&NvContext, sizeof(KNONVOLATILE_CONTEXT_POINTERS));
-		RtlVirtualUnwind(0, image_base, context.Rip, runtime_function, &context, &handler_data,
-		                 &establisher_frame, nullptr /*&NvContext*/);
-
-		if (context.Rip == 0u) {
-			break;
-		}
-	}
-
-	return frame;
-}
-
 void SysStackWalk(void** stack, int* depth) {
-	int n  = WalkStack(*depth, stack);
-	*depth = n;
-	//	unwind_info_t info = {stack, 0, *depth};
-	//	 _Unwind_Backtrace(&trace_fcn, &info);
-	//	 *depth = info.depth;
+	*depth = static_cast<int>(CaptureStackBackTrace(0, static_cast<DWORD>(*depth), stack, nullptr));
 }
 
 void SysStackUsagePrint(sys_dbg_stack_info_t& stack) {

src/common/profiler.cpp

@@ -1,7 +1,6 @@
 #include "common/profiler.h"
 
 #include "common/emulatorConfig.h"
-#include "common/subsystems.h"
 
 #include <algorithm>
 #include <common/TracyProtocol.hpp>
@@ -55,13 +54,7 @@ void SetThreadName(const char* name) {
 	}
 }
 
-void Close() {
-	if (tracy::ProfilerAvailable()) {
-		tracy::ShutdownProfiler();
-	}
-}
-
-KYTY_SUBSYSTEM_INIT(Profiler) {
+void Initialize() {
 	switch (Config::GetProfilerDirection()) {
 		case Config::ProfilerDirection::Network:
 			if (!tracy::ProfilerAvailable()) {
@@ -78,12 +71,10 @@ KYTY_SUBSYSTEM_INIT(Profiler) {
 	}
 }
 
-KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Profiler) {
-	Close();
-}
-
-KYTY_SUBSYSTEM_DESTROY(Profiler) {
-	Close();
+void Shutdown() {
+	if (tracy::ProfilerAvailable()) {
+		tracy::ShutdownProfiler();
+	}
 }
 
 } // namespace Profiler

src/common/profiler.h

@@ -2,7 +2,6 @@
 #define KYTY_COMMON_PROFILER_H_
 
 #include "common/common.h"
-#include "common/subsystems.h"
 
 #include <cstdint>
 #include <optional>
@@ -42,7 +41,15 @@ private:
 void EndBlock();
 void SetThreadName(const char* name);
 
-KYTY_SUBSYSTEM_DEFINE(Profiler);
+void Initialize();
+void Shutdown();
+
+struct Lifecycle {
+	static constexpr const char* name               = "Profiler";
+	static constexpr auto        initialize         = Profiler::Initialize;
+	static constexpr auto        shutdown           = Profiler::Shutdown;
+	static constexpr auto        emergency_shutdown = Profiler::Shutdown;
+};
 
 } // namespace Profiler
 

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/subsystems.cpp

@@ -1,232 +1,43 @@
 #include "common/subsystems.h"
 
-#include "common/assert.h"
-
-#include <cstdarg>
-#include <cstdio>
-#include <cstring>
-#include <memory>
-#include <string>
-
 namespace Common {
 
-class SubsystemPrivate {
-public:
-	SubsystemPrivate()          = default;
-	virtual ~SubsystemPrivate() = default;
+static Subsystems* g_active_subsystems = nullptr;
 
-	KYTY_CLASS_NO_COPY(SubsystemPrivate);
+Subsystems::Subsystems(bool print): m_print(print) {
+	m_active.reserve(16);
+	g_active_subsystems = this;
+}
 
-	bool        failed {false};
-	std::string fail_msg;
-};
-
-class SubsystemsListPrivate {
-public:
-	explicit SubsystemsListPrivate(SubsystemsList& p): parent(p) {}
-
-	virtual ~SubsystemsListPrivate() = default;
-
-	void SetArgs(int argc, char** argv) {
-		this->m_argc = argc;
-		this->m_argv = argv;
+Subsystems::~Subsystems() {
+	Destroy();
+	if (g_active_subsystems == this) {
+		g_active_subsystems = nullptr;
 	}
+}
 
-	void Add(Subsystem* s, std::initializer_list<Subsystem*> deps) {
-		EXIT_IF(!s);
-
-		const char* name = s->Id();
-
-		EXIT_IF(FindByName(name) != nullptr);
-
-		auto nl = std::make_unique<SubsListStruct>();
-
-		nl->s         = s;
-		nl->name      = name;
-		nl->deps      = nullptr;
-		nl->next      = std::move(list);
-		nl->prev_init = nullptr;
-
-		auto* node = nl.get();
-		list       = std::move(nl);
-
-		for (auto* dep: deps) {
-			const char* str = dep->Id();
-
-			auto l      = std::make_unique<DepsListStruct>();
-			l->dep_name = str;
-			l->next     = std::move(node->deps);
-
-			node->deps = std::move(l);
+void Subsystems::Destroy() {
+	for (auto it = m_active.rbegin(); it != m_active.rend(); ++it) {
+		if (it->shutdown != nullptr) {
+			it->shutdown();
 		}
-
-		node->initialized = false;
 	}
+	m_active.clear();
+}
 
-	bool InitAll(bool print_msg) {
-		while (SubsListStruct* n = FindNextToInitialize()) {
-			n->s->Init(&parent);
-
-			if (n->s->m_p->failed) {
-				fail_msg  = n->s->m_p->fail_msg.c_str();
-				fail_name = n->name;
-				return false;
-			}
-
-			if (print_msg) {
-				printf("Initialized: %s\n", n->name);
-			}
-
-			n->initialized = true;
-
-			SubsListStruct* last = last_init;
-			last_init            = n;
-			n->prev_init         = last;
+void Subsystems::EmergencyShutdown() {
+	for (auto it = m_active.rbegin(); it != m_active.rend(); ++it) {
+		if (it->emergency_shutdown != nullptr) {
+			it->emergency_shutdown();
 		}
-
-		return true;
 	}
+	m_active.clear();
+}
 
-	void DestroyAll(bool print_msg) {
-		for (SubsListStruct* n = last_init; n != nullptr; n = n->prev_init) {
-			n->s->Destroy(&parent);
-			n->initialized = false;
-
-			if (print_msg) {
-				printf("Destroyed: %s\n", n->name);
-			}
-		}
-
-		last_init = nullptr;
+void Subsystems::EmergencyShutdownActive() {
+	if (g_active_subsystems != nullptr) {
+		g_active_subsystems->EmergencyShutdown();
 	}
-
-	void ShutdownAll() {
-		for (SubsListStruct* n = last_init; n != nullptr; n = n->prev_init) {
-			n->s->UnexpectedShutdown(&parent);
-			n->initialized = false;
-		}
-
-		last_init = nullptr;
-	}
-
-	struct DepsListStruct {
-		const char*                     dep_name;
-		std::unique_ptr<DepsListStruct> next;
-	};
-
-	struct SubsListStruct {
-		Subsystem*                      s;
-		const char*                     name;
-		std::unique_ptr<DepsListStruct> deps;
-		std::unique_ptr<SubsListStruct> next;
-		SubsListStruct*                 prev_init;
-		bool                            initialized;
-	};
-
-	[[nodiscard]] SubsListStruct* FindByName(const char* name) const {
-		for (SubsListStruct* n = list.get(); n != nullptr; n = n->next.get()) {
-			if (std::strcmp(n->name, name) == 0) {
-				return n;
-			}
-		}
-
-		return nullptr;
-	}
-
-	[[nodiscard]] SubsListStruct* FindNextToInitialize() const {
-		for (SubsListStruct* n = list.get(); n != nullptr; n = n->next.get()) {
-			if (n->initialized) {
-				continue;
-			}
-
-			DepsListStruct* d = n->deps.get();
-			for (; d != nullptr; d = d->next.get()) {
-				SubsListStruct* s = FindByName(d->dep_name);
-				if ((s == nullptr) || !s->initialized) {
-					break;
-				}
-			}
-
-			if (d == nullptr) {
-				return n;
-			}
-		}
-
-		return nullptr;
-	}
-
-	KYTY_CLASS_NO_COPY(SubsystemsListPrivate);
-
-	std::unique_ptr<SubsListStruct> list;
-	SubsListStruct*                 last_init = nullptr;
-	int                             m_argc    = 0;
-	char**                          m_argv    = nullptr;
-	const char*                     fail_msg  = nullptr;
-	const char*                     fail_name = nullptr;
-	SubsystemsList&                 parent;
-};
-
-SubsystemsList::SubsystemsList(): m_p(std::make_unique<SubsystemsListPrivate>(*this)) {}
-
-SubsystemsList::~SubsystemsList() = default;
-
-void SubsystemsList::Add(Subsystem* s, std::initializer_list<Subsystem*> deps) {
-	m_p->Add(s, deps);
-}
-
-bool SubsystemsList::InitAll(bool print_msg) {
-	return m_p->InitAll(print_msg);
-}
-
-void SubsystemsList::DestroyAll(bool print_msg) {
-	m_p->DestroyAll(print_msg);
-}
-
-int* SubsystemsList::GetArgc() {
-	return &m_p->m_argc;
-}
-
-char** SubsystemsList::GetArgv() {
-	return m_p->m_argv;
-}
-
-Subsystem::Subsystem(): m_p(std::make_unique<SubsystemPrivate>()) {}
-
-Subsystem::~Subsystem() = default;
-
-void Subsystem::Fail(const char* format, ...) {
-	va_list args {};
-	va_start(args, format);
-
-	va_list args_copy {};
-	va_copy(args_copy, args);
-	int len = std::vsnprintf(nullptr, 0, format, args_copy);
-	va_end(args_copy);
-
-	if (len > 0) {
-		std::string msg(static_cast<size_t>(len) + 1, '\0');
-		std::vsnprintf(msg.data(), msg.size(), format, args);
-		m_p->fail_msg = msg.c_str();
-		m_p->failed   = true;
-	}
-
-	va_end(args);
-}
-
-const char* SubsystemsList::GetFailName() const {
-	return m_p->fail_name;
-}
-
-const char* SubsystemsList::GetFailMsg() const {
-	return m_p->fail_msg;
-}
-
-void SubsystemsList::SetArgs(int argc, char* argv[]) {
-	m_p->SetArgs(argc, argv);
-}
-
-void SubsystemsList::ShutdownAll() {
-	m_p->ShutdownAll();
 }
 
 } // namespace Common

src/common/subsystems.h

@@ -2,90 +2,61 @@
 #define KYTY_COMMON_SUBSYSTEMS_H_
 
 #include "common/common.h"
-#include "common/singleton.h"
 
-#include <initializer_list>
-#include <memory>
+#include <cstdio>
+#include <vector>
 
 namespace Common {
 
-class Subsystem;
-class SubsystemsListPrivate;
-class SubsystemPrivate;
-
-class SubsystemsList {
+class Subsystems {
 public:
-	SubsystemsList();
-	virtual ~SubsystemsList();
-	void SetArgs(int argc, char* argv[]);
+	using Callback = void (*)();
 
-	// void Add(Subsystem* s, const char* name, ...);
-	void Add(Subsystem* s, std::initializer_list<Subsystem*> deps);
+	explicit Subsystems(bool print = false);
+	~Subsystems();
 
-	bool InitAll(bool print_msg = false);
-	void DestroyAll(bool print_msg = false);
+	template <typename Lifecycle>
+	void Initialize() {
+		Lifecycle::initialize();
+		m_active.push_back({ShutdownCallback<Lifecycle>(), EmergencyCallback<Lifecycle>()});
+		if (m_print) {
+			std::printf("Initialized: %s\n", Lifecycle::name);
+		}
+	}
 
-	int*   GetArgc();
-	char** GetArgv();
+	void Destroy();
+	void EmergencyShutdown();
 
-	[[nodiscard]] const char* GetFailName() const;
-	[[nodiscard]] const char* GetFailMsg() const;
+	static void EmergencyShutdownActive();
 
-	void ShutdownAll();
-
-	static SubsystemsList* Instance() { return Common::Singleton<SubsystemsList>::Instance(); }
-
-	KYTY_CLASS_NO_COPY(SubsystemsList);
+	KYTY_CLASS_NO_COPY(Subsystems);
 
 private:
-	std::unique_ptr<SubsystemsListPrivate> m_p;
+	template <typename Lifecycle>
+	static consteval Callback ShutdownCallback() {
+		if constexpr (requires { Lifecycle::shutdown; }) {
+			return Lifecycle::shutdown;
+		}
+		return nullptr;
+	}
+
+	template <typename Lifecycle>
+	static consteval Callback EmergencyCallback() {
+		if constexpr (requires { Lifecycle::emergency_shutdown; }) {
+			return Lifecycle::emergency_shutdown;
+		}
+		return nullptr;
+	}
+
+	struct Entry {
+		Callback shutdown;
+		Callback emergency_shutdown;
+	};
+
+	std::vector<Entry> m_active;
+	bool               m_print;
 };
 
-using SubsystemsListSingleton = Common::Singleton<SubsystemsList>;
-
-class Subsystem {
-public:
-	Subsystem();
-	virtual ~Subsystem();
-
-	virtual const char* Id()                                       = 0;
-	virtual void        Init(SubsystemsList* parent)               = 0;
-	virtual void        Destroy(SubsystemsList* parent)            = 0;
-	virtual void        UnexpectedShutdown(SubsystemsList* parent) = 0;
-
-	friend class SubsystemsListPrivate;
-
-	KYTY_CLASS_NO_COPY(Subsystem);
-
-protected:
-	void Fail(const char* format, ...) KYTY_FORMAT_PRINTF(2, 3);
-
-private:
-	std::unique_ptr<SubsystemPrivate> m_p;
-};
-
-#define KYTY_SUBSYSTEM_DEFINE(s)                                                                   \
-	class s##Subsystem: public Common::Subsystem {                                                 \
-	public:                                                                                        \
-		static Subsystem* Instance() { return Common::Singleton<s##Subsystem>::Instance(); }       \
-		const char*       Id() { return #s; }                                                      \
-		void              Init(Common::SubsystemsList* parent);                                    \
-		void              Destroy(Common::SubsystemsList* parent);                                 \
-		void              UnexpectedShutdown(Common::SubsystemsList* parent);                      \
-	};                                                                                             \
-	typedef Common::Singleton<s##Subsystem> s##SubsystemSingleton;
-
-#define KYTY_SUBSYSTEM_INIT(s)                                                                     \
-	void s##Subsystem::Init([[maybe_unused]] Common::SubsystemsList* parent)
-#define KYTY_SUBSYSTEM_DESTROY(s)                                                                  \
-	void s##Subsystem::Destroy([[maybe_unused]] Common::SubsystemsList* parent)
-#define KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(s)                                                      \
-	void s##Subsystem::UnexpectedShutdown([[maybe_unused]] Common::SubsystemsList* parent)
-
-// #define KYTY_SUBSYSTEM_ADD(list, s, ...) list.push_back(s##SubsystemSingleton::Instance(), #s,
-// __VA_ARGS__); #define KYTY_SUBSYSTEM_ADD2(list, ...) list.Add2(s##SubsystemSingleton::Instance(),
-// __VA_ARGS__);
-
 } // namespace Common
 
 #endif /* KYTY_COMMON_SUBSYSTEMS_H_ */

src/common/threads.cpp

@@ -28,7 +28,8 @@
 #ifdef KYTY_WIN_CS
 #include <windows.h> // IWYU pragma: keep
 // IWYU pragma: no_include <winbase.h>
-constexpr DWORD KYTY_CS_SPIN_COUNT = 4000;
+constexpr DWORD    KYTY_CS_SPIN_COUNT          = 4000;
+constexpr uint64_t KYTY_SLEEP_SPIN_LIMIT_100NS = 500; // 50 us
 
 #ifndef CREATE_WAITABLE_TIMER_HIGH_RESOLUTION
 #define CREATE_WAITABLE_TIMER_HIGH_RESOLUTION 0x00000002
@@ -39,7 +40,10 @@ static void SleepHighResolution100ns(uint64_t units_100ns) {
 		return;
 	}
 
-	if (units_100ns <= 10000) {
+	// Keep spinning only where a kernel transition is
+	// likely to cost more than the requested delay; ordinary millisecond sleeps use the
+	// per-thread high-resolution waitable timer below.
+	if (units_100ns <= KYTY_SLEEP_SPIN_LIMIT_100NS) {
 		LARGE_INTEGER frequency {};
 		LARGE_INTEGER start {};
 		if (QueryPerformanceFrequency(&frequency) != 0 && QueryPerformanceCounter(&start) != 0 &&
@@ -248,15 +252,11 @@ static thread_id_t      g_main_thread;
 static int              g_main_thread_int;
 static std::atomic<int> g_thread_counter = 0;
 
-KYTY_SUBSYSTEM_INIT(Threads) {
+void InitializeThreads() {
 	g_main_thread     = std::this_thread::get_id();
 	g_main_thread_int = Thread::GetThreadIdUnique();
 }
 
-KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Threads) {}
-
-KYTY_SUBSYSTEM_DESTROY(Threads) {}
-
 Thread::Thread(thread_func_t func, void* arg)
     : m_thread(std::make_unique<ThreadPrivate>(func, arg)) {
 	while (!m_thread->started) {

src/common/threads.h

@@ -2,14 +2,13 @@
 #define KYTY_COMMON_THREADS_H_
 
 #include "common/common.h"
-#include "common/subsystems.h"
 
 #include <memory>
 #include <string>
 
 namespace Common {
 
-KYTY_SUBSYSTEM_DEFINE(Threads);
+void InitializeThreads();
 
 using thread_func_t    = void (*)(void*);
 using wait_poll_func_t = void (*)();

src/emulator.cpp

@@ -2,7 +2,6 @@
 
 #include "common/abi.h"
 #include "common/assert.h"
-#include "common/commonSubsystem.h"
 #include "common/emulatorConfig.h"
 #include "common/file.h"
 #include "common/logging/log.h"
@@ -16,6 +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"
@@ -27,13 +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() {
@@ -43,7 +61,7 @@ static void KytyClose() {
 
 	LOGF("done!\n");
 
-	Common::SubsystemsListSingleton::Instance()->ShutdownAll();
+	Common::Subsystems::EmergencyShutdownActive();
 }
 
 static void MountOrCreateDir(const std::filesystem::path& dir, const std::string& point) {
@@ -105,30 +123,13 @@ static void ClearDebugTextureFolder() {
 	}
 }
 
-static void Init(const Config::ConfigOptions& cfg, const std::filesystem::path& param_json) {
+static void Init(const Config::ConfigOptions& cfg, const std::filesystem::path& param_json,
+                 Common::Subsystems& subsystems) {
 	EXIT_IF(!Common::Thread::IsMainThread());
 
-	auto* slist = Common::SubsystemsList::Instance();
-
-	auto* audio       = Libs::Audio::AudioSubsystem::Instance();
-	auto* config      = Config::ConfigSubsystem::Instance();
-	auto* controller  = Libs::Controller::ControllerSubsystem::Instance();
-	auto* core        = Common::CommonSubsystem::Instance();
-	auto* file_system = Libs::LibKernel::FileSystem::FileSystemSubsystem::Instance();
-	auto* graphics    = Libs::Graphics::GraphicsSubsystem::Instance();
-	auto* log         = Log::LogSubsystem::Instance();
-	auto* memory      = Libs::LibKernel::Memory::MemorySubsystem::Instance();
-	auto* network     = Libs::Network::NetworkSubsystem::Instance();
-	auto* profiler    = Profiler::ProfilerSubsystem::Instance();
-	auto* pthread     = Libs::LibKernel::PthreadSubsystem::Instance();
-	auto* timer       = Loader::Timer::TimerSubsystem::Instance();
-
-	slist->Add(config, {core});
-	slist->InitAll(true);
-
+	subsystems.Initialize<Config::Lifecycle>();
 	Config::Load(cfg);
-	slist->Add(log, {core, config});
-	slist->InitAll(true);
+	subsystems.Initialize<Log::Lifecycle>();
 
 	if (Common::File::IsFileExisting(param_json)) {
 		Loader::SystemContentLoadParamSfo(param_json);
@@ -138,17 +139,16 @@ static void Init(const Config::ConfigOptions& cfg, const std::filesystem::path&
 		}
 	}
 
-	slist->Add(audio, {core, log, pthread, memory});
-	slist->Add(controller, {core, log, config});
-	slist->Add(file_system, {core, log, pthread});
-	slist->Add(graphics, {core, log, pthread, memory, config, profiler, controller});
-	slist->Add(memory, {core, log});
-	slist->Add(network, {core, log, pthread});
-	slist->Add(profiler, {core, config});
-	slist->Add(pthread, {core, log, timer});
-	slist->Add(timer, {core, log});
-
-	slist->InitAll(true);
+	// Initialization order is explicit; destruction is automatic and reversed.
+	subsystems.Initialize<Loader::Timer::Lifecycle>();
+	subsystems.Initialize<Libs::LibKernel::PthreadLifecycle>();
+	subsystems.Initialize<Profiler::Lifecycle>();
+	subsystems.Initialize<Libs::Network::Lifecycle>();
+	subsystems.Initialize<Libs::LibKernel::Memory::Lifecycle>();
+	subsystems.Initialize<Libs::LibKernel::FileSystem::Lifecycle>();
+	subsystems.Initialize<Libs::Controller::Lifecycle>();
+	subsystems.Initialize<Libs::Audio::Lifecycle>();
+	subsystems.Initialize<Libs::Graphics::Lifecycle>();
 }
 
 static void LoadElf(const std::filesystem::path& elf, bool dbg_print_reloc = false,
@@ -189,8 +189,9 @@ void Run(const RunOptions& options) {
 		EXIT("ELF is required\n");
 	}
 
-	const auto param_json = options.app0_dir / "sce_sys" / "param.json";
-	Init(options.config, param_json);
+	const auto         param_json = options.app0_dir / "sce_sys" / "param.json";
+	Common::Subsystems subsystems(true);
+	Init(options.config, param_json, subsystems);
 
 	ClearDebugTextureFolder();
 

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 {
@@ -15,6 +16,13 @@ bool TestWaitRegMemValue(uint64_t value, uint64_t ref, uint64_t mask, uint32_t f
 
 enum class Pm4ProcessResult { Complete, Blocked };
 
+enum class ContextStateOperation : uint32_t {
+	Clear     = 0,
+	Push      = 1,
+	Pop       = 2,
+	PushClear = 3,
+};
+
 class Pm4Execution {
 public:
 	[[nodiscard]] bool MadeProgress() const noexcept { return m_made_progress; }
@@ -23,10 +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;
@@ -43,30 +50,19 @@ 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);
 
 	void Reset();
+	void ApplyContextStateOperation(ContextStateOperation operation);
 
-	void BufferInit();
-	void BufferFlush();
-	void BufferFlushAndWait();
-	void BufferWait();
-	void BeginReadbackTransaction() {
-		if (m_readback_active) {
-			EXIT("nested command-processor readback transaction\n");
-		}
-		m_readback_active = true;
-	}
-	void EndReadbackTransaction() {
-		if (!m_readback_active) {
-			EXIT("command-processor readback transaction is not active\n");
-		}
-		m_readback_active = false;
-	}
-
+	void            BufferInit();
+	void            BufferFlush();
+	void            BufferFlushAndWait();
+	void            BufferWait();
 	HW::Context&    GetCtx() { return m_ctx; }
 	HW::UserConfig& GetUcfg() { return m_ucfg; }
 	HW::Shader&     GetShCtx() { return m_sh_ctx; }
@@ -108,7 +104,7 @@ 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; }
@@ -117,6 +113,7 @@ public:
 	void SetCeComplete(bool complete) { m_ce_complete = complete; }
 	void WaitCe();
 	void WaitDeDiff(uint32_t diff);
+	void WaitForRewind(bool valid);
 	void IncrementDe();
 	void IncrementCe();
 
@@ -135,13 +132,14 @@ 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>
@@ -155,13 +153,15 @@ private:
 	                          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;
+	HW::Context      m_saved_ctx;
+	bool             m_context_state_pushed = false;
 	HW::UserConfig   m_ucfg;
 	HW::Shader       m_sh_ctx;
 	HW::UserSgprType m_user_data_marker                 = HW::UserSgprType::Unknown;
@@ -173,16 +173,17 @@ private:
 	// Persistent draw state: indirect draws update it for subsequent draws.
 	uint32_t m_num_instances = 1;
 
-	uint32_t m_de_count        = 0;
-	uint32_t m_ce_count        = 0;
-	bool     m_ce_complete     = false;
-	bool     m_readback_active = false;
+	uint32_t m_de_count    = 0;
+	uint32_t m_ce_count    = 0;
+	bool     m_ce_complete = false;
 
 	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

@@ -35,7 +35,6 @@ 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;
@@ -82,7 +81,6 @@ constexpr auto MakeContextDispatchTable() {
 	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;
@@ -217,6 +215,7 @@ 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;
@@ -224,6 +223,7 @@ constexpr auto MakeOpcodeDispatchTable() {
 	g_cp_op_func[Pm4::IT_EVENT_WRITE_EOS]           = CpOpEventWriteEos;
 	g_cp_op_func[Pm4::IT_DMA_DATA]                  = CpOpDmaData;
 	g_cp_op_func[Pm4::IT_ACQUIRE_MEM]               = CpOpAcquireMem;
+	g_cp_op_func[Pm4::IT_REWIND]                    = CpOpRewind;
 	g_cp_op_func[Pm4::IT_SET_CONTEXT_REG]           = CpOpSetContextReg;
 	g_cp_op_func[Pm4::IT_SET_SH_REG]                = CpOpSetShaderReg;
 	g_cp_op_func[Pm4::IT_SET_UCONFIG_REG]           = CpOpSetUconfigReg;
@@ -238,6 +238,7 @@ 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;
 }
@@ -246,14 +247,13 @@ 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;
 
 	return g_cp_op_custom_func;
 }

src/graphics/guest_gpu/command_processor/pm4Dispatch.h

@@ -56,7 +56,6 @@ 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);
@@ -90,7 +89,6 @@ 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);
@@ -121,6 +119,7 @@ uint32_t HwUcSetGeIndexOffset(CommandProcessor&, uint32_t, uint32_t, const uint3
 uint32_t HwUcSetGdsOaRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 
 uint32_t CpOpNop(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
+uint32_t CpOpContextState(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpSetBase(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpClearState(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpIndexBase(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
@@ -132,6 +131,7 @@ uint32_t CpOpDrawIndirectMulti(CommandProcessor&, uint32_t, const uint32_t*, uin
 uint32_t CpOpDrawIndex(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpDrawIndexOffset(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpPfpSyncMe(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
+uint32_t CpOpRewind(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpIndexType(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpNumInstances(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);
 uint32_t CpOpDrawIndexAuto(CommandProcessor&, uint32_t, const uint32_t*, uint32_t, uint32_t);

src/graphics/guest_gpu/gpu_defs.h

@@ -3,15 +3,8 @@
 
 #include "common/common.h"
 
-#include <type_traits>
-
 namespace Libs::Graphics::Prospero {
 
-template <typename T>
-constexpr auto GpuEnumValue(T value) {
-	return static_cast<std::underlying_type_t<T>>(value);
-}
-
 enum class PrimitiveType : uint32_t {
 	kNone               = 0,
 	kPointList          = 1,
@@ -60,24 +53,34 @@ enum class GsOutputPrimitiveType : uint32_t {
 };
 
 enum class ChannelLayout : uint32_t {
-	kInvalid     = 0,
-	k8           = 1,
-	k16          = 2,
-	k8_8         = 3,
-	k32          = 4,
-	k16_16       = 5,
-	k11_11_10    = 6,
-	k10_10_10_2  = 9,
-	k8_8_8_8     = 10,
-	k32_32       = 11,
-	k16_16_16_16 = 12,
-	k32_32_32_32 = 14,
-	k5_6_5       = 16,
-	k5_5_5_1     = 17,
-	k4_4_4_4     = 19,
-	kBc1         = 35,
-	kBc2         = 36,
-	kBc3         = 37,
+	kInvalid         = 0,
+	k8               = 1,
+	k16              = 2,
+	k8_8             = 3,
+	k32              = 4,
+	k16_16           = 5,
+	k11_11_10        = 6,
+	k10_11_11        = 7,
+	k2_10_10_10      = 8,
+	k10_10_10_2      = 9,
+	k8_8_8_8         = 10,
+	k32_32           = 11,
+	k16_16_16_16     = 12,
+	k32_32_32        = 13,
+	k32_32_32_32     = 14,
+	k5_6_5           = 16,
+	k5_5_5_1         = 17,
+	k1_5_5_5         = 18,
+	k4_4_4_4         = 19,
+	k10_10_10_2Float = 31,
+	k9_9_9_5         = 34,
+	kBc1             = 35,
+	kBc2             = 36,
+	kBc3             = 37,
+	kBc4             = 38,
+	kBc5             = 39,
+	kBc6             = 40,
+	kBc7             = 41,
 };
 
 enum class ChannelType : uint32_t {
@@ -105,32 +108,52 @@ enum class ChannelOrder : uint32_t {
 	kAltReversed = 3,
 };
 
-// Maps a shader's logical RGBA export to the host color-attachment components. Two bits per
-// component keep this cheap to carry in shader and pipeline keys.
+// Selects the logical shader component written to each physical color-attachment component.
+// Two bits per component keep this cheap to carry in shader and pipeline keys.
 struct ColorComponentMapping {
-	uint8_t packed = 0xe4u; // RGBA: 0, 1, 2, 3
+	static constexpr uint8_t kIdentity = 0xe4u;
+
+	uint8_t packed = kIdentity;
 
 	[[nodiscard]] constexpr uint32_t Map(uint32_t component) const {
 		return component < 4u ? (packed >> (component * 2u)) & 0x3u : component;
 	}
 
-	[[nodiscard]] constexpr uint32_t ApplyMask(uint32_t mask) const {
+	[[nodiscard]] constexpr uint32_t ApplyMask(uint32_t logical_mask) const {
 		uint32_t mapped = 0;
-		for (uint32_t component = 0; component < 4u; component++) {
-			mapped |= ((mask >> component) & 1u) << Map(component);
+		for (uint32_t physical_component = 0; physical_component < 4u; physical_component++) {
+			mapped |= ((logical_mask >> Map(physical_component)) & 1u) << physical_component;
 		}
 		return mapped;
 	}
 
-	[[nodiscard]] constexpr bool IsIdentity() const { return packed == 0xe4u; }
+	// This mapping selects an intermediate component; next selects the final logical component.
+	[[nodiscard]] constexpr ColorComponentMapping Then(ColorComponentMapping next) const {
+		ColorComponentMapping result {0u};
+		for (uint32_t physical_component = 0; physical_component < 4u; physical_component++) {
+			result.packed |= static_cast<uint8_t>(next.Map(Map(physical_component))
+			                                      << (physical_component * 2u));
+		}
+		return result;
+	}
+
+	[[nodiscard]] constexpr bool IsIdentity() const { return packed == kIdentity; }
 
 	bool operator==(const ColorComponentMapping&) const = default;
 };
 
 static_assert(sizeof(ColorComponentMapping) == sizeof(uint8_t));
 
-inline constexpr ColorComponentMapping ColorMappingAbgr {0x1bu}; // 3, 2, 1, 0
+inline constexpr ColorComponentMapping ColorMappingRgba {ColorComponentMapping::kIdentity};
+inline constexpr ColorComponentMapping ColorMappingGr {0xe1u};   // 1, 0, 2, 3
+inline constexpr ColorComponentMapping ColorMappingRabg {0x6cu}; // 0, 3, 2, 1
+inline constexpr ColorComponentMapping ColorMappingRgab {0xb4u}; // 0, 1, 3, 2
 inline constexpr ColorComponentMapping ColorMappingBgra {0xc6u}; // 2, 1, 0, 3
+inline constexpr ColorComponentMapping ColorMappingAbgr {0x1bu}; // 3, 2, 1, 0
+inline constexpr ColorComponentMapping ColorMappingAgba {0x27u}; // 3, 1, 2, 0
+inline constexpr ColorComponentMapping ColorMappingArbg {0x63u}; // 3, 0, 2, 1
+inline constexpr ColorComponentMapping ColorMappingAgbr {0x87u}; // 3, 1, 0, 2
+inline constexpr ColorComponentMapping ColorMappingArgb {0x93u}; // 3, 0, 1, 2
 
 enum class DepthFormat : uint32_t {
 	kInvalid = 0,
@@ -378,9 +401,11 @@ enum class BufferFormat : uint32_t {
 	k8Srgb              = 128,
 	k8_8Srgb            = 129,
 	k8_8_8_8Srgb        = 130,
+	k10_10_10_2Float    = 131,
 	k9_9_9_5Float       = 132,
 	k5_6_5UNorm         = 133,
 	k5_5_5_1UNorm       = 134,
+	k1_5_5_5UNorm       = 135,
 	k4_4_4_4UNorm       = 136,
 	kFmask8_S4_F4       = 161,
 	kBc1UNorm           = 169,

src/graphics/guest_gpu/gpu_format.cpp

@@ -3,129 +3,221 @@
 #include "graphics/guest_gpu/gpu_defs.h"
 
 #include <array>
+#include <bit>
 
 namespace Libs::Graphics::Prospero {
 namespace {
 
 struct FormatInfo {
-	uint32_t 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;
 };
 
 constexpr FormatInfo kFormatInfo[] = {
-    {GpuEnumValue(BufferFormat::k8UNorm), 1, 0, 1, true, false},
-    {GpuEnumValue(BufferFormat::k8SNorm), 0, 0, 1, false, false},
-    {GpuEnumValue(BufferFormat::k8UInt), 1, 0, 1, true, true},
-    {GpuEnumValue(BufferFormat::k16UNorm), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::k16SNorm), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::k16UInt), 2, 0, 2, true, true},
-    {GpuEnumValue(BufferFormat::k16SInt), 2, 0, 2, false, false},
-    {GpuEnumValue(BufferFormat::k16Float), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::k8_8UNorm), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::k8_8SNorm), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::k8_8UInt), 2, 0, 2, true, true},
-    {GpuEnumValue(BufferFormat::k8_8SInt), 2, 0, 2, false, false},
-    {GpuEnumValue(BufferFormat::k32UInt), 4, 0, 4, true, true},
-    {GpuEnumValue(BufferFormat::k32SInt), 4, 0, 4, false, false},
-    {GpuEnumValue(BufferFormat::k32Float), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k16_16UNorm), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k16_16SNorm), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k16_16UInt), 4, 0, 4, true, true},
-    {GpuEnumValue(BufferFormat::k16_16SInt), 4, 0, 4, false, false},
-    {GpuEnumValue(BufferFormat::k16_16Float), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k11_11_10Float), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k10_10_10_2UNorm), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k10_10_10_2UInt), 4, 0, 4, true, true},
-    {GpuEnumValue(BufferFormat::k8_8_8_8UNorm), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k8_8_8_8SNorm), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k8_8_8_8UInt), 4, 0, 4, true, true},
-    {GpuEnumValue(BufferFormat::k8_8_8_8SInt), 4, 0, 4, false, false},
-    {GpuEnumValue(BufferFormat::k32_32UInt), 8, 0, 8, true, true},
-    {GpuEnumValue(BufferFormat::k32_32SInt), 8, 0, 8, false, false},
-    {GpuEnumValue(BufferFormat::k32_32Float), 8, 0, 8, true, false},
-    {GpuEnumValue(BufferFormat::k16_16_16_16UNorm), 8, 0, 8, true, false},
-    {GpuEnumValue(BufferFormat::k16_16_16_16SNorm), 8, 0, 8, true, false},
-    {GpuEnumValue(BufferFormat::k16_16_16_16UInt), 8, 0, 8, true, true},
-    {GpuEnumValue(BufferFormat::k16_16_16_16SInt), 8, 0, 8, false, false},
-    {GpuEnumValue(BufferFormat::k16_16_16_16Float), 8, 0, 8, true, false},
-    {GpuEnumValue(BufferFormat::k32_32_32UInt), 12, 0, 12, true, true},
-    {GpuEnumValue(BufferFormat::k32_32_32SInt), 12, 0, 12, false, false},
-    {GpuEnumValue(BufferFormat::k32_32_32Float), 12, 0, 12, true, false},
-    {GpuEnumValue(BufferFormat::k32_32_32_32UInt), 16, 0, 16, true, true},
-    {GpuEnumValue(BufferFormat::k32_32_32_32SInt), 16, 0, 16, false, false},
-    {GpuEnumValue(BufferFormat::k32_32_32_32Float), 16, 0, 16, true, false},
-    {GpuEnumValue(BufferFormat::k8Srgb), 1, 0, 0, true, false},
-    {GpuEnumValue(BufferFormat::k8_8Srgb), 2, 0, 0, true, false},
-    {GpuEnumValue(BufferFormat::k8_8_8_8Srgb), 4, 0, 4, true, false},
-    {GpuEnumValue(BufferFormat::k9_9_9_5Float), 4, 0, 0, true, false},
-    {GpuEnumValue(BufferFormat::k5_6_5UNorm), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::k5_5_5_1UNorm), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::k4_4_4_4UNorm), 2, 0, 2, true, false},
-    {GpuEnumValue(BufferFormat::kFmask8_S4_F4), 1, 0, 1, true, false},
-    {GpuEnumValue(BufferFormat::kBc1UNorm), 0, 8, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc1Srgb), 0, 8, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc2UNorm), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc2Srgb), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc3UNorm), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc3Srgb), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc4UNorm), 0, 8, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc4SNorm), 0, 8, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc5UNorm), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc5SNorm), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc6UFloat), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc6SFloat), 0, 16, 0, true, false},
-    {GpuEnumValue(BufferFormat::kBc7UNorm), 0, 16, 0, true, false},
-    {GpuEnumValue(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, 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_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, 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},
 };
 
+constexpr uint8_t RenderTargetTypeBit(ChannelType type) {
+	return static_cast<uint8_t>(1u << static_cast<uint32_t>(type));
+}
+
+constexpr auto kUNormType = RenderTargetTypeBit(ChannelType::kUNorm);
+constexpr auto kIntegerTypes =
+    RenderTargetTypeBit(ChannelType::kUInt) | RenderTargetTypeBit(ChannelType::kSInt);
+constexpr auto kNormalizedIntegerTypes =
+    kUNormType | RenderTargetTypeBit(ChannelType::kSNorm) | kIntegerTypes;
+
+struct RenderTargetLayoutInfo {
+	ChannelLayout       layout;
+	BufferFormat        first_linear_format;
+	uint8_t             linear_type_mask;
+	uint8_t             components;
+	BufferFormat        extra_format  = BufferFormat::kInvalid;
+	ChannelType         extra_type    = ChannelType::kUNorm;
+	ChannelOrderSupport order_support = ChannelOrderSupport::kAll;
+};
+
+constexpr RenderTargetLayoutInfo kRenderTargetLayouts[] = {
+    {ChannelLayout::k8, BufferFormat::k8UNorm, kNormalizedIntegerTypes, 1, BufferFormat::k8Srgb,
+     ChannelType::kSrgb},
+    {ChannelLayout::k16, BufferFormat::k16UNorm, kNormalizedIntegerTypes, 1, BufferFormat::k16Float,
+     ChannelType::kFloat},
+    {ChannelLayout::k8_8, BufferFormat::k8_8UNorm, kNormalizedIntegerTypes, 2,
+     BufferFormat::k8_8Srgb, ChannelType::kSrgb},
+    {ChannelLayout::k32, BufferFormat::k32UInt, kIntegerTypes, 1, BufferFormat::k32Float,
+     ChannelType::kFloat},
+    {ChannelLayout::k16_16, BufferFormat::k16_16UNorm, kNormalizedIntegerTypes, 2,
+     BufferFormat::k16_16Float, ChannelType::kFloat},
+    {ChannelLayout::k11_11_10, BufferFormat::k11_11_10UNorm, kNormalizedIntegerTypes, 3,
+     BufferFormat::k11_11_10Float, ChannelType::kFloat},
+    {ChannelLayout::k10_11_11, BufferFormat::k10_11_11UNorm, kNormalizedIntegerTypes, 3,
+     BufferFormat::k10_11_11Float, ChannelType::kFloat},
+    {ChannelLayout::k2_10_10_10, BufferFormat::k2_10_10_10UNorm, kNormalizedIntegerTypes, 4},
+    {ChannelLayout::k10_10_10_2, BufferFormat::k10_10_10_2UNorm, kNormalizedIntegerTypes, 4},
+    {ChannelLayout::k8_8_8_8, BufferFormat::k8_8_8_8UNorm, kNormalizedIntegerTypes, 4,
+     BufferFormat::k8_8_8_8Srgb, ChannelType::kSrgb},
+    {ChannelLayout::k32_32, BufferFormat::k32_32UInt, kIntegerTypes, 2, BufferFormat::k32_32Float,
+     ChannelType::kFloat},
+    {ChannelLayout::k16_16_16_16, BufferFormat::k16_16_16_16UNorm, kNormalizedIntegerTypes, 4,
+     BufferFormat::k16_16_16_16Float, ChannelType::kFloat},
+    {ChannelLayout::k32_32_32_32, BufferFormat::k32_32_32_32UInt, kIntegerTypes, 4,
+     BufferFormat::k32_32_32_32Float, ChannelType::kFloat},
+    {ChannelLayout::k5_6_5, BufferFormat::k5_6_5UNorm, kUNormType, 3},
+    {ChannelLayout::k5_5_5_1, BufferFormat::k5_5_5_1UNorm, kUNormType, 4},
+    {ChannelLayout::k1_5_5_5, BufferFormat::k1_5_5_5UNorm, kUNormType, 4},
+    {ChannelLayout::k4_4_4_4, BufferFormat::k4_4_4_4UNorm, kUNormType, 4},
+    {ChannelLayout::k10_10_10_2Float, BufferFormat::kInvalid, 0, 4, BufferFormat::k10_10_10_2Float,
+     ChannelType::kFloat, ChannelOrderSupport::kStandardOnly},
+};
+
+const RenderTargetLayoutInfo* FindRenderTargetLayout(ChannelLayout layout) {
+	for (const auto& info: kRenderTargetLayouts) {
+		if (info.layout == layout) {
+			return &info;
+		}
+	}
+	return nullptr;
+}
+
 constexpr auto MakeFormatInfoLookup() {
-	constexpr uint32_t                            kMaxFormat = GpuEnumValue(BufferFormat::kBc7Srgb);
+	constexpr size_t kMaxFormat = static_cast<size_t>(BufferFormat::kBc7Srgb);
 	std::array<const FormatInfo*, kMaxFormat + 1> lookup {};
 	for (const auto& info: kFormatInfo) {
-		lookup[info.format] = &info;
+		lookup[static_cast<size_t>(info.format)] = &info;
 	}
 	return lookup;
 }
 
 constexpr auto kFormatInfoLookup = MakeFormatInfoLookup();
 
-const FormatInfo* FindFormatInfo(uint32_t format) {
-	return format < kFormatInfoLookup.size() ? kFormatInfoLookup[format] : nullptr;
+const FormatInfo* FindFormatInfo(BufferFormat format) {
+	const auto index = static_cast<size_t>(format);
+	return index < kFormatInfoLookup.size() ? kFormatInfoLookup[index] : nullptr;
 }
 
 } // namespace
 
-uint32_t NumBytesPerElement(uint32_t format) {
+RenderTargetFormatEncoding ResolveRenderTargetFormat(ChannelLayout layout, ChannelType type) {
+	const auto* layout_info = FindRenderTargetLayout(layout);
+	const auto  raw_type    = static_cast<uint32_t>(type);
+	if (layout_info == nullptr || raw_type >= 8u) {
+		return {};
+	}
+
+	auto format = BufferFormat::kInvalid;
+	if (layout_info->extra_format != BufferFormat::kInvalid && type == layout_info->extra_type) {
+		format = layout_info->extra_format;
+	} else if ((layout_info->linear_type_mask & (1u << raw_type)) != 0) {
+		const auto first_type =
+		    std::countr_zero(static_cast<uint32_t>(layout_info->linear_type_mask));
+		format = static_cast<BufferFormat>(static_cast<uint32_t>(layout_info->first_linear_format) +
+		                                   raw_type - first_type);
+	}
+	if (format == BufferFormat::kInvalid) {
+		return {};
+	}
+	return {format, layout_info->components, layout_info->order_support};
+}
+
+uint32_t NumBytesPerElement(BufferFormat format) {
 	const auto* info = FindFormatInfo(format);
 	return info != nullptr ? info->bytes_per_element : 0;
 }
 
-uint32_t BlockCompressedBytesPerBlock(uint32_t format) {
+uint32_t BlockCompressedBytesPerBlock(BufferFormat format) {
 	const auto* info = FindFormatInfo(format);
 	return info != nullptr ? info->block_compressed_bytes_per_block : 0;
 }
 
-uint32_t RenderTargetBytesPerElement(uint32_t format) {
+uint32_t RenderTargetBytesPerElement(BufferFormat format) {
 	const auto* info = FindFormatInfo(format);
 	return info != nullptr ? info->render_target_bytes_per_element : 0;
 }
 
-bool IsSupportedTextureFormat(uint32_t format) {
-	const auto* info = FindFormatInfo(format);
-	return info != nullptr && info->supported_sampled_texture_format;
+bool IsFmaskTextureFormat(BufferFormat format) {
+	return format == BufferFormat::kFmask8_S4_F4;
 }
 
-bool IsUintTextureFormat(uint32_t format) {
+bool IsSampledTextureFormat(BufferFormat format) {
 	const auto* info = FindFormatInfo(format);
-	return info != nullptr && info->unsigned_integer_texture_format;
+	return info != nullptr && info->sampled_texture;
 }
 
-bool IsFmaskTextureFormat(uint32_t format) {
-	return format == GpuEnumValue(BufferFormat::kFmask8_S4_F4);
+bool IsUintTextureFormat(BufferFormat format) {
+	const auto* info = FindFormatInfo(format);
+	return info != nullptr && info->uint_texture;
+}
+
+BufferFormat RemapTextureFormat(BufferFormat format) {
+	return format == BufferFormat::k11_11_10UInt ? BufferFormat::k32UInt : format;
 }
 
 } // namespace Libs::Graphics::Prospero

src/graphics/guest_gpu/gpu_format.h

@@ -1,16 +1,50 @@
 #ifndef EMULATOR_INCLUDE_EMULATOR_GRAPHICS_GUEST_GPU_GPU_FORMAT_H_
 #define EMULATOR_INCLUDE_EMULATOR_GRAPHICS_GUEST_GPU_GPU_FORMAT_H_
 
-#include "common/common.h"
+#include "graphics/guest_gpu/gpu_defs.h"
 
 namespace Libs::Graphics::Prospero {
 
-uint32_t NumBytesPerElement(uint32_t format);
-uint32_t BlockCompressedBytesPerBlock(uint32_t format);
-uint32_t RenderTargetBytesPerElement(uint32_t format);
-bool     IsSupportedTextureFormat(uint32_t format);
-bool     IsUintTextureFormat(uint32_t format);
-bool     IsFmaskTextureFormat(uint32_t format);
+enum class ChannelOrderSupport : uint8_t {
+	kNone,
+	kStandardOnly,
+	kAll,
+};
+
+struct RenderTargetFormatEncoding {
+	BufferFormat        buffer_format = BufferFormat::kInvalid;
+	uint8_t             components    = 0;
+	ChannelOrderSupport order_support = ChannelOrderSupport::kNone;
+
+	[[nodiscard]] constexpr bool IsValid() const {
+		return buffer_format != BufferFormat::kInvalid && components >= 1u && components <= 4u &&
+		       order_support != ChannelOrderSupport::kNone;
+	}
+
+	[[nodiscard]] constexpr bool SupportsOrder(ChannelOrder order) const {
+		switch (order_support) {
+			case ChannelOrderSupport::kStandardOnly: return order == ChannelOrder::kStandard;
+			case ChannelOrderSupport::kAll:
+				switch (order) {
+					case ChannelOrder::kStandard:
+					case ChannelOrder::kAlt:
+					case ChannelOrder::kReversed:
+					case ChannelOrder::kAltReversed: return true;
+				}
+				return false;
+			default: return false;
+		}
+	}
+};
+
+RenderTargetFormatEncoding ResolveRenderTargetFormat(ChannelLayout layout, ChannelType type);
+uint32_t                   NumBytesPerElement(BufferFormat format);
+uint32_t                   BlockCompressedBytesPerBlock(BufferFormat format);
+uint32_t                   RenderTargetBytesPerElement(BufferFormat format);
+bool                       IsFmaskTextureFormat(BufferFormat format);
+bool                       IsSampledTextureFormat(BufferFormat format);
+bool                       IsUintTextureFormat(BufferFormat format);
+BufferFormat               RemapTextureFormat(BufferFormat format);
 
 } // namespace Libs::Graphics::Prospero
 

src/graphics/guest_gpu/graphicsRun.cpp

@@ -13,6 +13,7 @@
 #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"
@@ -26,6 +27,7 @@
 #include <cstdio>
 #include <deque>
 #include <memory>
+#include <mutex>
 #include <semaphore>
 #include <thread>
 #include <vector>
@@ -36,6 +38,7 @@ static thread_local CommandProcessor* g_current_processor = nullptr;
 static thread_local Pm4Execution*     g_current_execution = nullptr;
 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;
 
 class GpuMutexLock final {
 public:
@@ -58,113 +61,23 @@ private:
 	Common::Mutex& m_mutex;
 };
 
-struct OwnedCmdBuffer {
-	OwnedCmdBuffer() = default;
-	explicit OwnedCmdBuffer(const uint32_t* data, uint32_t count) {
-		EXIT_IF(data == nullptr && count != 0);
-		if (count != 0) {
-			m_words.assign(data, data + 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;
-};
-
-class GpuState {
-public:
-	static constexpr uint32_t ComputePipeCount     = 7;
-	static constexpr uint32_t QueuesPerComputePipe = 8;
-	static constexpr uint32_t ComputeQueueCount    = ComputePipeCount * QueuesPerComputePipe;
-	static constexpr uint32_t QueueCount           = 1 + ComputeQueueCount;
-
-	explicit GpuState(RenderContext& renderer): m_renderer(renderer) {
-		EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
-		m_gfx_cp = std::make_unique<CommandProcessor>(renderer);
-		m_thread = std::jthread(ThreadRun, this);
-	}
-	~GpuState();
-
-	KYTY_CLASS_NO_COPY(GpuState);
-
-	void 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);
-	void SubmitCompute(uint32_t queue, uint32_t* cmd_buffer, uint32_t num_dw,
-	                   bool trigger_agc_interrupt_on_done);
-	void SubmitFlipPreparation(uint64_t request_id);
-	void Done();
-	void Shutdown();
-	[[nodiscard]] bool IsStopping();
-	void               SendCommand(Common::UniqueFunction<void>&& command);
-	void               SendCommandSync(Common::UniqueFunction<void>&& command);
-	void SendCommandSyncWithProcessor(Common::UniqueFunction<void, CommandProcessor&>&& command);
-	int  GetFrameNum();
-	[[nodiscard]] static bool IsGpuThread() noexcept { return g_gpu_thread; }
-
-private:
-	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;
-	};
-
-	void              WaitLocked();
-	void              Enqueue(Submission submission);
-	void              WaitForIdle();
-	bool              Process(Submission& submission);
-	static void       ThreadRun(void* data);
-	CommandProcessor& GetProcessor(uint32_t queue_id);
-
-	RenderContext&                                 m_renderer;
-	Common::Mutex                                  m_submission_mutex;
-	Common::Mutex                                  m_queue_mutex;
-	std::mutex                                     m_shutdown_mutex;
-	Common::CondVar                                m_work_available;
-	Common::CondVar                                m_idle;
-	std::array<std::deque<Submission>, QueueCount> m_queues;
-	std::deque<Common::UniqueFunction<void>>       m_commands;
-	uint32_t                                       m_next_queue        = 0;
-	uint32_t                                       m_submission_count  = 0;
-	bool                                           m_processing        = false;
-	bool                                           m_graphics_done     = true;
-	bool                                           m_accepting         = true;
-	bool                                           m_stopping          = false;
-	bool                                           m_shutdown_complete = false;
-
-	std::unique_ptr<CommandProcessor>                                m_gfx_cp;
-	std::array<std::unique_ptr<CommandProcessor>, ComputeQueueCount> m_compute_cp;
-
-	uint64_t        m_submit_id = 0;
-	std::atomic_int m_done_num  = 0;
-	std::jthread    m_thread;
-};
-
 static bool GraphicsRunDebugDumpEnabled() {
 	return Config::GraphicsDebugDumpEnabled() &&
 	       Config::GetPrintfDirection() != Config::OutputDirection::Silent;
 }
 
-GpuState::~GpuState() {
+GuestGpu::GuestGpu(RenderContext& renderer): m_renderer(renderer) {
+	EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
+	GraphicsInitJmpTables();
+	m_gfx_cp = std::make_unique<CommandProcessor>(renderer, 0);
+	m_thread = std::jthread(ThreadRun, this);
+}
+
+GuestGpu::~GuestGpu() {
 	Shutdown();
 }
 
-void GpuState::Shutdown() {
+void GuestGpu::Shutdown() {
 	std::lock_guard shutdown_lock(m_shutdown_mutex);
 	if (m_shutdown_complete) {
 		return;
@@ -181,12 +94,12 @@ void GpuState::Shutdown() {
 	m_shutdown_complete = true;
 }
 
-bool GpuState::IsStopping() {
+bool GuestGpu::IsStopping() {
 	Common::LockGuard lock(m_queue_mutex);
 	return m_stopping;
 }
 
-void GpuState::SendCommand(Common::UniqueFunction<void>&& command) {
+void GuestGpu::SendCommand(Common::UniqueFunction<void>&& command) {
 	EXIT_IF(!command);
 	if (IsGpuThread()) {
 		command();
@@ -195,10 +108,26 @@ void GpuState::SendCommand(Common::UniqueFunction<void>&& command) {
 	Common::LockGuard lock(m_queue_mutex);
 	EXIT_IF(!m_accepting);
 	m_commands.push_back(std::move(command));
+	m_pending_commands.fetch_add(1, std::memory_order_release);
 	m_work_available.Signal();
 }
 
-void GpuState::SendCommandSync(Common::UniqueFunction<void>&& command) {
+void GuestGpu::ProcessCommands() {
+	EXIT_IF(!IsGpuThread());
+	while (m_pending_commands.load(std::memory_order_acquire) != 0) {
+		Common::UniqueFunction<void> command;
+		{
+			Common::LockGuard lock(m_queue_mutex);
+			EXIT_IF(m_commands.empty());
+			command = std::move(m_commands.front());
+			m_commands.pop_front();
+			EXIT_IF(m_pending_commands.fetch_sub(1, std::memory_order_acq_rel) == 0);
+		}
+		command();
+	}
+}
+
+void GuestGpu::SendCommandSync(Common::UniqueFunction<void>&& command) {
 	EXIT_IF(!command);
 	if (IsGpuThread()) {
 		command();
@@ -212,49 +141,37 @@ void GpuState::SendCommandSync(Common::UniqueFunction<void>&& command) {
 	done.acquire();
 }
 
-void GpuState::SendCommandSyncWithProcessor(
-    Common::UniqueFunction<void, CommandProcessor&>&& command) {
-	EXIT_IF(!command);
-	SendCommandSync([this, operation = std::move(command)]() mutable {
-		EXIT_IF(g_current_processor != nullptr);
-		g_current_processor = m_gfx_cp.get();
-		operation(*m_gfx_cp);
-		g_current_processor = nullptr;
-	});
-}
-
-void GpuState::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) {
+void GuestGpu::Submit(std::span<const uint32_t> draw_commands,
+                      std::span<const uint32_t> constant_commands) {
+	if (draw_commands.empty()) {
+		return;
+	}
 	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 GpuState::SubmitCompute(uint32_t queue, uint32_t* cmd_buffer, uint32_t num_dw,
-                             bool trigger_agc_interrupt_on_done) {
+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));
 }
 
-void GpuState::SubmitFlipPreparation(uint64_t request_id) {
+void GuestGpu::SubmitFlipPreparation(uint64_t request_id) {
 	GpuMutexLock lock(m_submission_mutex);
 	Submission   submission;
 	submission.type            = SubmissionType::FlipPreparation;
@@ -265,44 +182,37 @@ void GpuState::SubmitFlipPreparation(uint64_t request_id) {
 	Enqueue(std::move(submission));
 }
 
-void GpuState::Done() {
+void GuestGpu::Done() {
 	GpuMutexLock lock(m_submission_mutex);
-	if (IsGpuThread()) {
-		m_gfx_cp->BufferWait();
-	} else {
-		WaitLocked();
+	if (!IsGpuThread()) {
+		WaitForIdle();
 	}
 	m_graphics_done = true;
 	m_done_num++;
 }
 
-int GpuState::GetFrameNum() {
+int GuestGpu::GetFrameNum() const {
 	return m_done_num;
 }
 
-void GpuState::WaitLocked() {
-	WaitForIdle();
-	SendCommandSync([this] { m_gfx_cp->BufferWait(); });
-}
-
-CommandProcessor& GpuState::GetProcessor(uint32_t queue_id) {
+CommandProcessor& GuestGpu::GetProcessor(uint32_t queue_id) {
 	EXIT_IF(queue_id >= QueueCount);
 	if (queue_id == 0) {
 		return *m_gfx_cp;
 	}
 	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;
 }
 
 void CommandProcessor::Reset() {
-	BufferWait();
-
 	m_sh_ctx.Reset();
 	m_ucfg.Reset();
 	m_ctx.Reset();
+	m_saved_ctx.Reset();
+	m_context_state_pushed             = false;
 	m_index_type_and_size              = 0;
 	m_index_buffer_size                = 0;
 	m_user_data_marker                 = HW::UserSgprType::Unknown;
@@ -312,6 +222,30 @@ void CommandProcessor::Reset() {
 	std::memset(m_const_ram, 0, sizeof(m_const_ram));
 }
 
+void CommandProcessor::ApplyContextStateOperation(ContextStateOperation operation) {
+	switch (operation) {
+		case ContextStateOperation::Clear: m_ctx.Reset(); break;
+		case ContextStateOperation::Push:
+			EXIT_IF(m_context_state_pushed);
+			m_saved_ctx            = m_ctx;
+			m_context_state_pushed = true;
+			break;
+		case ContextStateOperation::Pop:
+			EXIT_IF(!m_context_state_pushed);
+			m_ctx                  = m_saved_ctx;
+			m_saved_ctx            = {};
+			m_context_state_pushed = false;
+			break;
+		case ContextStateOperation::PushClear:
+			EXIT_IF(m_context_state_pushed);
+			m_saved_ctx            = m_ctx;
+			m_context_state_pushed = true;
+			m_ctx.Reset();
+			break;
+		default: EXIT("unknown context state operation: %u\n", static_cast<uint32_t>(operation));
+	}
+}
+
 void CommandProcessor::BufferInit() {
 	GetScheduler().Begin(m_ctx, m_ucfg, m_sh_ctx);
 }
@@ -321,8 +255,7 @@ void CommandProcessor::BufferFlush() {
 }
 
 void CommandProcessor::BufferFlushAndWait() {
-	auto& submitted = GetScheduler().FlushAndGetSubmitted();
-	submitted.WaitForFenceOnly();
+	GetScheduler().FlushAndWait();
 }
 
 void CommandProcessor::BufferWait() {
@@ -349,8 +282,13 @@ void CommandProcessor::WaitDeDiff(uint32_t diff) {
 	}
 }
 
+void CommandProcessor::WaitForRewind(bool valid) {
+	if (!valid) {
+		SuspendPm4();
+	}
+}
+
 void CommandProcessor::IncrementDe() {
-	BufferWait();
 	m_de_count++;
 }
 
@@ -402,10 +340,8 @@ 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:
@@ -415,16 +351,17 @@ 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) {
@@ -449,7 +386,6 @@ void CommandProcessor::DmaData(uint8_t engine, uint8_t dst_sel, uint8_t dst_cach
 	if (num_bytes == 0) {
 		return;
 	}
-	EXIT_NOT_IMPLEMENTED((num_bytes & 3u) != 0);
 	EXIT_NOT_IMPLEMENTED(dst_cache_policy > 3);
 	EXIT_NOT_IMPLEMENTED(src_cache_policy > 3);
 	EXIT_NOT_IMPLEMENTED(wait_for_previous > 1);
@@ -466,6 +402,13 @@ void CommandProcessor::DmaData(uint8_t engine, uint8_t dst_sel, uint8_t dst_cach
 			default: return false;
 		}
 	};
+	if (dst_sel == 2) {
+		// kNowhere discards the GL2 prefetch destination without a guest-visible write.
+		if (src_sel != 3) {
+			EXIT("unsupported dmaData nowhere source selector 0x%02" PRIx8 "\n", src_sel);
+		}
+		return;
+	}
 	bool dst_gds = false;
 	if (!decode_gds(dst_sel, dst_gds)) {
 		EXIT("unsupported dmaData destination selector 0x%02" PRIx8 "\n", dst_sel);
@@ -488,7 +431,7 @@ void CommandProcessor::DmaData(uint8_t engine, uint8_t dst_sel, uint8_t dst_cach
 	                        dst_gds, src_gds);
 }
 
-void GpuState::Enqueue(Submission submission) {
+void GuestGpu::Enqueue(Submission submission) {
 	EXIT_IF(submission.queue_id >= QueueCount);
 	Common::LockGuard lock(m_queue_mutex);
 	EXIT_IF(!m_accepting);
@@ -497,18 +440,19 @@ void GpuState::Enqueue(Submission submission) {
 	m_work_available.Signal();
 }
 
-void GpuState::WaitForIdle() {
+void GuestGpu::WaitForIdle() {
 	Common::LockGuard lock(m_queue_mutex);
 	while (m_processing || !m_commands.empty() || m_submission_count != 0) {
 		m_idle.Wait(&m_queue_mutex);
 	}
 }
 
-void GpuState::ThreadRun(void* data) {
-	auto* gpu = static_cast<GpuState*>(data);
+void GuestGpu::ThreadRun(void* data) {
+	auto* gpu = static_cast<GuestGpu*>(data);
 	EXIT_IF(gpu == nullptr);
 	KYTY_PROFILER_THREAD("Thread_Gpu");
 	g_gpu_thread = true;
+	g_gpu_state  = gpu;
 
 	for (;;) {
 		Submission                   submission;
@@ -529,6 +473,7 @@ void GpuState::ThreadRun(void* data) {
 			} else if (!gpu->m_commands.empty()) {
 				command = std::move(gpu->m_commands.front());
 				gpu->m_commands.pop_front();
+				EXIT_IF(gpu->m_pending_commands.fetch_sub(1, std::memory_order_acq_rel) == 0);
 				gpu->m_processing = true;
 			} else {
 				int selected_queue = -1;
@@ -560,6 +505,7 @@ void GpuState::ThreadRun(void* data) {
 		}
 		if (should_stop) {
 			gpu->m_gfx_cp->BufferWait();
+			g_gpu_state  = nullptr;
 			g_gpu_thread = false;
 			return;
 		}
@@ -598,9 +544,13 @@ void GpuState::ThreadRun(void* data) {
 	}
 }
 
-bool GpuState::Process(Submission& submission) {
-	auto&      cp          = GetProcessor(submission.queue_id);
+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) {
 		cp.Reset();
@@ -619,21 +569,20 @@ bool GpuState::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;
@@ -650,14 +599,11 @@ bool GpuState::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",
@@ -669,7 +615,7 @@ bool GpuState::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) {
@@ -679,9 +625,6 @@ bool GpuState::Process(Submission& submission) {
 			} else if (complete) {
 				m_renderer.GetGpuResources().RunGarbageCollector();
 			}
-			if (complete && submission.trigger_agc_interrupt_on_done) {
-				Sync::TriggerAgcUserInterrupt();
-			}
 			break;
 		}
 		case SubmissionType::FlipPreparation:
@@ -693,12 +636,13 @@ bool GpuState::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;
@@ -723,14 +667,14 @@ 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);
 }
 
@@ -741,31 +685,33 @@ void CommandProcessor::SuspendPm4() {
 
 void CommandProcessor::ProcessPm4(Pm4Execution& execution, size_t stop_depth) {
 	while (execution.m_buffer_stack.size() > stop_depth) {
+		if (g_gpu_state != nullptr) {
+			g_gpu_state->ProcessCommands();
+		}
 		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;
 		}
@@ -800,8 +746,7 @@ 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;
 		}
@@ -834,8 +779,7 @@ 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;
 	}
 }
@@ -1134,10 +1078,12 @@ void CommandProcessor::DrawIndirectMulti(uint32_t data_offset, uint32_t max_coun
 
 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;
-	uint32_t local_z   = 1;
+	// uint32_t local_x   = 1;
+	// uint32_t local_y   = 1;
+	// uint32_t local_z   = 1;
 
 	{
 		CheckBuffer();
@@ -1161,9 +1107,9 @@ void CommandProcessor::DispatchDirect(uint32_t thread_group_x, uint32_t thread_g
 		}
 
 		const auto& cs = m_sh_ctx.GetCs().cs_regs;
-		local_x        = std::max(cs.num_thread_x, 1u);
-		local_y        = std::max(cs.num_thread_y, 1u);
-		local_z        = std::max(cs.num_thread_z, 1u);
+		// local_x        = std::max(cs.num_thread_x, 1u);
+		// local_y        = std::max(cs.num_thread_y, 1u);
+		// local_z        = std::max(cs.num_thread_z, 1u);
 		if (cs.wave_size == 64u) {
 			static std::atomic_bool logged_wave64_shader {false};
 			if (!logged_wave64_shader.exchange(true, std::memory_order_relaxed)) {
@@ -1179,27 +1125,27 @@ void CommandProcessor::DispatchDirect(uint32_t thread_group_x, uint32_t thread_g
 		                                              thread_group_y, thread_group_z, mode);
 	}
 
-	constexpr uint32_t DispatchInitiatorUseThreadDimensions = 1u << 5u;
+	/*constexpr uint32_t DispatchInitiatorUseThreadDimensions = 1u << 5u;
 	auto               group_count = [](uint32_t threads, uint32_t group_size) {
-		return (threads == 0
-		            ? 0u
-		            : (threads + std::max(group_size, 1u) - 1u) / std::max(group_size, 1u));
+	    return (threads == 0
+	                ? 0u
+	                : (threads + std::max(group_size, 1u) - 1u) / std::max(group_size, 1u));
 	};
 
 	auto groups_x = thread_group_x;
 	auto groups_y = thread_group_y;
 	auto groups_z = thread_group_z;
 	if ((mode & DispatchInitiatorUseThreadDimensions) != 0) {
-		groups_x = group_count(thread_group_x, local_x);
-		groups_y = group_count(thread_group_y, local_y);
-		groups_z = group_count(thread_group_z, local_z);
+	    groups_x = group_count(thread_group_x, local_x);
+	    groups_y = group_count(thread_group_y, local_y);
+	    groups_z = group_count(thread_group_z, local_z);
 	}
 
 	const uint64_t invocations =
 	    static_cast<uint64_t>(groups_x) * groups_y * groups_z * local_x * local_y * local_z;
 	if (invocations != 0) {
-		BufferFlushAndWait();
-	}
+	    BufferFlushAndWait();
+	}*/
 }
 
 void CommandProcessor::DispatchIndirect(uint32_t data_offset, uint32_t mode) {
@@ -1276,7 +1222,14 @@ 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");
 	}
@@ -1289,10 +1242,11 @@ 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);
@@ -1307,10 +1261,13 @@ 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.TriggerInterrupt(m_interrupt_event_id, interrupt_context_id);
+					}
 					return;
 				}
 			} else if (eop_event_type == 0x04 && cache_action == 0x00 && event_index == 0x05) {
@@ -1335,10 +1292,12 @@ 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,
@@ -1352,9 +1311,14 @@ void CommandProcessor::WriteAtEndOfPipe(uint32_t cache_policy, uint32_t event_wr
 					case 0x00:
 						switch (eop_event_type) {
 							case 0x04:
+								if (event_index == 0x05) {
+									write64(false);
+									return;
+								}
+								break;
+							case 0x14:
 							case 0x28:
-								if ((eop_event_type == 0x04 && event_index == 0x05) ||
-								    (eop_event_type == 0x28 && event_index == 0x00)) {
+								if (event_index == 0x00) {
 									write64(false);
 									return;
 								}
@@ -1412,26 +1376,35 @@ 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;
@@ -1488,15 +1461,17 @@ 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;
@@ -1513,7 +1488,6 @@ void CommandProcessor::TriggerEvent(uint32_t event_type, uint32_t event_index) {
 				     event_index);
 			}
 			EmitGlobalBarrier();
-			SynchronizeGpu();
 			break;
 		// DbDataWritebackInvalidate, DbMetadataWritebackInvalidate, CbMetadataWritebackInvalidate.
 		case 0x0000002a:
@@ -1534,12 +1508,36 @@ 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);
 	}
@@ -1602,7 +1600,7 @@ 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();
 }
 
@@ -1623,68 +1621,11 @@ void CommandProcessor::PrepareCpuFlip(uint64_t request_id) {
 }
 
 void CommandProcessor::SynchronizeGpu() {
-	GetScheduler().FinishCurrent();
+	GetScheduler().Finish();
 }
 
-Gpu::Gpu(RenderContext& renderer) {
-	EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
-	GraphicsInitJmpTables();
-	m_state = std::make_unique<GpuState>(renderer);
-}
-
-Gpu::~Gpu() = default;
-
-void Gpu::Shutdown() {
-	m_state->Shutdown();
-}
-
-bool Gpu::IsStopping() {
-	return m_state->IsStopping();
-}
-
-void Gpu::SendCommand(Common::UniqueFunction<void>&& command) {
-	m_state->SendCommand(std::move(command));
-}
-
-void Gpu::SendCommandSync(Common::UniqueFunction<void>&& command) {
-	m_state->SendCommandSync(std::move(command));
-}
-
-void Gpu::SendCommandSyncWithProcessor(Common::UniqueFunction<void, CommandProcessor&>&& command) {
-	m_state->SendCommandSyncWithProcessor(std::move(command));
-}
-
-void Gpu::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) {
-	EXIT_IF(draw_commands == nullptr || draw_size_dw == 0);
-	m_state->Submit(draw_commands, draw_size_dw, constant_commands, constant_size_dw,
-	                trigger_agc_interrupt_on_done);
-}
-
-void Gpu::SubmitCompute(uint32_t queue, uint32_t* commands, uint32_t size_dw,
-                        bool trigger_agc_interrupt_on_done) {
-	EXIT_IF(commands == nullptr || size_dw == 0);
-	m_state->SubmitCompute(queue, commands, size_dw, trigger_agc_interrupt_on_done);
-}
-
-void Gpu::SubmitFlipPreparation(uint64_t request_id) {
-	m_state->SubmitFlipPreparation(request_id);
-}
-
-void Gpu::Done() {
-	m_state->Done();
-}
-
-int Gpu::GetFrameNum() const {
-	return m_state->GetFrameNum();
-}
-
-bool Gpu::IsCommandProcessorThread() noexcept {
-	return g_current_processor != nullptr;
-}
-
-CommandProcessor* Gpu::CurrentCommandProcessor() noexcept {
-	return g_current_processor;
+bool GuestGpu::IsGpuThread() noexcept {
+	return g_gpu_thread;
 }
 
 } // namespace Libs::Graphics

src/graphics/guest_gpu/graphicsRun.h

@@ -3,41 +3,100 @@
 
 #include "common/abi.h"
 #include "common/common.h"
+#include "common/threads.h"
 #include "common/uniqueFunction.h"
+#include "graphics/guest_gpu/command_processor/commandProcessor.h"
 
+#include <array>
+#include <atomic>
+#include <cstdint>
+#include <deque>
 #include <memory>
+#include <mutex>
+#include <span>
+#include <thread>
 
 namespace Libs::Graphics {
 
-class CommandProcessor;
-class GpuState;
 class RenderContext;
 
-class Gpu final {
+class GuestGpu final {
 public:
-	explicit Gpu(RenderContext& renderer);
-	~Gpu();
-	KYTY_CLASS_NO_COPY(Gpu);
+	explicit GuestGpu(RenderContext& renderer);
+	~GuestGpu();
+	KYTY_CLASS_NO_COPY(GuestGpu);
 
 	void               Shutdown();
 	[[nodiscard]] bool IsStopping();
 	void               SendCommand(Common::UniqueFunction<void>&& command);
 	void               SendCommandSync(Common::UniqueFunction<void>&& command);
-	void SendCommandSyncWithProcessor(Common::UniqueFunction<void, CommandProcessor&>&& 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              IsCommandProcessorThread() noexcept;
-	[[nodiscard]] static CommandProcessor* CurrentCommandProcessor() noexcept;
+	[[nodiscard]] static bool IsGpuThread() noexcept;
 
 private:
-	std::unique_ptr<GpuState> m_state;
+	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;
+
+	enum class SubmissionType { Graphics, Compute, FlipPreparation };
+
+	struct Submission {
+		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);
+	void              WaitForIdle();
+	void              ProcessCommands();
+	bool              Process(Submission& submission);
+	static void       ThreadRun(void* data);
+	CommandProcessor& GetProcessor(uint32_t queue_id);
+
+	RenderContext&                                 m_renderer;
+	Common::Mutex                                  m_submission_mutex;
+	Common::Mutex                                  m_queue_mutex;
+	std::mutex                                     m_shutdown_mutex;
+	Common::CondVar                                m_work_available;
+	Common::CondVar                                m_idle;
+	std::array<std::deque<Submission>, QueueCount> m_queues;
+	std::deque<Common::UniqueFunction<void>>       m_commands;
+	std::atomic_uint32_t                           m_pending_commands {0};
+	uint32_t                                       m_next_queue        = 0;
+	uint32_t                                       m_submission_count  = 0;
+	bool                                           m_processing        = false;
+	bool                                           m_graphics_done     = true;
+	bool                                           m_accepting         = true;
+	bool                                           m_stopping          = false;
+	bool                                           m_shutdown_complete = false;
+
+	std::unique_ptr<CommandProcessor>                                m_gfx_cp;
+	std::array<std::unique_ptr<CommandProcessor>, ComputeQueueCount> m_compute_cp;
+
+	uint64_t        m_submit_id = 0;
+	std::atomic_int m_done_num  = 0;
+	std::jthread    m_thread;
+
+	friend class CommandProcessor;
 };
 } // namespace Libs::Graphics
 

src/graphics/guest_gpu/hardwareContext.h

@@ -29,27 +29,27 @@ struct ColorView {
 struct ColorInfo {
 	bool fmask_compression_enable = false;
 	// uint32_t fmask_compression_mode   = 0;
-	bool     fmask_data_compression_disable = false;
-	bool     fmask_one_frag_mode            = false;
-	bool     cmask_fast_clear_enable        = false;
-	bool     dcc_compression_enable         = false;
-	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;
-	uint32_t format                         = 0;
-	uint32_t channel_type                   = 0;
-	uint32_t channel_order                  = 0;
+	bool                    fmask_data_compression_disable = false;
+	bool                    fmask_one_frag_mode            = false;
+	bool                    cmask_fast_clear_enable        = false;
+	bool                    dcc_compression_enable         = false;
+	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;
-	uint32_t tile_mode               = 0;
-	uint32_t fmask_tile_mode         = 0;
-	uint32_t num_samples             = 0;
-	uint32_t num_fragments           = 0;
+	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;
 };
 
 struct ColorAttrib2 {
@@ -59,11 +59,11 @@ struct ColorAttrib2 {
 };
 
 struct ColorAttrib3 {
-	uint32_t depth              = 0;
-	uint32_t tile_mode          = 0;
-	uint32_t dimension          = 0;
-	bool     cmask_pipe_aligned = false;
-	bool     dcc_pipe_aligned   = false;
+	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;
 };
 
 struct ColorDccControl {
@@ -132,7 +132,7 @@ struct RenderTarget {
 };
 
 struct DepthZInfo {
-	uint32_t                                    format      = 0;
+	Prospero::DepthFormat                       format      = Prospero::DepthFormat::kInvalid;
 	uint32_t                                    num_samples = 0;
 	Prospero::TextureCompatiblePlaneCompression texture_compatibility =
 	    Prospero::TextureCompatiblePlaneCompression::kDisable;
@@ -144,16 +144,16 @@ struct DepthZInfo {
 
 	[[nodiscard]] static DepthZInfo Decode(uint32_t value) {
 		DepthZInfo info;
-		info.format                = value & 0x3u;
+		info.format                = static_cast<Prospero::DepthFormat>(value & 0x3u);
 		info.num_samples           = (value >> 2u) & 0x3u;
 		info.texture_compatibility = static_cast<Prospero::TextureCompatiblePlaneCompression>(
-		    value & Prospero::GpuEnumValue(Prospero::TextureCompatiblePlaneCompression::kBitMask));
+		    value & static_cast<uint32_t>(Prospero::TextureCompatiblePlaneCompression::kBitMask));
 		info.partially_resident = (value & 0x00001000u) != 0;
 		info.max_mip_level      = static_cast<uint8_t>((value >> 16u) & 0x0fu);
 		info.expclear_enabled   = (value & 0x08000000u) != 0;
 		info.htile_acceleration = (value & 0x20000000u) != 0;
 		info.z_compare_base     = static_cast<Prospero::ZCompareBase>(
-		    value & Prospero::GpuEnumValue(Prospero::ZCompareBase::kBitMask));
+		    value & static_cast<uint32_t>(Prospero::ZCompareBase::kBitMask));
 		return info;
 	}
 
@@ -167,7 +167,7 @@ struct DepthZInfo {
 };
 
 struct DepthStencilInfo {
-	uint32_t                           format = 0;
+	Prospero::StencilFormat            format = Prospero::StencilFormat::kInvalid;
 	Prospero::TextureCompatibleStencil texture_compatibility =
 	    Prospero::TextureCompatibleStencil::kDisable;
 	bool expclear_enabled       = false;
@@ -176,9 +176,9 @@ struct DepthStencilInfo {
 
 	[[nodiscard]] static DepthStencilInfo Decode(uint32_t value) {
 		DepthStencilInfo info;
-		info.format                = value & 0x1u;
+		info.format                = static_cast<Prospero::StencilFormat>(value & 0x1u);
 		info.texture_compatibility = static_cast<Prospero::TextureCompatibleStencil>(
-		    value & Prospero::GpuEnumValue(Prospero::TextureCompatibleStencil::kBitMask));
+		    value & static_cast<uint32_t>(Prospero::TextureCompatibleStencil::kBitMask));
 		info.partially_resident     = (value & 0x00001000u) != 0;
 		info.expclear_enabled       = (value & 0x08000000u) != 0;
 		info.htile_stencil_disabled = (value & 0x20000000u) != 0;
@@ -295,10 +295,7 @@ struct ClipControl {
 	bool    linear_attribute_clip_enable        = false;
 	bool    force_viewport_index_from_vs_enable = false;
 
-	[[nodiscard]] bool IsZClipModeRepresentable() const {
-		return min_z_clip_disable == max_z_clip_disable;
-	}
-	[[nodiscard]] bool IsZClipEnabled() const { return !min_z_clip_disable; }
+	[[nodiscard]] bool IsZClipEnabled() const { return !min_z_clip_disable && !max_z_clip_disable; }
 };
 
 struct DepthControl {
@@ -348,6 +345,16 @@ struct ModeControl {
 	bool    persp_corr_dis           = false;
 };
 
+struct PolyOffset {
+	int8_t neg_num_db_bits = -23;
+	bool   db_is_float_fmt = true;
+	float  clamp           = 0.0f;
+	float  front_scale     = 0.0f;
+	float  front_offset    = 0.0f;
+	float  back_scale      = 0.0f;
+	float  back_offset     = 0.0f;
+};
+
 struct BlendControl {
 	uint8_t color_srcblend       = 1;
 	uint8_t color_comb_fcn       = 0;
@@ -482,6 +489,26 @@ struct VsShaderResource2 {
 	uint8_t shared_vgprs      = 0;
 };
 
+struct HsShaderResource1 {
+	uint8_t vgprs                     = 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    threadgroup_configuration = false;
+	uint8_t ls_vgpr_component_count   = 0;
+	bool    fp16_overflow             = false;
+};
+
+struct HsShaderResource2 {
+	bool     scratch_en   = false;
+	uint8_t  user_sgpr    = 0;
+	uint16_t lds_size     = 0;
+	uint8_t  shared_vgprs = 0;
+};
+
 struct PsShaderResource1 {
 	uint8_t vgprs                    = 0;
 	uint8_t sgprs                    = 0;
@@ -500,7 +527,7 @@ struct PsShaderResource2 {
 	uint8_t user_sgpr              = 0;
 	bool    wave_cnt_en            = false;
 	uint8_t extra_lds_size         = 0;
-	bool    raster_ordered_shading = false;
+	uint8_t raster_ordered_shading = 0;
 	uint8_t shared_vgprs           = 0;
 };
 
@@ -513,22 +540,27 @@ struct PsStageRegisters {
 
 struct CsStageRegisters {
 
-	uint64_t data_addr      = 0;
-	uint32_t num_thread_x   = 0;
-	uint32_t num_thread_y   = 0;
-	uint32_t num_thread_z   = 0;
-	uint8_t  vgprs          = 0;
-	uint8_t  sgprs          = 0;
-	uint8_t  bulky          = 0;
-	uint8_t  wave_size      = 64;
-	uint8_t  scratch_en     = 0;
-	uint8_t  user_sgpr      = 0;
-	uint8_t  tgid_x_en      = 0;
-	uint8_t  tgid_y_en      = 0;
-	uint8_t  tgid_z_en      = 0;
-	uint8_t  tg_size_en     = 0;
-	uint8_t  tidig_comp_cnt = 0;
-	uint8_t  lds_size       = 0;
+	uint64_t data_addr                 = 0;
+	uint32_t num_thread_x              = 0;
+	uint32_t num_thread_y              = 0;
+	uint32_t num_thread_z              = 0;
+	uint8_t  vgprs                     = 0;
+	uint8_t  sgprs                     = 0;
+	uint8_t  float_mode                = 0;
+	bool     dx10_clamp                = false;
+	bool     ieee_mode                 = false;
+	bool     fp16_overflow             = false;
+	bool     bulky                     = false;
+	bool     threadgroup_configuration = false;
+	uint8_t  wave_size                 = 64;
+	bool     scratch_en                = false;
+	uint8_t  user_sgpr                 = 0;
+	bool     tgid_x_en                 = false;
+	bool     tgid_y_en                 = false;
+	bool     tgid_z_en                 = false;
+	bool     tg_size_en                = false;
+	uint8_t  tidig_comp_cnt            = 0;
+	uint16_t lds_size                  = 0;
 };
 
 struct EsStageRegisters {
@@ -543,24 +575,24 @@ struct LsStageRegisters {
 
 struct HsStageRegisters {
 	uint64_t          data_addr = 0;
-	VsShaderResource1 rsrc1;
-	VsShaderResource2 rsrc2;
+	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;
-	bool    debug_mode               = false;
-	bool    ieee_mode                = false;
-	bool    cu_group_enable          = false;
-	bool    require_forward_progress = false;
-	bool    lds_configuration        = false;
-	uint8_t gs_vgpr_component_count  = 0;
-	bool    fp16_overflow            = false;
+	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    cu_group_enable           = false;
+	bool    require_forward_progress  = false;
+	bool    threadgroup_configuration = false;
+	uint8_t gs_vgpr_component_count   = 0;
+	bool    fp16_overflow             = false;
 };
 
 struct GsShaderResource2 {
@@ -880,6 +912,8 @@ public:
 	void SetDepthControl(const DepthControl& control) { m_depth_control = control; }
 	[[nodiscard]] const ModeControl& GetModeControl() const { return m_mode_control; }
 	void SetModeControl(const ModeControl& control) { m_mode_control = control; }
+	[[nodiscard]] const PolyOffset& GetPolyOffset() const { return m_poly_offset; }
+	void SetPolyOffset(const PolyOffset& offset) { m_poly_offset = offset; }
 	[[nodiscard]] const EqaaControl& GetEqaaControl() const { return m_eqaa_control; }
 	void SetEqaaControl(const EqaaControl& control) { m_eqaa_control = control; }
 	[[nodiscard]] const StencilControl& GetStencilControl() const { return m_stencil_control; }
@@ -904,8 +938,10 @@ public:
 	[[nodiscard]] uint8_t GetStencilClearValue() const { return m_stencil_clear_value; }
 	void SetStencilClearValue(uint8_t clear_value) { m_stencil_clear_value = clear_value; }
 
-	[[nodiscard]] float GetLineWidth() const { return m_line_width; }
-	void                SetLineWidth(float width) { m_line_width = width; }
+	[[nodiscard]] float    GetLineWidth() const { return m_line_width; }
+	void                   SetLineWidth(float width) { m_line_width = width; }
+	[[nodiscard]] uint32_t GetPrimitiveResetIndex() const { return m_primitive_reset_index; }
+	void SetPrimitiveResetIndex(uint32_t index) { m_primitive_reset_index = index; }
 
 	[[nodiscard]] const ShaderRegisters& GetShaderRegisters() const { return m_sh_regs; }
 
@@ -950,7 +986,8 @@ public:
 	void SetScShaderControl(uint32_t value) { m_sh_regs.m_paScShaderControl = value; }
 
 private:
-	float m_line_width = 1.0f;
+	float    m_line_width            = 1.0f;
+	uint32_t m_primitive_reset_index = 0xffffffffu;
 
 	BlendControl    m_blend_control[8];
 	BlendColor      m_blend_color;
@@ -977,6 +1014,7 @@ private:
 	uint8_t           m_stencil_clear_value = 0;
 
 	ModeControl m_mode_control;
+	PolyOffset  m_poly_offset;
 	EqaaControl m_eqaa_control;
 
 	ShaderRegisters m_sh_regs;
@@ -991,12 +1029,14 @@ public:
 
 	void Reset() { *this = UserConfig(); }
 
-	void                   SetPrimitiveType(uint32_t prim_type) { m_prim_type = prim_type; }
-	[[nodiscard]] uint32_t GetPrimType() const { return m_prim_type; }
+	void SetPrimitiveType(Prospero::PrimitiveType prim_type) { m_prim_type = prim_type; }
+	[[nodiscard]] Prospero::PrimitiveType GetPrimType() const { return m_prim_type; }
 	void                   SetIndexOffset(uint32_t index_offset) { m_index_offset = index_offset; }
 	[[nodiscard]] uint32_t GetIndexOffset() const { return m_index_offset; }
 	void                   SetObjectId(uint32_t object_id) { m_object_id = object_id; }
 	[[nodiscard]] uint32_t GetObjectId() const { return m_object_id; }
+	void SetPrimitiveResetControl(uint32_t control) { m_primitive_reset_control = control; }
+	[[nodiscard]] uint32_t GetPrimitiveResetControl() const { return m_primitive_reset_control; }
 
 	[[nodiscard]] const GeControl& GetGeControl() const { return m_ge_cntl; }
 	void                           SetGeControl(const GeControl& control) { m_ge_cntl = control; }
@@ -1011,9 +1051,10 @@ public:
 	void SetGdsOaAddress(uint32_t value) { m_gds_oa.counters[m_gds_oa.GetIndex()].address = value; }
 
 private:
-	uint32_t m_prim_type    = 0;
-	uint32_t m_index_offset = 0;
-	uint32_t m_object_id    = 0;
+	Prospero::PrimitiveType m_prim_type               = Prospero::PrimitiveType::kNone;
+	uint32_t                m_index_offset            = 0;
+	uint32_t                m_object_id               = 0;
+	uint32_t                m_primitive_reset_control = 0;
 
 	GeControl    m_ge_cntl;
 	GeUserVgprEn m_ge_user_vgpr_en;
@@ -1034,8 +1075,8 @@ public:
 	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 VsShaderResource1& rsrc1) { m_vs.hs_regs.rsrc1 = rsrc1; }
-	void SetHsShaderResource2(const VsShaderResource2& rsrc2) { m_vs.hs_regs.rsrc2 = rsrc2; }
+	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 SetGsShaderResource1(const GsShaderResource1& rsrc1) { m_vs.gs_regs.rsrc1 = rsrc1; }
@@ -1048,6 +1089,7 @@ public:
 	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

@@ -21,15 +21,14 @@ 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";
+	names[R_CONTEXT_STATE]  = "R_CONTEXT_STATE";
 	return names;
 }
 
@@ -61,6 +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";
@@ -90,6 +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;
 }
@@ -99,7 +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();
 
@@ -142,9 +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 */
@@ -78,15 +80,14 @@ 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;
+constexpr uint32_t R_CONTEXT_STATE  = 0x1A;
 
 constexpr uint32_t R_NUM = 0x3F + 1;
 
@@ -316,8 +317,8 @@ 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;
 
@@ -371,6 +372,7 @@ constexpr uint32_t PA_CL_UCP_0_X          = 0x16F;
 constexpr uint32_t PA_CL_UCP_0_Y          = 0x170;
 constexpr uint32_t PA_CL_UCP_0_Z          = 0x171;
 constexpr uint32_t PA_CL_UCP_0_W          = 0x172;
+constexpr uint32_t PA_CL_UCP_5_W          = 0x186;
 constexpr uint32_t SPI_PS_INPUT_CNTL_0    = 0x191;
 constexpr uint32_t SPI_PS_INPUT_CNTL_31   = 0x1B0;
 constexpr uint32_t SPI_VS_OUT_CONFIG      = 0x1B1;
@@ -385,9 +387,6 @@ 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;
@@ -577,24 +576,28 @@ 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_SRESULTS_COMPARE_STATE0     = 0x2B0;
-constexpr uint32_t DB_SRESULTS_COMPARE_STATE1     = 0x2B1;
-constexpr uint32_t VGT_GS_MAX_VERT_OUT            = 0x2CE;
-constexpr uint32_t GE_NGG_SUBGRP_CNTL             = 0x2D3;
-constexpr uint32_t VGT_TESS_DISTRIBUTION          = 0x2D4;
-constexpr uint32_t VGT_SHADER_STAGES_EN           = 0x2D5;
-constexpr uint32_t VGT_LS_HS_CONFIG               = 0x2D6;
-constexpr uint32_t VGT_TF_PARAM                   = 0x2DB;
-constexpr uint32_t DB_ALPHA_TO_MASK               = 0x2DC;
-constexpr uint32_t PA_SU_POLY_OFFSET_DB_FMT_CNTL  = 0x2DE;
-constexpr uint32_t PA_SU_POLY_OFFSET_CLAMP        = 0x2DF;
-constexpr uint32_t PA_SU_POLY_OFFSET_FRONT_SCALE  = 0x2E0;
-constexpr uint32_t PA_SU_POLY_OFFSET_FRONT_OFFSET = 0x2E1;
-constexpr uint32_t PA_SU_POLY_OFFSET_BACK_SCALE   = 0x2E2;
-constexpr uint32_t PA_SU_POLY_OFFSET_BACK_OFFSET  = 0x2E3;
-constexpr uint32_t VGT_GS_INSTANCE_CNT            = 0x2E4;
-constexpr uint32_t PA_SC_CENTROID_PRIORITY_0      = 0x2F5;
-constexpr uint32_t PA_SC_CENTROID_PRIORITY_1      = 0x2F6;
+constexpr uint32_t DB_SRESULTS_COMPARE_STATE0                          = 0x2B0;
+constexpr uint32_t DB_SRESULTS_COMPARE_STATE1                          = 0x2B1;
+constexpr uint32_t VGT_GS_MAX_VERT_OUT                                 = 0x2CE;
+constexpr uint32_t GE_NGG_SUBGRP_CNTL                                  = 0x2D3;
+constexpr uint32_t VGT_TESS_DISTRIBUTION                               = 0x2D4;
+constexpr uint32_t VGT_SHADER_STAGES_EN                                = 0x2D5;
+constexpr uint32_t VGT_LS_HS_CONFIG                                    = 0x2D6;
+constexpr uint32_t VGT_TF_PARAM                                        = 0x2DB;
+constexpr uint32_t DB_ALPHA_TO_MASK                                    = 0x2DC;
+constexpr uint32_t PA_SU_POLY_OFFSET_DB_FMT_CNTL                       = 0x2DE;
+constexpr uint32_t PA_SU_POLY_OFFSET_DB_FMT_CNTL_NEG_NUM_DB_BITS_SHIFT = 0;
+constexpr uint32_t PA_SU_POLY_OFFSET_DB_FMT_CNTL_NEG_NUM_DB_BITS_MASK  = 0xFF;
+constexpr uint32_t PA_SU_POLY_OFFSET_DB_FMT_CNTL_DB_IS_FLOAT_FMT_SHIFT = 8;
+constexpr uint32_t PA_SU_POLY_OFFSET_DB_FMT_CNTL_DB_IS_FLOAT_FMT_MASK  = 0x1;
+constexpr uint32_t PA_SU_POLY_OFFSET_CLAMP                             = 0x2DF;
+constexpr uint32_t PA_SU_POLY_OFFSET_FRONT_SCALE                       = 0x2E0;
+constexpr uint32_t PA_SU_POLY_OFFSET_FRONT_OFFSET                      = 0x2E1;
+constexpr uint32_t PA_SU_POLY_OFFSET_BACK_SCALE                        = 0x2E2;
+constexpr uint32_t PA_SU_POLY_OFFSET_BACK_OFFSET                       = 0x2E3;
+constexpr uint32_t VGT_GS_INSTANCE_CNT                                 = 0x2E4;
+constexpr uint32_t PA_SC_CENTROID_PRIORITY_0                           = 0x2F5;
+constexpr uint32_t PA_SC_CENTROID_PRIORITY_1                           = 0x2F6;
 
 constexpr uint32_t PA_SC_AA_CONFIG                             = 0x2F8;
 constexpr uint32_t PA_SC_AA_CONFIG_MSAA_NUM_SAMPLES_SHIFT      = 0;
@@ -778,21 +781,21 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_FWD_PROGRESS_MASK      = 0x1;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_FP16_OVFL_SHIFT        = 29;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_FP16_OVFL_MASK         = 0x1;
 
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS                                = 0xB;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SCRATCH_EN_SHIFT               = 0;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SCRATCH_EN_MASK                = 0x1;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_SHIFT                = 1;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_MASK                 = 0x1F;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_WAVE_CNT_EN_SHIFT              = 7;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_WAVE_CNT_EN_MASK               = 0x1;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_EXTRA_LDS_SIZE_SHIFT           = 8;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_EXTRA_LDS_SIZE_MASK            = 0xFF;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_LOAD_INTRAWAVE_COLLISION_SHIFT = 26;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_LOAD_INTRAWAVE_COLLISION_MASK  = 0x1;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_MSB_SHIFT            = 27;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_MSB_MASK             = 0x1;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SHARED_VGPR_CNT_SHIFT          = 28;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SHARED_VGPR_CNT_MASK           = 0xF;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS                              = 0xB;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SCRATCH_EN_SHIFT             = 0;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SCRATCH_EN_MASK              = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_SHIFT              = 1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_MASK               = 0x1F;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_WAVE_CNT_EN_SHIFT            = 7;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_WAVE_CNT_EN_MASK             = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_EXTRA_LDS_SIZE_SHIFT         = 8;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_EXTRA_LDS_SIZE_MASK          = 0xFF;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_RASTER_ORDERED_SHADING_SHIFT = 25;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_RASTER_ORDERED_SHADING_MASK  = 0x3;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_MSB_SHIFT          = 27;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_USER_SGPR_MSB_MASK           = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_PS_SHARED_VGPR_CNT_SHIFT        = 28;
+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;
@@ -885,61 +888,109 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_GS_USER_SGPR_MSB_MASK     = 0x1;
 constexpr uint32_t SPI_SHADER_PGM_RSRC2_GS_SHARED_VGPR_CNT_SHIFT  = 28;
 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;
-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_SHADER_PGM_LO_HS            = 0x108;
-constexpr uint32_t SPI_SHADER_PGM_HI_HS            = 0x109;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS         = 0x10A;
-constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS         = 0x10B;
-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;
-constexpr uint32_t COMPUTE_NUM_THREAD_X            = 0x207;
-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 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;
+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_SHADER_PGM_LO_HS                           = 0x108;
+constexpr uint32_t SPI_SHADER_PGM_HI_HS                           = 0x109;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS                        = 0x10A;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS                        = 0x10B;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_VGPRS_SHIFT            = 0;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_VGPRS_MASK             = 0x3F;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_PRIORITY_SHIFT         = 10;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_PRIORITY_MASK          = 0x3;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_FLOAT_MODE_SHIFT       = 12;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_FLOAT_MODE_MASK        = 0xFF;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_DX10_CLAMP_SHIFT       = 21;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_DX10_CLAMP_MASK        = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_DEBUG_MODE_SHIFT       = 22;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_DEBUG_MODE_MASK        = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_IEEE_MODE_SHIFT        = 23;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_IEEE_MODE_MASK         = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_FWD_PROGRESS_SHIFT     = 25;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_FWD_PROGRESS_MASK      = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_WGP_MODE_SHIFT         = 26;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_WGP_MODE_MASK          = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_LS_VGPR_COMP_CNT_SHIFT = 28;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_LS_VGPR_COMP_CNT_MASK  = 0x3;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_FP16_OVFL_SHIFT        = 30;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS_FP16_OVFL_MASK         = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_SCRATCH_EN_SHIFT       = 0;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_SCRATCH_EN_MASK        = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_USER_SGPR_SHIFT        = 1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_USER_SGPR_MASK         = 0x1F;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_LDS_SIZE_SHIFT         = 18;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_LDS_SIZE_MASK          = 0x1FF;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_USER_SGPR_MSB_SHIFT    = 27;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_USER_SGPR_MSB_MASK     = 0x1;
+constexpr uint32_t SPI_SHADER_PGM_RSRC2_HS_SHARED_VGPR_CNT_SHIFT  = 28;
+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;
+constexpr uint32_t COMPUTE_NUM_THREAD_X                           = 0x207;
+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_BULKY_SHIFT  = 24;
-constexpr uint32_t COMPUTE_PGM_RSRC1_BULKY_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                  = 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_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_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_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;
@@ -1002,6 +1053,7 @@ 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 GE_USER_VGPR_EN                     = 0x262;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR1_SHIFT = 0;
@@ -1036,7 +1088,7 @@ constexpr uint32_t UC_NOP                       = 0x80003FFF;
 
 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.h

@@ -3,6 +3,7 @@
 
 #include "common/abi.h"
 #include "common/common.h"
+#include "graphics/guest_gpu/gpu_defs.h"
 
 namespace Libs::Graphics {
 
@@ -39,7 +40,7 @@ enum class TileBlockFamily : uint32_t {
 };
 
 struct TileBlockLayout {
-	TileBlockFamily family            = TileBlockFamily::Standard256B;
+	TileBlockFamily family            = TileBlockFamily::Count;
 	uint32_t        bytes_per_element = 0;
 	uint32_t        block_size        = 0;
 	uint32_t        block_width       = 0;
@@ -47,42 +48,73 @@ struct TileBlockLayout {
 	uint32_t        block_depth       = 0;
 };
 
-struct TileVolumeLayout {
-	TileBlockFamily family            = TileBlockFamily::Count;
-	uint32_t        bytes_per_element = 0;
-	uint32_t        texel_width       = 1;
-	uint32_t        texel_height      = 1;
-	uint32_t        first_tail_level  = 0;
-	uint32_t        block_depth       = 1;
-	uint64_t        block_slice_size  = 0;
-	uint64_t        total_size        = 0;
-	uint64_t        level_offsets[16] = {};
-	uint64_t        level_sizes[16]   = {};
-	uint32_t        tail_x[16]        = {};
-	uint32_t        tail_y[16]        = {};
-	uint32_t        level_widths[16]  = {};
-	uint32_t        level_heights[16] = {};
+struct TileTextureElementLayout {
+	uint32_t bytes        = 0;
+	uint32_t texel_width  = 1;
+	uint32_t texel_height = 1;
+};
+
+struct TileTextureBlockLayout {
+	TileBlockLayout block;
+	uint32_t        texel_width  = 1;
+	uint32_t        texel_height = 1;
+};
+
+enum class TileSurfaceDimension : uint32_t {
+	Dim2D,
+	Dim3D,
+};
+
+struct TileSurfaceDescription {
+	Prospero::BufferFormat format    = Prospero::BufferFormat::kInvalid;
+	Prospero::TileMode     tile_mode = Prospero::TileMode::kLinear;
+	TileSurfaceDimension   dimension = TileSurfaceDimension::Dim2D;
+	uint32_t               width     = 0;
+	uint32_t               height    = 0;
+	uint32_t               depth     = 1;
+	uint32_t               levels    = 1;
+	uint32_t               layers    = 1;
+};
+
+// Width/height fields are measured in addressable elements, not texture texels.
+struct TileMipLayout {
+	uint64_t offset        = 0;
+	uint64_t size          = 0;
+	uint32_t width         = 0;
+	uint32_t height        = 0;
+	uint32_t padded_width  = 0;
+	uint32_t padded_height = 0;
+	uint32_t tail_x        = 0;
+	uint32_t tail_y        = 0;
+};
+
+struct TileSurfaceLayout {
+	TileSurfaceDescription description;
+	TileTextureBlockLayout texture;
+	uint32_t               first_tail_level = 0;
+	uint64_t               block_slice_size = 0;
+	uint64_t               total_size       = 0;
+	TileMipLayout          mips[16]         = {};
 };
 
 bool TileGetBlockLayout(TileBlockFamily family, uint32_t bytes_per_element,
                         TileBlockLayout& layout);
+bool TileGetTextureElementLayout(Prospero::BufferFormat format, TileTextureElementLayout& layout);
+bool TileGetTextureBlockLayout(Prospero::BufferFormat format, Prospero::TileMode tile, bool volume,
+                               TileTextureBlockLayout& layout);
+bool TileGetTiledTextureLayout(const TileSurfaceDescription& description,
+                               TileSurfaceLayout&            layout);
 bool TileGetBlockOffset(const TileBlockLayout& layout, uint32_t x, uint32_t y, uint32_t z,
                         uint32_t& byte_offset);
 bool TileGetBlockXor(const TileBlockLayout& layout, uint32_t block_x, uint32_t block_y,
                      uint32_t& byte_offset);
 bool TileGetBlockXor(const TileBlockLayout& layout, uint32_t block_x, uint32_t block_y,
                      uint32_t block_z, uint32_t& byte_offset);
-bool TileIsStandard256BTextureSupported(uint32_t format);
-bool TileIsStandard4KBTextureSupported(uint32_t format);
-bool TileIsStandard64KBTextureSupported(uint32_t format);
-bool TileGetTextureVolumeLayout(uint32_t format, uint32_t width, uint32_t height, uint32_t depth,
-                                uint32_t levels, uint32_t tile, TileVolumeLayout& layout);
 
-bool     TileGetHtileSize(uint32_t width, uint32_t height, TileSizeAlign& htile_size);
-bool     TileGetDepthSize(uint32_t width, uint32_t height, uint32_t pitch, uint32_t z_format,
-                          uint32_t stencil_format, bool htile, TileSizeAlign& stencil_size,
-                          TileSizeAlign& htile_size, TileSizeAlign& depth_size,
-                          uint32_t num_fragments_log2 = 0);
+bool     TileGetDepthSize(uint32_t width, uint32_t height, uint32_t pitch,
+                          Prospero::DepthFormat z_format, Prospero::StencilFormat stencil_format,
+                          bool htile, TileSizeAlign& stencil_size, TileSizeAlign& htile_size,
+                          TileSizeAlign& depth_size, uint32_t num_fragments_log2 = 0);
 uint32_t TileGetRenderTargetPitch(uint32_t width, uint32_t bytes_per_element,
                                   uint32_t num_fragments_log2 = 0);
 uint32_t TileGetDepthPitch(uint32_t width, uint32_t bytes_per_element,
@@ -94,13 +126,14 @@ bool     TileGetRenderTargetMipLayout(uint32_t width, uint32_t height, uint32_t
                                       uint32_t bytes_per_element, uint32_t levels,
                                       TileSizeAlign& total_size, TileSizeOffset* level_sizes,
                                       TilePaddedSize* padded_size);
-void     TileGetTextureSize(uint32_t format, uint32_t width, uint32_t height, uint32_t pitch,
-                            uint32_t levels, uint32_t tile, TileSizeAlign* total_size,
+void     TileGetTextureSize(Prospero::BufferFormat format, uint32_t width, uint32_t height,
+                            uint32_t levels, Prospero::TileMode tile, TileSizeAlign* total_size,
                             TileSizeOffset* level_sizes, TilePaddedSize* padded_size);
-void TileGetTextureTotalSize(uint32_t format, uint32_t width, uint32_t height, uint32_t depth,
-                             uint32_t pitch, uint32_t levels, uint32_t tile, bool volume_texture,
-                             TileSizeAlign& total_size);
-uint32_t TileGetTexturePitch(uint32_t format, uint32_t width, uint32_t levels, uint32_t tile);
+void     TileGetTextureTotalSize(Prospero::BufferFormat format, uint32_t width, uint32_t height,
+                                 uint32_t depth, uint32_t levels, Prospero::TileMode tile,
+                                 bool volume_texture, TileSizeAlign& total_size);
+uint32_t TileGetTexturePitch(Prospero::BufferFormat format, uint32_t width,
+                             Prospero::TileMode tile);
 
 } // namespace Libs::Graphics
 

src/graphics/host_gpu/graphicContext.h

@@ -3,20 +3,88 @@
 
 #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                               compute_wave64_supported              = false;
+	bool                               sample_rate_shading_enabled           = false;
+	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]] 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;
@@ -24,7 +92,6 @@ struct GraphicContext: public VulkanInstance {
 	[[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);
@@ -36,6 +103,15 @@ 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 {

src/graphics/host_gpu/memoryTracker.cpp

@@ -79,7 +79,6 @@ RegionManager* MemoryTracker::GetOrCreateRegion(uint64_t index) {
 
 bool MemoryTracker::IsRegionCpuModified(uint64_t vaddr, uint64_t size) {
 	CheckNotInUploadCallback();
-	std::lock_guard access(m_access_mutex);
 	return Iterate<true>(vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 		std::scoped_lock lock(manager->lock);
 		return manager->IsModified<DirtySource::Cpu>(offset, bytes);
@@ -88,7 +87,6 @@ bool MemoryTracker::IsRegionCpuModified(uint64_t vaddr, uint64_t size) {
 
 bool MemoryTracker::IsRegionGpuModified(uint64_t vaddr, uint64_t size) {
 	CheckNotInUploadCallback();
-	std::lock_guard access(m_access_mutex);
 	return Iterate<false>(vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 		std::scoped_lock lock(manager->lock);
 		return manager->IsModified<DirtySource::Gpu>(offset, bytes);
@@ -97,7 +95,6 @@ bool MemoryTracker::IsRegionGpuModified(uint64_t vaddr, uint64_t size) {
 
 void MemoryTracker::MarkRegionAsCpuModified(uint64_t vaddr, uint64_t size) {
 	CheckNotInUploadCallback();
-	std::lock_guard access(m_access_mutex);
 	Iterate<true>(vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 		std::scoped_lock lock(manager->lock);
 		manager->ChangeState<DirtySource::Cpu, true>(manager->GetCpuAddr() + offset, bytes);
@@ -106,7 +103,6 @@ void MemoryTracker::MarkRegionAsCpuModified(uint64_t vaddr, uint64_t size) {
 
 void MemoryTracker::MarkRegionAsGpuModified(uint64_t vaddr, uint64_t size) {
 	CheckNotInUploadCallback();
-	std::lock_guard access(m_access_mutex);
 	Iterate<true>(vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 		std::scoped_lock lock(manager->lock);
 		manager->ChangeState<DirtySource::Gpu, true>(manager->GetCpuAddr() + offset, bytes);
@@ -115,14 +111,13 @@ void MemoryTracker::MarkRegionAsGpuModified(uint64_t vaddr, uint64_t size) {
 
 void MemoryTracker::UnmarkRegionAsGpuModified(uint64_t vaddr, uint64_t size) {
 	CheckNotInUploadCallback();
-	std::lock_guard access(m_access_mutex);
 	Iterate<false>(vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 		std::scoped_lock lock(manager->lock);
 		manager->ChangeState<DirtySource::Gpu, false>(manager->GetCpuAddr() + offset, bytes);
 	});
 }
 
-void MemoryTracker::UntrackMemoryLocked(uint64_t vaddr, uint64_t size) {
+void MemoryTracker::UntrackMemoryImpl(uint64_t vaddr, uint64_t size) {
 	std::vector<RegionManager*> managers;
 	managers.reserve((vaddr % TRACKER_REGION_SIZE + size + TRACKER_REGION_SIZE - 1) /
 	                 TRACKER_REGION_SIZE);
@@ -148,8 +143,7 @@ void MemoryTracker::UntrackMemoryLocked(uint64_t vaddr, uint64_t size) {
 
 void MemoryTracker::UntrackMemory(uint64_t vaddr, uint64_t size) {
 	CheckNotInUploadCallback();
-	std::lock_guard access(m_access_mutex);
-	UntrackMemoryLocked(vaddr, size);
+	UntrackMemoryImpl(vaddr, size);
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/memoryTracker.h

@@ -29,45 +29,29 @@ public:
 	void               MarkRegionAsGpuModified(uint64_t vaddr, uint64_t size);
 	void               UnmarkRegionAsGpuModified(uint64_t vaddr, uint64_t size);
 	void               UntrackMemory(uint64_t vaddr, uint64_t size);
+	// Removes protection from a range and flushes GPU-owned data when required.
 	template <typename Flush>
-	void InvalidateRegion(uint64_t vaddr, uint64_t size, Flush&& on_flush) {
+	void InvalidateRegion(uint64_t vaddr, uint64_t size, Flush&& on_flush) noexcept {
 		static_assert(std::is_invocable_v<Flush&>);
 		CheckNotInUploadCallback();
 		ValidateRange(vaddr, size);
 
-		const auto update_cpu_state = [this, vaddr, size] {
-			std::lock_guard             access(m_access_mutex);
-			std::vector<RegionManager*> managers;
-			Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t, uint64_t) {
-				managers.push_back(manager);
-			});
-			std::vector<std::unique_lock<TrackingSpinLock>> locks;
-			locks.reserve(managers.size());
-			for (auto* manager: managers) {
-				locks.emplace_back(manager->lock);
+		Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t offset, uint64_t bytes) {
+			const bool should_flush = [&] {
+				// Perform both the GPU modification check and CPU state change with the lock in
+				// case the GPU thread is racing to mark the page modified. If a flush is needed,
+				// on_flush performs the CPU state change.
+				std::scoped_lock lock(manager->lock);
+				if (manager->IsModified<DirtySource::Gpu>(offset, bytes)) {
+					return true;
+				}
+				manager->ChangeState<DirtySource::Cpu, true>(manager->GetCpuAddr() + offset, bytes);
+				return false;
+			}();
+			if (should_flush) {
+				on_flush();
 			}
-			const bool gpu_modified = Iterate<false>(
-			    vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
-				    return manager->IsModified<DirtySource::Gpu>(offset, bytes);
-			    });
-			if (gpu_modified) {
-				return true;
-			}
-			Iterate<false>(vaddr, size,
-			               [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
-				               manager->ChangeState<DirtySource::Cpu, true>(
-				                   manager->GetCpuAddr() + offset, bytes);
-			               });
-			return false;
-		};
-
-		if (!update_cpu_state()) {
-			return;
-		}
-		std::forward<Flush>(on_flush)();
-		if (update_cpu_state()) {
-			EXIT("memory invalidation retained GPU-owned pages\n");
-		}
+		});
 	}
 #if KYTY_BUILD == KYTY_BUILD_DEBUG
 	void ValidateGpuDirtyPages(const RangeSet& dirty, uint64_t vaddr, uint64_t size,
@@ -84,7 +68,6 @@ public:
 		static_assert(std::is_nothrow_invocable_v<Preflight&, uint64_t, uint64_t>);
 		static_assert(std::is_nothrow_invocable_v<Func&, uint64_t, uint64_t>);
 		CheckNotInUploadCallback();
-		std::lock_guard             access(m_access_mutex);
 		std::vector<RegionManager*> managers;
 		Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t, uint64_t) {
 			managers.push_back(manager);
@@ -125,7 +108,6 @@ public:
 		static_assert(std::is_nothrow_invocable_v<RangeFunc&, uint64_t, uint64_t>);
 		static_assert(std::is_nothrow_invocable_v<UploadFunc&>);
 		CheckNotInUploadCallback();
-		std::unique_lock access(m_access_mutex);
 		Iterate<true>(vaddr, size, [](RegionManager*, uint64_t, uint64_t) {});
 		const auto* previous_upload_owner = std::exchange(s_upload_owner, this);
 		Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t offset, uint64_t bytes) {
@@ -189,13 +171,12 @@ private:
 	}
 
 	static void    ValidateRange(uint64_t vaddr, uint64_t size);
-	void           UntrackMemoryLocked(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;
 	std::vector<std::unique_ptr<RegionManager>>    m_region_storage;
 	std::mutex                                     m_region_mutex;
-	std::mutex                                     m_access_mutex;
 	PageManager&                                   m_page_manager;
 };
 

src/graphics/host_gpu/pageManager.cpp

@@ -1,5 +1,6 @@
 #include "graphics/host_gpu/pageManager.h"
 
+#include "common/platform/sysDbg.h"
 #include "graphics/host_gpu/regionDefinitions.h"
 #include "kernel/memory.h"
 
@@ -52,20 +53,20 @@ constexpr uint32_t READ_WRITE_PROTECTION = PAGE_READWRITE;
 	std::fputs("PageManager fail-fast: ", stderr);
 	std::fputs(reason != nullptr ? reason : "invalid page state", stderr);
 	std::fputc('\n', stderr);
+#if !defined(__APPLE__)
+	void* frames[16] {};
+	int   frame_count = static_cast<int>(std::size(frames));
+	SysStackWalk(frames, &frame_count);
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
-	void*      frames[16] {};
-	const auto frame_count =
-	    CaptureStackBackTrace(0, static_cast<DWORD>(std::size(frames)), frames, nullptr);
 	const auto image_base = reinterpret_cast<uintptr_t>(GetModuleHandleW(nullptr));
-	for (uint16_t i = 0; i < frame_count; i++) {
+	for (int i = 0; i < frame_count; i++) {
 		const auto address = reinterpret_cast<uintptr_t>(frames[i]);
-		std::fprintf(stderr, "  frame[%u]=0x%016" PRIxPTR " image_rva=0x%016" PRIxPTR "\n", i,
+		std::fprintf(stderr, "  frame[%d]=0x%016" PRIxPTR " image_rva=0x%016" PRIxPTR "\n", i,
 		             address, address >= image_base ? address - image_base : 0);
 	}
-#elif !defined(__APPLE__)
-	void*     frames[16] {};
-	const int frame_count = ::backtrace(frames, static_cast<int>(std::size(frames)));
+#else
 	::backtrace_symbols_fd(frames, frame_count, STDERR_FILENO);
+#endif
 #endif
 	std::fflush(stderr);
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS

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/renderer/cache/bufferCache.h

@@ -3,13 +3,15 @@
 
 #include "common/abi.h"
 #include "common/common.h"
-#include "common/threads.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>
@@ -17,93 +19,95 @@
 namespace Libs::Graphics {
 
 struct GraphicContext;
-class CommandBuffer;
 class CommandScheduler;
 class TextureCache;
-class ResourceMutex;
 
-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, ResourceMutex& resource_mutex);
+	            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);
-	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);
-	[[nodiscard]] bool HasPageOverlap(uint64_t vaddr, uint64_t size);
+	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;
-	};
-	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 bool PageOverlaps(uint64_t left, uint64_t left_size, uint64_t right,
-	                                       uint64_t right_size) noexcept;
 	[[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);
+	void WriteDataBuffer(Buffer& buffer, uint64_t address, const void* source, uint64_t size);
+	void TouchBuffer(const Buffer& buffer);
+	[[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 DownloadBufferMemory(std::span<const DownloadCopy> copies);
 	void WriteHostMemory(uint64_t vaddr, std::span<const uint8_t> data);
+	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;
-	Common::Mutex                                     m_mutex;
-	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;
+	std::map<uint64_t, BufferId>                      m_buffers;
+	PageTable                                         m_page_table;
 	RangeSet                                          m_gpu_modified_ranges;
 	MemoryTracker                                     m_memory_tracker;
 	StreamBuffer                                      m_staging_buffer;
@@ -111,8 +115,7 @@ private:
 	StreamBuffer                                      m_download_buffer;
 	StreamBuffer                                      m_device_buffer;
 	TextureCache&                                     m_texture_cache;
-	ResourceMutex&                                    m_resource_mutex;
-	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 +1,167 @@
+#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, uint32_t caching_pagebits,
+                           uint64_t caching_num_pages)
+    : m_graphics(graphics), m_scheduler(scheduler), m_buffer_cache(buffer_cache),
+      m_caching_pagesize(uint64_t {1} << caching_pagebits),
+      m_caching_num_pages(caching_num_pages), m_fault_buffer_size(caching_num_pages / 8),
+      m_fault_buffer(graphics, scheduler, MemoryUsage::DeviceLocal, 0, AllFlags,
+                     m_fault_buffer_size),
+      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.sType        = vk::StructureType::eDescriptorSetLayoutCreateInfo;
+	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.sType    = vk::StructureType::eShaderModuleCreateInfo;
+	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.sType          = vk::StructureType::ePipelineLayoutCreateInfo;
+	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.sType  = vk::StructureType::ePipelineShaderStageCreateInfo;
+	stage.stage  = vk::ShaderStageFlagBits::eCompute;
+	stage.module = module;
+	stage.pName  = "main";
+	vk::ComputePipelineCreateInfo pipeline_info {};
+	pipeline_info.sType  = vk::StructureType::eComputePipelineCreateInfo;
+	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.sType         = vk::StructureType::eBufferMemoryBarrier2;
+	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].sType           = vk::StructureType::eWriteDescriptorSet;
+		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.sType                    = vk::StructureType::eDependencyInfo;
+	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    = m_caching_num_pages / 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);
+		m_fault_ranges.Clear();
+		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) {
+			m_fault_ranges.Add(faults[index], m_caching_pagesize);
+			LOGF("Accessed non-GPU cached memory at 0x%016" PRIx64 "\n", faults[index]);
+		}
+		m_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 +1,46 @@
+#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/rangeSet.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,
+	             uint32_t caching_pagebits, uint64_t caching_num_pages);
+	~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;
+	RangeSet                                   m_fault_ranges;
+	uint64_t                                   m_caching_pagesize;
+	uint64_t                                   m_caching_num_pages;
+	uint64_t                                   m_fault_buffer_size;
+	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

@@ -1,15 +1,13 @@
 #include "graphics/host_gpu/renderer/cache/gpuResourceManager.h"
 
 #include "common/assert.h"
-#include "graphics/guest_gpu/command_processor/commandProcessor.h"
 #include "graphics/guest_gpu/graphicsRun.h"
 #include "graphics/host_gpu/renderer/commandScheduler.h"
 namespace Libs::Graphics {
 
 GpuResourceManager::GpuResourceManager(GraphicContext& graphics, CommandScheduler& scheduler)
-    : m_scheduler(scheduler),
-      m_buffer_cache(graphics, scheduler, m_page_manager, m_texture_cache, m_resource_mutex),
-      m_texture_cache(graphics, scheduler, m_page_manager, m_buffer_cache, m_resource_mutex) {}
+    : m_scheduler(scheduler), m_buffer_cache(graphics, scheduler, m_page_manager, m_texture_cache),
+      m_texture_cache(graphics, scheduler, m_page_manager, m_buffer_cache) {}
 
 GpuResourceManager::~GpuResourceManager() = default;
 
@@ -18,69 +16,21 @@ bool GpuResourceManager::HandleFault(PageFaultAccess access, uint64_t fault_vadd
 	if (!IsMapped(fault_vaddr, fault_size)) {
 		return false;
 	}
-	if (CommandScheduler::InDeferredOperation()) {
-		EXIT("unsupported guest-memory fault from an asynchronous GPU completion, "
-		     "addr=0x%016" PRIx64 " access=%u\n",
-		     fault_vaddr, static_cast<uint32_t>(access));
+	if (access == PageFaultAccess::Write) {
+		m_buffer_cache.InvalidateMemory(fault_vaddr, fault_size);
+		m_texture_cache.InvalidateMemory(fault_vaddr, fault_size);
+	} else {
+		m_buffer_cache.ReadMemory(fault_vaddr, fault_size);
 	}
-	bool       handled = false;
-	const auto resolve = [this, access, fault_vaddr, &handled](CommandProcessor& cp) {
-		cp.BeginReadbackTransaction();
-		{
-			ResourceMutex::FaultScope fault(m_resource_mutex);
-			if (access == PageFaultAccess::Write) {
-				m_buffer_cache.InvalidateMemory(fault_vaddr, fault_size);
-				m_texture_cache.InvalidateMemory(fault_vaddr, fault_size);
-			} else {
-				m_buffer_cache.ReadMemory(fault_vaddr, fault_size);
-			}
-			handled = true;
-		}
-		cp.EndReadbackTransaction();
-	};
-	if (auto* cp = Gpu::CurrentCommandProcessor(); cp != nullptr) {
-		resolve(*cp);
-		return handled;
-	}
-	if (m_resource_mutex.IsOwnedByCurrentThread()) {
-		EXIT("unsupported page fault from a pre-owned resource transaction, addr=0x%016" PRIx64
-		     " access=%u\n",
-		     fault_vaddr, static_cast<uint32_t>(access));
-	}
-	EXIT_IF(m_gpu == nullptr);
-	m_gpu->SendCommandSyncWithProcessor(resolve);
-	return handled;
+	return true;
 }
 
 bool GpuResourceManager::InvalidateMemory(uint64_t vaddr, uint64_t size) {
 	if (!IsMapped(vaddr, size)) {
 		return false;
 	}
-	if (CommandScheduler::InDeferredOperation()) {
-		EXIT("unsupported memory invalidation from an asynchronous GPU completion, "
-		     "addr=0x%016" PRIx64 " size=0x%016" PRIx64 "\n",
-		     vaddr, size);
-	}
-	const auto resolve = [this, vaddr, size](CommandProcessor& cp) {
-		cp.BeginReadbackTransaction();
-		{
-			ResourceMutex::FaultScope fault(m_resource_mutex);
-			m_buffer_cache.InvalidateMemory(vaddr, size);
-			m_texture_cache.InvalidateMemory(vaddr, size);
-		}
-		cp.EndReadbackTransaction();
-	};
-	if (auto* cp = Gpu::CurrentCommandProcessor(); cp != nullptr) {
-		resolve(*cp);
-		return true;
-	}
-	if (m_resource_mutex.IsOwnedByCurrentThread()) {
-		EXIT("unsupported memory invalidation from a pre-owned resource transaction, "
-		     "addr=0x%016" PRIx64 " size=0x%016" PRIx64 "\n",
-		     vaddr, size);
-	}
-	EXIT_IF(m_gpu == nullptr);
-	m_gpu->SendCommandSyncWithProcessor(resolve);
+	m_buffer_cache.InvalidateMemory(vaddr, size);
+	m_texture_cache.InvalidateMemory(vaddr, size);
 	return true;
 }
 
@@ -107,18 +57,13 @@ void GpuResourceManager::UnmapMemory(uint64_t vaddr, uint64_t size) {
 		     "addr=0x%016" PRIx64 " size=0x%016" PRIx64 "\n",
 		     vaddr, size);
 	}
-	if (m_resource_mutex.IsOwnedByCurrentThread()) {
-		EXIT("unsupported memory unmap from a pre-owned resource transaction, "
-		     "addr=0x%016" PRIx64 " size=0x%016" PRIx64 "\n",
-		     vaddr, 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);
@@ -131,7 +76,19 @@ 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

@@ -5,7 +5,6 @@
 #include "common/common.h"
 #include "graphics/host_gpu/pageManager.h"
 #include "graphics/host_gpu/renderer/cache/bufferCache.h"
-#include "graphics/host_gpu/renderer/cache/resourceMutex.h"
 #include "graphics/host_gpu/renderer/cache/textureCache.h"
 
 #include <cstdint>
@@ -14,7 +13,7 @@
 namespace Libs::Graphics {
 
 class CommandScheduler;
-class Gpu;
+class GuestGpu;
 
 class GpuResourceManager {
 public:
@@ -24,24 +23,25 @@ public:
 
 	[[nodiscard]] BufferCache&  GetBufferCache() { return m_buffer_cache; }
 	[[nodiscard]] TextureCache& GetTextureCache() { return m_texture_cache; }
-	void                        SetGpu(Gpu* gpu) noexcept { m_gpu = gpu; }
+	void                        SetGpu(GuestGpu* gpu) noexcept { m_gpu = gpu; }
 
 	[[nodiscard]] bool HandleFault(PageFaultAccess access, uint64_t fault_vaddr) noexcept;
 	[[nodiscard]] bool InvalidateMemory(uint64_t vaddr, uint64_t size);
 	[[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:
 	PageManager               m_page_manager;
-	ResourceMutex             m_resource_mutex;
 	CommandScheduler&         m_scheduler;
 	BufferCache               m_buffer_cache;
 	TextureCache              m_texture_cache;
 	mutable std::shared_mutex m_mapped_ranges_mutex;
 	RangeSet                  m_mapped_ranges;
-	Gpu*                      m_gpu = nullptr;
+	GuestGpu*                 m_gpu = nullptr;
+	bool                      m_fault_process_pending = false;
 };
 
 } // namespace Libs::Graphics

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

@@ -1,93 +0,0 @@
-#include "graphics/host_gpu/renderer/cache/resourceMutex.h"
-
-#include "common/assert.h"
-
-namespace Libs::Graphics {
-
-ResourceMutex::FaultScope::FaultScope(ResourceMutex& mutex): m_mutex(mutex) {
-	m_resource_preowned = m_mutex.BeginFault();
-}
-
-ResourceMutex::FaultScope::~FaultScope() {
-	m_mutex.EndFault(m_resource_preowned);
-}
-
-ResourceMutex::~ResourceMutex() {
-	std::lock_guard state(m_state);
-	if (m_resource_owner != std::thread::id {} || m_fault_owner != std::thread::id {}) {
-		EXIT("ResourceMutex destroyed with an active owner or fault transaction\n");
-	}
-}
-
-void ResourceMutex::lock() {
-	const auto current = std::this_thread::get_id();
-	{
-		std::lock_guard state(m_state);
-		if (m_resource_owner == current) {
-			EXIT("recursive resource transaction\n");
-		}
-		if (m_fault_owner == current) {
-			EXIT("resource transaction re-entered from a page-fault callback\n");
-		}
-	}
-	m_resource.lock();
-	std::lock_guard state(m_state);
-	if (m_resource_owner != std::thread::id {}) {
-		EXIT("resource mutex acquired with a stale owner\n");
-	}
-	m_resource_owner = current;
-}
-
-void ResourceMutex::unlock() {
-	const auto current = std::this_thread::get_id();
-	{
-		std::lock_guard state(m_state);
-		if (m_resource_owner != current) {
-			EXIT("resource transaction released without ownership\n");
-		}
-		m_resource_owner = {};
-	}
-	m_resource.unlock();
-}
-
-bool ResourceMutex::IsOwnedByCurrentThread() {
-	std::lock_guard state(m_state);
-	return m_resource_owner == std::this_thread::get_id();
-}
-
-bool ResourceMutex::BeginFault() {
-	const auto current = std::this_thread::get_id();
-	{
-		std::lock_guard state(m_state);
-		if (m_fault_owner == current) {
-			EXIT("nested resource page-fault transaction\n");
-		}
-		if (m_resource_owner == current) {
-			m_fault_owner = current;
-			return true;
-		}
-	}
-	lock();
-	std::lock_guard state(m_state);
-	if (m_resource_owner != current || m_fault_owner != std::thread::id {}) {
-		EXIT("resource fault transaction acquired inconsistent ownership\n");
-	}
-	m_fault_owner = current;
-	return false;
-}
-
-void ResourceMutex::EndFault(bool resource_preowned) {
-	const auto current = std::this_thread::get_id();
-	{
-		std::lock_guard state(m_state);
-		if (m_fault_owner != current || m_resource_owner != current) {
-			EXIT("resource page-fault transaction ended with inconsistent ownership\n");
-		}
-		m_fault_owner = {};
-	}
-	if (!resource_preowned) {
-		unlock();
-	}
-}
-
-} // namespace Libs::Graphics

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

@@ -1,48 +0,0 @@
-#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_RESOURCEMUTEX_H_
-#define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_RESOURCEMUTEX_H_
-
-#include "common/common.h"
-
-#include <mutex>
-#include <thread>
-
-namespace Libs::Graphics {
-
-// Owner-tracked shared buffer/image transaction. External faults pause GPU submissions first;
-// command-processor faults drain pending guest processors before entering this transaction.
-class ResourceMutex final {
-public:
-	class FaultScope final {
-	public:
-		explicit FaultScope(ResourceMutex& mutex);
-		~FaultScope();
-		KYTY_CLASS_NO_COPY(FaultScope);
-
-	private:
-		ResourceMutex& m_mutex;
-		bool           m_resource_preowned = false;
-	};
-
-	ResourceMutex() = default;
-	~ResourceMutex();
-	KYTY_CLASS_NO_COPY(ResourceMutex);
-
-	void               lock();
-	void               unlock();
-	[[nodiscard]] bool IsOwnedByCurrentThread();
-
-private:
-	friend class FaultScope;
-
-	[[nodiscard]] bool BeginFault();
-	void               EndFault(bool resource_preowned);
-
-	std::mutex      m_resource;
-	std::mutex      m_state;
-	std::thread::id m_resource_owner;
-	std::thread::id m_fault_owner;
-};
-
-} // namespace Libs::Graphics
-
-#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_RESOURCEMUTEX_H_

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

@@ -73,8 +73,12 @@ Buffer::Buffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsag
 	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 {}
@@ -99,6 +103,13 @@ Buffer::Buffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsag
 	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.sType  = vk::StructureType::eBufferDeviceAddressInfo;
+		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);
@@ -122,6 +133,11 @@ vk::Buffer Buffer::Handle() const noexcept {
 	return m_buffer->buffer;
 }
 
+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;
 }
@@ -141,6 +157,15 @@ void Buffer::Flush(uint64_t offset, uint64_t size) {
 	}
 }
 
+void Buffer::Invalidate(uint64_t offset, uint64_t size) {
+	EXIT_IF(m_usage != MemoryUsage::Download || offset > m_size || size > m_size - offset);
+	if (!m_is_coherent && 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) {
@@ -310,16 +335,6 @@ void StreamBuffer::Commit() {
 	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);

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

@@ -48,12 +48,14 @@ public:
 	[[nodiscard]] bool               IsCoherent() const noexcept { return m_is_coherent; }
 	[[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               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,
@@ -65,6 +67,10 @@ 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;
+	size_t lru_id     = 0;
+
 protected:
 	[[nodiscard]] GraphicContext&   Graphics() const noexcept { return *m_graphics; }
 	[[nodiscard]] CommandScheduler& Scheduler() const noexcept { return *m_scheduler; }
@@ -80,6 +86,7 @@ private:
 	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;
@@ -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:

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

@@ -4,19 +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 <optional>
-#include <set>
-#include <span>
-#include <utility>
+#include <unordered_map>
+#include <unordered_set>
 #include <vector>
 
 namespace Libs::Graphics {
@@ -26,10 +24,7 @@ class Buffer;
 class BufferCache;
 class CommandBuffer;
 class CommandScheduler;
-class ResourceMutex;
 class RenderExecutor;
-class StreamBuffer;
-class TileManager;
 struct TextureCacheTestAccess;
 
 class TextureCache {
@@ -49,30 +44,37 @@ public:
 	};
 
 	TextureCache(GraphicContext& graphics, CommandScheduler& scheduler, PageManager& page_manager,
-	             BufferCache& buffer_cache, ResourceMutex& resource_mutex);
+	             BufferCache& buffer_cache);
 	~TextureCache();
 	KYTY_CLASS_NO_COPY(TextureCache);
 
 	[[nodiscard]] ImageId       FindImage(ImageDesc& desc, bool exact_format = false);
+	void                        UpdateImage(ImageId id);
 	[[nodiscard]] ImageId       FindImageFromRange(uint64_t address, uint64_t size,
 	                                               bool ensure_valid = true);
 	[[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);
+	void               InvalidateMemoryFromGPU(uint64_t address, uint64_t size);
 	[[nodiscard]] RegionInfo QueryRegion(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);
+	// Returns true when registered DCC absorbed the fill and the caller may skip the dispatch.
+	// False may still record PendingDcc state, but the guest dispatch must execute.
+	[[nodiscard]] bool TryConsumeDccFill(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);
@@ -84,13 +86,18 @@ private:
 	struct ColorTransferPlan;
 	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 FindRenderTarget 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 {
@@ -102,25 +109,21 @@ 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 DeleteImages(std::span<const ImageId> ids, std::optional<ImageId> native_source = {});
-	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 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);
@@ -149,32 +152,24 @@ private:
 	[[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;
-	ResourceMutex&                                    m_resource_mutex;
-	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

@@ -9,20 +9,42 @@
 #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 void ResolveDccClearInfo(RenderColorInfo& info, vk::Format format, bool has_dcc,
+                                uint32_t packed_clear) {
+	// Register-backed DCC clears use the target's packed clear value. Decode one-word guest
+	// formats here; unsupported encodings remain tracked without unsafe materialization.
+	info.metadata_clear_supported =
+	    has_dcc && DecodePackedColorClear(format, packed_clear, info.color_clear_value);
+	switch (format) {
+		case vk::Format::eR8G8B8A8Unorm:
+		case vk::Format::eR8G8B8A8Srgb:
+		case vk::Format::eB8G8R8A8Unorm:
+		case vk::Format::eB8G8R8A8Srgb:
+		case vk::Format::eA2B10G10R10UnormPack32:
+		case vk::Format::eA2R10G10B10UnormPack32:
+			info.metadata_fixed_clear_supported = has_dcc;
+			break;
+		default: info.metadata_fixed_clear_supported = false; break;
+	}
+	if (!info.metadata_clear_supported) {
+		info.color_clear_value = {};
+	}
+}
+
 // 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,
@@ -55,20 +77,22 @@ 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  = {};
+		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.metadata_clear_supported       = false;
+		r.metadata_fixed_clear_supported = false;
+		r.color_clear_value              = {};
 		return;
 	}
 	const auto samples = render_sample_count(rt.attrib.num_fragments);
@@ -101,14 +125,15 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 			     " attrib3_tile=0x%08" PRIx32 " attrib3_dim=0x%08" PRIx32 " fmt=0x%08" PRIx32
 			     " nfmt=0x%08" PRIx32 " order=0x%08" PRIx32 "\n",
 			     rt_slot, rt.base.addr, mask, rt.attrib2.width, rt.attrib2.height,
-			     rt.attrib3.tile_mode, rt.attrib3.dimension, rt.info.format, rt.info.channel_type,
-			     rt.info.channel_order);
+			     static_cast<uint32_t>(rt.attrib3.tile_mode), rt.attrib3.dimension,
+			     static_cast<uint32_t>(rt.info.format), static_cast<uint32_t>(rt.info.channel_type),
+			     static_cast<uint32_t>(rt.info.channel_order));
 		}
 	}
 
-	// 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;
@@ -119,27 +144,42 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	uint32_t   pitch  = 0;
 	uint64_t   size   = 0;
 	bool       tile   = false;
-	const bool volume = rt.attrib3.dimension == 2;
-	if (rt.attrib3.dimension != 1 && !volume) {
+	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("2D render target has nonzero depth: %u\n", rt.attrib3.depth);
+		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;
-	const bool     standard64 =
-	    rt.attrib3.tile_mode == Prospero::GpuEnumValue(Prospero::TileMode::kStandard64KB);
+	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;
+	const bool     texture_tile = standard4 || standard64 || depth_tile;
 
 	switch (rt.attrib3.tile_mode) {
-		case Prospero::GpuEnumValue(Prospero::TileMode::kLinear):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kStandard64KB):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kRenderTarget):
+		case Prospero::TileMode::kLinear:
+		case Prospero::TileMode::kStandard4KB:
+		case Prospero::TileMode::kStandard64KB:
+		case Prospero::TileMode::kDepth:
+		case Prospero::TileMode::kRenderTarget:
 			tile = !RenderIsColorTileModeLinear(rt.attrib3.tile_mode);
 			break;
-		default: EXIT("unknown tile mode: %u\n", rt.attrib3.tile_mode);
+		default: EXIT("unknown tile mode: %u\n", static_cast<uint32_t>(rt.attrib3.tile_mode));
 	}
 	if (!tile && levels > 1) {
 		EXIT("linear mipmapped render targets are unsupported\n");
@@ -147,6 +187,9 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	if (samples > 1 && (!tile || levels != 1)) {
 		EXIT("multisampled render targets require a single-mip tiled surface\n");
 	}
+	if (texture_tile && samples != 1) {
+		EXIT("texture-tiled color render targets do not support multisampling\n");
+	}
 
 	width  = rt.attrib2.width + 1;
 	height = rt.attrib2.height + 1;
@@ -157,31 +200,38 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		EXIT("render-target format has no valid element size\n");
 	}
 	const auto transfer_format = ImageOps::RenderTargetTransferFormat(bytes_per_element);
-	if (standard64 &&
-	    (rt.attrib3.dimension != 1 || rt.attrib3.depth != 0 || levels != 1 ||
-	     rt.view.current_mip_level != 0 || view.base_layer != 0 || view.image_layers != 1 ||
-	     samples != 1 || bytes_per_element != 4 || rt.pitch.pitch_div8_minus1 != 0 ||
-	     (rt.base.addr & 0xffffu) != 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)) {
-		EXIT("unsupported Standard64KB render target: addr=0x%016" PRIx64
+	TileTextureBlockLayout texture_tile_layout {};
+	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)) {
+		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",
-		     rt.base.addr, rt.attrib3.dimension, rt.attrib3.depth, levels, view.base_layer,
-		     view.image_layers, rt.attrib.num_samples, rt.attrib.num_fragments, bytes_per_element,
-		     rt.cmask.addr, rt.fmask.addr, rt.dcc_addr.addr);
+		     rt.base.addr, static_cast<uint32_t>(rt.attrib3.tile_mode), rt.attrib3.dimension,
+		     rt.attrib3.depth, levels, view.base_layer, view.image_layers, rt.attrib.num_samples,
+		     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)) {
+		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, levels, rt.attrib3.tile_mode);
-		} else if (standard64) {
-			pitch = TileGetTexturePitch(Prospero::GpuEnumValue(Prospero::BufferFormat::k32Float),
-			                            width, levels, rt.attrib3.tile_mode);
+			pitch = TileGetTexturePitch(transfer_format, width, rt.attrib3.tile_mode);
+		} else if (texture_tile) {
+			pitch = TileGetTexturePitch(transfer_format, width, rt.attrib3.tile_mode);
 		} else {
 			pitch = TileGetRenderTargetPitch(width, bytes_per_element, rt.attrib.num_fragments);
 		}
@@ -192,26 +242,32 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		pitch = width;
 	}
 
-	TileSizeOffset   mip_sizes[16] {};
-	TilePaddedSize   mip_padded[16] {};
-	TileVolumeLayout volume_layout {};
-	uint64_t         backing_size = 0;
+	TileSizeOffset    mip_sizes[16] {};
+	TilePaddedSize    mip_padded[16] {};
+	TileSurfaceLayout volume_layout {};
+	uint64_t          backing_size = 0;
 	if (volume) {
-		if (!tile || !TileGetTextureVolumeLayout(transfer_format, width, height, depth, levels,
-		                                         rt.attrib3.tile_mode, volume_layout)) {
+		const TileSurfaceDescription description {transfer_format,
+		                                          rt.attrib3.tile_mode,
+		                                          TileSurfaceDimension::Dim3D,
+		                                          width,
+		                                          height,
+		                                          depth,
+		                                          levels,
+		                                          1};
+		if (!tile || !TileGetTiledTextureLayout(description, volume_layout)) {
 			EXIT("unsupported 3D render-target layout: %ux%ux%u levels=%u tile=%u\n", width, height,
-			     depth, levels, rt.attrib3.tile_mode);
+			     depth, levels, static_cast<uint32_t>(rt.attrib3.tile_mode));
 		}
 		size         = volume_layout.block_slice_size;
 		backing_size = volume_layout.total_size;
 	} else if (tile) {
 		TileSizeAlign layout {};
 		bool          valid_layout = false;
-		if (standard64) {
-			TileGetTextureSize(Prospero::GpuEnumValue(Prospero::BufferFormat::k32Float), width,
-			                   height, pitch, levels, rt.attrib3.tile_mode, &layout, mip_sizes,
-			                   mip_padded);
-			valid_layout = layout.size != 0 && layout.align == 65536;
+		if (texture_tile) {
+			TileGetTextureSize(transfer_format, width, height, levels, rt.attrib3.tile_mode,
+			                   &layout, mip_sizes, mip_padded);
+			valid_layout = layout.size != 0 && layout.align == texture_tile_layout.block.block_size;
 		} else {
 			valid_layout =
 			    levels == 1 ? TileGetRenderTargetSize(width, height, pitch, bytes_per_element,
@@ -271,8 +327,9 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 		     " extent=%ux%ux%u view_mip=%u view_extent=%ux%u levels=%u pitch=%u"
 		     " fmt=0x%08" PRIx32 " nfmt=0x%08" PRIx32 " order=0x%08" PRIx32 " samples=%u tile=%s\n",
 		     rt_slot, rt.base.addr, backing_size, width, height, depth, rt.view.current_mip_level,
-		     view_extent.width, view_extent.height, levels, pitch, rt.info.format,
-		     rt.info.channel_type, rt.info.channel_order, samples, tile ? "tiled" : "linear");
+		     view_extent.width, view_extent.height, levels, pitch,
+		     static_cast<uint32_t>(rt.info.format), static_cast<uint32_t>(rt.info.channel_type),
+		     static_cast<uint32_t>(rt.info.channel_order), samples, tile ? "tiled" : "linear");
 	}
 
 	TextureCache::ImageDesc desc {};
@@ -280,20 +337,28 @@ 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) {
+		// DCC uses a separate metadata allocation. Carry its address through ImageInfo so
+		// TextureCache can associate metadata fills observed before target registration.
+		desc.info.metadata.kind          = ImageMetadataKind::Dcc;
+		desc.info.metadata.range.address = rt.dcc_addr.addr;
+	}
 	for (uint32_t level = 0; level < levels; level++) {
 		if (volume) {
+			const auto& mip             = volume_layout.mips[level];
 			desc.info.mip_layout[level] = {
-			    volume_layout.level_offsets[level],
-			    volume_layout.level_sizes[level],
-			    volume_layout.level_widths[level],
-			    volume_layout.level_heights[level],
+			    mip.offset,
+			    mip.size,
+			    mip.padded_width,
+			    mip.padded_height,
 			};
 			continue;
 		}
@@ -311,8 +376,13 @@ 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;
@@ -332,7 +402,7 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	r.samples                  = samples;
 	r.export_mapping           = target_format.export_mapping;
 	r.color_clear_enable       = false;
-	r.color_clear_value        = {};
+	ResolveDccClearInfo(r, target_format.format, has_dcc, rt.clear_word0.word0);
 	BindRenderTarget(r.image_id);
 }
 

src/graphics/host_gpu/renderer/colorRenderTarget.h

@@ -10,8 +10,6 @@
 
 namespace Libs::Graphics {
 
-class RenderCommandBuffer;
-
 enum class RenderColorType {
 	NoColorOutput,
 	RenderTexture,
@@ -31,7 +29,9 @@ struct RenderColorInfo {
 	uint32_t                        target_slot      = 0;
 	uint32_t                        samples          = 1;
 	Prospero::ColorComponentMapping export_mapping;
-	bool                            color_clear_enable = false;
+	bool                            color_clear_enable             = false;
+	bool                            metadata_clear_supported       = false;
+	bool                            metadata_fixed_clear_supported = false;
 	vk::ClearColorValue             color_clear_value {};
 };
 

src/graphics/host_gpu/renderer/commandScheduler.cpp

@@ -1,96 +1,97 @@
 #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, VulkanToString(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, VulkanToString(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;
+	if (m_hint == m_ticks.size()) {
+		m_hint = 0;
+	}
+	return m_buffers[*found];
 }
 
 bool CommandScheduler::InDeferredOperation() noexcept {
@@ -99,6 +100,7 @@ bool CommandScheduler::InDeferredOperation() noexcept {
 
 CommandScheduler::CommandScheduler(RenderContext& context, GraphicContext& graphics)
     : m_master(graphics), m_context(context), m_graphics(graphics),
+      m_command_pool(graphics, m_master), m_command(*this),
       m_priority_thread([this](std::stop_token stop) { PriorityOperationsThread(stop); }) {}
 
 CommandScheduler::~CommandScheduler() {
@@ -125,9 +127,11 @@ 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();
@@ -153,17 +157,10 @@ void CommandScheduler::Begin(HW::Context& registers, HW::UserConfig& user_config
 	m_user_config = &user_config;
 	m_shaders     = &shaders;
 
-	if (!Active()) {
-		for (auto& buffer: m_buffers) {
-			buffer = std::make_unique<RenderCommandBuffer>(*this);
-		}
-		m_current = 0;
-	}
-
-	BindCurrent();
-	if (!m_recording) {
-		Current().Begin();
-		m_recording = true;
+	if (m_command.IsInvalid()) {
+		BeginNext();
+	} else {
+		BindCurrent();
 	}
 }
 
@@ -172,7 +169,7 @@ void CommandScheduler::BeginRendering(const RenderState& state) {
 }
 
 void CommandScheduler::EndRendering() {
-	if (Active() && m_recording) {
+	if (Active() && !m_command.IsInvalid()) {
 		Current().EndRendering();
 	}
 }
@@ -183,71 +180,63 @@ 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() {
-	m_master.Refresh();
+	PopPendingOperations(true);
+}
+
+void CommandScheduler::PopPendingOperations(bool refresh_gpu_tick) {
+	if (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));
 	}
 }
 
@@ -353,49 +342,81 @@ 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() || m_current >= BufferCount);
+	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 {
+void CommandScheduler::BindCurrent() {
 	EXIT_IF(m_registers == nullptr || m_user_config == nullptr || m_shaders == nullptr);
-	Current().Bind(*m_registers, *m_user_config, *m_shaders);
+	m_command.Bind(*m_registers, *m_user_config, *m_shaders);
 }
 
-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_recording = false;
-	return submitted;
+CommandBuffer& CommandScheduler::BeginCommand() {
+	EXIT_IF(!m_command.IsInvalid());
+	m_command.m_buffer = m_command_pool.Commit();
+	m_command.Begin();
+	return m_command;
+}
+
+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);
+
+	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.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();
+
+		result = graphics.queue.submit(1, &submit_info, nullptr);
+	}
+
+	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);
+	}
+	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
+
+	m_command.m_buffer = nullptr;
+	return tick;
 }
 
 void CommandScheduler::BeginNext() {
-	EXIT_IF(m_recording);
-	m_current = (m_current + 1) % BufferCount;
-	Current().WaitForFenceAndReset();
-	PopPendingOperations();
+	EXIT_IF(!m_command.IsInvalid());
 	BindCurrent();
-	Current().Begin();
-	m_recording = true;
+	BeginCommand();
 }
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/renderer/commandScheduler.h

@@ -6,33 +6,18 @@
 #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>
 
 #include <thread>
+#include <vector>
 
 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:
-	static constexpr int BufferCount = 8;
-
 	CommandScheduler(RenderContext& context, GraphicContext& graphics);
 	~CommandScheduler();
 	KYTY_CLASS_NO_COPY(CommandScheduler);
@@ -42,9 +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();
@@ -56,32 +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_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:
 	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 };
@@ -91,35 +80,28 @@ private:
 		uint64_t                     tick = 0;
 	};
 
-	void                       BindCurrent() const;
-	CommandBuffer&             SubmitCurrent(SubmitInfo& submit);
-	void                       BeginNext();
-	void                       PriorityOperationsThread(std::stop_token stop);
-	void                       RunOperation(Common::UniqueFunction<void>&& operation);
-	[[nodiscard]] CommandSlot* AllocateCommandBuffer();
-	[[nodiscard]] uint64_t     NextSubmitSequence() noexcept;
+	void BindCurrent();
+	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::array<std::unique_ptr<RenderCommandBuffer>, BufferCount> m_buffers;
-	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;
+	HW::Context*                 m_registers            = nullptr;
+	HW::UserConfig*              m_user_config          = nullptr;
+	HW::Shader*                  m_shaders              = nullptr;
 };
 
 } // 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"
@@ -18,98 +15,28 @@
 #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.flags            = vk::CommandBufferUsageFlagBits::eOneTimeSubmit;
 	begin_info.pInheritanceInfo = nullptr;
 
 	auto result = buffer.begin(&begin_info);
@@ -137,125 +64,6 @@ 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);

src/graphics/host_gpu/renderer/debug.cpp

@@ -10,6 +10,7 @@
 #include "graphics/host_gpu/renderer/render.h"
 
 #include <algorithm>
+#include <array>
 #include <atomic>
 #include <cmath>
 #include <fmt/format.h>
@@ -26,7 +27,7 @@ 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();
 
@@ -44,7 +45,7 @@ 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++) {
@@ -70,14 +71,16 @@ void uc_print(const char* func, const HW::UserConfig& uc) {
 	LOGF("\t GetPrimType()         = 0x%08" PRIx32 "\n"
 	     "\t GetIndexOffset()      = 0x%08" PRIx32 "\n"
 	     "\t GetObjectId()         = 0x%08" PRIx32 "\n"
+	     "\t primitive_reset       = 0x%08" PRIx32 "\n"
 	     "\t primitive_group_size  = 0x%04" PRIx16 "\n"
 	     "\t vertex_group_size     = 0x%04" PRIx16 "\n"
 	     "\t en_user_vgpr1         = %s\n"
 	     "\t en_user_vgpr2         = %s\n"
 	     "\t en_user_vgpr3         = %s\n",
-	     uc.GetPrimType(), uc.GetIndexOffset(), uc.GetObjectId(), ge_cntl.primitive_group_size,
-	     ge_cntl.vertex_group_size, user_en.vgpr1 ? "true" : "false",
-	     user_en.vgpr2 ? "true" : "false", user_en.vgpr3 ? "true" : "false");
+	     static_cast<uint32_t>(uc.GetPrimType()), uc.GetIndexOffset(), uc.GetObjectId(),
+	     uc.GetPrimitiveResetControl(), ge_cntl.primitive_group_size, ge_cntl.vertex_group_size,
+	     user_en.vgpr1 ? "true" : "false", user_en.vgpr2 ? "true" : "false",
+	     user_en.vgpr3 ? "true" : "false");
 }
 
 void uc_check(const HW::UserConfig& uc) {
@@ -140,11 +143,12 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	    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", rt.info.format));
-	dst.push_back(
-	    fmt::format("\t info.channel_type               = 0x{:08x}\n", rt.info.channel_type));
-	dst.push_back(
-	    fmt::format("\t info.channel_order              = 0x{:08x}\n", rt.info.channel_order));
+	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",
+	                          static_cast<uint32_t>(rt.info.channel_type)));
+	dst.push_back(fmt::format("\t info.channel_order              = 0x{:08x}\n",
+	                          static_cast<uint32_t>(rt.info.channel_order)));
 	dst.push_back(fmt::format("\t info.blend_bypa                 = {}\n",
 	                          rt.info.blend_bypass ? "true" : "false"));
 	dst.push_back(fmt::format("\t info.blend_clamp                = {}\n",
@@ -153,10 +157,10 @@ 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", rt.attrib.tile_mode));
-	dst.push_back(
-	    fmt::format("\t attrib.fmask_tile_mode          = 0x{:08x}\n", rt.attrib.fmask_tile_mode));
+	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(
@@ -167,8 +171,8 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	dst.push_back(
 	    fmt::format("\t attrib2.num_mip_levels          = 0x{:08x}\n", rt.attrib2.num_mip_levels));
 	dst.push_back(fmt::format("\t attrib3.depth                   = 0x{:08x}\n", rt.attrib3.depth));
-	dst.push_back(
-	    fmt::format("\t attrib3.tile_mode               = 0x{:08x}\n", rt.attrib3.tile_mode));
+	dst.push_back(fmt::format("\t attrib3.tile_mode               = 0x{:08x}\n",
+	                          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",
@@ -211,22 +215,22 @@ std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt)
 	return dst;
 }
 
-static bool RenderIsColorTileMode(uint32_t tile_mode) {
+static bool RenderIsColorTileMode(Prospero::TileMode tile_mode) {
 	// AGC CxRenderTarget::TileMode shifted by CB_COLOR*_ATTRIB3.COLOR_SW_MODE.
 	switch (tile_mode) {
-		case Prospero::GpuEnumValue(Prospero::TileMode::kLinear):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kStandard256B):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kStandard4KB):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kStandard64KB):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kPrt):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kDepth):
-		case Prospero::GpuEnumValue(Prospero::TileMode::kRenderTarget): return true;
+		case Prospero::TileMode::kLinear:
+		case Prospero::TileMode::kStandard256B:
+		case Prospero::TileMode::kStandard4KB:
+		case Prospero::TileMode::kStandard64KB:
+		case Prospero::TileMode::kPrt:
+		case Prospero::TileMode::kDepth:
+		case Prospero::TileMode::kRenderTarget: return true;
 		default: return false;
 	}
 }
 
-bool RenderIsColorTileModeLinear(uint32_t tile_mode) {
-	return tile_mode == Prospero::GpuEnumValue(Prospero::TileMode::kLinear);
+bool RenderIsColorTileModeLinear(Prospero::TileMode tile_mode) {
+	return tile_mode == Prospero::TileMode::kLinear;
 }
 
 static bool RenderIsColorDimension(uint32_t dimension) {
@@ -262,20 +266,23 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
+		const bool native_2x_one_frag =
+		    rt.info.fmask_compression_enable && !rt.info.fmask_data_compression_disable &&
+		    rt.info.cmask_fast_clear_enable && rt.fmask.addr != 0 && rt.cmask.addr != 0 &&
+		    rt.attrib.num_samples == 1 && rt.attrib.num_fragments == 1;
+		EXIT_NOT_IMPLEMENTED(rt.info.fmask_one_frag_mode && !native_2x_one_frag);
 		if (rt.info.fmask_compression_enable) {
 			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;
@@ -360,7 +367,8 @@ static void RtCheck(const HW::RenderTarget& rt) {
 			}
 		}
 		if (!RenderIsColorTileMode(rt.attrib3.tile_mode)) {
-			EXIT("unknown PS5 render-target tile mode: 0x%08" PRIx32 "\n", rt.attrib3.tile_mode);
+			EXIT("unknown PS5 render-target tile mode: 0x%08" PRIx32 "\n",
+			     static_cast<uint32_t>(rt.attrib3.tile_mode));
 		}
 		if (!RenderIsColorDimension(rt.attrib3.dimension)) {
 			EXIT("unknown PS5 render-target dimension: 0x%08" PRIx32 "\n", rt.attrib3.dimension);
@@ -455,13 +463,14 @@ static void ZPrint(const char* func, const HW::DepthRenderTarget& z) {
 	     "\t height                                = 0x%08" PRIx32 "\n"
 	     "\t size.x_max                            = 0x%04" PRIx16 "\n"
 	     "\t size.y_max                            = 0x%04" PRIx16 "\n",
-	     z.z_info.format, z.z_info.num_samples,
-	     Prospero::GpuEnumValue(z.z_info.texture_compatibility),
+	     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",
 	     z.z_info.expclear_enabled ? "true" : "false",
-	     Prospero::GpuEnumValue(z.z_info.z_compare_base),
+	     static_cast<uint32_t>(z.z_info.z_compare_base),
 	     z.z_info.partially_resident ? "true" : "false", z.z_info.max_mip_level,
-	     z.stencil_info.format, Prospero::GpuEnumValue(z.stencil_info.texture_compatibility),
+	     static_cast<uint32_t>(z.stencil_info.format),
+	     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,
@@ -490,15 +499,16 @@ static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
 
 	EXIT_NOT_IMPLEMENTED(!z.z_info.HasValidTextureCompatibility());
 	EXIT_NOT_IMPLEMENTED(!z.stencil_info.HasValidTextureCompatibility());
-	if (z.z_info.format == 0) {
-		EXIT_NOT_IMPLEMENTED(z.z_info.format != 0);
+	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 != 0x00000001 && z.z_info.format != 0x00000003);
+		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) {
@@ -512,16 +522,17 @@ static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
 		EXIT_NOT_IMPLEMENTED(z.z_info.max_mip_level != 0);
 	}
 
-	if (z.stencil_info.format == 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 != 0x00000001);
+		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 != 0 || z.stencil_info.format != 0) {
+	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);
@@ -553,13 +564,11 @@ static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
 				     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 != 0 && 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.depth_view.depth_write_disable &&
+		                     z.z_read_base_addr != z.z_write_base_addr);
+		EXIT_NOT_IMPLEMENTED(!z.depth_view.stencil_write_disable &&
+		                     z.stencil_read_base_addr != z.stencil_write_base_addr);
+		EXIT_NOT_IMPLEMENTED(!z.depth_view.depth_write_disable && z.z_write_base_addr == 0);
 		// EXIT_NOT_IMPLEMENTED(z.pitch_div8_minus1 != 0x000000ff);
 		// EXIT_NOT_IMPLEMENTED(z.height_div8_minus1 != 0x0000008f);
 		// EXIT_NOT_IMPLEMENTED(z.slice_div64_minus1 != 0x00008fff);
@@ -608,8 +617,7 @@ 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.IsZClipModeRepresentable() ||
-	                     c.user_clip_plane_negate_y || c.clip_disable ||
+	                     c.vertex_kill_any || c.user_clip_plane_negate_y || c.clip_disable ||
 	                     c.user_clip_plane_cull_only || c.cull_on_clipping_error_disable ||
 	                     c.force_viewport_index_from_vs_enable);
 }
@@ -633,17 +641,6 @@ static void RcPrint(const char* func, const HW::RenderControl& c) {
 static void RcCheck(const HW::RenderControl& c) {
 	// EXIT_NOT_IMPLEMENTED(c.depth_clear_enable != false);
 	// EXIT_NOT_IMPLEMENTED(c.stencil_clear_enable != false);
-	if ((c.resummarize_enable || c.copy_centroid || c.copy_sample != 0)) {
-		static std::atomic<uint32_t> log_count {0};
-		if (log_count.fetch_add(1) < 16) {
-			LOGF("\t warning: unsupported PS5 render-control metadata state, continuing: "
-			     "resummarize=%s copy_centroid=%s copy_sample=%" PRIu8 "\n",
-			     c.resummarize_enable ? "true" : "false", c.copy_centroid ? "true" : "false",
-			     c.copy_sample);
-		}
-		return;
-	}
-	EXIT_NOT_IMPLEMENTED(c.resummarize_enable != false);
 	// EXIT_NOT_IMPLEMENTED(c.stencil_compress_disable != false);
 	// EXIT_NOT_IMPLEMENTED(c.depth_compress_disable != false);
 	EXIT_NOT_IMPLEMENTED(c.copy_centroid != false);
@@ -686,14 +683,6 @@ static void McCheck(const HW::ModeControl& c) {
 	}
 	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.poly_offset_front_enable || c.poly_offset_back_enable) {
-		static bool logged = false;
-		if (!logged) {
-			LOGF("\t temporary: PA_SU_SC_MODE_CNTL.POLY_OFFSET_*_ENABLE is not implemented; "
-			     "continuing without depth bias\n");
-			logged = true;
-		}
-	}
 	if (c.vtx_window_offset_enable) {
 		static bool logged = false;
 		if (!logged) {
@@ -899,12 +888,10 @@ 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);
 	}
 }
 
@@ -1067,18 +1054,24 @@ static ScissorRect ScissorRectClamp(ScissorRect r, uint32_t width, uint32_t heig
 	return r;
 }
 
+static constexpr std::array<uint16_t, 16> MakeScissorIntersectionRules() {
+	std::array<uint16_t, 16> rules {};
+	for (uint32_t candidate = 0; candidate < rules.size(); candidate++) {
+		for (uint32_t combination = 0; combination < 16; combination++) {
+			if ((combination & candidate) == candidate) {
+				rules[candidate] |= static_cast<uint16_t>(1u << combination);
+			}
+		}
+	}
+	return rules;
+}
+
 static bool ScissorClipRuleToIntersectionMask(uint16_t rule, uint8_t* mask) {
 	EXIT_IF(mask == nullptr);
 
-	for (uint32_t candidate = 0; candidate < 16; candidate++) {
-		uint16_t expected = 0;
-		for (uint32_t combination = 0; combination < 16; combination++) {
-			if ((combination & candidate) == candidate) {
-				expected |= static_cast<uint16_t>(1u << combination);
-			}
-		}
-
-		if (expected == rule) {
+	static constexpr auto rules = MakeScissorIntersectionRules();
+	for (uint32_t candidate = 0; candidate < rules.size(); candidate++) {
+		if (rules[candidate] == rule) {
 			*mask = static_cast<uint8_t>(candidate);
 			return true;
 		}
@@ -1162,7 +1155,7 @@ 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);
@@ -1219,8 +1212,9 @@ void hw_check(const RenderCommandBuffer& buffer) {
 			const auto& mrt = hw.GetRenderTarget(i);
 			LOGF("\t RT%u addr=0x%010" PRIx64 " mask=0x%x fmt=0x%08" PRIx32
 			     " width=%u height=%u tile=%u\n",
-			     i, mrt.base.addr, (hw.GetRenderTargetMask() >> (i * 4u)) & 0x0fu, mrt.info.format,
-			     mrt.attrib2.width + 1, mrt.attrib2.height + 1, mrt.attrib3.tile_mode);
+			     i, mrt.base.addr, (hw.GetRenderTargetMask() >> (i * 4u)) & 0x0fu,
+			     static_cast<uint32_t>(mrt.info.format), mrt.attrib2.width + 1,
+			     mrt.attrib2.height + 1, static_cast<uint32_t>(mrt.attrib3.tile_mode));
 		}
 	}
 	if (rc.depth_clear_enable && hw.GetDepthClearValue() != 0.0f &&
@@ -1234,7 +1228,7 @@ 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);
@@ -1259,9 +1253,10 @@ void hw_print(const RenderCommandBuffer& buffer) {
 		     "\t GetRenderTargetMask()   = 0x%08" PRIx32 "\n"
 		     "\t GetDepthClearValue()    = %f\n"
 		     "\t GetStencilClearValue()  = %" PRIu8 "\n"
-		     "\t GetLineWidth()          = %f\n",
+		     "\t GetLineWidth()          = %f\n"
+		     "\t primitive_reset_index   = 0x%08" PRIx32 "\n",
 		     hw.GetRenderTargetMask(), hw.GetDepthClearValue(), hw.GetStencilClearValue(),
-		     hw.GetLineWidth());
+		     hw.GetLineWidth(), hw.GetPrimitiveResetIndex());
 
 		LOGF("%s", Common::Concat(rt_print("RenderTraget:", rt), "").c_str());
 

src/graphics/host_gpu/renderer/debug.h

@@ -9,7 +9,7 @@
 
 namespace Libs::Graphics {
 
-class RenderCommandBuffer;
+class CommandBuffer;
 
 namespace HW {
 class Context;
@@ -28,15 +28,15 @@ 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(uint32_t tile_mode);
-void                     hw_print(const RenderCommandBuffer& buffer);
-void                     hw_check(const RenderCommandBuffer& buffer);
+bool                     RenderIsColorTileModeLinear(Prospero::TileMode tile_mode);
+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);

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"
@@ -38,7 +37,10 @@ namespace Libs::Graphics {
 	EXIT("unsupported render state; details were printed above\n");
 }
 
-static vk::StencilOp ConvertStencilOp(uint8_t value) {
+static vk::StencilOp ConvertStencilOp(uint8_t value, uint8_t write_mask, uint8_t op_value) {
+	if (write_mask == 0) {
+		return vk::StencilOp::eKeep;
+	}
 	switch (static_cast<Prospero::StencilOp>(value)) {
 		case Prospero::StencilOp::kKeep: return vk::StencilOp::eKeep;
 		case Prospero::StencilOp::kZero: return vk::StencilOp::eZero;
@@ -49,18 +51,27 @@ static vk::StencilOp ConvertStencilOp(uint8_t value) {
 		case Prospero::StencilOp::kInvert: return vk::StencilOp::eInvert;
 		case Prospero::StencilOp::kAddWrap: return vk::StencilOp::eIncrementAndWrap;
 		case Prospero::StencilOp::kSubWrap: return vk::StencilOp::eDecrementAndWrap;
-		default: DepthFatal("unsupported stencil operation");
+		case Prospero::StencilOp::kXor:
+			if ((write_mask & op_value) == 0) {
+				return vk::StencilOp::eKeep;
+			}
+			if ((write_mask & ~op_value) != 0) {
+				DepthFatal("unsupported stencil XOR operands: write mask=0x%02" PRIx8
+				           ", operation value=0x%02" PRIx8,
+				           write_mask, op_value);
+			}
+			return vk::StencilOp::eInvert;
+		default: DepthFatal("unsupported stencil operation: 0x%02" PRIx8, value);
 	}
 }
 
 static bool UsesStencilOpValue(uint8_t fail, uint8_t pass, uint8_t depth_fail) {
-	return fail == Prospero::GpuEnumValue(Prospero::StencilOp::kReplaceOp) ||
-	       pass == Prospero::GpuEnumValue(Prospero::StencilOp::kReplaceOp) ||
-	       depth_fail == Prospero::GpuEnumValue(Prospero::StencilOp::kReplaceOp);
+	constexpr auto replace_op = static_cast<uint8_t>(Prospero::StencilOp::kReplaceOp);
+	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();
@@ -86,19 +97,18 @@ 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;
-	const auto& hw = buffer.GetRegisters();
-	const auto& z  = hw.GetDepthRenderTarget();
-	const auto& rc = hw.GetRenderControl();
-	const auto& dc = hw.GetDepthControl();
-	const auto& sc = hw.GetStencilControl();
-	const auto& sm = hw.GetStencilMask();
-	const bool  has_stencil =
-	    z.stencil_info.format != Prospero::GpuEnumValue(Prospero::StencilFormat::kInvalid);
-	const bool depth_active =
+	const auto& hw          = buffer.GetRegisters();
+	const auto& z           = hw.GetDepthRenderTarget();
+	const auto& rc          = hw.GetRenderControl();
+	const auto& dc          = hw.GetDepthControl();
+	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);
 	if (!depth_active && !stencil_active) {
@@ -109,9 +119,8 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 		DepthFatal("invalid PS5 depth texture-compatibility encoding");
 	}
 	const bool attachment_unbound =
-	    z.z_info.format == Prospero::GpuEnumValue(Prospero::DepthFormat::kInvalid) &&
-	    z.stencil_info.format == Prospero::GpuEnumValue(Prospero::StencilFormat::kInvalid) &&
-	    z.z_info.num_samples == 0 &&
+	    z.z_info.format == Prospero::DepthFormat::kInvalid &&
+	    z.stencil_info.format == Prospero::StencilFormat::kInvalid && z.z_info.num_samples == 0 &&
 	    z.z_info.texture_compatibility == Prospero::TextureCompatiblePlaneCompression::kDisable &&
 	    !z.z_info.expclear_enabled && !z.z_info.partially_resident && z.z_info.max_mip_level == 0 &&
 	    z.stencil_info.texture_compatibility == Prospero::TextureCompatibleStencil::kDisable &&
@@ -156,27 +165,28 @@ 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.resummarize_enable || 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 ||
+	if (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.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::GpuEnumValue(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) {
+		if (z.stencil_info.format != Prospero::StencilFormat::k8UInt || !htile_stencil_compat ||
+		    z.stencil_read_base_addr == 0 ||
+		    (!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 ||
@@ -192,8 +202,6 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 		if (z.depth_view.slice_max >= 32) {
 			DepthFatal("HTile clear tracking supports at most 32 slices");
 		}
-	} else if (z.htile_data_base_addr != 0) {
-		DepthFatal("HTile address without an enabled tile surface");
 	}
 	const bool size_xy_valid = z.size.valid;
 	const bool wh_valid      = z.width_height_valid && z.width != 0 && z.height != 0;
@@ -218,7 +226,7 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 		DepthFatal("no host depth/stencil format supports required usage for %s",
 		           VulkanToString(ideal_format).c_str());
 	}
-	const uint32_t guest_format = Prospero::GpuEnumValue(policy->guest_format);
+	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 ||
@@ -279,34 +287,39 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	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 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");
 		}
 		r.stencil_static_front = {
-		    ConvertStencilOp(sc.stencil_fail), ConvertStencilOp(sc.stencil_zpass),
-		    ConvertStencilOp(sc.stencil_zfail), static_cast<vk::CompareOp>(dc.stencilfunc)};
-		r.stencil_dynamic_front = {sm.stencil_mask,
-		                           rc.stencil_clear_enable ? 0u : sm.stencil_writemask,
-		                           sm.stencil_testval};
+		    ConvertStencilOp(sc.stencil_fail, front_write_mask, sm.stencil_opval),
+		    ConvertStencilOp(sc.stencil_zpass, front_write_mask, sm.stencil_opval),
+		    ConvertStencilOp(sc.stencil_zfail, front_write_mask, sm.stencil_opval),
+		    static_cast<vk::CompareOp>(dc.stencilfunc)};
+		r.stencil_dynamic_front = {sm.stencil_mask, front_write_mask, sm.stencil_testval};
 		if (dc.backface_enable) {
-			r.stencil_static_back  = {ConvertStencilOp(sc.stencil_fail_bf),
-			                          ConvertStencilOp(sc.stencil_zpass_bf),
-			                          ConvertStencilOp(sc.stencil_zfail_bf),
-			                          static_cast<vk::CompareOp>(dc.stencilfunc_bf)};
-			r.stencil_dynamic_back = {sm.stencil_mask_bf,
-			                          rc.stencil_clear_enable ? 0u : sm.stencil_writemask_bf,
-			                          sm.stencil_testval_bf};
+			r.stencil_static_back = {
+			    ConvertStencilOp(sc.stencil_fail_bf, back_write_mask, sm.stencil_opval_bf),
+			    ConvertStencilOp(sc.stencil_zpass_bf, back_write_mask, sm.stencil_opval_bf),
+			    ConvertStencilOp(sc.stencil_zfail_bf, back_write_mask, sm.stencil_opval_bf),
+			    static_cast<vk::CompareOp>(dc.stencilfunc_bf)};
+			r.stencil_dynamic_back = {sm.stencil_mask_bf, back_write_mask, sm.stencil_testval_bf};
 		} else {
 			r.stencil_static_back  = r.stencil_static_front;
 			r.stencil_dynamic_back = r.stencil_dynamic_front;
@@ -331,7 +344,7 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	desc.info.pitch           = pitch;
 	desc.info.bytes_per_block = bytes;
 	desc.info.samples         = samples;
-	desc.info.tile_mode       = Prospero::GpuEnumValue(Prospero::TileMode::kDepth);
+	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.metadata.kind   = has_htile ? ImageMetadataKind::Htile : ImageMetadataKind::None;

src/graphics/host_gpu/renderer/depthRenderTarget.h

@@ -11,8 +11,6 @@
 
 namespace Libs::Graphics {
 
-class RenderCommandBuffer;
-
 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);

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

@@ -646,13 +646,13 @@ void Validate(const ImageInfo& info) {
 	}
 }
 
-uint32_t RenderTargetTransferFormat(uint32_t bytes_per_element) {
+Prospero::BufferFormat RenderTargetTransferFormat(uint32_t bytes_per_element) {
 	switch (bytes_per_element) {
-		case 1: return Prospero::GpuEnumValue(Prospero::BufferFormat::k8UNorm);
-		case 2: return Prospero::GpuEnumValue(Prospero::BufferFormat::k16UNorm);
-		case 4: return Prospero::GpuEnumValue(Prospero::BufferFormat::k32Float);
-		case 8: return Prospero::GpuEnumValue(Prospero::BufferFormat::k16_16_16_16Float);
-		case 16: return Prospero::GpuEnumValue(Prospero::BufferFormat::k32_32_32_32Float);
+		case 1: return Prospero::BufferFormat::k8UNorm;
+		case 2: return Prospero::BufferFormat::k16UNorm;
+		case 4: return Prospero::BufferFormat::k32Float;
+		case 8: return Prospero::BufferFormat::k16_16_16_16Float;
+		case 16: return Prospero::BufferFormat::k32_32_32_32Float;
 		default: EXIT("unsupported render-target element size: %u\n", bytes_per_element);
 	}
 }
@@ -663,7 +663,8 @@ Image::Image(GraphicContext& graphics, CommandScheduler& scheduler, const ImageI
     : info(image_info), m_graphics(&graphics), m_scheduler(&scheduler) {
 	KYTY_PROFILER_FUNCTION();
 	ImageOps::Validate(info);
-	m_cpu_dirty = !info.data.Empty();
+	m_cpu_dirty =
+	    !info.data.Empty() && info.metadata.compression == VideoOutCompression::Uncompressed;
 	if (info.pixel_format == vk::Format::eUndefined) {
 		return;
 	}
@@ -735,15 +736,10 @@ Image::~Image() {
 	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);

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;
@@ -55,6 +39,7 @@ struct ImageBinding {
 	bool is_target     = false;
 	bool needs_rebind  = false;
 	bool force_general = false;
+	bool shader_write  = false;
 };
 
 class Image final {
@@ -156,7 +141,7 @@ public:
 
 	ImageInfo        info;
 	VulkanImage      backing;
-	ImageViewCache   views;
+	std::vector<CachedImageView> views;
 	ImageUsage       usage;
 	ImageBinding     binding;
 	bool             registered     = false;
@@ -188,8 +173,8 @@ private:
 
 namespace ImageOps {
 
-void                   Validate(const ImageInfo& info);
-[[nodiscard]] uint32_t RenderTargetTransferFormat(uint32_t bytes_per_element);
+void                                 Validate(const ImageInfo& info);
+[[nodiscard]] Prospero::BufferFormat RenderTargetTransferFormat(uint32_t bytes_per_element);
 
 } // namespace ImageOps
 

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

@@ -67,14 +67,14 @@ struct ImageInfo {
 	ImageMetadataInfo            metadata;
 	uint32_t                     htile_clear_mask = UINT32_MAX;
 	vk::Format                   pixel_format     = vk::Format::eUndefined;
-	uint32_t                     guest_format     = 0;
+	Prospero::BufferFormat       guest_format     = Prospero::BufferFormat::kInvalid;
 	Prospero::ImageType          type             = Prospero::ImageType::kColor2D;
 	vk::Extent3D                 extent           = {1, 1, 1};
 	ImageSubresources            resources;
 	uint32_t                     pitch           = 0;
 	uint32_t                     bytes_per_block = 0;
 	uint32_t                     samples         = 1;
-	uint32_t                     tile_mode       = 0;
+	Prospero::TileMode           tile_mode       = Prospero::TileMode::kLinear;
 	bool                         bgra16          = false;
 	std::array<ImageMipInfo, 16> mip_layout {};
 
@@ -86,9 +86,7 @@ struct ImageInfo {
 	[[nodiscard]] bool IsBlock() const noexcept {
 		return Prospero::BlockCompressedBytesPerBlock(guest_format) != 0;
 	}
-	[[nodiscard]] bool IsTiled() const noexcept {
-		return tile_mode != Prospero::GpuEnumValue(Prospero::TileMode::kLinear);
-	}
+	[[nodiscard]] bool IsTiled() const noexcept { return tile_mode != Prospero::TileMode::kLinear; }
 	[[nodiscard]] constexpr bool IsVolume() const noexcept {
 		return type == Prospero::ImageType::kColor3D;
 	}
@@ -214,7 +212,6 @@ struct DepthFormatPolicy {
 	Prospero::DepthFormat     depth_format;
 	Prospero::BufferFormat    guest_format;
 	uint32_t                  bytes_per_element;
-	vk::Format                sampled_view_format;
 	vk::Format                depth_attachment_format;
 	std::array<vk::Format, 3> stencil_attachment_formats;
 };
@@ -223,21 +220,19 @@ inline constexpr std::array<DepthFormatPolicy, 2> DEPTH_FORMAT_POLICIES {{
     {Prospero::DepthFormat::kZ16,
      Prospero::BufferFormat::k16UNorm,
      2,
-     vk::Format::eR16Unorm,
      vk::Format::eD16Unorm,
      {vk::Format::eD16UnormS8Uint, vk::Format::eD24UnormS8Uint, vk::Format::eD32SfloatS8Uint}},
     {Prospero::DepthFormat::kZ32F,
      Prospero::BufferFormat::k32Float,
      4,
-     vk::Format::eR32Sfloat,
      vk::Format::eD32Sfloat,
      {vk::Format::eD32SfloatS8Uint, vk::Format::eUndefined, vk::Format::eUndefined}},
 }};
 
 [[nodiscard]] inline constexpr const DepthFormatPolicy*
-FindDepthFormatPolicy(uint32_t depth_format) noexcept {
+FindDepthFormatPolicy(Prospero::DepthFormat depth_format) noexcept {
 	for (const auto& policy: DEPTH_FORMAT_POLICIES) {
-		if (Prospero::GpuEnumValue(policy.depth_format) == depth_format) {
+		if (policy.depth_format == depth_format) {
 			return &policy;
 		}
 	}
@@ -245,9 +240,9 @@ FindDepthFormatPolicy(uint32_t depth_format) noexcept {
 }
 
 [[nodiscard]] inline constexpr const DepthFormatPolicy*
-FindGuestDepthFormatPolicy(uint32_t guest_format) noexcept {
+FindGuestDepthFormatPolicy(Prospero::BufferFormat guest_format) noexcept {
 	for (const auto& policy: DEPTH_FORMAT_POLICIES) {
-		if (Prospero::GpuEnumValue(policy.guest_format) == guest_format) {
+		if (policy.guest_format == guest_format) {
 			return &policy;
 		}
 	}
@@ -269,10 +264,11 @@ FindGuestDepthFormatPolicy(uint32_t guest_format) noexcept {
 	return has_stencil ? policy.stencil_attachment_formats.front() : policy.depth_attachment_format;
 }
 
-[[nodiscard]] inline constexpr vk::Format DepthAttachmentFormat(uint32_t depth_format,
-                                                                uint32_t stencil_format) noexcept {
+[[nodiscard]] inline constexpr vk::Format
+DepthAttachmentFormat(Prospero::DepthFormat   depth_format,
+                      Prospero::StencilFormat stencil_format) noexcept {
 	bool has_stencil = false;
-	switch (static_cast<Prospero::StencilFormat>(stencil_format)) {
+	switch (stencil_format) {
 		case Prospero::StencilFormat::kInvalid: break;
 		case Prospero::StencilFormat::k8UInt: has_stencil = true; break;
 		default: return vk::Format::eUndefined;
@@ -318,26 +314,6 @@ inline bool ImageInfo::IsDepth() const noexcept {
 	return std::bit_cast<uint32_t>(static_cast<float>(value) / 65535.0f);
 }
 
-[[nodiscard]] inline constexpr bool IsSupportedSampledDepthFormat(vk::Format image_format,
-                                                                  uint32_t   guest_format,
-                                                                  vk::Format view_format) noexcept {
-	const auto* policy = FindGuestDepthFormatPolicy(guest_format);
-	return policy != nullptr && view_format == policy->sampled_view_format &&
-	       (image_format == policy->depth_attachment_format ||
-	        IsStencilAttachmentFormat(*policy, image_format));
-}
-
-[[nodiscard]] inline constexpr bool IsSupportedSampledDepthFormat(vk::Format image_format,
-                                                                  vk::Format view_format) noexcept {
-	for (const auto& policy: DEPTH_FORMAT_POLICIES) {
-		if (IsSupportedSampledDepthFormat(image_format, Prospero::GpuEnumValue(policy.guest_format),
-		                                  view_format)) {
-			return true;
-		}
-	}
-	return false;
-}
-
 [[nodiscard]] inline constexpr bool IsSupportedDepthTargetFormat(const ImageInfo& info) {
 	const auto* policy = FindGuestDepthFormatPolicy(info.guest_format);
 	return policy != nullptr && info.bytes_per_block == policy->bytes_per_element &&
@@ -383,10 +359,10 @@ CanUseVideoOutNativeWithoutUpload(VideoOutCompression compression, bool render_t
 }
 
 struct VideoOutPixelFormatInfo {
-	vk::Format format            = vk::Format::eUndefined;
-	uint32_t   guest_format      = 0;
-	uint32_t   bytes_per_element = 0;
-	bool       bgra16            = false;
+	vk::Format             format            = vk::Format::eUndefined;
+	Prospero::BufferFormat guest_format      = Prospero::BufferFormat::kInvalid;
+	uint32_t               bytes_per_element = 0;
+	bool                   bgra16            = false;
 };
 
 struct VideoOutFormatPolicy {
@@ -396,23 +372,17 @@ struct VideoOutFormatPolicy {
 
 inline constexpr std::array<VideoOutFormatPolicy, 6> VIDEO_OUT_FORMAT_POLICIES {{
     {0x8000000022000000ull,
-     {vk::Format::eR8G8B8A8Srgb, Prospero::GpuEnumValue(Prospero::BufferFormat::k8_8_8_8Srgb), 4,
-      false}},
+     {vk::Format::eR8G8B8A8Srgb, Prospero::BufferFormat::k8_8_8_8Srgb, 4, false}},
     {0x8000000000000000ull,
-     {vk::Format::eB8G8R8A8Srgb, Prospero::GpuEnumValue(Prospero::BufferFormat::k8_8_8_8Srgb), 4,
-      false}},
+     {vk::Format::eB8G8R8A8Srgb, Prospero::BufferFormat::k8_8_8_8Srgb, 4, false}},
     {0x8100000022000000ull,
-     {vk::Format::eA2B10G10R10UnormPack32,
-      Prospero::GpuEnumValue(Prospero::BufferFormat::k10_10_10_2UNorm), 4, false}},
+     {vk::Format::eA2B10G10R10UnormPack32, Prospero::BufferFormat::k10_10_10_2UNorm, 4, false}},
     {0x8100000000000000ull,
-     {vk::Format::eA2R10G10B10UnormPack32,
-      Prospero::GpuEnumValue(Prospero::BufferFormat::k10_10_10_2UNorm), 4, false}},
+     {vk::Format::eA2R10G10B10UnormPack32, Prospero::BufferFormat::k10_10_10_2UNorm, 4, false}},
     {0xc001000622000000ull,
-     {vk::Format::eR16G16B16A16Sfloat,
-      Prospero::GpuEnumValue(Prospero::BufferFormat::k16_16_16_16Float), 8, false}},
+     {vk::Format::eR16G16B16A16Sfloat, Prospero::BufferFormat::k16_16_16_16Float, 8, false}},
     {0xc001000600000000ull,
-     {vk::Format::eR16G16B16A16Sfloat,
-      Prospero::GpuEnumValue(Prospero::BufferFormat::k16_16_16_16Float), 8, true}},
+     {vk::Format::eR16G16B16A16Sfloat, Prospero::BufferFormat::k16_16_16_16Float, 8, true}},
 }};
 
 [[nodiscard]] inline bool DecodeVideoOutPixelFormat(uint64_t                 pixel_format,
@@ -439,15 +409,15 @@ inline constexpr std::array<VideoOutFormatPolicy, 6> VIDEO_OUT_FORMAT_POLICIES {
 }
 
 [[nodiscard]] inline constexpr bool
-IsSupportedDisplayRenderTargetTileMode(uint32_t tile_mode) noexcept {
-	return tile_mode == Prospero::GpuEnumValue(Prospero::TileMode::kRenderTarget);
+IsSupportedDisplayRenderTargetTileMode(Prospero::TileMode tile_mode) noexcept {
+	return tile_mode == Prospero::TileMode::kRenderTarget;
 }
 
 [[nodiscard]] inline constexpr bool IsSupportedStandard64RenderTarget(const ImageInfo& info) {
-	if (info.tile_mode != Prospero::GpuEnumValue(Prospero::TileMode::kStandard64KB) ||
-	    info.data.address == 0 || (info.data.address & 0xffffu) != 0 || info.extent.width == 0 ||
-	    info.extent.height == 0 || info.bytes_per_block != 4 || info.resources.levels != 1 ||
-	    info.resources.layers != 1 || info.samples != 1) {
+	if (info.tile_mode != Prospero::TileMode::kStandard64KB || info.data.address == 0 ||
+	    (info.data.address & 0xffffu) != 0 || info.extent.width == 0 || info.extent.height == 0 ||
+	    info.bytes_per_block != 4 || info.resources.levels != 1 || info.resources.layers != 1 ||
+	    info.samples != 1) {
 		return false;
 	}
 	const auto expected_pitch =
@@ -460,7 +430,7 @@ IsSupportedDisplayRenderTargetTileMode(uint32_t tile_mode) noexcept {
 }
 
 [[nodiscard]] inline constexpr bool IsTiledRenderTarget(const ImageInfo& info) noexcept {
-	return info.tile_mode == Prospero::GpuEnumValue(Prospero::TileMode::kRenderTarget) ||
+	return info.tile_mode == Prospero::TileMode::kRenderTarget ||
 	       IsSupportedStandard64RenderTarget(info);
 }
 

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 {
@@ -36,17 +34,6 @@ namespace {
 	}
 }
 
-[[nodiscard]] bool IsDepthViewFormat(vk::Format format) {
-	switch (format) {
-		case vk::Format::eD16Unorm:
-		case vk::Format::eR16Unorm:
-		case vk::Format::eD32Sfloat:
-		case vk::Format::eR32Sfloat:
-		case vk::Format::eR32Uint: return true;
-		default: return false;
-	}
-}
-
 [[nodiscard]] bool IsValidViewType(const VulkanImage& image, const ImageViewInfo& info) {
 	switch (image.image_type) {
 		case vk::ImageType::e1D:
@@ -327,13 +314,13 @@ vk::ImageView Image::FindView(const ImageViewInfo& view_info) {
 	const auto& image      = backing;
 	auto        normalized = view_info;
 	const bool  is_storage = static_cast<bool>(normalized.usage & vk::ImageUsageFlagBits::eStorage);
-	normalized.aspect      = FullAspectMask(image.format);
-	if (normalized.aspect & vk::ImageAspectFlagBits::eDepth &&
-	    IsDepthViewFormat(normalized.format)) {
+	const auto  image_aspect = FullAspectMask(image.format);
+	if (image_aspect & vk::ImageAspectFlagBits::eDepth &&
+	    ImageViewOps::IsFormatDepthCompatible(normalized.format)) {
 		normalized.format = image.format;
 		normalized.aspect = vk::ImageAspectFlagBits::eDepth;
 	}
-	if (normalized.aspect & vk::ImageAspectFlagBits::eStencil &&
+	if (image_aspect & vk::ImageAspectFlagBits::eStencil &&
 	    IsStencilViewFormat(normalized.format)) {
 		normalized.format = image.format;
 		normalized.aspect = vk::ImageAspectFlagBits::eStencil;
@@ -368,8 +355,7 @@ 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;
 		}
@@ -405,7 +391,7 @@ 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

@@ -12,8 +12,25 @@ namespace ImageViewOps {
 
 [[nodiscard]] vk::ImageAspectFlags DepthAspectMask(vk::Format format);
 [[nodiscard]] bool                 FormatsCompatible(vk::Format base, vk::Format view) noexcept;
+
+[[nodiscard]] inline bool IsFormatDepthCompatible(vk::Format format) noexcept {
+	switch (format) {
+		case vk::Format::eD32Sfloat:
+		case vk::Format::eR32Sfloat:
+		case vk::Format::eR32Uint:
+		case vk::Format::eD16Unorm:
+		case vk::Format::eR16Unorm: return true;
+		default: return false;
+	}
+}
 } // namespace ImageViewOps
 
+[[nodiscard]] inline bool IsSupportedSampledDepthFormat(vk::Format image_format,
+                                                        vk::Format view_format) noexcept {
+	return DepthAspectTransferFormat(image_format) != vk::Format::eUndefined &&
+	       ImageViewOps::IsFormatDepthCompatible(view_format);
+}
+
 [[nodiscard]] inline bool IsValidImageSwizzle(uint32_t swizzle) noexcept {
 	if ((swizzle & ~0xfffu) != 0) {
 		return false;
@@ -86,8 +103,14 @@ SelectSampledDepthView(vk::Format image_format, vk::Format view_format, uint32_t
 
 [[nodiscard]] inline bool
 IsSupportedSampledDepthResource(const ShaderRecompiler::IR::ImageResource& resource) noexcept {
-	return resource.kind == ShaderRecompiler::IR::ResourceKind::Image &&
-	       (resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2D ||
+	if (resource.kind != ShaderRecompiler::IR::ResourceKind::Image &&
+	    resource.kind != ShaderRecompiler::IR::ResourceKind::ImageUint) {
+		return false;
+	}
+	if (resource.kind == ShaderRecompiler::IR::ResourceKind::ImageUint && resource.depth_compare) {
+		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) &&
@@ -95,14 +118,6 @@ IsSupportedSampledDepthResource(const ShaderRecompiler::IR::ImageResource& resou
 	       !resource.written && !resource.atomic;
 }
 
-[[nodiscard]] inline bool
-IsSupportedSampledDepthUintResource(const ShaderRecompiler::IR::ImageResource& resource) noexcept {
-	return resource.kind == ShaderRecompiler::IR::ResourceKind::ImageUint &&
-	       resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2D &&
-	       resource.mip_mode == ShaderRecompiler::IR::ImageMipMode::None && resource.read &&
-	       !resource.written && !resource.atomic && !resource.depth_compare;
-}
-
 inline void ValidateStorageColorView(vk::Format image_format, vk::Format view_format,
                                      uint32_t swizzle) noexcept {
 	if (!ImageViewOps::FormatsCompatible(image_format, view_format) ||
@@ -120,7 +135,11 @@ IsSupportedStorageImageResource(const ShaderRecompiler::IR::ImageResource& resou
 	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2D ||
 	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim3D ||
 	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DArray) &&
-	       resource.mip_mode == ShaderRecompiler::IR::ImageMipMode::None && resource.written &&
+	       ((resource.mip_mode == ShaderRecompiler::IR::ImageMipMode::None &&
+	         resource.mip_count == 1u) ||
+	        (resource.mip_mode == ShaderRecompiler::IR::ImageMipMode::DynamicStorage &&
+	         resource.mip_count != 0u)) &&
+	       resource.written &&
 	       (!resource.atomic ||
 	        (resource.kind == ShaderRecompiler::IR::ResourceKind::StorageImageUint &&
 	         resource.read)) &&

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

@@ -14,164 +14,71 @@
 namespace Libs::Graphics {
 namespace {
 
-struct RenderTargetFormatMapping {
-	Prospero::ChannelLayout layout;
-	Prospero::ChannelType   type;
-	Prospero::ChannelOrder  order;
-	RenderTargetFormatInfo  info;
+// Rows are channel order; columns are the number of physical components minus one. Unused
+// selectors complete each entry to a permutation so logical write masks can be inverted.
+constexpr Prospero::ColorComponentMapping kRenderTargetColorMappings[4][4] = {
+    {Prospero::ColorMappingRgba, Prospero::ColorMappingRgba, Prospero::ColorMappingRgba,
+     Prospero::ColorMappingRgba},
+    {Prospero::ColorMappingGr, Prospero::ColorMappingRabg, Prospero::ColorMappingRgab,
+     Prospero::ColorMappingBgra},
+    {Prospero::ColorMappingBgra, Prospero::ColorMappingGr, Prospero::ColorMappingBgra,
+     Prospero::ColorMappingAbgr},
+    {Prospero::ColorMappingAgba, Prospero::ColorMappingArbg, Prospero::ColorMappingAgbr,
+     Prospero::ColorMappingArgb},
 };
 
-constexpr RenderTargetFormatMapping kRenderTargetFormats[] = {
-    {Prospero::ChannelLayout::k8_8,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR8G8Unorm, 2}},
-    {Prospero::ChannelLayout::k8_8_8_8,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR8G8B8A8Unorm, 4}},
-    {Prospero::ChannelLayout::k8_8_8_8,
-     Prospero::ChannelType::kSNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR8G8B8A8Snorm, 4}},
-    {Prospero::ChannelLayout::k8_8_8_8,
-     Prospero::ChannelType::kSrgb,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR8G8B8A8Srgb, 4}},
-    {Prospero::ChannelLayout::k8_8_8_8,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kAlt,
-     {vk::Format::eB8G8R8A8Unorm, 4}},
-    {Prospero::ChannelLayout::k8_8_8_8,
-     Prospero::ChannelType::kSNorm,
-     Prospero::ChannelOrder::kAlt,
-     {vk::Format::eB8G8R8A8Snorm, 4}},
-    {Prospero::ChannelLayout::k8_8_8_8,
-     Prospero::ChannelType::kSrgb,
-     Prospero::ChannelOrder::kAlt,
-     {vk::Format::eB8G8R8A8Srgb, 4}},
-    {Prospero::ChannelLayout::k5_5_5_1,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR5G5B5A1UnormPack16, 2}},
-    {Prospero::ChannelLayout::k4_4_4_4,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kReversed,
-     {vk::Format::eB4G4R4A4UnormPack16, 2}},
-    {Prospero::ChannelLayout::k10_10_10_2,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eA2B10G10R10UnormPack32, 4}},
-    {Prospero::ChannelLayout::k10_10_10_2,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kAlt,
-     {vk::Format::eA2R10G10B10UnormPack32, 4}},
-    {Prospero::ChannelLayout::k11_11_10,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eB10G11R11UfloatPack32, 4}},
-    {Prospero::ChannelLayout::k5_6_5,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eB5G6R5UnormPack16, 2}},
-    {Prospero::ChannelLayout::k16,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16Unorm, 2}},
-    {Prospero::ChannelLayout::k16,
-     Prospero::ChannelType::kUInt,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16Uint, 2}},
-    {Prospero::ChannelLayout::k16,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16Sfloat, 2}},
-    {Prospero::ChannelLayout::k16_16,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16G16Unorm, 4}},
-    {Prospero::ChannelLayout::k16_16,
-     Prospero::ChannelType::kSNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16G16Snorm, 4}},
-    {Prospero::ChannelLayout::k16_16,
-     Prospero::ChannelType::kUInt,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16G16Uint, 4}},
-    {Prospero::ChannelLayout::k16_16,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16G16Sfloat, 4}},
-    {Prospero::ChannelLayout::k16_16_16_16,
-     Prospero::ChannelType::kUNorm,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16G16B16A16Unorm, 8}},
-    {Prospero::ChannelLayout::k16_16_16_16,
-     Prospero::ChannelType::kUInt,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16G16B16A16Uint, 8}},
-    {Prospero::ChannelLayout::k16_16_16_16,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR16G16B16A16Sfloat, 8}},
-    {Prospero::ChannelLayout::k16_16_16_16,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kAlt,
-     {vk::Format::eR16G16B16A16Sfloat, 8, Prospero::ColorMappingBgra}},
-    {Prospero::ChannelLayout::k16_16_16_16,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kReversed,
-     {vk::Format::eR16G16B16A16Sfloat, 8, Prospero::ColorMappingAbgr}},
-    {Prospero::ChannelLayout::k32,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR32Sfloat, 4}},
-    {Prospero::ChannelLayout::k32_32,
-     Prospero::ChannelType::kUInt,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR32G32Uint, 8}},
-    {Prospero::ChannelLayout::k32_32,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR32G32Sfloat, 8}},
-    {Prospero::ChannelLayout::k32_32_32_32,
-     Prospero::ChannelType::kFloat,
-     Prospero::ChannelOrder::kStandard,
-     {vk::Format::eR32G32B32A32Sfloat, 16}},
+struct RenderTargetHostFormatInfo {
+	vk::Format                      format = vk::Format::eUndefined;
+	Prospero::ColorComponentMapping host_to_storage;
 };
 
+RenderTargetHostFormatInfo ResolveRenderTargetHostFormat(Prospero::BufferFormat guest_format,
+                                                         Prospero::ChannelOrder order) {
+	if (order == Prospero::ChannelOrder::kAlt) {
+		switch (guest_format) {
+			case Prospero::BufferFormat::k8_8_8_8UNorm:
+				return {vk::Format::eB8G8R8A8Unorm, Prospero::ColorMappingBgra};
+			case Prospero::BufferFormat::k8_8_8_8SNorm:
+				return {vk::Format::eB8G8R8A8Snorm, Prospero::ColorMappingBgra};
+			case Prospero::BufferFormat::k8_8_8_8Srgb:
+				return {vk::Format::eB8G8R8A8Srgb, Prospero::ColorMappingBgra};
+			case Prospero::BufferFormat::k10_10_10_2UNorm:
+				return {vk::Format::eA2R10G10B10UnormPack32, Prospero::ColorMappingBgra};
+			default: break;
+		}
+	}
+	switch (guest_format) {
+		case Prospero::BufferFormat::k5_5_5_1UNorm:
+			return {vk::Format::eA1R5G5B5UnormPack16, Prospero::ColorMappingBgra};
+		case Prospero::BufferFormat::k1_5_5_5UNorm:
+			return {vk::Format::eR5G5B5A1UnormPack16, Prospero::ColorMappingAbgr};
+		case Prospero::BufferFormat::k4_4_4_4UNorm:
+			return {vk::Format::eR4G4B4A4UnormPack16, Prospero::ColorMappingAbgr};
+		default: return {VulkanFormat(guest_format), {}};
+	}
+}
+
 } // namespace
 
-// TODO: cleanup!
-RenderTargetFormatInfo TextureGetRenderTargetFormat(uint32_t raw_layout, uint32_t raw_type,
-                                                    uint32_t raw_order) {
-	const auto layout = static_cast<Prospero::ChannelLayout>(raw_layout);
-	const auto type   = static_cast<Prospero::ChannelType>(raw_type);
-	const auto order  = static_cast<Prospero::ChannelOrder>(raw_order);
-
-	if (layout == Prospero::ChannelLayout::k8 && type == Prospero::ChannelType::kUNorm &&
-	    raw_order <= Prospero::GpuEnumValue(Prospero::ChannelOrder::kAltReversed)) {
-		return {vk::Format::eR8Unorm, 1};
-	}
-	for (const auto& mapping: kRenderTargetFormats) {
-		if (mapping.layout == layout && mapping.type == type && mapping.order == order) {
-			return mapping.info;
+RenderTargetFormatInfo TextureGetRenderTargetFormat(Prospero::ChannelLayout layout,
+                                                    Prospero::ChannelType   type,
+                                                    Prospero::ChannelOrder  order) {
+	const auto encoding = Prospero::ResolveRenderTargetFormat(layout, type);
+	if (encoding.IsValid() && encoding.SupportsOrder(order)) {
+		const auto host  = ResolveRenderTargetHostFormat(encoding.buffer_format, order);
+		const auto bytes = Prospero::RenderTargetBytesPerElement(encoding.buffer_format);
+		if (host.format != vk::Format::eUndefined && bytes != 0) {
+			const auto order_mapping =
+			    kRenderTargetColorMappings[static_cast<size_t>(order)][encoding.components - 1u];
+			return {host.format, bytes, host.host_to_storage.Then(order_mapping)};
 		}
 	}
-	EXIT("unsupported render-target format combination: layout=%u type=%u order=%u\n", raw_layout,
-	     raw_type, raw_order);
+	EXIT("unsupported render-target format combination: layout=%u type=%u order=%u\n",
+	     static_cast<uint32_t>(layout), static_cast<uint32_t>(type), static_cast<uint32_t>(order));
 }
 
 namespace {
 
-static uint64_t GetLevelSrcOffset(const TileSizeOffset& level_size) {
-	return (level_size.src_size != 0 ? level_size.src_offset : level_size.offset);
-}
-
-static uint64_t GetLevelSrcSize(const TileSizeOffset& level_size) {
-	return (level_size.src_size != 0 ? level_size.src_size : level_size.size);
-}
-
 static uint32_t GetTextureLevelDepth(uint32_t depth, uint32_t level, bool volume_texture) {
 	return volume_texture ? std::max(depth >> level, 1u) : depth;
 }
@@ -184,20 +91,6 @@ static size_t GetTextureRegionCount(uint32_t depth, uint64_t levels, bool volume
 	return count;
 }
 
-uint64_t TextureUploadSliceSourceOffset(const TextureUploadLayout& layout, uint32_t level,
-                                        uint32_t slice) {
-	if (level >= 16 || layout.level_sizes[level].size == 0) {
-		EXIT("invalid texture upload slice source, level=%u slice=%u\n", level, slice);
-	}
-	const auto level_offset = GetLevelSrcOffset(layout.level_sizes[level]);
-	const auto slice_stride =
-	    layout.source_slice_stride != 0 ? layout.source_slice_stride : layout.slice_stride;
-	if (slice_stride != 0 && slice > (UINT64_MAX - level_offset) / slice_stride) {
-		EXIT("texture upload slice source offset overflow, level=%u slice=%u\n", level, slice);
-	}
-	return level_offset + static_cast<uint64_t>(slice) * slice_stride;
-}
-
 vk::ComponentSwizzle TextureGetComponentSwizzle(uint8_t s) {
 	switch (static_cast<Prospero::CompSwizzle>(s)) {
 		case Prospero::CompSwizzle::kZero: return vk::ComponentSwizzle::eZero;
@@ -226,23 +119,25 @@ vk::ComponentMapping TextureGetComponentMapping(uint32_t swizzle) {
 	return components;
 }
 
-vk::Format TextureGetFormat(uint32_t fmt) {
-	const auto vk_format = VulkanFormat(fmt);
+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", fmt);
-	return vk::Format::eUndefined;
+	EXIT("unknown format: fmt = %u\n", static_cast<uint32_t>(format));
+	return SurfaceFormatInfo(vk::Format::eUndefined, Prospero::BufferFormat::kInvalid);
 }
 
 namespace {
 
-uint64_t CalcTextureSliceStride(const TileSizeOffset* level_sizes, uint64_t levels,
+uint64_t CalcTextureSliceStride(const TextureUploadMipLayout* mips, uint64_t levels,
                                 uint64_t total_size, uint32_t depth) {
 	uint64_t stride = 0;
 	for (uint32_t i = 0; i < levels; i++) {
-		stride =
-		    std::max(stride, static_cast<uint64_t>(level_sizes[i].offset) + level_sizes[i].size);
+		stride = std::max(stride, mips[i].offset + mips[i].size);
 	}
 
 	if (depth > 1 && total_size != 0 && total_size % depth == 0) {
@@ -255,39 +150,26 @@ uint64_t CalcTextureSliceStride(const TileSizeOffset* level_sizes, uint64_t leve
 	return stride;
 }
 
-uint64_t CalcLinearUploadLevelSize(uint32_t fmt, uint32_t pitch, uint32_t height) {
-	if (const uint32_t bytes_per_element = Prospero::NumBytesPerElement(fmt);
-	    bytes_per_element != 0) {
-		return static_cast<uint64_t>(pitch) * height * bytes_per_element;
-	}
-
-	if (const uint32_t bytes_per_block = Prospero::BlockCompressedBytesPerBlock(fmt);
-	    bytes_per_block != 0) {
-		const uint32_t blocks_w = std::max((pitch + 3u) / 4u, 1u);
-		const uint32_t blocks_h = std::max((height + 3u) / 4u, 1u);
-		return static_cast<uint64_t>(blocks_w) * blocks_h * bytes_per_block;
-	}
-
-	return 0;
+uint64_t CalcLinearUploadLevelSize(const TileTextureElementLayout& element, uint32_t pitch,
+                                   uint32_t height) {
+	const uint32_t elements_w =
+	    std::max((pitch + element.texel_width - 1u) / element.texel_width, 1u);
+	const uint32_t elements_h =
+	    std::max((height + element.texel_height - 1u) / element.texel_height, 1u);
+	return static_cast<uint64_t>(elements_w) * elements_h * element.bytes;
 }
 
-uint64_t SetLinearUploadLevels(TileSizeOffset* level_sizes, uint32_t fmt, uint64_t height,
+uint64_t SetLinearUploadLevels(TextureUploadMipLayout*         mips,
+                               const TileTextureElementLayout& element, uint64_t height,
                                uint64_t levels, uint32_t base_pitch) {
 	uint64_t offset = 0;
 	auto     pitch  = base_pitch;
 	auto     h      = static_cast<uint32_t>(height);
 
 	for (uint32_t i = 0; i < levels; i++) {
-		const auto size = CalcLinearUploadLevelSize(fmt, pitch, h);
-		EXIT_NOT_IMPLEMENTED(size > 0xffffffffull);
-		EXIT_NOT_IMPLEMENTED(offset > 0xffffffffull);
-
-		level_sizes[i].size       = static_cast<uint32_t>(size);
-		level_sizes[i].offset     = static_cast<uint32_t>(offset);
-		level_sizes[i].src_size   = 0;
-		level_sizes[i].src_offset = 0;
-		level_sizes[i].x          = 0;
-		level_sizes[i].y          = 0;
+		const auto size = CalcLinearUploadLevelSize(element, pitch, h);
+		mips[i].size    = size;
+		mips[i].offset  = offset;
 
 		offset += size;
 		if (pitch > 1) {
@@ -303,141 +185,126 @@ uint64_t SetLinearUploadLevels(TileSizeOffset* level_sizes, uint32_t fmt, uint64
 
 } // namespace
 
-TextureUploadLayout TextureCalcUploadLayout(uint32_t fmt, uint64_t width, uint64_t height,
-                                            uint64_t levels, uint32_t depth, uint64_t pitch,
-                                            uint64_t tile, uint64_t upload_size,
+TextureUploadLayout TextureCalcUploadLayout(Prospero::BufferFormat format, uint32_t width,
+                                            uint32_t height, uint32_t levels, uint32_t depth,
+                                            Prospero::TileMode tile, uint64_t upload_size,
                                             bool allow_depth_tile, bool volume_texture,
                                             const char* owner) {
 	TextureUploadLayout layout {};
-	layout.tile           = static_cast<uint32_t>(tile);
-	layout.pitch          = static_cast<uint32_t>(pitch);
-	layout.texel_block    = Prospero::BlockCompressedBytesPerBlock(fmt) != 0 ? 4u : 1u;
-	layout.volume_texture = volume_texture;
+	layout.surface.description = {
+	    format,
+	    tile,
+	    volume_texture ? TileSurfaceDimension::Dim3D : TileSurfaceDimension::Dim2D,
+	    width,
+	    height,
+	    volume_texture ? depth : 1u,
+	    levels,
+	    volume_texture ? 1u : depth,
+	};
+	const auto&              description = layout.surface.description;
+	const auto               tile_mode   = description.tile_mode;
+	TileTextureElementLayout element {};
 
-	if (fmt != 0) {
-		if (layout.tile != 0) {
-			const auto tile_mode = static_cast<Prospero::TileMode>(layout.tile);
-			switch (tile_mode) {
-				case Prospero::TileMode::kStandard256B:
-					if (TileIsStandard256BTextureSupported(fmt)) {
-						layout.tile_family = TileBlockFamily::Standard256B;
-					}
-					break;
-				case Prospero::TileMode::kStandard4KB:
-					if (TileIsStandard4KBTextureSupported(fmt)) {
-						layout.tile_family = TileBlockFamily::Standard4KB;
-					}
-					break;
-				case Prospero::TileMode::kStandard64KB:
-					if (TileIsStandard64KBTextureSupported(fmt)) {
-						layout.tile_family = TileBlockFamily::Standard64KB;
-					}
-					break;
-				case Prospero::TileMode::kPrt:
-					if (TileIsStandard64KBTextureSupported(fmt)) {
-						layout.tile_family = TileBlockFamily::Prt64KB;
-					}
-					break;
-				case Prospero::TileMode::kRenderTarget:
-					if (Prospero::RenderTargetBytesPerElement(fmt) != 0) {
-						layout.tile_family = TileBlockFamily::RenderTarget64KB;
-					}
-					break;
-				case Prospero::TileMode::kDepth:
-					if (allow_depth_tile && Prospero::RenderTargetBytesPerElement(fmt) != 0) {
-						layout.tile_family = TileBlockFamily::Depth64KB;
-					}
-					break;
-				default: break;
-			}
-			if (layout.tile_family == TileBlockFamily::Count) {
-				EXIT("%s: unsupported typed tiled upload: fmt=%u tile=%u "
-				     "size=%" PRIu64 " extent=%" PRIu64 "x%" PRIu64 " pitch=%" PRIu64
-				     " levels=%" PRIu64 "\n",
-				     owner, static_cast<uint32_t>(fmt), layout.tile, upload_size, width, height,
-				     pitch, levels);
-			}
-		}
-
-		layout.pitch = TileGetTexturePitch(fmt, width, levels, layout.tile);
-
-		TileGetTextureSize(fmt, width, height, layout.pitch, levels, layout.tile, nullptr,
-		                   layout.level_sizes, layout.padded_sizes);
-
-		if (static_cast<Prospero::TileMode>(layout.tile) != Prospero::TileMode::kLinear) {
-			if (layout.volume_texture) {
-				layout.slice_stride = SetLinearUploadLevels(layout.level_sizes, fmt, height, levels,
-				                                            static_cast<uint32_t>(width));
-			} else {
-				TileSizeOffset tiled_levels[16] {};
-				std::copy_n(layout.level_sizes, levels, tiled_levels);
-				layout.source_slice_stride =
-				    CalcTextureSliceStride(tiled_levels, levels, upload_size, depth);
-				SetLinearUploadLevels(layout.level_sizes, fmt, height, levels, layout.pitch);
-				for (uint32_t i = 0; i < levels; ++i) {
-					if (tiled_levels[i].src_size > tiled_levels[i].size) {
-						layout.first_tail_level = std::min(layout.first_tail_level, i);
-					}
-					layout.level_sizes[i].src_offset = GetLevelSrcOffset(tiled_levels[i]);
-					layout.level_sizes[i].src_size   = GetLevelSrcSize(tiled_levels[i]);
-					layout.level_sizes[i].x          = tiled_levels[i].x;
-					layout.level_sizes[i].y          = tiled_levels[i].y;
-				}
-			}
-		} else if (layout.volume_texture) {
-			layout.slice_stride =
-			    CalcTextureSliceStride(layout.level_sizes, levels, upload_size, depth);
-		}
-	} else {
+	if (format == Prospero::BufferFormat::kInvalid) {
 		EXIT("%s: legacy texture upload format unsupported: fmt=0 tile=%u size=%" PRIu64
-		     " extent=%" PRIu64 "x%" PRIu64 " pitch=%" PRIu64 " levels=%" PRIu64 "\n",
-		     owner, layout.tile, upload_size, width, height, pitch, levels);
+		     " extent=%ux%u levels=%u\n",
+		     owner, static_cast<uint32_t>(description.tile_mode), upload_size, width, height,
+		     levels);
 	}
 
-	if (!layout.volume_texture) {
-		layout.slice_stride =
-		    CalcTextureSliceStride(layout.level_sizes, levels, upload_size, depth);
+	switch (tile_mode) {
+		case Prospero::TileMode::kLinear:
+			if (!TileGetTextureElementLayout(format, element)) {
+				EXIT("%s: unsupported linear texture format: fmt=%u\n", owner,
+				     static_cast<uint32_t>(format));
+			}
+			break;
+		case Prospero::TileMode::kDepth:
+			if (!allow_depth_tile || !TileGetTiledTextureLayout(description, layout.surface)) {
+				EXIT("%s: unsupported typed tiled upload: fmt=%u tile=%u "
+				     "size=%" PRIu64 " extent=%ux%u levels=%u\n",
+				     owner, static_cast<uint32_t>(format),
+				     static_cast<uint32_t>(description.tile_mode), upload_size, width, height,
+				     levels);
+			}
+			break;
+		default:
+			if (!TileGetTiledTextureLayout(description, layout.surface)) {
+				EXIT("%s: unsupported typed tiled upload: fmt=%u tile=%u "
+				     "size=%" PRIu64 " extent=%ux%u levels=%u\n",
+				     owner, static_cast<uint32_t>(format),
+				     static_cast<uint32_t>(description.tile_mode), upload_size, width, height,
+				     levels);
+			}
+			break;
+	}
+
+	if (tile_mode != Prospero::TileMode::kLinear) {
+		element = {layout.surface.texture.block.bytes_per_element,
+		           layout.surface.texture.texel_width, layout.surface.texture.texel_height};
+	}
+	layout.pitch = TileGetTexturePitch(format, width, description.tile_mode);
+	if (tile_mode == Prospero::TileMode::kLinear) {
+		TileSizeOffset level_sizes[16] {};
+		TilePaddedSize padded_sizes[16] {};
+		TileGetTextureSize(format, width, height, levels, description.tile_mode, nullptr,
+		                   level_sizes, padded_sizes);
+		for (uint32_t level = 0; level < levels; ++level) {
+			layout.mips[level] = {level_sizes[level].offset, level_sizes[level].size,
+			                      padded_sizes[level].width, padded_sizes[level].height};
+		}
+	} else if (volume_texture) {
+		layout.slice_stride = SetLinearUploadLevels(layout.mips, element, height, levels, width);
+	} else {
+		layout.source_slice_stride = layout.surface.block_slice_size;
+		if (depth > 1 && upload_size != 0 && upload_size % depth == 0) {
+			layout.source_slice_stride = std::max(layout.source_slice_stride, upload_size / depth);
+		}
+		SetLinearUploadLevels(layout.mips, element, height, levels, layout.pitch);
+	}
+
+	if (!volume_texture || tile_mode == Prospero::TileMode::kLinear) {
+		layout.slice_stride = CalcTextureSliceStride(layout.mips, levels, upload_size, depth);
 	}
 	return layout;
 }
 
-std::vector<vk::BufferImageCopy> TextureBuildImageCopies(const TextureUploadLayout& layout,
-                                                         uint32_t width, uint32_t height,
-                                                         uint32_t depth, uint64_t levels,
-                                                         bool array_texture, bool volume_texture) {
-	uint32_t mip_width  = width;
-	uint32_t mip_height = height;
-	uint32_t mip_pitch  = volume_texture && static_cast<Prospero::TileMode>(layout.tile) !=
-	                                            Prospero::TileMode::kLinear
-	                          ? width
-	                          : layout.pitch;
+std::vector<vk::BufferImageCopy> TextureBuildImageCopies(const TextureUploadLayout& layout) {
+	const auto& description    = layout.surface.description;
+	const bool  volume_texture = description.dimension == TileSurfaceDimension::Dim3D;
+	const auto  depth          = volume_texture ? description.depth : description.layers;
+	uint32_t    mip_width      = description.width;
+	uint32_t    mip_height     = description.height;
+	uint32_t    mip_pitch = volume_texture && description.tile_mode != Prospero::TileMode::kLinear
+	                            ? description.width
+	                            : layout.pitch;
 
 	std::vector<vk::BufferImageCopy> regions;
-	regions.reserve(GetTextureRegionCount(depth, levels, volume_texture));
-	for (uint32_t i = 0; i < levels; i++) {
-		EXIT_NOT_IMPLEMENTED(layout.level_sizes[i].size == 0);
+	regions.reserve(GetTextureRegionCount(depth, description.levels, volume_texture));
+	for (uint32_t i = 0; i < description.levels; i++) {
+		EXIT_NOT_IMPLEMENTED(layout.mips[i].size == 0);
 
 		const auto mip_depth = GetTextureLevelDepth(depth, i, volume_texture);
 
 		for (uint32_t z = 0; z < mip_depth; z++) {
 			const auto          slice_offset = z * layout.slice_stride;
 			vk::BufferImageCopy region {};
-			region.bufferOffset     = layout.level_sizes[i].offset + slice_offset;
-			region.imageSubresource = {vk::ImageAspectFlagBits::eColor, i, array_texture ? z : 0,
+			region.bufferOffset     = layout.mips[i].offset + slice_offset;
+			region.imageSubresource = {vk::ImageAspectFlagBits::eColor, i, volume_texture ? 0u : z,
 			                           1};
 			region.imageOffset.z    = volume_texture ? static_cast<int>(z) : 0;
 			region.imageExtent      = {mip_width, mip_height, 1};
-			const bool linear =
-			    static_cast<Prospero::TileMode>(layout.tile) == Prospero::TileMode::kLinear;
+			const bool linear       = description.tile_mode == Prospero::TileMode::kLinear;
 			if (linear) {
-				region.bufferRowLength   = layout.padded_sizes[i].width;
-				region.bufferImageHeight = layout.padded_sizes[i].height;
+				region.bufferRowLength   = layout.mips[i].row_length;
+				region.bufferImageHeight = layout.mips[i].image_height;
 			} else {
-				const auto align = [](uint32_t value, uint32_t block) {
+				const auto texel_width = layout.surface.texture.texel_width;
+				const auto align       = [](uint32_t value, uint32_t block) {
 					return ((value + block - 1u) / block) * block;
 				};
-				const auto pitch       = align(mip_pitch, layout.texel_block);
-				region.bufferRowLength = pitch > align(mip_width, layout.texel_block) ? pitch : 0;
+				const auto pitch       = align(mip_pitch, texel_width);
+				region.bufferRowLength = pitch > align(mip_width, texel_width) ? pitch : 0;
 			}
 			regions.push_back(region);
 		}
@@ -456,24 +323,6 @@ std::vector<vk::BufferImageCopy> TextureBuildImageCopies(const TextureUploadLayo
 	return regions;
 }
 
-struct GpuTileElementLayout {
-	uint32_t bytes = 0;
-	uint32_t wide  = 1;
-	uint32_t tall  = 1;
-};
-
-static bool GetGpuTileElementLayout(uint32_t fmt, GpuTileElementLayout& out) {
-	if (const auto bytes = Prospero::NumBytesPerElement(fmt); bytes != 0) {
-		out = {bytes, 1, 1};
-		return true;
-	}
-	if (const auto bytes = Prospero::BlockCompressedBytesPerBlock(fmt); bytes != 0) {
-		out = {bytes, 4, 4};
-		return true;
-	}
-	return false;
-}
-
 static bool SetGpuTileSize(uint64_t offset, uint64_t length, uint64_t capacity, uint64_t& size) {
 	if (offset > capacity || length > capacity - offset) {
 		return false;
@@ -483,42 +332,32 @@ static bool SetGpuTileSize(uint64_t offset, uint64_t length, uint64_t capacity,
 }
 
 bool TextureBuildGpuTileInfos(uint64_t tiled_size, const std::vector<vk::BufferImageCopy>& regions,
-                              const TextureUploadLayout& layout, uint32_t fmt, uint32_t depth,
-                              uint64_t levels, std::vector<GpuTileInfo>& out_infos) {
+                              const TextureUploadLayout& layout, uint32_t levels,
+                              std::vector<GpuTileInfo>& out_infos) {
+	const auto& description    = layout.surface.description;
+	const bool  volume_texture = description.dimension == TileSurfaceDimension::Dim3D;
+	const auto  depth          = volume_texture ? description.depth : description.layers;
 	if (tiled_size == 0 || levels == 0 || levels > 16 || depth == 0 ||
-	    regions.size() != GetTextureRegionCount(depth, levels, layout.volume_texture) ||
-	    Prospero::IsFmaskTextureFormat(fmt)) {
+	    regions.size() != GetTextureRegionCount(depth, levels, volume_texture) ||
+	    Prospero::IsFmaskTextureFormat(description.format)) {
 		return false;
 	}
 
-	GpuTileElementLayout element {};
-	if (layout.tile_family == TileBlockFamily::RenderTarget64KB ||
-	    layout.tile_family == TileBlockFamily::Depth64KB) {
-		element.bytes = Prospero::RenderTargetBytesPerElement(fmt);
-	} else if (!GetGpuTileElementLayout(fmt, element)) {
-		return false;
-	}
-	if (element.bytes == 0) {
+	const auto& surface = layout.surface;
+	const auto& texture = surface.texture;
+	const auto& block   = texture.block;
+	if (surface.description.levels < levels || block.bytes_per_element == 0 ||
+	    block.family == TileBlockFamily::Count) {
 		return false;
 	}
 
 	std::vector<GpuTileInfo> infos;
 	infos.reserve(regions.size());
-	if (layout.volume_texture) {
-		TileVolumeLayout volume {};
-		if (!TileGetTextureVolumeLayout(fmt, regions[0].imageExtent.width,
-		                                regions[0].imageExtent.height, depth,
-		                                static_cast<uint32_t>(levels), layout.tile, volume)) {
-			return false;
-		}
-		element = {volume.bytes_per_element, volume.texel_width, volume.texel_height};
-		TileBlockLayout block {};
-		if (!TileGetBlockLayout(volume.family, element.bytes, block)) return false;
-
+	if (volume_texture) {
 		size_t region_base = 0;
 		for (uint32_t level = 0; level < levels; ++level) {
 			const uint32_t mip_depth     = GetTextureLevelDepth(depth, level, true);
-			const bool     tail          = level >= volume.first_tail_level;
+			const bool     tail          = level >= surface.first_tail_level;
 			const uint64_t linear_stride = layout.slice_stride;
 			for (uint32_t z = 0; z < mip_depth; z += block.block_depth) {
 				const uint32_t copy_depth = std::min(block.block_depth, mip_depth - z);
@@ -533,50 +372,43 @@ bool TextureBuildGpuTileInfos(uint64_t tiled_size, const std::vector<vk::BufferI
 				info.bytes_per_element = block.bytes_per_element;
 				info.linear_offset     = region.bufferOffset;
 				info.tiled_offset =
-				    static_cast<uint64_t>(z / block.block_depth) * volume.block_slice_size +
-				    volume.level_offsets[level];
+				    static_cast<uint64_t>(z / block.block_depth) * surface.block_slice_size +
+				    surface.mips[level].offset;
 				const uint64_t linear_span =
 				    static_cast<uint64_t>(copy_depth - 1u) * linear_stride +
-				    layout.level_sizes[level].size;
+				    layout.mips[level].size;
 				if (!SetGpuTileSize(info.linear_offset, linear_span, UINT64_MAX,
 				                    info.linear_size) ||
-				    !SetGpuTileSize(info.tiled_offset, volume.level_sizes[level], tiled_size,
+				    !SetGpuTileSize(info.tiled_offset, surface.mips[level].size, tiled_size,
 				                    info.tiled_size)) {
 					return false;
 				}
 				info.linear_slice_stride = linear_stride;
-				info.width =
-				    std::max((region.imageExtent.width + element.wide - 1u) / element.wide, 1u);
-				info.height = std::max((logical_height + element.tall - 1u) / element.tall, 1u);
-				info.depth  = copy_depth;
+				info.width  = std::max((region.imageExtent.width + texture.texel_width - 1u) /
+				                           texture.texel_width,
+				                       1u);
+				info.height = std::max(
+				    (logical_height + texture.texel_height - 1u) / texture.texel_height, 1u);
+				info.depth = copy_depth;
 				info.surface_z =
 				    block.block_depth == 1 ? static_cast<uint32_t>(region.imageOffset.z) : 0;
-				info.pitch        = std::max((pitch + element.wide - 1u) / element.wide, 1u);
-				info.tail_x       = tail ? volume.tail_x[level] : 0;
-				info.tail_y       = tail ? volume.tail_y[level] : 0;
+				info.pitch = std::max((pitch + texture.texel_width - 1u) / texture.texel_width, 1u);
+				info.tail_x       = tail ? surface.mips[level].tail_x : 0;
+				info.tail_y       = tail ? surface.mips[level].tail_y : 0;
 				info.tail         = tail;
-				info.tiled_width  = volume.level_widths[level];
-				info.tiled_height = volume.level_heights[level];
+				info.tiled_width  = surface.mips[level].padded_width;
+				info.tiled_height = surface.mips[level].padded_height;
 				infos.push_back(info);
 			}
 			region_base += mip_depth;
 		}
 	} else {
-		const auto base_family = layout.tile_family;
-		if (base_family == TileBlockFamily::Count) {
-			return false;
-		}
-
 		size_t region_index = 0;
 		for (uint32_t level = 0; level < levels; level++) {
-			const auto&     level_size = layout.level_sizes[level];
-			const bool      tail       = level >= layout.first_tail_level;
-			const auto      family     = base_family;
-			TileBlockLayout block {};
-			if (!TileGetBlockLayout(family, element.bytes, block)) {
-				return false;
-			}
-			const auto level_depth = GetTextureLevelDepth(depth, level, layout.volume_texture);
+			const auto& level_size  = layout.mips[level];
+			const auto& mip         = surface.mips[level];
+			const bool  tail        = level >= surface.first_tail_level;
+			const auto  level_depth = GetTextureLevelDepth(depth, level, false);
 			for (uint32_t z = 0; z < level_depth; z++) {
 				const auto& region = regions[region_index++];
 				const auto  pitch =
@@ -585,39 +417,36 @@ bool TextureBuildGpuTileInfos(uint64_t tiled_size, const std::vector<vk::BufferI
 				                                 ? region.bufferImageHeight
 				                                 : region.imageExtent.height;
 				GpuTileInfo info {};
-				info.family            = block.family;
-				info.bytes_per_element = block.bytes_per_element;
-				info.linear_offset     = region.bufferOffset;
-				info.tiled_offset      = TextureUploadSliceSourceOffset(layout, level, z);
-				if (!SetGpuTileSize(info.linear_offset, level_size.size, UINT64_MAX,
-				                    info.linear_size) ||
-				    !SetGpuTileSize(info.tiled_offset, GetLevelSrcSize(level_size), tiled_size,
-				                    info.tiled_size)) {
+				info.family              = block.family;
+				info.bytes_per_element   = block.bytes_per_element;
+				info.linear_offset       = region.bufferOffset;
+				const auto source_stride = layout.source_slice_stride != 0
+				                               ? layout.source_slice_stride
+				                               : surface.block_slice_size;
+				if (z > (UINT64_MAX - mip.offset) / source_stride) {
 					return false;
 				}
-				info.width =
-				    std::max((region.imageExtent.width + element.wide - 1u) / element.wide, 1u);
-				info.height    = std::max((logical_height + element.tall - 1u) / element.tall, 1u);
-				info.surface_z = base_family == TileBlockFamily::RenderTarget64KB ||
-				                         base_family == TileBlockFamily::Depth64KB
+				info.tiled_offset = mip.offset + static_cast<uint64_t>(z) * source_stride;
+				if (!SetGpuTileSize(info.linear_offset, level_size.size, UINT64_MAX,
+				                    info.linear_size) ||
+				    !SetGpuTileSize(info.tiled_offset, mip.size, tiled_size, info.tiled_size)) {
+					return false;
+				}
+				info.width  = std::max((region.imageExtent.width + texture.texel_width - 1u) /
+				                           texture.texel_width,
+				                       1u);
+				info.height = std::max(
+				    (logical_height + texture.texel_height - 1u) / texture.texel_height, 1u);
+				info.surface_z = block.family == TileBlockFamily::RenderTarget64KB ||
+				                         block.family == TileBlockFamily::Depth64KB
 				                     ? region.imageSubresource.baseArrayLayer
 				                     : 0;
-				info.pitch     = std::max((pitch + element.wide - 1u) / element.wide, 1u);
-				info.tail      = tail;
-				info.tail_x    = tail ? level_size.x : 0;
-				info.tail_y    = tail ? level_size.y : 0;
-				info.tiled_width =
-				    layout.padded_sizes[level].width != 0
-				        ? std::max((layout.padded_sizes[level].width + element.wide - 1u) /
-				                       element.wide,
-				                   1u)
-				        : info.pitch;
-				info.tiled_height =
-				    layout.padded_sizes[level].height != 0
-				        ? std::max((layout.padded_sizes[level].height + element.tall - 1u) /
-				                       element.tall,
-				                   1u)
-				        : info.height;
+				info.pitch = std::max((pitch + texture.texel_width - 1u) / texture.texel_width, 1u);
+				info.tail  = tail;
+				info.tail_x       = tail ? mip.tail_x : 0;
+				info.tail_y       = tail ? mip.tail_y : 0;
+				info.tiled_width  = mip.padded_width;
+				info.tiled_height = mip.padded_height;
 				infos.push_back(info);
 			}
 		}

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"
 
@@ -19,34 +19,44 @@ 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;
+	uint32_t row_length   = 0;
+	uint32_t image_height = 0;
+};
+
 struct TextureUploadLayout {
-	uint32_t        tile                = 0;
-	uint32_t        pitch               = 0;
-	uint32_t        texel_block         = 1;
-	uint64_t        slice_stride        = 0;
-	uint64_t        source_slice_stride = 0;
-	uint32_t        first_tail_level    = 16;
-	TileBlockFamily tile_family         = TileBlockFamily::Count;
-	bool            volume_texture      = false;
-	TileSizeOffset  level_sizes[16]     = {};
-	TilePaddedSize  padded_sizes[16]    = {};
+	uint32_t               pitch               = 0;
+	uint64_t               slice_stride        = 0;
+	uint64_t               source_slice_stride = 0;
+	TileSurfaceLayout      surface;
+	TextureUploadMipLayout mips[16] = {};
 };
 
 vk::ComponentMapping   TextureGetComponentMapping(uint32_t swizzle);
-vk::Format             TextureGetFormat(uint32_t fmt);
-RenderTargetFormatInfo TextureGetRenderTargetFormat(uint32_t layout, uint32_t type, uint32_t order);
-TextureUploadLayout    TextureCalcUploadLayout(uint32_t fmt, uint64_t width, uint64_t height,
-                                               uint64_t levels, uint32_t depth, uint64_t pitch,
-                                               uint64_t tile, uint64_t upload_size,
+SurfaceFormatInfo      TextureGetSurfaceFormatInfo(Prospero::BufferFormat format);
+RenderTargetFormatInfo TextureGetRenderTargetFormat(Prospero::ChannelLayout layout,
+                                                    Prospero::ChannelType   type,
+                                                    Prospero::ChannelOrder  order);
+TextureUploadLayout    TextureCalcUploadLayout(Prospero::BufferFormat format, uint32_t width,
+                                               uint32_t height, uint32_t levels, uint32_t depth,
+                                               Prospero::TileMode tile, uint64_t upload_size,
                                                bool allow_depth_tile, bool volume_texture,
                                                const char* owner);
-std::vector<vk::BufferImageCopy> TextureBuildImageCopies(const TextureUploadLayout& layout,
-                                                         uint32_t width, uint32_t height,
-                                                         uint32_t depth, uint64_t levels,
-                                                         bool array_texture, bool volume_texture);
+std::vector<vk::BufferImageCopy> TextureBuildImageCopies(const TextureUploadLayout& layout);
 bool TextureBuildGpuTileInfos(uint64_t tiled_size, const std::vector<vk::BufferImageCopy>& regions,
-                              const TextureUploadLayout& layout, uint32_t fmt, uint32_t depth,
-                              uint64_t levels, std::vector<GpuTileInfo>& infos);
+                              const TextureUploadLayout& layout, uint32_t levels,
+                              std::vector<GpuTileInfo>& infos);
 
 } // namespace Libs::Graphics
 

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

@@ -1,300 +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) {
-	EXIT_IF(!texture.image_id || texture.image_view == nullptr ||
-	        texture.layout == vk::ImageLayout::eUndefined);
-	return {nullptr, texture.image_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);
-
-	const auto descriptor_count = program.info.buffers.size() + program.info.images.size() +
-	                              program.info.samplers.size() + program.info.addresses.size() + 3u;
-	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());
-
-	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));
-				}
-				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);
-	}
-	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

@@ -1,114 +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;
-	};
-
-	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);
-	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 +1,97 @@
+#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.sType          = vk::StructureType::eDescriptorSetAllocateInfo;
+	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.sType         = vk::StructureType::eDescriptorPoolCreateInfo;
+	create.flags         = {};
+	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 +12,54 @@
 #include <cstdint>
 #include <cstring>
 #include <type_traits>
+#include <vector>
 
 namespace Libs::Graphics {
 
+namespace ShaderRecompiler::IR {
+struct ResourceSnapshot;
+}
+
+struct BufferView {
+	vk::Buffer     buffer = nullptr;
+	vk::DeviceSize offset = 0;
+	vk::DeviceSize range  = VK_WHOLE_SIZE;
+};
+
+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;
+	BufferView                  gds;
+	BufferView                  flattened_srt;
+	BufferView                  user_data;
+};
+
+struct PreparedBindings {
+	const ShaderRecompiler::IR::Program*          program  = nullptr;
+	const ShaderRecompiler::IR::ResourceSnapshot* snapshot = nullptr;
+	NativeDescriptors                             resources;
+	std::vector<BufferId>                         buffer_ids;
+	std::vector<uint32_t>                         flattened_srt;
+	std::vector<uint32_t>                         user_data;
+	bool                                          committed = false;
+};
+
+[[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>);
@@ -34,9 +81,9 @@ 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);
+                                                    const Image& image, bool r128 = false);
 [[nodiscard]] bool IsSupportedDepthTextureEncoding(const ShaderTextureResource& descriptor,
-                                                   const Image&                 image);
+                                                   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

@@ -10,11 +10,14 @@
 #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/shaderCompiler.h"
 
 #include <atomic>
+#include <cstdio>
 #include <cstring>
 #include <span>
 #include <utility>
+#include <vector>
 
 namespace Libs::Graphics {
 
@@ -36,16 +39,132 @@ void NormalizeStaticParamsForDynamicState(PipelineStaticParameters& static_param
 
 } // namespace
 
+struct PipelineCache::ProgramCache {
+	struct SourceKey {
+		ShaderType stage = ShaderType::Unknown;
+		uint64_t   hash  = 0;
+
+		bool operator==(const SourceKey&) const = default;
+	};
+
+	struct Permutation {
+		std::vector<uint32_t>                               static_key;
+		std::shared_ptr<const ShaderRecompiler::IR::Program> program;
+		ShaderProgram                                       handle;
+	};
+
+	struct SourceKeyHash {
+		std::size_t operator()(const SourceKey& 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));
+			}
+			return hash;
+		}
+	};
+
+	static uint64_t MixId(uint64_t hash, uint64_t value) {
+		return hash ^ (value + 0x9e3779b97f4a7c15ull + (hash << 6u) + (hash >> 2u));
+	}
+
+	static constexpr std::size_t MaxStaticKeyWords =
+	    5 + ShaderVertexInputInfo::RES_MAX * 17;
+
+	template <typename InputInfo>
+	ShaderProgram Get(const ShaderParams& params, InputInfo& input_info) {
+		constexpr ShaderType stage = [] {
+			if constexpr (std::is_same_v<InputInfo, ShaderVertexInputInfo>) {
+				return ShaderType::Vertex;
+			} else if constexpr (std::is_same_v<InputInfo, ShaderPixelInputInfo>) {
+				return ShaderType::Pixel;
+			} else {
+				static_assert(std::is_same_v<InputInfo, ShaderComputeInputInfo>);
+				return ShaderType::Compute;
+			}
+		}();
+
+		BuildStageStaticKey(input_info, key_scratch);
+		auto& permutations = programs[{stage, params.hash}];
+		for (const auto& permutation: permutations) {
+			if (permutation.static_key == key_scratch &&
+			    MaterializeProgram(permutation.program, params, input_info)) {
+				return permutation.handle;
+			}
+		}
+
+		const auto module = CompileProgram(device, params, input_info);
+		EXIT_IF(module == nullptr || !input_info.stage);
+		uint64_t id = MixId(MixId(params.hash, static_cast<uint64_t>(stage)), permutations.size());
+		if (id == 0) {
+			id = 1;
+		}
+		const ShaderProgram handle {.id = id, .module = module};
+		permutations.push_back({key_scratch, input_info.stage.program, handle});
+
+		std::printf("Num compiled %u shaders\n", ++num_compiled);
+		return handle;
+	}
+
+	explicit ProgramCache(vk::Device device): device(device) {
+		key_scratch.reserve(MaxStaticKeyWords);
+	}
+
+	std::unordered_map<SourceKey, std::vector<Permutation>, SourceKeyHash> programs;
+	std::vector<uint32_t> key_scratch;
+	vk::Device device;
+	uint32_t   num_compiled = 0;
+};
+
+PipelineCache::PipelineCache(GraphicContext& graphics)
+    : m_graphics(graphics), m_program_cache(std::make_unique<ProgramCache>(graphics.device)) {
+	EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
+}
+
 PipelineCache::~PipelineCache() {
 	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);
+	for (const auto& [key, permutations]: m_program_cache->programs) {
+		(void)key;
+		for (const auto& permutation: permutations) {
+			m_graphics.device.destroyShaderModule(permutation.handle.module, nullptr);
+		}
+	}
+}
+
+ShaderProgram PipelineCache::GetVertexProgram(const HW::VertexShaderInfo& regs,
+                                              const HW::ShaderRegisters&  sh,
+                                              ShaderVertexInputInfo&      input_info) {
+	const auto params = PrepareProgram(regs, sh, input_info);
+	Common::LockGuard lock(m_mutex);
+	return m_program_cache->Get(params, input_info);
+}
+
+ShaderProgram PipelineCache::GetPixelProgram(
+    const HW::PixelShaderInfo& regs, const HW::ShaderRegisters& sh,
+    const ShaderVertexInputInfo&                        vertex_info,
+    std::span<const Prospero::ColorComponentMapping, 8> target_export_mapping,
+    ShaderPixelInputInfo&                               input_info) {
+	const auto params = PrepareProgram(regs, sh, vertex_info, target_export_mapping, input_info);
+	Common::LockGuard lock(m_mutex);
+	return m_program_cache->Get(params, input_info);
+}
+
+ShaderProgram PipelineCache::GetComputeProgram(const HW::ComputeShaderInfo& regs,
+                                               const HW::ShaderRegisters&   sh,
+                                               ShaderComputeInputInfo&      input_info) {
+	input_info.needs_lds_barriers = !m_graphics.compute_wave64_supported;
+	const auto params             = PrepareProgram(regs, sh, input_info);
+	Common::LockGuard lock(m_mutex);
+	return m_program_cache->Get(params, input_info);
 }
 
 bool PipelineStaticParameters::operator==(const PipelineStaticParameters& other) const noexcept {
@@ -53,23 +172,22 @@ bool PipelineStaticParameters::operator==(const PipelineStaticParameters& other)
 }
 
 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 ps_active,
-    std::span<const uint32_t> vs_spirv, std::span<const uint32_t> ps_spirv) {
+    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++) {
@@ -80,11 +198,8 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 	}
 	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 {};
@@ -120,6 +235,12 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 			     depth.samples);
 		}
 	}
+	if (color_count == 0 && !with_depth) {
+		attachment_samples = render_sample_count(ctx.GetAaConfig().msaa_num_samples);
+		EXIT_IF(!static_cast<bool>(
+		    m_graphics.GetPhysicalDeviceProperties().limits.framebufferNoAttachmentsSampleCounts &
+		    vulkan_sample_count(attachment_samples)));
+	}
 	EXIT_IF(attachment_samples == 0 ||
 	        vulkan_sample_count(attachment_samples) == vk::SampleCountFlagBits {});
 
@@ -130,12 +251,12 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 		}
 	}
 
-	const auto& clip_control = ctx.GetClipControl();
-	EXIT_NOT_IMPLEMENTED(!clip_control.IsZClipModeRepresentable());
-	static_params.negative_one_to_one = !clip_control.dx_clip_space;
-	static_params.depth_clip_enable   = clip_control.IsZClipEnabled();
-	static_params.topology            = topology;
-	static_params.samples             = attachment_samples;
+	const auto& clip_control               = ctx.GetClipControl();
+	static_params.negative_one_to_one      = !clip_control.dx_clip_space;
+	static_params.depth_clip_enable        = clip_control.IsZClipEnabled();
+	static_params.topology                 = topology;
+	static_params.primitive_restart_enable = primitive_restart_enable;
+	static_params.samples                  = attachment_samples;
 	static_params.sample_shading_enable =
 	    ps_active && attachment_samples > 1 && ps_input_info->ps_sample_shading;
 	if (static_params.sample_shading_enable && !m_graphics.sample_rate_shading_enabled) {
@@ -194,20 +315,17 @@ 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);
+	LogPipelineTrace("CreatePipelineInternal begin", vs_id, ps_id);
+	CreatePipelineInternal(m_graphics, *cached, rendering, vs_input_info, vertex_program.module,
+	                       ps_input_info, pixel_program.module, static_params);
+	LogPipelineTrace("CreatePipelineInternal done", vs_id, ps_id);
 
 	EXIT_NOT_IMPLEMENTED(cached->pipeline == nullptr);
 	EXIT_NOT_IMPLEMENTED(cached->pipeline_layout == nullptr);
@@ -219,19 +337,16 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 }
 
 PipelineCache::ComputePipeline&
-PipelineCache::CreateComputePipeline(ShaderComputeInputInfo&      input_info,
-                                     const HW::ComputeShaderInfo& cs_regs,
-                                     std::span<const uint32_t>    cs_spirv) {
+PipelineCache::CreateComputePipeline(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;
+	p.cs_shader_id = compute_program.id;
 
 	ComputePipelineKey key {};
 	key.cs_shader_id = p.cs_shader_id;
@@ -245,7 +360,7 @@ PipelineCache::CreateComputePipeline(ShaderComputeInputInfo&      input_info,
 	}
 
 	auto cached = std::make_unique<ComputePipeline>(p);
-	CreatePipelineInternal(m_graphics, m_descriptor_cache, *cached, input_info, cs_spirv);
+	CreatePipelineInternal(m_graphics, *cached, input_info, compute_program.module);
 
 	EXIT_NOT_IMPLEMENTED(cached->pipeline == nullptr);
 	EXIT_NOT_IMPLEMENTED(cached->pipeline_layout == nullptr);

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

@@ -20,8 +20,7 @@ namespace Libs::Graphics {
 struct GraphicContext;
 struct RenderColorInfo;
 struct RenderDepthInfo;
-class RenderCommandBuffer;
-class DescriptorCache;
+class CommandBuffer;
 
 namespace HW {
 class Context;
@@ -38,6 +37,7 @@ struct PipelineStaticParameters {
 	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;
@@ -79,9 +79,9 @@ 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(uint32_t) + sizeof(bool) * 4 +
-                  sizeof(vk::CompareOp) + sizeof(bool) + sizeof(float) * 2 + sizeof(bool) +
-                  sizeof(PipelineStencilStaticState) * 2 + sizeof(uint32_t) +
+                  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);
@@ -95,44 +95,65 @@ struct PipelineRenderingState {
 	bool operator==(const PipelineRenderingState&) 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);
 
 	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;
+		uint64_t vs_shader_id = 0;
+		uint64_t ps_shader_id = 0;
 	};
 
 	struct ComputePipeline: Pipeline {
-		ShaderId cs_shader_id;
+		uint64_t cs_shader_id = 0;
 	};
 
-	GraphicsPipeline&
-	CreateGraphicsPipeline(RenderColorInfo* colors, uint32_t color_count, RenderDepthInfo& depth,
-	                       ShaderVertexInputInfo& vs_input_info, RenderCommandBuffer& command,
-	                       ShaderPixelInputInfo* ps_input_info, vk::PrimitiveTopology topology,
-	                       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);
+	ShaderProgram GetVertexProgram(const HW::VertexShaderInfo& regs,
+	                               const HW::ShaderRegisters&  sh,
+	                               ShaderVertexInputInfo&      input_info);
+	ShaderProgram
+	GetPixelProgram(const HW::PixelShaderInfo& regs, const HW::ShaderRegisters& sh,
+	                const ShaderVertexInputInfo&                        vertex_info,
+	                std::span<const Prospero::ColorComponentMapping, 8> target_export_mapping,
+	                ShaderPixelInputInfo&                               input_info);
+	ShaderProgram GetComputeProgram(const HW::ComputeShaderInfo& regs,
+	                                const HW::ShaderRegisters& sh,
+	                                ShaderComputeInputInfo& input_info);
+
+	GraphicsPipeline& 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);
+	ComputePipeline& CreateComputePipeline(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;
 		PipelineStaticParameters static_params;
 
 		bool operator==(const GraphicsPipelineKey& other) const {
@@ -142,7 +163,7 @@ private:
 	};
 
 	struct ComputePipelineKey {
-		ShaderId cs_shader_id;
+		uint64_t cs_shader_id = 0;
 
 		bool operator==(const ComputePipelineKey& other) const {
 			return cs_shader_id == other.cs_shader_id;
@@ -155,15 +176,6 @@ 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++) {
@@ -185,8 +197,8 @@ 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::MixStaticParams(hash, key.static_params);
 			return hash;
 		}
@@ -195,13 +207,13 @@ private:
 	struct ComputePipelineKeyHash {
 		std::size_t operator()(const ComputePipelineKey& key) const {
 			std::size_t hash = 0;
-			PipelineKeyHash::MixShaderId(hash, key.cs_shader_id);
+			PipelineKeyHash::Mix(hash, key.cs_shader_id);
 			return hash;
 		}
 	};
 
-	GraphicContext&  m_graphics;
-	DescriptorCache& m_descriptor_cache;
+	GraphicContext&                m_graphics;
+	std::unique_ptr<ProgramCache> m_program_cache;
 	std::unordered_map<GraphicsPipelineKey, std::unique_ptr<GraphicsPipeline>,
 	                   GraphicsPipelineKeyHash>
 	    m_graphics_pipelines;
@@ -210,19 +222,15 @@ private:
 	Common::Mutex m_mutex;
 };
 
-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::GraphicsPipeline& pipeline,
+    const PipelineRenderingState& rendering, const ShaderVertexInputInfo& vs_input_info,
+    vk::ShaderModule vertex_module, const ShaderPixelInputInfo* ps_input_info,
+	vk::ShaderModule pixel_module, const PipelineStaticParameters& static_params);
+void CreatePipelineInternal(GraphicContext& graphics, PipelineCache::ComputePipeline& pipeline,
+                            const ShaderComputeInputInfo& input_info,
+                            vk::ShaderModule              compute_module);
 
 } // namespace Libs::Graphics
 

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

@@ -36,6 +36,22 @@ VulkanMemoryBarrier MakeShaderWriteDependency() {
 	return barrier;
 }
 
+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;
+}
+
+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;
+}
+
 vk::BufferMemoryBarrier MakeGdsDependency(vk::Buffer buffer) {
 	EXIT_IF(buffer == nullptr);
 
@@ -84,6 +100,21 @@ bool HasShaderBufferWrites(const ShaderStageRuntime& runtime) {
 	return !CollectShaderBufferWrites(*runtime.program, *runtime.resources).empty();
 }
 
+void ShaderAccessBarrier(vk::CommandBuffer vk_buffer, vk::PipelineStageFlags source_stages) {
+	EXIT_IF(vk_buffer == nullptr || !source_stages);
+	const auto barrier = MakeShaderAccessDependency();
+	vk_buffer.pipelineBarrier(source_stages, vk::PipelineStageFlagBits::eAllCommands,
+	                          vk::DependencyFlags {}, 1, &barrier, 0, nullptr, 0, nullptr);
+}
+
+void ShaderWriteHazardBarrier(vk::CommandBuffer      vk_buffer,
+                              vk::PipelineStageFlags destination_stages) {
+	EXIT_IF(vk_buffer == nullptr || !destination_stages);
+	const auto barrier = MakeShaderWriteHazardDependency();
+	vk_buffer.pipelineBarrier(vk::PipelineStageFlagBits::eAllCommands, destination_stages,
+	                          vk::DependencyFlags {}, 1, &barrier, 0, nullptr, 0, nullptr);
+}
+
 void ShaderWriteBarrier(vk::CommandBuffer vk_buffer, vk::PipelineStageFlags source_stages) {
 	EXIT_IF(vk_buffer == nullptr || !source_stages);
 	const auto barrier = MakeShaderWriteDependency();

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

@@ -2,7 +2,7 @@
 #define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_SHADERRESOURCEBARRIER_H_
 
 #include "graphics/host_gpu/graphicContext.h"
-#include "graphics/shader/recompiler/ir/ResourceMaterialization.h"
+#include "graphics/shader/recompiler/ir/passes/ResourceMaterialization.h"
 
 #include <vector>
 
@@ -18,12 +18,17 @@ struct ShaderBufferWriteRange {
 };
 
 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,
+                              vk::PipelineStageFlags destination_stages);
 void ShaderWriteBarrier(vk::CommandBuffer vk_buffer, vk::PipelineStageFlags source_stages);
 
 } // namespace Libs::Graphics

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

@@ -1,60 +0,0 @@
-#include "graphics/host_gpu/renderer/pipeline/shaderSubgroup.h"
-
-#include "graphics/shader/recompiler/emitter/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"
@@ -20,10 +19,13 @@
 #include <algorithm>
 #include <limits>
 #include <span>
+#include <vector>
 
 namespace Libs::Graphics {
 
 // IDK: maybe we can remove it?
+constexpr uint8_t kTemporaryVertexAttribFormat113 =
+    static_cast<uint8_t>(Prospero::VertexAttribFormat::k16_16SInt);
 constexpr uint32_t kTemporaryPs5BufferFormat121 = 121u;
 
 static bool NarrowInputFormat(vk::Format& format, uint32_t& size, uint32_t used_components) {
@@ -88,8 +90,9 @@ static bool NarrowInputFormat(vk::Format& format, uint32_t& size, uint32_t used_
 
 static void GetInputFormat(const ShaderBufferResource& res, vk::Format& format, uint32_t& size,
                            uint32_t used_components) {
-	auto fmt = res.Format();
-	if (fmt == Prospero::GpuEnumValue(Prospero::VertexAttribFormat::k16_16SInt)) {
+	const auto fmt        = res.Format();
+	const auto raw_format = res.RawFormat();
+	if (raw_format == kTemporaryVertexAttribFormat113) {
 		static bool logged_113 = false;
 		if (!logged_113) {
 			LOGF("InputFormat: temporary: accepting invalid PS5 buffer format 113 as "
@@ -99,12 +102,12 @@ static void GetInputFormat(const ShaderBufferResource& res, vk::Format& format,
 		format = vk::Format::eR32G32B32A32Sfloat;
 		size   = 4;
 		if (NarrowInputFormat(format, size, used_components)) {
-			LOGF("InputFormat: narrowing fmt=%u to %s for used_components=%u\n", fmt,
+			LOGF("InputFormat: narrowing fmt=%u to %s for used_components=%u\n", raw_format,
 			     VulkanToString(format).c_str(), used_components);
 		}
 		return;
 	}
-	if (fmt == kTemporaryPs5BufferFormat121) {
+	if (raw_format == kTemporaryPs5BufferFormat121) {
 		static bool logged_121 = false;
 		if (!logged_121) {
 			LOGF("InputFormat: accepting PS5 buffer format 121 as vk::Format::eR16G16Sfloat\n");
@@ -113,13 +116,13 @@ static void GetInputFormat(const ShaderBufferResource& res, vk::Format& format,
 		format = vk::Format::eR16G16Sfloat;
 		size   = 2;
 		if (NarrowInputFormat(format, size, used_components)) {
-			LOGF("InputFormat: narrowing fmt=%u to %s for used_components=%u\n", fmt,
+			LOGF("InputFormat: narrowing fmt=%u to %s for used_components=%u\n", raw_format,
 			     VulkanToString(format).c_str(), used_components);
 		}
 		return;
 	}
 
-	switch (static_cast<Prospero::BufferFormat>(fmt)) {
+	switch (fmt) {
 		case Prospero::BufferFormat::k32_32_32_32Float:
 			format = vk::Format::eR32G32B32A32Sfloat;
 			size   = 4;
@@ -329,7 +332,7 @@ static void GetInputFormat(const ShaderBufferResource& res, vk::Format& format,
 			size   = 1;
 			break;
 		default:
-			EXIT("unknown format: fmt = %u\n", fmt);
+			EXIT("unknown format: fmt = %u\n", raw_format);
 			format = vk::Format::eUndefined;
 			size   = 4;
 			break;
@@ -386,116 +389,52 @@ 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::Program& 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.sType        = vk::StructureType::eDescriptorSetLayoutCreateInfo;
+	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::GraphicsPipeline& pipeline,
+    const PipelineRenderingState& rendering, const ShaderVertexInputInfo& vs_input_info,
+    vk::ShaderModule vertex_module, const ShaderPixelInputInfo* ps_input_info,
+	vk::ShaderModule pixel_module, const PipelineStaticParameters& static_params) {
+	const bool ps_active = ps_input_info != nullptr;
+	EXIT_IF(vertex_module == nullptr || (ps_active && pixel_module == nullptr));
 
 	const bool rect_list = 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);
@@ -520,40 +459,27 @@ void CreatePipelineInternal(
 		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.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.module              = vertex_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 {};
+	vk::PipelineShaderStageCreateInfo frag_shader_stage_info {};
 	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.module              = pixel_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 tess_control_shader_stage_info {};
 	tess_control_shader_stage_info.sType  = vk::StructureType::ePipelineShaderStageCreateInfo;
@@ -596,9 +522,10 @@ void CreatePipelineInternal(
 
 			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);
+			const auto compiled_components =
+			    vs_input_info.stage.program->info.vertex_fetch_components[index];
+			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));
 
@@ -723,7 +650,8 @@ void CreatePipelineInternal(
 	input_assembly.pNext    = nullptr;
 	input_assembly.flags    = {};
 	input_assembly.topology = static_params.topology;
-	input_assembly.primitiveRestartEnable = VK_FALSE;
+	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];
@@ -873,40 +801,36 @@ void CreatePipelineInternal(
 	color_blending.blendConstants[2] = static_params.blend_color_blue;
 	color_blending.blendConstants[3] = static_params.blend_color_alpha;
 
-	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,
+	                  vk::ShaderStageFlagBits::eVertex);
 	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);
+	constexpr auto GraphicsStages =
+	    vk::ShaderStageFlagBits::eVertex | 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",
+		     pipeline.vs_shader_id, pipeline.ps_shader_id, 1u);
 	}
 	result = graphics.device.createPipelineLayout(&pipeline_layout_info, nullptr,
 	                                              &pipeline.pipeline_layout);
@@ -947,15 +871,23 @@ void CreatePipelineInternal(
 	    vk::DynamicState::eViewport,
 	    vk::DynamicState::eScissor,
 	    vk::DynamicState::eLineWidth,
+	    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
 	};
-	const auto dynamic_states_count =
+	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--;
+	}
+#endif
 
 	vk::PipelineDynamicStateCreateInfo dynamic_state {};
 	dynamic_state.sType             = vk::StructureType::ePipelineDynamicStateCreateInfo;
@@ -997,11 +929,12 @@ void CreatePipelineInternal(
 	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
+		LOGF("PipelineTrace: vkCreateGraphicsPipelines begin VS=%" PRIu64 " PS=%" PRIu64
+		     " 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),
+		     pipeline.vs_shader_id, pipeline.ps_shader_id,
+		     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,
@@ -1017,40 +950,19 @@ void CreatePipelineInternal(
 
 	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::ComputePipeline& pipeline,
+                            const ShaderComputeInputInfo& input_info,
+                            vk::ShaderModule              compute_module) {
+	EXIT_IF(compute_module == nullptr);
 
 	vk::PipelineShaderStageCreateInfo                     comp_shader_stage_info {};
 	vk::PipelineShaderStageRequiredSubgroupSizeCreateInfo comp_subgroup_size {};
@@ -1058,40 +970,41 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	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.module              = compute_module;
 	comp_shader_stage_info.pName               = "main";
 	comp_shader_stage_info.pSpecializationInfo = nullptr;
 	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.sType =
+		    vk::StructureType::ePipelineShaderStageRequiredSubgroupSizeCreateInfo;
+		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));
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
@@ -1117,8 +1030,6 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	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;
@@ -49,20 +48,6 @@ enum class CommandBufferDebugOp : uint32_t {
 	Unknown,
 };
 
-class FenceResourceRetainer {
-public:
-	FenceResourceRetainer() = default;
-	~FenceResourceRetainer();
-	KYTY_CLASS_NO_COPY(FenceResourceRetainer);
-
-	void               Retain(std::shared_ptr<void> resource);
-	void               ReleaseAfterFence() noexcept;
-	[[nodiscard]] bool Empty() const noexcept { return m_resources.empty(); }
-
-private:
-	std::vector<std::shared_ptr<void>> m_resources;
-};
-
 struct SubmitInfo {
 	static constexpr uint32_t MaxSemaphores = 3;
 
@@ -91,78 +76,52 @@ 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 {
@@ -170,66 +129,69 @@ 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,
+	void DrawIndex(uint64_t submit_id, CommandBuffer& 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 DrawAuto(uint64_t submit_id, CommandBuffer& 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, 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:
 	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 log_pipeline_phase,
+	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,
+	[[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);
 
-	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;
+	std::array<uint32_t, ShaderRecompiler::IR::NativePushConstantSize / sizeof(uint32_t)>
+	    m_push_constants {};
 
 	friend struct RenderExecutorTestAccess;
 };

src/graphics/host_gpu/renderer/renderCompute.cpp

@@ -11,16 +11,14 @@
 #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"
-#include "graphics/shader/recompiler/ir/ResourceMaterialization.h"
 #include "graphics/shader/recompiler/ir/ShaderIR.h"
+#include "graphics/shader/recompiler/ir/passes/ResourceMaterialization.h"
 #include "graphics/shader/shader.h"
 #include "kernel/eventQueue.h"
 #include "kernel/pthread.h"
@@ -46,7 +44,7 @@ static uint64_t BufferDescriptorSize(const ShaderBufferResource& descriptor) {
 }
 
 bool RenderExecutor::TryConsumeComputeMetaClear(const ShaderComputeInputInfo& input,
-                                                const RenderCommandBuffer&    buffer) {
+                                                const CommandBuffer&          buffer) {
 	const auto& program   = *input.stage.program;
 	const auto& resources = *input.stage.resources;
 	if (resources.buffers.size() != program.info.buffers.size()) {
@@ -83,9 +81,8 @@ bool ResolveComputeImageClear(const ShaderComputeInputInfo& input, uint32_t grou
 	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()) {
+	    !program.info.images.empty() || !program.info.samplers.empty() || program.info.uses_dma ||
+	    !resources.images.empty() || !resources.samplers.empty()) {
 		return false;
 	}
 	const auto& resource   = program.info.buffers.front();
@@ -93,7 +90,7 @@ bool ResolveComputeImageClear(const ShaderComputeInputInfo& input, uint32_t grou
 	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::GpuEnumValue(Prospero::BufferFormat::k32_32_32_32UInt) ||
+	    descriptor.Format() != Prospero::BufferFormat::k32_32_32_32UInt ||
 	    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) {
@@ -115,7 +112,7 @@ bool ResolveComputeImageClear(const ShaderComputeInputInfo& input, uint32_t grou
 	    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 &&
+	    input.thread_ids_num == 1 && input.wave_size == 64 && !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) {
@@ -139,6 +136,21 @@ static bool TryConsumeComputeImageClear(const ShaderComputeInputInfo& input, Com
 	}
 	auto& cache = command.GetContext().GetTextureCache();
 	if (!cache.ClearImageFromBuffer(command, descriptor.Base48(), size, packed_clear)) {
+		// Recognized metadata-fill shaders access DCC as an ordinary storage buffer and may run
+		// before the render target is bound. TryConsumeDccFill either consumes registered state
+		// or retains a PendingDcc fill while allowing the dispatch to run.
+		const bool registered_metadata =
+		    cache.TryConsumeDccFill(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: %s metadata clear shader=0x%016" PRIx64
+			     " addr=0x%016" PRIx64 " size=0x%016" PRIx64 " value=0x%08" PRIx32 "\n",
+			     registered_metadata ? "tracked" : "deferred", input.stage.program->shader_hash,
+			     descriptor.Base48(), size, packed_clear);
+		}
+		if (registered_metadata) {
+			return true;
+		}
 		return false;
 	}
 	static std::atomic<uint32_t> logged_clears {0};
@@ -150,10 +162,11 @@ 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 +207,12 @@ 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;
@@ -246,7 +255,7 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 			LOGF("  CS buffer[%u]: source=%u usage=%s addr=0x%012" PRIx64
 			     " stride=%u records=%u format=%u\n",
 			     i, buffer.source, buffer.written ? "read-write" : "read-only", r.Base48(),
-			     r.Stride(), r.NumRecords(), r.Format());
+			     r.Stride(), r.NumRecords(), r.RawFormat());
 		}
 		for (uint32_t i = 0; i < program.info.images.size(); i++) {
 			const auto& image = program.info.images[i];
@@ -258,13 +267,14 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 			      image.kind == ShaderRecompiler::IR::ResourceKind::ImageUint)
 			         ? "true"
 			         : "false",
-			     r.Base40(), static_cast<uint32_t>(r.Type()), r.Format(),
+			     r.Base40(), static_cast<uint32_t>(r.Type()), static_cast<uint32_t>(r.Format()),
 			     static_cast<uint32_t>(r.Width5()) + 1u, static_cast<uint32_t>(r.Height5()) + 1u,
 			     static_cast<uint32_t>(r.Depth()) + 1u,
-			     std::max<uint32_t>(static_cast<uint32_t>(r.LastLevel()),
-			                        static_cast<uint32_t>(r.MaxMip())) +
-			         1u,
-			     r.TileMode());
+			     r.Type() == Prospero::ImageType::kColor2DMsaa ||
+			             r.Type() == Prospero::ImageType::kColor2DMsaaArray
+			         ? 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++) {
 			const auto r = DecodeNativeDescriptor<ShaderSamplerResource>(resources.samplers[i]);
@@ -317,17 +327,19 @@ 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);
-
-	auto vk_buffer = buffer.Handle();
-	CommitBindings(buffer, vk::PipelineBindPoint::eCompute, pipeline.pipeline_layout, bindings);
-	vk_buffer.bindPipeline(vk::PipelineBindPoint::eCompute, pipeline.pipeline);
-	vk_buffer.dispatch(thread_group_x, thread_group_y, thread_group_z);
+	    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();
+	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(),
@@ -338,8 +350,15 @@ void RenderExecutor::DispatchDirect(uint64_t submit_id, RenderCommandBuffer& buf
 	                }) ||
 	    has_storage_writes;
 	if (has_storage_writes) {
-		ShaderWriteBarrier(vk_buffer, vk::PipelineStageFlagBits::eComputeShader);
+		// A host fence used to serialize every dispatch. Preserve its read-before-write ordering
+		// while allowing the queue to execute asynchronously.
+		ShaderWriteHazardBarrier(vk_buffer, vk::PipelineStageFlagBits::eComputeShader);
 	}
+	vk_buffer.bindPipeline(vk::PipelineBindPoint::eCompute, pipeline.pipeline);
+	vk_buffer.dispatch(thread_group_x, thread_group_y, thread_group_z);
+
+	// The removed host fence also ordered read-only dispatches before later writers.
+	ShaderAccessBarrier(vk_buffer, vk::PipelineStageFlagBits::eComputeShader);
 	ResetBindings();
 }
 

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());
 }
 
@@ -26,7 +26,7 @@ RenderContext::~RenderContext() {
 void RenderContext::InitializeGpu(VideoOut::VideoOutDriver* video_out) {
 	EXIT_IF(m_gpu != nullptr);
 	m_video_out = video_out;
-	m_gpu       = std::make_unique<Gpu>(*this);
+	m_gpu       = std::make_unique<GuestGpu>(*this);
 	m_gpu_resources.SetGpu(m_gpu.get());
 }
 
@@ -38,14 +38,14 @@ 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;
 	}
 }
 
-Gpu& RenderContext::GetGpu() const {
+GuestGpu& RenderContext::GetGpu() const {
 	EXIT_IF(m_gpu == nullptr);
 	return *m_gpu;
 }
@@ -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"
 
@@ -23,8 +23,7 @@ class VideoOutDriver;
 
 namespace Libs::Graphics {
 
-constexpr int AGC_USER_INTERRUPT_EVENT = 0x1800;
-class Gpu;
+class GuestGpu;
 
 class RenderContext {
 public:
@@ -35,43 +34,42 @@ public:
 	[[nodiscard]] GraphicContext&           GetGraphics() const noexcept { return m_graphics; }
 	void                                    InitializeGpu(VideoOut::VideoOutDriver* video_out);
 	void                                    ShutdownGpu();
-	[[nodiscard]] Gpu&                      GetGpu() const;
+	[[nodiscard]] GuestGpu&                 GetGpu() const;
 	[[nodiscard]] VideoOut::VideoOutDriver& GetVideoOut() const;
 
 	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<Gpu>      m_gpu;
+	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

@@ -1,43 +1,12 @@
 #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;
-
-	float viewport_scale[3]  = {};
-	float viewport_offset[3] = {};
-	int   scissor_ltrb[4]    = {0};
-
-	float    line_width                                       = 1.0f;
-	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);
-static_assert(sizeof(PipelineDynamicParameters) ==
-              sizeof(bool) + sizeof(float[3]) + sizeof(float[3]) + sizeof(int[4]) + sizeof(float) +
-                  sizeof(uint32_t) + sizeof(bool[RENDER_COLOR_ATTACHMENTS_MAX]) +
-                  sizeof(PipelineStencilDynamicState) * 2);
-
 [[nodiscard]] int32_t ResolveVertexOffset(uint32_t                     index_offset,
                                           const ShaderVertexInputInfo& vs_input_info);
 

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

@@ -72,7 +72,10 @@ 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) {
@@ -199,20 +202,6 @@ void GraphicContext::CreateBuffer(uint64_t size, VulkanBuffer& buffer) {
 	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);

src/graphics/host_gpu/vulkanCommon.cpp

@@ -83,11 +83,11 @@ constexpr FormatMapping kFormatMappings[] = {
 };
 
 constexpr auto MakeFormatLookup() {
-	constexpr auto kMaxFormat = Prospero::GpuEnumValue(Prospero::BufferFormat::kBc7Srgb);
+	constexpr auto kMaxFormat = static_cast<size_t>(Prospero::BufferFormat::kBc7Srgb);
 	std::array<vk::Format, kMaxFormat + 1> lookup {};
 	lookup.fill(vk::Format::eUndefined);
 	for (const auto& mapping: kFormatMappings) {
-		lookup[Prospero::GpuEnumValue(mapping.guest)] = mapping.host;
+		lookup[static_cast<size_t>(mapping.guest)] = mapping.host;
 	}
 	return lookup;
 }
@@ -112,9 +112,9 @@ std::string VulkanToString(vk::QueueFlags value) {
 	return vk::to_string(value);
 }
 
-vk::Format VulkanFormat(uint32_t guest_format) {
-	return guest_format < kFormatLookup.size() ? kFormatLookup[guest_format]
-	                                           : vk::Format::eUndefined;
+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;
 }
 
 void RequireVulkanSuccess(vk::Result result, const char* operation) {

src/graphics/host_gpu/vulkanCommon.h

@@ -10,8 +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>
@@ -20,18 +25,31 @@ namespace Libs::Graphics {
 
 using VulkanMemoryBarrier = vk::MemoryBarrier;
 
-template <typename Handle>
-[[nodiscard]] void* VulkanHandleToPointer(Handle handle) {
-	return reinterpret_cast<void*>(static_cast<typename Handle::CType>(handle));
-}
-
 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(uint32_t guest_format);
+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 (;;) {

src/graphics/host_gpu/vulkanInstance.h

@@ -1,93 +0,0 @@
-#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_VULKANINSTANCE_H_
-#define EMULATOR_SRC_GRAPHICS_HOST_GPU_VULKANINSTANCE_H_
-
-#include "common/abi.h"
-#include "common/common.h"
-#include "common/threads.h"
-#include "graphics/host_gpu/vulkanCommon.h"
-
-#include <map>
-#include <mutex>
-#include <tuple>
-#include <vk_mem_alloc.h>
-
-namespace Libs::Graphics {
-
-inline constexpr uint32_t VULKAN_TARGET_API_VERSION = VK_API_VERSION_1_3;
-
-struct VulkanInstance {
-	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                               subgroup_size_control_enabled     = false;
-	bool                               sample_rate_shading_enabled       = false;
-	uint32_t                           subgroup_size                     = 0;
-	uint32_t                           min_subgroup_size                 = 0;
-	uint32_t                           max_subgroup_size                 = 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]] vk::DeviceSize StorageMinAlignment() const {
-		const auto alignment = physical_device_properties.limits.minStorageBufferOffsetAlignment;
-		return alignment != 0 ? alignment : 1;
-	}
-
-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;
-};
-
-} // namespace Libs::Graphics
-
-#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_VULKANINSTANCE_H_