kernel: preserve guest pthread priorities

Extracted from Stefanos Costa's pthread fix in #147.
2026-10-29 08:41:20 +00:00 · 2026-08-03 01:05:00 +02:00
418 changed files with 70442 additions and 101527 deletions
@@ -1,6 +1,6 @@
-name: Game Emulation Status Report
-description: Submit a KytyPS5 game compatibility result, including boot, rendering, audio, input, or gameplay behavior.
-title: "[GAME STATUS]: "
+name: Game Emulation Bug Report
+description: Report KytyPS5 game boot, crash, freeze, rendering, audio, input, or gameplay behavior problems.
+title: "[GAME BUG]: "
 body:
  - type: markdown
    attributes:
@@ -10,7 +10,6 @@ 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
@@ -45,28 +44,6 @@ 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:
@@ -77,22 +54,22 @@ body:
  - type: markdown
    attributes:
      value: |
-        ## Test result
+        ## Problem details

  - type: textarea
    id: what-happened
    attributes:
-      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."
+      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."
    validations:
      required: true

  - type: textarea
    id: steps-to-reproduce
    attributes:
-      label: Steps to reproduce the result
-      description: List the shortest path to the reported status or failure from a fresh KytyPS5 start.
+      label: Steps to reproduce
+      description: List the shortest path that triggers the problem from a fresh KytyPS5 start.
      placeholder: |
        1. Open KytyPS5.
        2. Boot the game.
@@ -104,7 +81,7 @@ body:
    id: expected-behavior
    attributes:
      label: Expected behavior
-      description: Explain what should happen at the same point in the game. If there is no known problem, write "No known issue".
+      description: Explain what should happen at the same point in the game.
      placeholder: "Example: The save should load and gameplay should continue normally."
    validations:
      required: true
@@ -6,32 +6,9 @@ on:
    branches: [ main, master ]
  pull_request:

-concurrency:
-  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
-  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
-
-env:
-  SCCACHE_GHA_ENABLED: "true"
-  KYTY_BUILD_ORIGIN: ${{ github.repository != 'KytyPS5/KytyPS5' && 'Fork' || github.event_name == 'push' && 'Official' || 'Source' }}
-  KYTY_BUILD_REPOSITORY: ${{ github.repository }}
-
 jobs:
-  metadata:
-    name: Prepare build metadata
-    runs-on: ubuntu-latest
-    outputs:
-      release_tag: ${{ steps.release_tag.outputs.value }}
-
-    steps:
-      - name: Generate release tag
-        id: release_tag
-        shell: bash
-        run: |
-          echo "value=KytyPS5-$(date -u +'%Y-%m-%d')-${GITHUB_SHA::7}" >> "$GITHUB_OUTPUT"
-
  windows:
    name: Build KytyPS5 (Windows)
-    needs: metadata
    runs-on: windows-2022

    steps:
@@ -83,9 +60,6 @@ jobs:
          arch: win64_msvc2022_64
          cache: true

-      - name: Setup sccache
-        uses: mozilla-actions/sccache-action@v0.0.11
-
      - name: Verify toolchain
        shell: cmd
        run: |
@@ -102,25 +76,19 @@ 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 }} ^
            -DCMAKE_C_COMPILER=clang-cl ^
            -DCMAKE_CXX_COMPILER=clang-cl ^
-            -DCMAKE_C_COMPILER_LAUNCHER=sccache ^
-            -DCMAKE_CXX_COMPILER_LAUNCHER=sccache ^
            -DCMAKE_PREFIX_PATH="%Qt6_DIR%"

      - name: Build
        shell: cmd
        run: |
-          cmake --build _Build/windows --target launcher audio_out2_port_tests ime_dialog_tests virtual_memory_allocation_tests --parallel
+          cmake --build _Build/windows --target launcher audio_out2_port_tests virtual_memory_allocation_tests --parallel

      - name: Test
        shell: cmd
        run: |
-          ctest --test-dir _Build/windows --output-on-failure -R "^(audio_out2_port|ime_dialog|virtual_memory_allocation)$"
+          ctest --test-dir _Build/windows --output-on-failure -R "^(audio_out2_port|virtual_memory_allocation)$"

      - name: Install
        shell: cmd
@@ -136,7 +104,6 @@ jobs:

  macos:
    name: Build KytyPS5 (macOS)
-    needs: metadata
    runs-on: macos-15

    steps:
@@ -166,9 +133,6 @@ jobs:
          arch: clang_64
          cache: true

-      - name: Setup sccache
-        uses: mozilla-actions/sccache-action@v0.0.11
-
      - name: Verify toolchain
        shell: bash
        run: |
@@ -186,29 +150,23 @@ 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="${{ needs.metadata.outputs.release_tag }}" \
            -DCMAKE_OSX_ARCHITECTURES=x86_64 \
            -DCMAKE_C_COMPILER=clang \
            -DCMAKE_CXX_COMPILER=clang++ \
-            -DCMAKE_C_COMPILER_LAUNCHER=sccache \
-            -DCMAKE_CXX_COMPILER_LAUNCHER=sccache \
            -DCMAKE_PREFIX_PATH="$QT_ROOT_DIR"

      - name: Build
        shell: bash
        run: |
          cmake --build _Build/macos \
-            --target launcher audio_out2_port_tests ime_dialog_tests virtual_memory_allocation_tests \
+            --target launcher audio_out2_port_tests virtual_memory_allocation_tests \
            --parallel

      - name: Test
        shell: bash
        run: |
          ctest --test-dir _Build/macos --output-on-failure \
-            -R '^(audio_out2_port|ime_dialog|virtual_memory_allocation)$'
+            -R '^(audio_out2_port|virtual_memory_allocation)$'

      - name: Install
        shell: bash
@@ -223,71 +181,38 @@ jobs:
        run: |
          archive="$RUNNER_TEMP/MoltenVK-macos.tar"
          package="$RUNNER_TEMP/MoltenVK"
-          install_dir="_Build/macos/install"
-          app_dir="$install_dir/KytyPS5.app"
-          moltenvk="$install_dir/libMoltenVK.dylib"
          curl --fail --location --retry 3 \
            --output "$archive" \
            "https://github.com/KhronosGroup/MoltenVK/releases/download/$MOLTENVK_VERSION/MoltenVK-macos.tar"
          echo "$MOLTENVK_SHA256  $archive" | shasum -a 256 --check
          tar -xf "$archive" -C "$RUNNER_TEMP"
-          test -d "$app_dir"
          install -m 755 \
            "$package/MoltenVK/dynamic/dylib/macOS/libMoltenVK.dylib" \
-            "$moltenvk"
-          install -m 644 "$package/LICENSE" "$install_dir/LICENSE.MoltenVK"
-          codesign --force --sign - --timestamp=none "$moltenvk"
-          install -d -m 755 "$app_dir/Contents/Frameworks" "$app_dir/Contents/Resources"
-          install -m 755 "$moltenvk" "$app_dir/Contents/Frameworks/libMoltenVK.dylib"
-          install -m 644 "$package/LICENSE" "$app_dir/Contents/Resources/LICENSE.MoltenVK"
-          codesign --force --sign - --timestamp=none "$app_dir"
+            _Build/macos/install/libMoltenVK.dylib
+          install -m 644 "$package/LICENSE" _Build/macos/install/LICENSE.MoltenVK
+          codesign --force --sign - --timestamp=none _Build/macos/install/libMoltenVK.dylib

      - name: Verify artifacts
        shell: bash
        run: |
-          install_dir="_Build/macos/install"
-          app_dir="$install_dir/KytyPS5.app"
-          test -d "$app_dir"
-          for binary in \
-            "$install_dir/kyty_emulator" \
-            "$install_dir/libMoltenVK.dylib" \
-            "$app_dir/Contents/MacOS/KytyPS5" \
-            "$app_dir/Contents/MacOS/kyty_emulator" \
-            "$app_dir/Contents/Frameworks/libMoltenVK.dylib"; do
-            file "$binary"
-            lipo "$binary" -verify_arch x86_64
-          done
-          for binary in \
-            "$install_dir/kyty_emulator" \
-            "$install_dir/libMoltenVK.dylib" \
-            "$app_dir/Contents/MacOS/KytyPS5" \
-            "$app_dir/Contents/MacOS/kyty_emulator" \
-            "$app_dir/Contents/Frameworks/libMoltenVK.dylib"; do
-            codesign --verify --strict "$binary"
-          done
-          codesign --verify --deep --strict --verbose=2 "$app_dir"
-
-      - name: Package macOS artifacts
-        shell: bash
-        run: |
-          install_dir="_Build/macos/install"
-          archive="$RUNNER_TEMP/KytyPS5-macOS-x86_64.zip"
-          (
-            cd "$install_dir"
-            zip -qry "$archive" .
-          )
-          unzip -t "$archive"
+          file _Build/macos/install/launcher
+          file _Build/macos/install/kyty_emulator
+          file _Build/macos/install/libMoltenVK.dylib
+          lipo _Build/macos/install/launcher -verify_arch x86_64
+          lipo _Build/macos/install/kyty_emulator -verify_arch x86_64
+          lipo _Build/macos/install/libMoltenVK.dylib -verify_arch x86_64
+          codesign --verify --strict _Build/macos/install/kyty_emulator
+          codesign --verify --strict _Build/macos/install/libMoltenVK.dylib

      - name: Upload macOS artifacts
        uses: actions/upload-artifact@v4
        with:
          name: KytyPS5-macOS-x86_64
-          path: ${{ runner.temp }}/KytyPS5-macOS-x86_64.zip
+          path: _Build/macos/install/**
          if-no-files-found: error

  linux:
    name: Build KytyPS5 (Linux)
-    needs: metadata
    runs-on: ubuntu-24.04

    steps:
@@ -319,7 +244,6 @@ jobs:
            libxss-dev \
            lld \
            ninja-build \
-            pkg-config \
            wayland-protocols

      - name: Install Qt 6.10.3
@@ -331,9 +255,6 @@ jobs:
          arch: linux_gcc_64
          cache: true

-      - name: Setup sccache
-        uses: mozilla-actions/sccache-action@v0.0.11
-
      - name: Verify toolchain
        shell: bash
        run: |
@@ -352,14 +273,8 @@ 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="${{ needs.metadata.outputs.release_tag }}" \
            -DCMAKE_C_COMPILER=clang \
            -DCMAKE_CXX_COMPILER=clang++ \
-            -DCMAKE_C_COMPILER_LAUNCHER=sccache \
-            -DCMAKE_CXX_COMPILER_LAUNCHER=sccache \
            -DCMAKE_PREFIX_PATH="$Qt6_DIR" 2>&1 | tee _Build/configure.log
          exit "${PIPESTATUS[0]}"

@@ -382,14 +297,14 @@ jobs:
        run: |
          cmake --build _Build/linux \
            --target launcher page_manager_tests memory_tracker_tests \
-              audio_out2_port_tests ime_dialog_tests virtual_memory_allocation_tests \
+              audio_out2_port_tests virtual_memory_allocation_tests \
            --parallel

      - name: Test
        shell: bash
        run: |
          ctest --test-dir _Build/linux --output-on-failure \
-            -R '^(audio_out2_port|ime_dialog|page_manager|memory_tracker|virtual_memory_allocation)$'
+            -R '^(audio_out2_port|page_manager|memory_tracker|virtual_memory_allocation)$'

      - name: Install
        shell: bash
@@ -426,70 +341,46 @@ jobs:

  release:
    name: Release 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]
+    if: github.event_name == 'push' && github.repository == 'KytyPS5/KytyPS5'
+    needs: [windows, macos, linux]
    runs-on: ubuntu-latest
    permissions:
      contents: write
-    env:
-      RELEASE_NAME: ${{ needs.metadata.outputs.release_tag }}

    steps:
-      - name: Download Windows build
-        if: needs.windows.result == 'success'
+      - name: Download builds
        uses: actions/download-artifact@v4
        with:
-          name: KytyPS5-Windows-x64
-          path: artifacts/KytyPS5-Windows-x64
+          path: artifacts

-      - 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: Set release name
+        shell: bash
+        run: |
+          echo "RELEASE_NAME=KytyPS5-$(date -u +'%Y-%m-%d')-${GITHUB_SHA::7}" >> "$GITHUB_ENV"

-      - 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'
+      - name: Package builds
        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/$RELEASE_NAME-Windows-x64.zip" .
          )
-
-      - name: Package macOS build
-        if: needs.macos.result == 'success'
-        shell: bash
-        run: |
-          macos_archive="artifacts/KytyPS5-macOS-x86_64/KytyPS5-macOS-x86_64.zip"
-          test -f "$macos_archive"
-          cp "$macos_archive" "$RELEASE_NAME-macOS-x86_64.zip"
-
-      - name: Package Linux build
-        if: needs.linux.result == 'success'
-        shell: bash
-        run: |
-          linux_dir="artifacts/KytyPS5-Linux-x86_64"
-          test -d "$linux_dir"
-          chmod a+x \
-            "$linux_dir/launcher" \
-            "$linux_dir/kyty_emulator"
+          (
+            cd "$macos_dir"
+            zip -r "$GITHUB_WORKSPACE/$RELEASE_NAME-macOS-x86_64.zip" .
+          )
          tar -C "$linux_dir" -czf \
            "$GITHUB_WORKSPACE/$RELEASE_NAME-Linux-x86_64.tar.gz" .

@@ -498,16 +389,11 @@ jobs:
        env:
          GH_TOKEN: ${{ github.token }}
        run: |
-          assets=()
-          for artifact in \
-            "$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 "${#assets[@]}" -gt 0
+          assets=(
+            "$RELEASE_NAME-Windows-x64.zip"
+            "$RELEASE_NAME-macOS-x86_64.zip"
+            "$RELEASE_NAME-Linux-x86_64.tar.gz"
+          )
          if gh release view "$RELEASE_NAME" --repo "$GITHUB_REPOSITORY" > /dev/null 2>&1; then
            gh release upload "$RELEASE_NAME" "${assets[@]}" \
              --clobber \
@@ -5,4 +5,3 @@ build/
 _Build/vscode-clang/
 _Build/
 cmake-build-debug/
-_PipelineCache/
@@ -22,7 +22,7 @@
 	url = https://github.com/GPUOpen-LibrariesAndSDKs/VulkanMemoryAllocator.git
 [submodule "3rdparty/ffmpeg-core"]
 	path = 3rdparty/ffmpeg-core
-	url = https://github.com/KytyPS5/ext-ffmpeg-core.git
+	url = https://github.com/shadps4-emu/ext-ffmpeg-core.git
 	shallow = true
 [submodule "3rdparty/fmt"]
 	path = 3rdparty/fmt
@@ -39,6 +39,3 @@
 [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
@@ -10,14 +10,7 @@
            "stopAtEntry": false,
            "cwd": "${workspaceFolder}\\_Build\\vscode-clang\\launcher",
            "environment": [],
-            "externalConsole": false,
-            "linux": {
-                "type": "cppdbg",
-                "program": "${workspaceFolder}/_Build/vscode-clang/launcher/launcher",
-                "cwd": "${workspaceFolder}/_Build/vscode-clang/launcher",
-                "MIMode": "gdb",
-                "miDebuggerPath": "/usr/bin/gdb"
-            }
+            "externalConsole": false
        },
        {
            "name": "Debug kyty_emulator",
@@ -32,40 +25,13 @@
            "stopAtEntry": false,
            "cwd": "${workspaceFolder}\\_Build\\vscode-clang",
            "environment": [],
-            "externalConsole": true,
-            "linux": {
-                "type": "cppdbg",
-                "program": "${workspaceFolder}/_Build/vscode-clang/kyty_emulator",
-                "cwd": "${workspaceFolder}/_Build/vscode-clang",
-                "externalConsole": false,
-                "MIMode": "gdb",
-                "miDebuggerPath": "/usr/bin/gdb",
-                "setupCommands": [
-                    {
-                        "description": "Pass Kyty's internal signals to its handlers",
-                        "text": "handle SIG37 SIGSEGV SIGBUS SIGILL nostop noprint pass",
-                        "ignoreFailures": false
-                    }
-                ]
-            }
+            "externalConsole": true
        },
        {
            "name": "Attach to running kyty_emulator",
            "type": "cppvsdbg",
            "request": "attach",
-            "processId": "${command:pickProcess}",
-            "linux": {
-                "type": "cppdbg",
-                "MIMode": "gdb",
-                "miDebuggerPath": "/usr/bin/gdb",
-                "setupCommands": [
-                    {
-                        "description": "Pass Kyty's internal signals to its handlers",
-                        "text": "handle SIG37 SIGSEGV SIGBUS SIGILL nostop noprint pass",
-                        "ignoreFailures": false
-                    }
-                ]
-            }
+            "processId": "${command:pickProcess}"
        }
    ]
 }
@@ -2,32 +2,6 @@ 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)

@@ -37,25 +11,6 @@ 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)
@@ -1,875 +0,0 @@
-/******************************************************************************
- * 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
@@ -4,12 +4,6 @@ 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")
@@ -34,29 +28,6 @@ include(CTest)

 option(KYTY_ENABLE_CLANG_TIDY "Run clang-tidy checks during builds" OFF)

-set(KYTY_BUILD_ORIGIN "Source" CACHE STRING "Build provenance shown in diagnostics")
-set_property(CACHE KYTY_BUILD_ORIGIN PROPERTY STRINGS Source Official Fork)
-if(NOT KYTY_BUILD_ORIGIN STREQUAL "Source" AND NOT KYTY_BUILD_ORIGIN STREQUAL "Official" AND
-   NOT KYTY_BUILD_ORIGIN STREQUAL "Fork")
-	message(FATAL_ERROR "KYTY_BUILD_ORIGIN must be Source, Official, or Fork")
-endif()
-
-set(KYTY_RELEASE_TAG "" CACHE STRING "Release tag embedded in official builds")
-set(KYTY_BUILD_REPOSITORY "" CACHE STRING "GitHub repository embedded in fork builds")
-set(KYTY_OFFICIAL_BUILD OFF)
-set(KYTY_FORK_BUILD OFF)
-if(KYTY_BUILD_ORIGIN STREQUAL "Official")
-	if(KYTY_RELEASE_TAG STREQUAL "")
-		message(FATAL_ERROR "Official builds require KYTY_RELEASE_TAG")
-	endif()
-	set(KYTY_OFFICIAL_BUILD ON)
-elseif(KYTY_BUILD_ORIGIN STREQUAL "Fork")
-	if(KYTY_BUILD_REPOSITORY STREQUAL "")
-		message(FATAL_ERROR "Fork builds require KYTY_BUILD_REPOSITORY")
-	endif()
-	set(KYTY_FORK_BUILD ON)
-endif()
-
 set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)

 include(TestBigEndian)
@@ -126,7 +97,7 @@ if (KYTY_LINKER STREQUAL LD AND NOT LINUX)
 	set(KYTY_LD_OPTIONS "-Wl,--image-base=0x100000000000")
 endif()

-project(Kyty${KYTY_PROJECT_NAME}${CMAKE_BUILD_TYPE}${KYTY_COMPILER} VERSION 0.3.0)
+project(Kyty${KYTY_PROJECT_NAME}${CMAKE_BUILD_TYPE}${KYTY_COMPILER} VERSION 0.2.2)

 include_directories(
 	${KYTY_THIRD_PARTY_DIR}/gtest/include
@@ -146,20 +117,16 @@ configure_file(

 find_package(Git)

-add_custom_target(KytyGitVersion
-	COMMAND ${CMAKE_COMMAND}
-		-D INPUT_FILE=${KYTY_SOURCE_DIR}/kytyGitVersion.h.in
-		-D OUTPUT_FILE=${PROJECT_BINARY_DIR}/kytyGitVersion.h
-		-D GIT_EXECUTABLE=${GIT_EXECUTABLE}
-		-D GIT_WORKING_DIRECTORY=${CMAKE_CURRENT_SOURCE_DIR}
-		-P ${KYTY_SOURCE_DIR}/generate_version.cmake
-	BYPRODUCTS ${PROJECT_BINARY_DIR}/kytyGitVersion.h
+add_custom_target( KytyGitVersion
+    COMMAND ${CMAKE_COMMAND}
+	-D INPUT_FILE=${KYTY_SOURCE_DIR}/kytyGitVersion.h.in
+	-D OUTPUT_FILE=${PROJECT_BINARY_DIR}/kytyGitVersion.h
+	-D GIT_EXECUTABLE=${GIT_EXECUTABLE}
+	-D GIT_WORKING_DIRECTORY=${CMAKE_CURRENT_SOURCE_DIR}
+	-P ${KYTY_SOURCE_DIR}/generate_version.cmake
 	COMMENT "Generate kytyGitVersion.h"
 )

-add_library(kyty_git_version INTERFACE)
-add_dependencies(kyty_git_version KytyGitVersion)
-
 if (CMAKE_CXX_COMPILER_VERSION VERSION_GREATER_EQUAL 12.0.0)
 	list(APPEND KYTY_IWYU
 		kyty_emulator
@@ -203,21 +170,14 @@ 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/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/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/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"
 	"${KYTY_SOURCE_DIR}/graphics/presentation/window/*.cpp"
@@ -254,7 +214,7 @@ foreach(shader_name IN LISTS gpu_tiler_shader_names)
 	add_custom_command(
 		OUTPUT "${shader_header}"
 		COMMAND ${CMAKE_COMMAND} -E make_directory "${gpu_tiler_generated_dir}"
-		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.3 -Os
+		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.0 -Os
 		        "-I${gpu_tiler_shader_dir}" -o "${shader_spv}" "${shader_source}"
 		COMMAND ${CMAKE_COMMAND} -DINPUT=${shader_spv} -DOUTPUT=${shader_header}
 		        -DSYMBOL=${shader_symbol} -P "${KYTY_SOURCE_DIR}/embed_spirv.cmake"
@@ -263,22 +223,6 @@ foreach(shader_name IN LISTS gpu_tiler_shader_names)
 	)
 	list(APPEND gpu_tiler_shader_headers "${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 "${PROJECT_BINARY_DIR}/gpu_blit_shaders")
@@ -294,7 +238,7 @@ foreach(shader_source IN LISTS gpu_blit_shader_sources)
 	add_custom_command(
 		OUTPUT "${shader_header}"
 		COMMAND ${CMAKE_COMMAND} -E make_directory "${gpu_blit_generated_dir}"
-		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.3 -Os
+		COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.0 -Os
 		        "-I${gpu_tiler_shader_dir}" -o "${shader_spv}" "${shader_source}"
 		COMMAND ${CMAKE_COMMAND} -DINPUT=${shader_spv} -DOUTPUT=${shader_header}
 		        -DSYMBOL=${shader_symbol} -P "${KYTY_SOURCE_DIR}/embed_spirv.cmake"
@@ -315,12 +259,26 @@ if (KYTY_CLANG_CL)
 	set_source_files_properties(${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 imgui xxhash FFmpeg::ffmpeg fmt::fmt nlohmann_json::nlohmann_json LibAtrac9 kyty_git_version Zydis::Zydis xbyak::xbyak)
+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)

-# Linux system libraries.
+# Linux system libraries required by the static FFmpeg archive.
 if(LINUX)
 	find_package(Threads REQUIRED)
 	list(APPEND kyty_emulator_link_libraries m ${CMAKE_DL_LIBS} Threads::Threads)
+
+	# Optional FFmpeg dependencies.
+	find_package(ZLIB)
+	if(ZLIB_FOUND)
+		list(APPEND kyty_emulator_link_libraries ZLIB::ZLIB)
+	endif()
+	find_library(KYTY_BZ2_LIBRARY bz2)
+	if(KYTY_BZ2_LIBRARY)
+		list(APPEND kyty_emulator_link_libraries ${KYTY_BZ2_LIBRARY})
+	endif()
+	find_library(KYTY_LZMA_LIBRARY lzma)
+	if(KYTY_LZMA_LIBRARY)
+		list(APPEND kyty_emulator_link_libraries ${KYTY_LZMA_LIBRARY})
+	endif()
 endif()

 set(inc_headers
@@ -328,11 +286,9 @@ set(inc_headers
 	${KYTY_THIRD_PARTY_DIR}/SDL2/include
 	${KYTY_THIRD_PARTY_DIR}/VulkanMemoryAllocator/include
 	${KYTY_THIRD_PARTY_DIR}/SPIRV-Tools/include
-	${KYTY_THIRD_PARTY_DIR}/SPIRV-Headers/include
 	${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
 )

@@ -378,8 +334,9 @@ 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/passes/ReadLaneElimination.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/SrtWalker.cpp"
+	"${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"
 )
 target_link_libraries(scalar_provenance_tests fmt::fmt)
 target_include_directories(scalar_provenance_tests PRIVATE ${inc_headers})
@@ -388,7 +345,6 @@ add_executable(page_manager_tests EXCLUDE_FROM_ALL
 	"${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
@@ -396,13 +352,6 @@ 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"
@@ -418,30 +367,40 @@ add_executable(shader_vertex_metadata_tests EXCLUDE_FROM_ALL
 )
 target_include_directories(shader_vertex_metadata_tests PRIVATE ${inc_headers})

-add_executable(resource_materialization_tests EXCLUDE_FROM_ALL
-	"${KYTY_TESTS_DIR}/ResourceMaterializationTests.cpp"
+add_executable(shader_stage_runtime_tests EXCLUDE_FROM_ALL
+	"${KYTY_TESTS_DIR}/ShaderStageRuntimeTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/guest_gpu/gpu_format.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/hostMemory.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ResourceMaterialization.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/SrtWalker.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"
 )
-target_link_libraries(resource_materialization_tests fmt::fmt)
-target_include_directories(resource_materialization_tests PRIVATE ${inc_headers})
+target_link_libraries(shader_stage_runtime_tests fmt::fmt)
+target_include_directories(shader_stage_runtime_tests PRIVATE ${inc_headers})

 add_executable(resource_tracking_tests EXCLUDE_FROM_ALL
 	"${KYTY_TESTS_DIR}/ResourceTrackingTests.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/guest_gpu/gpu_format.cpp"
 	"${KYTY_SOURCE_DIR}/graphics/host_gpu/hostMemory.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/shaderPixelParameter.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/SrtWalker.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ResourceTracking.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ResourceMaterialization.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/ShaderInfoCollection.cpp"
-	"${KYTY_SOURCE_DIR}/graphics/shader/recompiler/ir/passes/BindingLayout.cpp"
+	"${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"
 )
 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"
@@ -458,16 +417,6 @@ 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"
 )
@@ -486,7 +435,7 @@ 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 "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.3 -Os
+	COMMAND "${KYTY_GLSLANG_VALIDATOR}" -V --target-env vulkan1.0 -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
@@ -510,17 +459,6 @@ if(NOT KYTY_CLANG_CL)
 endif()

 if(BUILD_TESTING)
-	add_test(NAME emulator_user_name_cli
-	         COMMAND $<TARGET_FILE:kyty_emulator> --help --user-name 1234567890123456)
-	add_test(NAME emulator_user_name_too_long
-	         COMMAND $<TARGET_FILE:kyty_emulator> --help --user-name 12345678901234567)
-	set_tests_properties(emulator_user_name_too_long PROPERTIES WILL_FAIL TRUE)
-	add_test(NAME emulator_present_mode_cli
-	         COMMAND $<TARGET_FILE:kyty_emulator> --help --present-mode Immediate)
-	add_test(NAME emulator_present_mode_invalid
-	         COMMAND $<TARGET_FILE:kyty_emulator> --help --present-mode Invalid)
-	set_tests_properties(emulator_present_mode_invalid PROPERTIES WILL_FAIL TRUE)
-	add_test(NAME ime_dialog COMMAND $<TARGET_FILE:ime_dialog_tests>)
 	add_test(NAME shader_cfg COMMAND $<TARGET_FILE:shader_cfg_tests>)
 	add_test(NAME scalar_provenance COMMAND $<TARGET_FILE:scalar_provenance_tests>)
 	add_test(NAME image_page_table COMMAND $<TARGET_FILE:image_page_table_tests>)
@@ -528,46 +466,31 @@ if(BUILD_TESTING)
 	add_test(NAME page_manager COMMAND $<TARGET_FILE:page_manager_tests>)
 	add_test(NAME bit_array COMMAND $<TARGET_FILE:bit_array_tests>)
 	add_test(NAME shader_vertex_metadata COMMAND $<TARGET_FILE:shader_vertex_metadata_tests>)
-	add_test(NAME resource_materialization COMMAND $<TARGET_FILE:resource_materialization_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 graphics_pipeline_rasterization
-		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --polygon-mode-only)
-	add_test(NAME graphics_draw_offsets
-		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --draw-offset-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)

 	add_test(NAME texture_cache_layered_image
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --layered-image-only)
-	add_test(NAME host_image_allocation
-		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --host-image-allocation-only)
 	add_test(NAME texture_cache_image_views
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --image-view-cache-only)
 	add_test(NAME texture_cache_storage_sampled
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --storage-sampled-only)
 	add_test(NAME texture_cache_depth_readback
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --depth-readback-only)
-	add_test(NAME compute_meta_clear_classification
-		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --compute-meta-clear-only)
 	add_test(NAME buffer_cache_dirty_gc
 		COMMAND $<TARGET_FILE:shader_recompiler_compute_tests> --buffer-cache-gc-only)

@@ -582,7 +505,6 @@ if(BUILD_TESTING)
 	endif()

 	add_custom_target(kyty_tests DEPENDS
-		ime_dialog_tests
 		shader_cfg_tests
 		scalar_provenance_tests
 		image_page_table_tests
@@ -590,14 +512,12 @@ if(BUILD_TESTING)
 		page_manager_tests
 		bit_array_tests
 		shader_vertex_metadata_tests
-		resource_materialization_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()

@@ -643,6 +563,8 @@ elseif(WIN32 OR LINUX)
 	set_target_properties(kyty_emulator PROPERTIES LINK_FLAGS "${KYTY_LD_OPTIONS} -Wl,-Map=${KYTY_EMULATOR_MAP_LINK_PATH}")
 endif()

+add_dependencies(kyty_emulator KytyGitVersion)
+
 if(APPLE)
 	# The emulator writes x86-64 trampolines/PLT stubs into memory and executes them, and
 	# runs guest code under Rosetta. That requires the JIT / unsigned-executable-memory
@@ -31,9 +31,6 @@ 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,
@@ -55,7 +52,7 @@ 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="Neptunia ReVerse running in KytyPS5">
+      <img src="docs/screenshots/ps5-04.png" width="300" alt="Minecraft Legends running in KytyPS5">
    </td>
    <td align="center">
      <strong>SILENT HILL: The Short Message</strong><br>
@@ -65,11 +62,11 @@ 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="Hellboy running in KytyPS5">
+      <img src="docs/screenshots/ps5-02.png" width="300" alt="Disgaea 6 running in KytyPS5">
    </td>
    <td align="center">
      <strong>Paleo Pines</strong><br>
-      <img src="docs/screenshots/ps5-06.png" width="300" alt="Paleo Pines running in KytyPS5">
+      <img src="docs/screenshots/ps5-06.png" width="300" alt="Dreaming Sarah running in KytyPS5">
    </td>
  </tr>
 </table>
@@ -90,14 +87,8 @@ starting a large change.

 Set up the clang-format hook after cloning:

-Install `pre-commit` using the method appropriate for your platform:
-
- **Arch Linux / CachyOS:** `sudo pacman -S pre-commit`
- **Other Linux / macOS / Windows:** `python -m pip install pre-commit`
-
-Then install the Git hook:
-
-```bash
+```powershell
+python -m pip install pre-commit
 python -m pre_commit install --install-hooks
 ```

@@ -171,7 +162,7 @@ build has no working sound and no gamepad hotplug:

 ```bash
 sudo apt-get install --no-install-recommends \
-  clang lld ninja-build cmake git glslang-tools pkg-config \
+  clang lld ninja-build cmake git glslang-tools \
  libgl1-mesa-dev libx11-dev libxcursor-dev libxext-dev libxfixes-dev \
  libxi-dev libxrandr-dev libxss-dev libxkbcommon-dev \
  libasound2-dev libpulse-dev libudev-dev libdbus-1-dev libwayland-dev wayland-protocols
@@ -192,37 +183,13 @@ cmake --install _Build/linux --prefix _Build/linux/install
 ```

 The install step copies the Qt libraries and plugins next to the binaries, so
-`_Build/linux/install` runs without a matching system Qt. FFmpeg is linked statically
-from the pinned [KytyPS5 FFmpeg core](https://github.com/KytyPS5/ext-ffmpeg-core)
-release, including VP9 and WebM support. System FFmpeg packages are not required.
+`_Build/linux/install` runs without a matching system Qt.

 As on Windows, the MSVC compiler is not used; Clang is required. `cl.exe` is rejected at configure
 time.

 The CMake source root is the repository root.

-### Building on NixOS
-
-A development shell provides Clang, CMake, Ninja, Qt 6, the Vulkan headers, and the SDL2 backend
-libraries. Enter it and configure exactly as on other Linux distributions; the shell exports
-`CMAKE_PREFIX_PATH` and `QT_PLUGIN_PATH`, so the `-DCMAKE_PREFIX_PATH="$Qt6_DIR"` argument is not
-needed:
-
-```bash
-nix-shell # or: nix develop
-git submodule update --init --recursive
-
-cmake -S . -B _Build/linux -G Ninja -DCMAKE_BUILD_TYPE=Release \
-  -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++
-
-cmake --build _Build/linux --target launcher --parallel
-cmake --install _Build/linux --prefix _Build/linux/install
-```
-
-The configure step downloads the FFmpeg prebuilts and the `xbyak`/`zydis` sources, so it needs
-network access; a fully sandboxed `nix build` would require vendoring those inputs. A Vulkan 1.3
-driver must be available at runtime (on NixOS, `hardware.graphics.enable = true`).
-
 ### Building on macOS

 macOS builds target x86-64 and run under Rosetta 2 on Apple Silicon, so the PS5's x86-64 game
@@ -250,23 +217,17 @@ cmake --install _Build/macos --prefix _Build/macos/install
 ```

 The build re-signs `kyty_emulator` with the JIT entitlements it needs to execute translated
-guest code; no manual signing step is required. When the launcher is built, the install
-also produces `_Build/macos/install/KytyPS5.app` — double-click to launch the GUI.
-A flat `kyty_emulator` is kept for CLI usage.
+guest code; no manual signing step is required.

 Vulkan comes from MoltenVK. Download `MoltenVK-macos.tar` from the
 [MoltenVK releases](https://github.com/KhronosGroup/MoltenVK/releases), then copy
-`MoltenVK/dynamic/dylib/macOS/libMoltenVK.dylib` next to the flat `kyty_emulator`
-(and, for the bundle, into `KytyPS5.app/Contents/Frameworks/`) and ad-hoc sign it:
+`MoltenVK/dynamic/dylib/macOS/libMoltenVK.dylib` next to `kyty_emulator` and ad-hoc sign it:

 ```bash
 codesign --force --sign - _Build/macos/install/libMoltenVK.dylib
-# For the bundle (if present):
-codesign --force --sign - _Build/macos/install/KytyPS5.app/Contents/Frameworks/libMoltenVK.dylib
-codesign --force --sign - _Build/macos/install/KytyPS5.app
 ```

-Release archives already include a signed `libMoltenVK.dylib` (both flat and inside the bundle).
+Release archives already include a signed `libMoltenVK.dylib`.

 ### Regression tests

@@ -310,10 +271,6 @@ To use the graphical launcher:
 ./_Build/linux/install/launcher
 ```

-```bash
-open _Build/macos/install/KytyPS5.app  # or double-click in Finder
-```
-
 On first launch, add one or more game folders in the global settings. The launcher searches those
 folders recursively for game directories containing `eboot.bin`. Select a detected game and run it
 from the game list.
@@ -328,17 +285,12 @@ The emulator can also be started directly with a legally obtained game directory
 ./_Build/linux/install/kyty_emulator --game "/games/ExampleGame"
 ```

-On macOS, the adjacent flat or app-bundled `libMoltenVK.dylib` is found automatically; no
-environment variable is required:
+On macOS, point SDL at the MoltenVK library explicitly; the hardened runtime prevents it from
+being picked up from the executable's directory:

 ```bash
-./_Build/macos/install/kyty_emulator --game "/games/ExampleGame"
-```
-
-To override the Vulkan loader, set `SDL_VULKAN_LIBRARY`:
-
-```bash
-SDL_VULKAN_LIBRARY=/path/to/libMoltenVK.dylib ./kyty_emulator --game "/games/ExampleGame"
+cd _Build/macos/install
+SDL_VULKAN_LIBRARY="$PWD/libMoltenVK.dylib" ./kyty_emulator --game "/games/ExampleGame"
 ```

 Run `kyty_emulator --help` to see the available graphics, logging, validation, profiling, and
@@ -1,27 +0,0 @@
-{
-  "nodes": {
-    "nixpkgs": {
-      "locked": {
-        "lastModified": 1789426278,
-        "narHash": "sha256-GnotcvKtbTMnoVx6G4E7ZdUX6487Dky74dqq3pU7iRk=",
-        "owner": "NixOS",
-        "repo": "nixpkgs",
-        "rev": "efe6f071ede9d21c37462d2d6682d5e670099684",
-        "type": "github"
-      },
-      "original": {
-        "owner": "NixOS",
-        "ref": "nixos-unstable",
-        "repo": "nixpkgs",
-        "type": "github"
-      }
-    },
-    "root": {
-      "inputs": {
-        "nixpkgs": "nixpkgs"
-      }
-    }
-  },
-  "root": "root",
-  "version": 7
-}
@@ -1,20 +0,0 @@
-{
-  description = "KytyPS5 development shell";
-
-  inputs.nixpkgs.url = "github:NixOS/nixpkgs/nixos-unstable";
-
-  outputs = { self, nixpkgs }:
-    let
-      systems = [ "x86_64-linux" ];
-      forAllSystems = nixpkgs.lib.genAttrs systems;
-    in
-    {
-      devShells = forAllSystems (system:
-        let
-          pkgs = import nixpkgs { inherit system; };
-        in
-        {
-          default = import ./nix/devshell.nix { inherit pkgs; };
-        });
-    };
-}
@@ -1,84 +0,0 @@
-{ pkgs }:
-
-let
-  # Build-time dependencies. Every -dev package the bundled SDL2 probes for must
-  # be listed here, otherwise SDL2 quietly drops the audio/Wayland/udev backends.
-  buildDeps = with pkgs; [
-    qt6.qtbase # Concurrent, Network, Widgets (launcher)
-    vulkan-headers
-    vulkan-loader
-    vulkan-tools
-    vulkan-validation-layers
-    libglvnd
-    # X11 / input
-    libX11
-    libXcursor
-    libXext
-    libXfixes
-    libXi
-    libXrandr
-    libXScrnSaver
-    libxkbcommon
-    # Wayland
-    wayland
-    wayland-protocols
-    libdecor
-    # audio + device/dbus
-    alsa-lib
-    libpulseaudio
-    systemd # libudev
-    dbus
-  ];
-
-  # Libraries the built binaries need at runtime.
-  runtimeLibs = with pkgs; [
-    stdenv.cc.cc.lib # libstdc++ / libgcc_s
-    libgcc.lib
-    qt6.qtbase # libQt6Concurrent/Network/Widgets
-    vulkan-loader
-    libglvnd
-    alsa-lib
-    libpulseaudio
-    libX11
-    libXcursor
-    libXext
-    libXfixes
-    libXi
-    libXrandr
-    libXScrnSaver
-    libxkbcommon
-    wayland
-    libdecor
-    systemd
-    dbus
-  ];
-in
-pkgs.mkShell {
-  name = "kytyps5";
-
-  nativeBuildInputs = with pkgs; [
-    clang
-    lld
-    cmake
-    ninja
-    pkg-config
-    git
-    glslang # provides glslangValidator for the bundled shaders
-  ];
-
-  buildInputs = buildDeps;
-
-  shellHook = ''
-    export CMAKE_PREFIX_PATH="${pkgs.qt6.qtbase}:''${CMAKE_PREFIX_PATH:-}"
-    export QT_PLUGIN_PATH="${pkgs.qt6.qtbase}/lib/qt-6/plugins:${pkgs.qt6.qtbase}/lib/qt6/plugins:''${QT_PLUGIN_PATH:-}"
-    export LD_LIBRARY_PATH="${pkgs.lib.makeLibraryPath runtimeLibs}:''${LD_LIBRARY_PATH:-}"
-
-    if [ -z "''${KYTY_SHELL_QUIET:-}" ]; then
-      echo "KytyPS5 dev shell"
-      echo "  cmake -S . -B _Build/linux -G Ninja -DCMAKE_BUILD_TYPE=Release \\"
-      echo "    -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++"
-      echo "  cmake --build _Build/linux --target launcher --parallel"
-      echo "  cmake --install _Build/linux --prefix _Build/linux/install"
-    fi
-  '';
-}
@@ -1,4 +0,0 @@
-let
-  pkgs = import <nixpkgs> { };
-in
-import ./nix/devshell.nix { inherit pkgs; }
@@ -1,8 +1,4 @@
 #define KYTY_VERSION "@KYTY_VERSION@"
-#define KYTY_BUILD_REPOSITORY "@KYTY_BUILD_REPOSITORY@"
-#define KYTY_RELEASE_TAG "@KYTY_RELEASE_TAG@"
-#cmakedefine KYTY_OFFICIAL_BUILD
-#cmakedefine KYTY_FORK_BUILD
 #define KYTY_ENDIAN @KYTY_ENDIAN@
 #define KYTY_PLATFORM @KYTY_PLATFORM@
 #define KYTY_COMPILER KYTY_COMPILER_@KYTY_COMPILER@
@@ -24,7 +24,6 @@ target_link_libraries(common
 		Tracy::TracyClient
 	PRIVATE
 		cpuinfo
-		kyty_git_version
 		spdlog::spdlog
 )

@@ -1,25 +0,0 @@
-// SPDX-FileCopyrightText: 2014 Jannik Vogel <email@jannikvogel.de>
-// SPDX-License-Identifier: CC0-1.0
-
-#pragma once
-
-#include <cstddef>
-#include <type_traits>
-
-namespace Common {
-
-template <typename T>
-[[nodiscard]] constexpr T AlignUp(T value, std::size_t size) {
-	static_assert(std::is_unsigned_v<T>, "T must be an unsigned value.");
-	auto mod{static_cast<T>(value % size)};
-	value -= mod;
-	return static_cast<T>(mod == T{0} ? value : value + size);
-}
-
-template <typename T>
-[[nodiscard]] constexpr T AlignDown(T value, std::size_t size) {
-	static_assert(std::is_unsigned_v<T>, "T must be an unsigned value.");
-	return static_cast<T>(value - value % size);
-}
-
-} // namespace Common
@@ -1,8 +1,8 @@
 #include "common/assert.h"

+#include "common/debug.h"
 #include "common/logging/log.h"
 #include "common/subsystems.h"
-#include "kytyGitVersion.h"

 #include <cstdio>
 #include <cstdlib>
@@ -11,15 +11,30 @@

 namespace Common {

+#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS && KYTY_BUILD == KYTY_BUILD_DEBUG &&                    \
+    KYTY_COMPILER == KYTY_COMPILER_CLANG
+constexpr int PRINT_STACK_FROM = 4;
+#else
+constexpr int PRINT_STACK_FROM = 2;
+#endif
+
 static std::string BuildFatalReport(const char* title, std::string_view text, const char* file,
                                    int line) {
-	return fmt::format("--- Build ---\n{}\n{}\n{} in {}:{}\n", KYTY_BUILD_LABEL, title, text, file,
-	                   line);
+	DebugStack stack;
+	DebugStack::Trace(&stack);
+
+	std::string report = "--- 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)));
+	}
+	report += fmt::format("{}\n{} in {}:{}\n", title, text, file, line);
+	return report;
 }

 static int DbgReport(const char* title, std::string_view text, const char* file, int line) {
 	Log::WriteFatal(BuildFatalReport(title, text, file, line));
-	Subsystems::EmergencyShutdownActive();
+	SubsystemsListSingleton::Instance()->ShutdownAll();
 	return 1;
 }

@@ -43,7 +58,6 @@ int DbgExitHandler(const char* file, int line, fmt::text_style style, std::strin
 }

 void DbgExit(int status) {
-	Subsystems::EmergencyShutdownActive();
 	std::fflush(nullptr);
 	std::_Exit(status);
 }
@@ -29,7 +29,11 @@ void DbgExit(int status);

 } // namespace Common

+#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS || KYTY_PLATFORM == KYTY_PLATFORM_LINUX
 #define EXIT_HALT() (Common::DbgExit(321), 1)
+#else
+#define EXIT_HALT() (std::_Exit(321), 1)
+#endif

 #ifndef KYTY_FINAL
 #define EXIT_IF(x)                                                                                 \
@@ -0,0 +1,21 @@
+#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
@@ -0,0 +1,12 @@
+#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_ */
@@ -1,7 +1,20 @@
 #include "common/debug.h"

+#include "common/platform/sysDbg.h"
+
 namespace Common {

+void DebugStack::Print(int from, [[maybe_unused]] bool with_name) const {
+	for (int i = from; i < depth; i++) {
+		printf("[%d] %016" PRIx64 "\n", i - from, static_cast<uint64_t>(GetAddr(i)));
+	}
+}
+
+void DebugStack::Trace(DebugStack* stack) {
+	stack->depth = DEBUG_MAX_STACK_DEPTH;
+	SysStackWalk(stack->stack.data(), &stack->depth);
+}
+
 std::string Debug::GetCompiler() {
 #if KYTY_COMPILER == KYTY_COMPILER_CLANG
 	return "clang";
@@ -2,17 +2,42 @@
 #define KYTY_COMMON_DEBUG_H_

 #include "common/common.h"
+#include "common/stringUtils.h"

-#include <string>
+#include <array>
+#include <cstring>

 namespace Common {

+#if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS && KYTY_BUILD == KYTY_BUILD_DEBUG &&                    \
+    KYTY_COMPILER == KYTY_COMPILER_CLANG
+constexpr int DEBUG_MAX_STACK_DEPTH = 20;
+#else
+constexpr int DEBUG_MAX_STACK_DEPTH = 15;
+#endif
+
 namespace Debug {
 std::string GetCompiler();
 std::string GetLinker();
 std::string GetBitness();
 } // namespace Debug

+struct DebugStack {
+	int                                      depth {0};
+	std::array<void*, DEBUG_MAX_STACK_DEPTH> stack {};
+
+	[[nodiscard]] uintptr_t GetAddr(int i) const { return reinterpret_cast<uintptr_t>(stack[i]); }
+
+	void Print(int from, bool with_name = true) const;
+
+	void CopyTo(DebugStack* s) const {
+		std::memcpy(s->stack.data(), stack.data(), sizeof(void*) * depth);
+		s->depth = depth;
+	}
+
+	static void Trace(DebugStack* stack);
+};
+
 } // namespace Common

 #endif /* KYTY_COMMON_DEBUG_H_ */
@@ -9,20 +9,18 @@ namespace Config {

 static std::unique_ptr<ConfigOptions> g_config;

-void Initialize() {
+KYTY_SUBSYSTEM_INIT(Config) {
 	EXIT_IF(g_config != nullptr);

 	g_config = std::make_unique<ConfigOptions>();
 }

-void Shutdown() {
-	g_config.reset();
-}
+KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Config) {}
+
+KYTY_SUBSYSTEM_DESTROY(Config) {}

 void Load(const ConfigOptions& cfg) {
 	EXIT_IF(g_config == nullptr);
-	EXIT_IF(cfg.user_name.empty() || cfg.user_name.size() > MAX_USER_NAME_LENGTH);
-	EXIT_IF(!IsConfiguredUserIdValid(cfg.user_id));

 	*g_config = cfg;
 }
@@ -35,46 +33,10 @@ uint32_t GetScreenHeight() {
 	return g_config->screen_height;
 }

-const std::string& GetUserName() {
-	return g_config->user_name;
-}
-
-int32_t GetUserId() {
-	return g_config->user_id;
-}
-
-const std::string& GetAudioInputDevice() {
-	return g_config->audio_input_device;
-}
-
-PresentMode GetPresentMode() {
-	return g_config->present_mode;
-}
-
-int32_t GetGpuIndex() {
-	return g_config->gpu_index;
-}
-
-bool FullscreenEnabled() {
-	return g_config->fullscreen_enabled;
-}
-
-bool VrEnabled() {
-	return g_config->vr_enabled;
-}
-
-bool AmdCpuEnabled() {
-	return g_config->amd_cpu_enabled;
-}
-
 uint32_t GetVblankFrequency() {
 	return std::clamp(g_config->vblank_frequency, 30u, 360u);
 }

-uint32_t GetConsoleLanguage() {
-	return g_config->console_language;
-}
-
 bool VulkanValidationEnabled() {
 	return g_config->vulkan_validation_enabled;
 }
@@ -87,7 +49,7 @@ ShaderOptimizationType GetShaderOptimizationType() {
 	return g_config->shader_optimization_type;
 }

-LogDirection GetShaderLogDirection() {
+ShaderLogDirection GetShaderLogDirection() {
 	return g_config->shader_log_direction;
 }

@@ -107,7 +69,7 @@ bool GraphicsDebugDumpEnabled() {
 	return g_config->graphics_debug_dump_enabled;
 }

-LogDirection GetPrintfDirection() {
+OutputDirection GetPrintfDirection() {
 	return g_config->printf_direction;
 }

@@ -115,18 +77,14 @@ std::filesystem::path GetPrintfOutputFile() {
 	return g_config->printf_output_file;
 }

-bool ProfilerEnabled() {
-	return g_config->profiler_enabled;
+ProfilerDirection GetProfilerDirection() {
+	return g_config->profiler_direction;
 }

 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;
 }
@@ -135,18 +93,4 @@ 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
@@ -2,96 +2,52 @@
 #define KYTY_COMMON_EMULATOR_CONFIG_H_

 #include "common/common.h"
+#include "common/subsystems.h"

-#include <cstddef>
 #include <filesystem>
-#include <string>
-#include <vector>

 namespace 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;
-};
+KYTY_SUBSYSTEM_DEFINE(Config);

 enum class ShaderOptimizationType { None, Size, Performance };

-enum class LogDirection { Silent, Console, File };
+enum class ShaderLogDirection { Silent, Console, File };

-enum class PresentMode { Fifo, Mailbox, Immediate };
+enum class ProfilerDirection { None, Network };

-using Keymap = std::vector<std::string>;
-
-constexpr uint32_t DEFAULT_CONSOLE_LANGUAGE = 1;
-constexpr uint32_t MAX_CONSOLE_LANGUAGE     = 29;
-constexpr std::size_t MAX_USER_NAME_LENGTH = 16;
-constexpr int32_t DEFAULT_USER_ID           = 1000;
-
-constexpr bool IsConfiguredUserIdValid(int32_t user_id) {
-	constexpr int32_t USER_ID_EVERYONE = 0xfe;
-	constexpr int32_t USER_ID_SYSTEM   = 0xff;
-	return user_id >= 0 && user_id != USER_ID_EVERYONE && user_id != USER_ID_SYSTEM;
-}
+enum class OutputDirection { Silent, Console, File };

 struct ConfigOptions {
 	uint32_t               screen_width                = 1280;
 	uint32_t               screen_height               = 720;
-	std::string            user_name                   = "Kyty";
-	int32_t                user_id                     = DEFAULT_USER_ID;
-	std::string            audio_input_device;
-	PresentMode            present_mode                = PresentMode::Mailbox;
-	int32_t                gpu_index                   = -1;
-	bool                   fullscreen_enabled          = false;
-	bool                   vr_enabled                  = false;
-	bool                   amd_cpu_enabled             = false;
 	uint32_t               vblank_frequency            = 60;
-	uint32_t               console_language            = DEFAULT_CONSOLE_LANGUAGE;
 	bool                   vulkan_validation_enabled   = false;
 	bool                   shader_validation_enabled   = false;
 	ShaderOptimizationType shader_optimization_type    = ShaderOptimizationType::None;
-	LogDirection           shader_log_direction        = LogDirection::Silent;
+	ShaderLogDirection     shader_log_direction        = ShaderLogDirection::Silent;
 	std::filesystem::path  shader_log_folder           = "_Shaders";
 	bool                   command_buffer_dump_enabled = false;
 	std::filesystem::path  command_buffer_dump_folder  = "_Buffers";
 	bool                   graphics_debug_dump_enabled = false;
-	LogDirection           printf_direction            = LogDirection::Silent;
+	OutputDirection        printf_direction            = OutputDirection::Console;
 	std::filesystem::path  printf_output_file          = "_kyty.txt";
-	bool                   profiler_enabled            = false;
+	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();
-const std::string& GetUserName();
-int32_t  GetUserId();
-const std::string& GetAudioInputDevice();
-PresentMode GetPresentMode();
-int32_t GetGpuIndex();
-bool     FullscreenEnabled();
-bool     VrEnabled();
-bool     AmdCpuEnabled();
 uint32_t GetVblankFrequency();
-uint32_t GetConsoleLanguage();
 bool     VulkanValidationEnabled();

 bool                   ShaderValidationEnabled();
 ShaderOptimizationType GetShaderOptimizationType();
-LogDirection           GetShaderLogDirection();
+ShaderLogDirection     GetShaderLogDirection();
 std::filesystem::path  GetShaderLogFolder();

 bool                  CommandBufferDumpEnabled();
@@ -99,23 +55,15 @@ std::filesystem::path GetCommandBufferDumpFolder();

 bool GraphicsDebugDumpEnabled();

-LogDirection          GetPrintfDirection();
+OutputDirection       GetPrintfDirection();
 std::filesystem::path GetPrintfOutputFile();

-bool ProfilerEnabled();
+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

@@ -25,6 +25,9 @@
 #ifdef CopyFile
 #undef CopyFile
 #endif
+#ifdef MoveFile
+#undef MoveFile
+#endif
 #endif

 namespace Common {
@@ -69,7 +72,7 @@ File::File(): m_p(std::make_unique<FilePrivate>()) {
 }

 File::~File() {
-	Close();
+	EXIT_IF(m_p->f != nullptr);
 }

 File::File(const std::filesystem::path& name): m_p(std::make_unique<FilePrivate>()) {
@@ -541,14 +544,14 @@ bool File::CopyFile(const std::filesystem::path& src,
 	return SysFileCopyFile(src, dst);
 }

-bool File::RenameFile(const std::filesystem::path& src,
-                      const std::filesystem::path& dst) // @suppress("Member declaration not found")
+bool File::MoveFile(const std::filesystem::path& src,
+                    const std::filesystem::path& dst) // @suppress("Member declaration not found")
 {
 	if (IsFileExisting(dst)) {
 		DeleteFile(dst); // @suppress("Invalid arguments")
 	}

-	return SysFileRenameFile(src, dst);
+	return SysFileMoveFile(src, dst);
 }

 void File::RemoveReadonly(const std::filesystem::path& name) {
@@ -20,6 +20,9 @@
 #ifdef CopyFile
 #undef CopyFile
 #endif
+#ifdef MoveFile
+#undef MoveFile
+#endif
 #endif

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

 	KYTY_CLASS_NO_COPY(File);
@@ -2,20 +2,18 @@

 #include <atomic>
 #include <cstdio>
+#include <cstdlib>

 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
 #include <windows.h> // IWYU pragma: keep
-#else
-#include <algorithm>
+#elif defined(__APPLE__)
 #include <csignal>
-#include <cstdlib>
-#include <initializer_list>
-#include <unistd.h>
-#if defined(__APPLE__)
 #include <sys/ucontext.h>
 #else
+#include <csignal>
+#include <initializer_list>
 #include <ucontext.h> // IWYU pragma: keep
-#endif
+#include <unistd.h>
 #endif

 // IWYU pragma: no_include <errhandlingapi.h>
@@ -30,59 +28,54 @@ 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
-
-#if KYTY_PLATFORM == KYTY_PLATFORM_LINUX
-
-// macOS uses the same POSIX platform setting and needs a signal stack too.
-class ThreadSignalStack {
-public:
-	ThreadSignalStack() {
-		const auto page_size = static_cast<size_t>(::getpagesize());
-		const auto stack_size =
-		    (std::max<size_t>(64 * 1024, MINSIGSTKSZ) + page_size - 1) & ~(page_size - 1);
-		if (::posix_memalign(&m_memory, page_size, stack_size) != 0) {
-			return;
-		}
-
-		stack_t stack {};
-		stack.ss_sp   = m_memory;
-		stack.ss_size = stack_size;
-		if (::sigaltstack(&stack, &m_previous) != 0) {
-			std::free(m_memory);
-			m_memory = nullptr;
-		}
-	}
-
-	~ThreadSignalStack() {
-		if (m_memory != nullptr && ::sigaltstack(&m_previous, nullptr) == 0) {
-			std::free(m_memory);
-		}
-	}
-
-	[[nodiscard]] bool IsInitialized() const { return m_memory != nullptr; }
-
-	KYTY_CLASS_NO_COPY(ThreadSignalStack)
-
-private:
-	void*   m_memory = nullptr;
-	stack_t m_previous {};
-};
-
-bool InitializeThreadSignalStack() {
-	// Keep fault handling off guest stacks, which GPU tracking can make read-only.
-	thread_local ThreadSignalStack signal_stack;
-	return signal_stack.IsInitialized();
+	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) noexcept {
+static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) {
+	FilterScope filter_scope;
+
 	auto* exception_record = exception->ExceptionRecord;

 	if (exception_record->ExceptionCode == DBG_PRINTEXCEPTION_C ||
@@ -112,6 +105,12 @@ static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) noexcept {
 	} 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;
 	}

@@ -132,21 +131,28 @@ static LONG WINAPI ExceptionFilter(PEXCEPTION_POINTERS exception) noexcept {
 	info.r14 = exception->ContextRecord->R14;
 	info.r15 = exception->ContextRecord->R15;

-	const auto handler = g_handler.load(std::memory_order_acquire);
-	if (handler != nullptr && handler(info)) {
-		return EXCEPTION_CONTINUE_EXECUTION;
-	}
-	return EXCEPTION_CONTINUE_SEARCH;
+	const auto handler = LoadInstalledHandler();
+
+	return handler(info) ? EXCEPTION_CONTINUE_EXECUTION : 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) {
@@ -164,6 +170,11 @@ 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;
@@ -199,7 +210,14 @@ 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 && handler(info)) {
+	if (handler == nullptr) {
+		FailFast("host exception callback is null");
+	}
+
+	const bool resolved   = handler(info);
+	g_in_exception_filter = false;
+
+	if (resolved) {
 		return; // retry the faulting instruction against the fixed mapping
 	}

@@ -226,6 +244,8 @@ 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;

@@ -269,8 +289,9 @@ 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 = g_handler.load(std::memory_order_acquire);
-	if (handler != nullptr && handler(info)) {
+	const auto handler = LoadInstalledHandler();
+
+	if (handler(info)) {
 		return;
 	}

@@ -292,7 +313,7 @@ bool InstallHandler(Handler handler) {
 	g_handler.store(handler, std::memory_order_release);

 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
-	if (AddVectoredExceptionHandler(0, ExceptionFilter) == nullptr) {
+	if (AddVectoredExceptionHandler(1, ExceptionFilter) == nullptr) {
 		g_handler.store(nullptr, std::memory_order_release);
 		g_install_state.store(0, std::memory_order_release);
 		printf("AddVectoredExceptionHandler() failed\n");
@@ -301,7 +322,7 @@ bool InstallHandler(Handler handler) {
 #elif defined(__APPLE__)
 	struct sigaction sa {};
 	sa.sa_sigaction = SignalHandler;
-	sa.sa_flags     = SA_SIGINFO | SA_ONSTACK;
+	sa.sa_flags     = SA_SIGINFO;
 	sigemptyset(&sa.sa_mask);
 	// The guest signal-dispatch path (KernelRaiseException) interrupts threads with
 	// SIGUSR1; block it while a fault is being resolved so a stop-the-world request
@@ -322,7 +343,8 @@ bool InstallHandler(Handler handler) {
 	struct sigaction action {};
 	action.sa_sigaction = SignalHandler;
 	sigemptyset(&action.sa_mask);
-	action.sa_flags = SA_SIGINFO | SA_RESTART | SA_ONSTACK;
+	// Fault resolution needs the normal thread stack.
+	action.sa_flags = SA_SIGINFO | SA_RESTART;

 	for (const int signal_number: {SIGSEGV, SIGBUS, SIGILL}) {
 		if (::sigaction(signal_number, &action, nullptr) != 0) {
@@ -39,10 +39,6 @@ using Handler = bool (*)(const ExceptionInfo&);

 bool InstallHandler(Handler handler);

-#if KYTY_PLATFORM == KYTY_PLATFORM_LINUX
-bool InitializeThreadSignalStack();
-#endif
-
 } // namespace Common::HostException

 #endif /* KYTY_COMMON_HOST_EXCEPTION_H_ */
@@ -129,14 +129,6 @@ static void WriteImpl(std::string_view text, fmt::text_style style = {}) {
 	}
 }

-void WriteToConsoleAndLog(std::string_view text) {
-	WriteImpl(text);
-	if (g_initialized && g_direction != Direction::Console) {
-		WriteStdout(text);
-	}
-	Flush();
-}
-
 void WriteFatal(std::string_view text) {
 	if (g_direction == Direction::Silent || !g_initialized) {
 		WriteStdout(text);
@@ -157,19 +149,25 @@ void WriteFatal(fmt::text_style style, std::string_view text) {
 	Flush();
 }

-void Initialize() {
+KYTY_SUBSYSTEM_INIT(Log) {
 	g_initialized = true;
 	switch (Config::GetPrintfDirection()) {
-		case Config::LogDirection::Silent: g_direction = Direction::Silent; break;
-		case Config::LogDirection::Console: g_direction = Direction::Console; break;
-		case Config::LogDirection::File: g_direction = Direction::File; break;
+		case Config::OutputDirection::Silent: g_direction = Direction::Silent; break;
+		case Config::OutputDirection::Console: g_direction = Direction::Console; break;
+		case Config::OutputDirection::File: g_direction = Direction::File; break;
 	}
 	g_output_file =
 	    (g_direction == Direction::File ? Config::GetPrintfOutputFile() : std::filesystem::path {});
 	SetupLogger();
 }

-void Shutdown() {
+KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Log) {
+	Flush();
+	std::lock_guard lock(g_logger_mutex);
+	g_logger.reset();
+}
+
+KYTY_SUBSYSTEM_DESTROY(Log) {
 	Flush();
 	std::lock_guard lock(g_logger_mutex);
 	g_logger.reset();
@@ -2,6 +2,7 @@
 #define KYTY_COMMON_LOGGING_LOG_H_

 #include "common/common.h"
+#include "common/subsystems.h"

 #include <fmt/color.h>
 #include <fmt/printf.h>
@@ -9,15 +10,7 @@

 namespace 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;
-};
+KYTY_SUBSYSTEM_DEFINE(Log);

 enum class Direction { Silent, Console, File };

@@ -25,7 +18,6 @@ Direction GetDirection();
 bool      IsSilent();
 void      Write(std::string_view text);
 void      Write(fmt::text_style style, std::string_view text);
-void      WriteToConsoleAndLog(std::string_view text);
 void      WriteFatal(std::string_view text);
 void      WriteFatal(fmt::text_style style, std::string_view text);
 void      Flush();
@@ -30,6 +30,8 @@ struct sys_dbg_stack_info_t {
 #endif
 };

-void SysStackUsage(sys_dbg_stack_info_t& s); // NOLINT(google-runtime-references)
+void SysStackWalk(void** stack, int* depth);
+void SysStackUsage(sys_dbg_stack_info_t& s);          // NOLINT(google-runtime-references)
+void SysStackUsagePrint(sys_dbg_stack_info_t& stack); // NOLINT(google-runtime-references)

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

 #endif /* KYTY_COMMON_PLATFORM_SYSFILEIO_H_ */
@@ -8,6 +8,7 @@

 #include <cstdlib>
 #include <cstring>
+#include <execinfo.h>
 #include <pthread.h>
 #include <sys/param.h>
 #include <sys/types.h>
@@ -16,6 +17,56 @@
 #include <libgen.h> // POSIX basename() lives here on macOS, not in <cstring>
 #endif

+// Avoid unwinding a guest-owned stack.
+static bool OnOwnStack() {
+	const char* probe = reinterpret_cast<const char*>(&probe);
+
+	pthread_attr_t attr {};
+#if defined(__APPLE__)
+	const auto* top  = static_cast<const char*>(pthread_get_stackaddr_np(pthread_self()));
+	const auto  size = pthread_get_stacksize_np(pthread_self());
+	(void)attr;
+	return top != nullptr && size != 0 && probe < top && probe >= top - size;
+#else
+	if (pthread_getattr_np(pthread_self(), &attr) != 0) {
+		return false;
+	}
+	void*      base = nullptr;
+	size_t     size = 0;
+	const bool ok = pthread_attr_getstack(&attr, &base, &size) == 0 && base != nullptr && size != 0;
+	pthread_attr_destroy(&attr);
+	if (!ok) {
+		return false;
+	}
+	const auto* low = static_cast<const char*>(base);
+	return probe >= low && probe < low + size;
+#endif
+}
+
+void SysStackWalk(void** stack, int* depth) {
+	if (stack == nullptr || depth == nullptr || *depth <= 0) {
+		if (depth != nullptr) {
+			*depth = 0;
+		}
+		return;
+	}
+
+	if (!OnOwnStack()) {
+		*depth = 0;
+		return;
+	}
+
+	const int n = ::backtrace(stack, *depth);
+	*depth      = (n < 0 ? 0 : n);
+}
+
+void SysStackUsagePrint(sys_dbg_stack_info_t& stack) {
+	printf("stack: (0x%" PRIx64 ", %" PRIu64 ")\n", static_cast<uint64_t>(stack.commited_addr),
+	       static_cast<uint64_t>(stack.commited_size));
+	printf("code: (0x%" PRIx64 ", %" PRIu64 ")\n", static_cast<uint64_t>(stack.code_addr),
+	       static_cast<uint64_t>(stack.code_size));
+}
+
 void SysStackUsage(sys_dbg_stack_info_t& s) {
 	pid_t pid = getpid();

@@ -693,7 +693,7 @@ bool SysFileCopyFile(const std::filesystem::path& src, const std::filesystem::pa
 	       !error;
 }

-bool SysFileRenameFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
+bool SysFileMoveFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
 	auto real_src = get_internal_name(src);
 	auto real_dst = get_internal_name(dst);

@@ -31,7 +31,8 @@
 namespace Common {

 static pthread_mutex_t              g_virtual_mutex {};
-static std::map<uintptr_t, size_t>* g_allocs = nullptr;
+static std::map<uintptr_t, size_t>* g_allocs   = nullptr;
+static std::map<uintptr_t, int>*    g_protects = nullptr;

 void SysVirtualInit() {
 	pthread_mutexattr_t attr {};
@@ -45,7 +46,8 @@ void SysVirtualInit() {
 	pthread_mutex_init(&g_virtual_mutex, &attr);
 	pthread_mutexattr_destroy(&attr);

-	g_allocs = new std::map<uintptr_t, size_t>;
+	g_allocs   = new std::map<uintptr_t, size_t>;
+	g_protects = new std::map<uintptr_t, int>;
 }

 static int get_protection_flag(VirtualMemory::Mode mode) {
@@ -66,6 +68,21 @@ static int get_protection_flag(VirtualMemory::Mode mode) {
 	return protect;
 }

+static VirtualMemory::Mode get_protection_flag(int mode) {
+	switch (mode) {
+		case PROT_NONE: return VirtualMemory::Mode::NoAccess;
+		case PROT_READ: return VirtualMemory::Mode::Read;
+		case PROT_WRITE: return VirtualMemory::Mode::Write;
+		case PROT_READ | PROT_WRITE: return VirtualMemory::Mode::ReadWrite; // NOLINT
+		case PROT_EXEC: return VirtualMemory::Mode::Execute;
+		case PROT_EXEC | PROT_WRITE: return VirtualMemory::Mode::ExecuteWrite; // NOLINT
+		case PROT_EXEC | PROT_READ: return VirtualMemory::Mode::ExecuteRead;   // NOLINT
+		case PROT_EXEC | PROT_WRITE | PROT_READ:
+			return VirtualMemory::Mode::ExecuteReadWrite; // NOLINT
+		default: return VirtualMemory::Mode::NoAccess;
+	}
+}
+
 // Keep automatic mappings inside the guest and GPU-addressable low window.
 #ifdef KYTY_FIXED_NOREPLACE
 static constexpr uintptr_t LOW_ARENA_LIMIT = 0x000000FC00000000ULL; // libc mspace window ceiling
@@ -144,6 +161,11 @@ uint64_t SysVirtualAlloc(uint64_t address, uint64_t size, VirtualMemory::Mode mo
 	if (ptr != MAP_FAILED) {
 		pthread_mutex_lock(&g_virtual_mutex);
 		record_alloc(ret_addr, size);
+		uintptr_t page_start = ret_addr >> 12u;
+		uintptr_t page_end   = (ret_addr + size - 1) >> 12u;
+		for (uintptr_t page = page_start; page <= page_end; page++) {
+			(*g_protects)[page] = protect;
+		}
 		pthread_mutex_unlock(&g_virtual_mutex);
 	}

@@ -229,6 +251,11 @@ uint64_t SysVirtualAllocAligned(uint64_t address, uint64_t size, VirtualMemory::

 	pthread_mutex_lock(&g_virtual_mutex);
 	record_alloc(ret_addr, size);
+	uintptr_t page_start = ret_addr >> 12u;
+	uintptr_t page_end   = (ret_addr + size - 1) >> 12u;
+	for (uintptr_t page = page_start; page <= page_end; page++) {
+		(*g_protects)[page] = protect;
+	}
 	pthread_mutex_unlock(&g_virtual_mutex);

 	return ret_addr;
@@ -310,6 +337,11 @@ bool SysVirtualAllocFixed(uint64_t address, uint64_t size, VirtualMemory::Mode m
 	if (ptr != MAP_FAILED) {
 		pthread_mutex_lock(&g_virtual_mutex);
 		record_alloc(ret_addr, size);
+		uintptr_t page_start = ret_addr >> 12u;
+		uintptr_t page_end   = (ret_addr + size - 1) >> 12u;
+		for (uintptr_t page = page_start; page <= page_end; page++) {
+			(*g_protects)[page] = protect;
+		}
 		pthread_mutex_unlock(&g_virtual_mutex);

 		return true;
@@ -484,7 +516,18 @@ bool SysVirtualFree(uint64_t address) {
 		return false;
 	}

-	return munmap(reinterpret_cast<void*>(addr), size) == 0;
+	if (munmap(reinterpret_cast<void*>(addr), size) == 0) {
+		uintptr_t page_start = addr >> 12u;
+		uintptr_t page_end   = (addr + size - 1) >> 12u;
+		pthread_mutex_lock(&g_virtual_mutex);
+		for (uintptr_t page = page_start; page <= page_end; page++) {
+			g_protects->erase(page);
+		}
+		pthread_mutex_unlock(&g_virtual_mutex);
+		return true;
+	}
+
+	return false;
 }

 bool SysVirtualFreeRange(uint64_t address, uint64_t size) {
@@ -539,16 +582,40 @@ bool SysVirtualFreeRange(uint64_t address, uint64_t size) {
 	if (end < alloc_end) {
 		(*g_allocs)[end] = alloc_end - end;
 	}
+	for (uintptr_t page = addr >> 12u; page <= (end - 1u) >> 12u; page++) {
+		g_protects->erase(page);
+	}
 	pthread_mutex_unlock(&g_virtual_mutex);
 	return true;
 }

-bool SysVirtualProtect(uint64_t address, uint64_t size, VirtualMemory::Mode mode) {
-	const auto addr       = static_cast<uintptr_t>(address);
-	const auto page_start = addr >> 12u;
-	const auto page_end   = (addr + size - 1) >> 12u;
-	return mprotect(reinterpret_cast<void*>(page_start << 12u), (page_end - page_start + 1) << 12u,
-	                get_protection_flag(mode)) == 0;
+bool SysVirtualProtect(uint64_t address, uint64_t size, VirtualMemory::Mode mode,
+                       VirtualMemory::Mode* old_mode) {
+	auto addr = static_cast<uintptr_t>(address);
+
+	pthread_mutex_lock(&g_virtual_mutex);
+	if (old_mode != nullptr) {
+		if (auto s = g_protects->find(addr >> 12u); s != g_protects->end()) {
+			*old_mode = get_protection_flag(s->second);
+		} else {
+			*old_mode = VirtualMemory::Mode::NoAccess;
+		}
+	}
+	pthread_mutex_unlock(&g_virtual_mutex);
+
+	uintptr_t page_start = addr >> 12u;
+	uintptr_t page_end   = (addr + size - 1) >> 12u;
+	if (mprotect(reinterpret_cast<void*>(page_start << 12u), (page_end - page_start + 1) << 12u,
+	             get_protection_flag(mode)) == 0) {
+		pthread_mutex_lock(&g_virtual_mutex);
+		for (uintptr_t page = page_start; page <= page_end; page++) {
+			(*g_protects)[page] = get_protection_flag(mode);
+		}
+		pthread_mutex_unlock(&g_virtual_mutex);
+		return true;
+	}
+
+	return false;
 }

 bool SysVirtualFlushInstructionCache(uint64_t /*address*/, uint64_t /*size*/) {
@@ -19,7 +19,8 @@ bool     SysVirtualReserveFixed(uint64_t address, uint64_t size);
 bool     SysVirtualDecommit(uint64_t address, uint64_t size);
 bool     SysVirtualFree(uint64_t address);
 bool     SysVirtualFreeRange(uint64_t address, uint64_t size);
-bool     SysVirtualProtect(uint64_t address, uint64_t size, VirtualMemory::Mode mode);
+bool     SysVirtualProtect(uint64_t address, uint64_t size, VirtualMemory::Mode mode,
+                           VirtualMemory::Mode* old_mode = nullptr);
 bool     SysVirtualFlushInstructionCache(uint64_t address, uint64_t size);

 } // namespace Common
@@ -15,6 +15,66 @@
 #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;
+}
+
+void SysStackUsagePrint(sys_dbg_stack_info_t& stack) {
+	printf("stack: (0x%" PRIx64 ", %" PRIu64 ") + (0x%" PRIx64 ", %" PRIu64 ") + (0x%" PRIx64
+	       ", %" PRIu64 ")\n",
+	       static_cast<uint64_t>(stack.reserved_addr), static_cast<uint64_t>(stack.reserved_size),
+	       static_cast<uint64_t>(stack.guard_addr), static_cast<uint64_t>(stack.guard_size),
+	       static_cast<uint64_t>(stack.commited_addr), static_cast<uint64_t>(stack.commited_size));
+}
+
 void SysStackUsage(sys_dbg_stack_info_t& s) {
 	MEMORY_BASIC_INFORMATION mbi {};
 	[[maybe_unused]] size_t  ss = VirtualQuery(&mbi, &mbi, sizeof(mbi));
@@ -565,7 +565,7 @@ bool SysFileCopyFile(const std::filesystem::path& src, const std::filesystem::pa
 	return CopyFileW(src_wide.c_str(), dst_wide.c_str(), FALSE) != 0;
 }

-bool SysFileRenameFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
+bool SysFileMoveFile(const std::filesystem::path& src, const std::filesystem::path& dst) {
 	auto src_wide = src.wstring();
 	auto dst_wide = dst.wstring();
 	return MoveFileW(src_wide.c_str(), dst_wide.c_str()) != 0;
@@ -45,6 +45,18 @@ static DWORD GetProtectionFlag(VirtualMemory::Mode mode) {
 	return protect;
 }

+static VirtualMemory::Mode GetProtectionFlag(DWORD mode) {
+	switch (mode) {
+		case PAGE_NOACCESS: return VirtualMemory::Mode::NoAccess;
+		case PAGE_READONLY: return VirtualMemory::Mode::Read;
+		case PAGE_READWRITE: return VirtualMemory::Mode::ReadWrite;
+		case PAGE_EXECUTE: return VirtualMemory::Mode::Execute;
+		case PAGE_EXECUTE_READ: return VirtualMemory::Mode::ExecuteRead;
+		case PAGE_EXECUTE_READWRITE: return VirtualMemory::Mode::ExecuteReadWrite;
+		default: return VirtualMemory::Mode::NoAccess;
+	}
+}
+
 void SysVirtualInit() {}

 uint64_t SysVirtualAlloc(uint64_t address, uint64_t size, VirtualMemory::Mode mode) {
@@ -316,13 +328,17 @@ bool SysVirtualFreeRange(uint64_t address, uint64_t size) {
 	return current - address == size && SysVirtualFree(address);
 }

-bool SysVirtualProtect(uint64_t address, uint64_t size, VirtualMemory::Mode mode) {
+bool SysVirtualProtect(uint64_t address, uint64_t size, VirtualMemory::Mode mode,
+                       VirtualMemory::Mode* old_mode) {
 	DWORD old_protect = 0;
 	if (VirtualProtect(reinterpret_cast<LPVOID>(static_cast<uintptr_t>(address)), size,
 	                   GetProtectionFlag(mode), &old_protect) == 0) {
 		printf("VirtualProtect() failed: 0x%08" PRIx32 "\n", static_cast<uint32_t>(GetLastError()));
 		return false;
 	}
+	if (old_mode != nullptr) {
+		*old_mode = GetProtectionFlag(old_protect);
+	}
 	return true;
 }

@@ -1,6 +1,7 @@
 #include "common/profiler.h"

 #include "common/emulatorConfig.h"
+#include "common/subsystems.h"

 #include <algorithm>
 #include <common/TracyProtocol.hpp>
@@ -54,21 +55,35 @@ void SetThreadName(const char* name) {
 	}
 }

-void Initialize() {
-	if (Config::ProfilerEnabled() && !tracy::ProfilerAvailable()) {
-		tracy::StartupProfiler();
-		TracySetProgramName("KytyPS5");
-		::printf("Tracy profiler enabled: client %d.%d.%d, protocol %u, "
-		         "broadcast %u, connect to 127.0.0.1:8086\n",
-		         tracy::Version::Major, tracy::Version::Minor, tracy::Version::Patch,
-		         tracy::ProtocolVersion, tracy::BroadcastVersion);
-	}
-}
-
-void Shutdown() {
+void Close() {
 	if (tracy::ProfilerAvailable()) {
 		tracy::ShutdownProfiler();
 	}
 }

+KYTY_SUBSYSTEM_INIT(Profiler) {
+	switch (Config::GetProfilerDirection()) {
+		case Config::ProfilerDirection::Network:
+			if (!tracy::ProfilerAvailable()) {
+				tracy::StartupProfiler();
+				TracySetProgramName("KytyPS5");
+				::printf("Tracy profiler enabled: client %d.%d.%d, protocol %u, "
+				         "broadcast %u, connect to 127.0.0.1:8086\n",
+				         tracy::Version::Major, tracy::Version::Minor, tracy::Version::Patch,
+				         tracy::ProtocolVersion, tracy::BroadcastVersion);
+			}
+			break;
+		case Config::ProfilerDirection::None:
+		default: break;
+	}
+}
+
+KYTY_SUBSYSTEM_UNEXPECTED_SHUTDOWN(Profiler) {
+	Close();
+}
+
+KYTY_SUBSYSTEM_DESTROY(Profiler) {
+	Close();
+}
+
 } // namespace Profiler
@@ -2,6 +2,7 @@
 #define KYTY_COMMON_PROFILER_H_

 #include "common/common.h"
+#include "common/subsystems.h"

 #include <cstdint>
 #include <optional>
@@ -41,15 +42,7 @@ private:
 void EndBlock();
 void SetThreadName(const char* name);

-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;
-};
+KYTY_SUBSYSTEM_DEFINE(Profiler);

 } // namespace Profiler

@@ -1,132 +0,0 @@
-#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_
@@ -316,12 +316,10 @@ inline std::string SafeCsv(std::string_view text) {
 	return std::string(text);
 }

-inline std::string Utf16ToUtf8(std::u16string_view utf16) {
-	if (utf16.empty()) {
-		return {};
-	}
+inline std::string Utf16ToUtf8(const char16_t* utf16) {
+	std::u16string                                                    input(utf16);
 	std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> convert;
-	return convert.to_bytes(utf16.data(), utf16.data() + utf16.size());
+	return convert.to_bytes(input);
 }

 inline ByteBuffer HexToBin(std::string_view text) {
@@ -1,43 +1,232 @@
 #include "common/subsystems.h"

+#include "common/assert.h"
+
+#include <cstdarg>
+#include <cstdio>
+#include <cstring>
+#include <memory>
+#include <string>
+
 namespace Common {

-static Subsystems* g_active_subsystems = nullptr;
+class SubsystemPrivate {
+public:
+	SubsystemPrivate()          = default;
+	virtual ~SubsystemPrivate() = default;

-Subsystems::Subsystems(bool print): m_print(print) {
-	m_active.reserve(16);
-	g_active_subsystems = this;
-}
+	KYTY_CLASS_NO_COPY(SubsystemPrivate);

-Subsystems::~Subsystems() {
-	Destroy();
-	if (g_active_subsystems == this) {
-		g_active_subsystems = nullptr;
+	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;
 	}
-}

-void Subsystems::Destroy() {
-	for (auto it = m_active.rbegin(); it != m_active.rend(); ++it) {
-		if (it->shutdown != nullptr) {
-			it->shutdown();
+	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);
 		}
-	}
-	m_active.clear();
-}

-void Subsystems::EmergencyShutdown() {
-	for (auto it = m_active.rbegin(); it != m_active.rend(); ++it) {
-		if (it->emergency_shutdown != nullptr) {
-			it->emergency_shutdown();
+		node->initialized = false;
+	}
+
+	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;
 		}
+
+		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 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);
 }

-void Subsystems::EmergencyShutdownActive() {
-	if (g_active_subsystems != nullptr) {
-		g_active_subsystems->EmergencyShutdown();
+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
@@ -2,61 +2,90 @@
 #define KYTY_COMMON_SUBSYSTEMS_H_

 #include "common/common.h"
+#include "common/singleton.h"

-#include <cstdio>
-#include <vector>
+#include <initializer_list>
+#include <memory>

 namespace Common {

-class Subsystems {
+class Subsystem;
+class SubsystemsListPrivate;
+class SubsystemPrivate;
+
+class SubsystemsList {
 public:
-	using Callback = void (*)();
+	SubsystemsList();
+	virtual ~SubsystemsList();
+	void SetArgs(int argc, char* argv[]);

-	explicit Subsystems(bool print = false);
-	~Subsystems();
+	// void Add(Subsystem* s, const char* name, ...);
+	void Add(Subsystem* s, std::initializer_list<Subsystem*> deps);

-	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);
-		}
-	}
+	bool InitAll(bool print_msg = false);
+	void DestroyAll(bool print_msg = false);

-	void Destroy();
-	void EmergencyShutdown();
+	int*   GetArgc();
+	char** GetArgv();

-	static void EmergencyShutdownActive();
+	[[nodiscard]] const char* GetFailName() const;
+	[[nodiscard]] const char* GetFailMsg() const;

-	KYTY_CLASS_NO_COPY(Subsystems);
+	void ShutdownAll();
+
+	static SubsystemsList* Instance() { return Common::Singleton<SubsystemsList>::Instance(); }
+
+	KYTY_CLASS_NO_COPY(SubsystemsList);

 private:
-	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;
+	std::unique_ptr<SubsystemsListPrivate> m_p;
 };

+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_ */
@@ -1,6 +1,7 @@
 #include "common/threads.h"

 #include "common/assert.h"
+#include "common/debug.h"

 #include <algorithm>
 #include <atomic>
@@ -8,6 +9,7 @@
 #include <chrono>             // IWYU pragma: keep
 #include <condition_variable> // IWYU pragma: keep
 #include <mutex>
+#include <vector>

 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS && KYTY_COMPILER == KYTY_COMPILER_CLANG
 #define KYTY_WIN_CS
@@ -26,8 +28,7 @@
 #ifdef KYTY_WIN_CS
 #include <windows.h> // IWYU pragma: keep
 // IWYU pragma: no_include <winbase.h>
-constexpr DWORD    KYTY_CS_SPIN_COUNT          = 4000;
-constexpr uint64_t KYTY_SLEEP_SPIN_LIMIT_100NS = 500; // 50 us
+constexpr DWORD KYTY_CS_SPIN_COUNT = 4000;

 #ifndef CREATE_WAITABLE_TIMER_HIGH_RESOLUTION
 #define CREATE_WAITABLE_TIMER_HIGH_RESOLUTION 0x00000002
@@ -38,10 +39,7 @@ static void SleepHighResolution100ns(uint64_t units_100ns) {
 		return;
 	}

-	// 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) {
+	if (units_100ns <= 10000) {
 		LARGE_INTEGER frequency {};
 		LARGE_INTEGER start {};
 		if (QueryPerformanceFrequency(&frequency) != 0 && QueryPerformanceCounter(&start) != 0 &&
@@ -196,7 +194,37 @@ struct CondVarPrivate {
 #endif
 };

-static wait_poll_func_t g_cond_wait_poll_callback = nullptr;
+static std::recursive_mutex                         g_cond_waiters_mutex;
+static std::vector<std::pair<int, CondVarPrivate*>> g_cond_waiters;
+static wait_poll_func_t                             g_cond_wait_poll_callback = nullptr;
+
+static void WakeCondVar(CondVarPrivate* cond_var) {
+#ifdef KYTY_WIN_CS
+	static auto func = ResolveWakeAllConditionVariable();
+	EXIT_NOT_IMPLEMENTED(func == nullptr);
+	func(&cond_var->m_cv);
+#else
+	cond_var->m_cv.notify_all();
+#endif
+}
+
+static void RegisterCondWaiter(CondVarPrivate* cond_var) {
+	std::lock_guard lock(g_cond_waiters_mutex);
+	g_cond_waiters.emplace_back(Thread::GetThreadIdUnique(), cond_var);
+}
+
+static void UnregisterCondWaiter(CondVarPrivate* cond_var) {
+	const auto      thread_id = Thread::GetThreadIdUnique();
+	std::lock_guard lock(g_cond_waiters_mutex);
+
+	const auto it = std::find_if(g_cond_waiters.begin(), g_cond_waiters.end(),
+	                             [thread_id, cond_var](const auto& waiter) {
+		                             return waiter.first == thread_id && waiter.second == cond_var;
+	                             });
+	if (it != g_cond_waiters.end()) {
+		g_cond_waiters.erase(it);
+	}
+}

 struct ThreadPrivate {
 	ThreadPrivate(thread_func_t f, void* a): func(f), arg(a), m_thread(&Run, this) {}
@@ -220,11 +248,15 @@ static thread_id_t      g_main_thread;
 static int              g_main_thread_int;
 static std::atomic<int> g_thread_counter = 0;

-void InitializeThreads() {
+KYTY_SUBSYSTEM_INIT(Threads) {
 	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) {
@@ -334,6 +366,7 @@ CondVar::~CondVar() {
 }

 void CondVar::Wait(Mutex* mutex) {
+	RegisterCondWaiter(m_cond_var.get());
 #ifndef KYTY_WIN_CS
 	std::unique_lock<std::recursive_mutex> cpp_lock(mutex->m_mutex->m_mutex, std::adopt_lock_t());
 #endif
@@ -374,6 +407,7 @@ void CondVar::Wait(Mutex* mutex) {
 	}
 	cpp_lock.release();
 #endif
+	UnregisterCondWaiter(m_cond_var.get());
 }

 void CondVar::SetWaitPollCallback(wait_poll_func_t callback) {
@@ -382,6 +416,7 @@ void CondVar::SetWaitPollCallback(wait_poll_func_t callback) {

 bool CondVar::WaitFor(Mutex* mutex, uint32_t micros) {
 	bool ok = false;
+	RegisterCondWaiter(m_cond_var.get());
 #ifndef KYTY_WIN_CS
 	std::unique_lock<std::recursive_mutex> cpp_lock(mutex->m_mutex->m_mutex, std::adopt_lock_t());
 #endif
@@ -396,6 +431,7 @@ bool CondVar::WaitFor(Mutex* mutex, uint32_t micros) {
 	      std::cv_status::no_timeout);
 	cpp_lock.release();
 #endif
+	UnregisterCondWaiter(m_cond_var.get());
 	return ok;
 }

@@ -410,13 +446,16 @@ void CondVar::Signal() {
 }

 void CondVar::SignalAll() {
-#ifdef KYTY_WIN_CS
-	static auto func = ResolveWakeAllConditionVariable();
-	EXIT_NOT_IMPLEMENTED(func == nullptr);
-	func(&m_cond_var->m_cv);
-#else
-	m_cond_var->m_cv.notify_all();
-#endif
+	WakeCondVar(m_cond_var.get());
+}
+
+void CondVar::SignalThread(int thread_id) {
+	std::lock_guard lock(g_cond_waiters_mutex);
+	for (const auto& waiter: g_cond_waiters) {
+		if (waiter.first == thread_id) {
+			WakeCondVar(waiter.second);
+		}
+	}
 }

 int Thread::GetThreadIdUnique() {
@@ -2,13 +2,14 @@
 #define KYTY_COMMON_THREADS_H_

 #include "common/common.h"
+#include "common/subsystems.h"

 #include <memory>
 #include <string>

 namespace Common {

-void InitializeThreads();
+KYTY_SUBSYSTEM_DEFINE(Threads);

 using thread_func_t    = void (*)(void*);
 using wait_poll_func_t = void (*)();
@@ -77,6 +78,7 @@ public:
 	void Signal();
 	void SignalAll();

+	static void SignalThread(int thread_id);
 	static void SetWaitPollCallback(wait_poll_func_t callback);

 	KYTY_CLASS_NO_COPY(CondVar);
@@ -50,8 +50,8 @@ bool FreeRange(uint64_t address, uint64_t size) {
 	return SysVirtualFreeRange(address, size);
 }

-bool Protect(uint64_t address, uint64_t size, Mode mode) {
-	return SysVirtualProtect(address, size, mode);
+bool Protect(uint64_t address, uint64_t size, Mode mode, Mode* old_mode) {
+	return SysVirtualProtect(address, size, mode, old_mode);
 }

 bool FlushInstructionCache(uint64_t address, uint64_t size) {
@@ -35,7 +35,7 @@ bool     ReserveFixed(uint64_t address, uint64_t size);
 bool     Decommit(uint64_t address, uint64_t size);
 bool     Free(uint64_t address);
 bool     FreeRange(uint64_t address, uint64_t size);
-bool     Protect(uint64_t address, uint64_t size, Mode mode);
+bool     Protect(uint64_t address, uint64_t size, Mode mode, Mode* old_mode = nullptr);
 bool     FlushInstructionCache(uint64_t address, uint64_t size);

 } // namespace VirtualMemory
@@ -2,6 +2,7 @@

 #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"
@@ -15,7 +16,6 @@
 #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"
@@ -27,31 +27,13 @@

 #include <cstdlib>
 #include <filesystem>
-#include <thread>

 namespace Emulator {

 static void PrintSystemInfo() {
-	const Common::SystemInfo info = Common::GetSystemInfo();
+	Common::SystemInfo info = Common::GetSystemInfo();

-#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());
+	LOGF("ProcessorName = %s\n", info.ProcessorName.c_str());
 }

 static void KytyClose() {
@@ -61,7 +43,7 @@ static void KytyClose() {

 	LOGF("done!\n");

-	Common::Subsystems::EmergencyShutdownActive();
+	Common::SubsystemsListSingleton::Instance()->ShutdownAll();
 }

 static void MountOrCreateDir(const std::filesystem::path& dir, const std::string& point) {
@@ -123,13 +105,30 @@ static void ClearDebugTextureFolder() {
 	}
 }

-static void Init(const Config::ConfigOptions& cfg, const std::filesystem::path& param_json,
-                 Common::Subsystems& subsystems) {
+static void Init(const Config::ConfigOptions& cfg, const std::filesystem::path& param_json) {
 	EXIT_IF(!Common::Thread::IsMainThread());

-	subsystems.Initialize<Config::Lifecycle>();
+	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);
+
 	Config::Load(cfg);
-	subsystems.Initialize<Log::Lifecycle>();
+	slist->Add(log, {core, config});
+	slist->InitAll(true);

 	if (Common::File::IsFileExisting(param_json)) {
 		Loader::SystemContentLoadParamSfo(param_json);
@@ -139,16 +138,17 @@ static void Init(const Config::ConfigOptions& cfg, const std::filesystem::path&
 		}
 	}

-	// 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>();
+	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);
 }

 static void LoadElf(const std::filesystem::path& elf, bool dbg_print_reloc = false,
@@ -189,25 +189,16 @@ void Run(const RunOptions& options) {
 		EXIT("ELF is required\n");
 	}

-	const auto         param_json = options.app0_dir / "sce_sys" / "param.json";
-	Common::Subsystems subsystems(true);
-	Init(options.config, param_json, subsystems);
+	const auto param_json = options.app0_dir / "sce_sys" / "param.json";
+	Init(options.config, param_json);

 	ClearDebugTextureFolder();

 	PrintSystemInfo();
-	std::string title_id;
-	if (Loader::SystemContentParamSfoGetString("TITLE_ID", &title_id) && !title_id.empty()) {
-		Log::WriteToConsoleAndLog(fmt::format("Title ID: {}\n", title_id));
-	}

 	int ok = atexit(KytyClose);
 	EXIT_NOT_IMPLEMENTED(ok != 0);

-	// Guest threads are still running, so skip KytyClose() and only flush emergency state.
-	ok = at_quick_exit(Common::Subsystems::EmergencyShutdownActive);
-	EXIT_NOT_IMPLEMENTED(ok != 0);
-
 	Libs::LibKernel::FileSystem::Mount(options.app0_dir, "/app0");
 	Libs::LibKernel::FileSystem::Mount(options.app0_dir, "/hostapp");

@@ -1,9 +1,7 @@
 set(KYTY_GIT_VERSION "unknown")
-set(KYTY_GIT_HASH "unknown")
-set(KYTY_GIT_REVISION "unknown")
 if(GIT_EXECUTABLE)
 	execute_process(
-		COMMAND "${GIT_EXECUTABLE}" describe --tags --always --dirty
+		COMMAND "${GIT_EXECUTABLE}" describe --tags --always
 		WORKING_DIRECTORY "${GIT_WORKING_DIRECTORY}"
 		OUTPUT_VARIABLE KYTY_GIT_VERSION
 		OUTPUT_STRIP_TRAILING_WHITESPACE
@@ -13,29 +11,5 @@ if(GIT_EXECUTABLE)
 	if(NOT GIT_RESULT EQUAL 0)
 		set(KYTY_GIT_VERSION "unknown")
 	endif()
-
-	execute_process(
-		COMMAND "${GIT_EXECUTABLE}" rev-parse HEAD
-		WORKING_DIRECTORY "${GIT_WORKING_DIRECTORY}"
-		OUTPUT_VARIABLE KYTY_GIT_REVISION
-		OUTPUT_STRIP_TRAILING_WHITESPACE
-		RESULT_VARIABLE GIT_HASH_RESULT
-		ERROR_QUIET
-	)
-	if(NOT GIT_HASH_RESULT EQUAL 0)
-		set(KYTY_GIT_HASH "unknown")
-		set(KYTY_GIT_REVISION "unknown")
-	else()
-		string(SUBSTRING "${KYTY_GIT_REVISION}" 0 7 KYTY_GIT_HASH)
-		execute_process(
-			COMMAND "${GIT_EXECUTABLE}" diff-index --quiet HEAD --
-			WORKING_DIRECTORY "${GIT_WORKING_DIRECTORY}"
-			RESULT_VARIABLE GIT_DIRTY_RESULT
-			ERROR_QUIET
-		)
-		if(NOT GIT_DIRTY_RESULT EQUAL 0)
-			string(APPEND KYTY_GIT_HASH "-dirty")
-		endif()
-	endif()
 endif()
 configure_file("${INPUT_FILE}" "${OUTPUT_FILE}")
@@ -7,7 +7,6 @@
 #include "graphics/host_gpu/renderer/renderContext.h"

 #include <cstdint>
-#include <span>
 #include <vector>

 namespace Libs::Graphics {
@@ -16,13 +15,6 @@ 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; }
@@ -31,13 +23,13 @@ private:
 	friend class CommandProcessor;

 	struct BufferCursor {
-		std::span<const uint32_t> commands;
-		uint32_t                  offset_dw = 0;
+		uint32_t* next_packet         = nullptr;
+		uint32_t  remaining_dw        = 0;
+		uint32_t  total_dw            = 0;
+		uint32_t  deferred_advance_dw = 0;
 	};

 	std::vector<BufferCursor> m_buffer_stack;
-	std::span<const uint32_t> m_next_buffer;
-	bool                      m_chain         = false;
 	bool                      m_suspended     = false;
 	bool                      m_made_progress = false;
 };
@@ -51,19 +43,30 @@ public:
 		int64_t flip_arg  = 0;
 	};

-	CommandProcessor(RenderContext& renderer, int interrupt_event_id)
-	    : m_renderer(renderer), m_interrupt_event_id(interrupt_event_id) {}
+	explicit CommandProcessor(RenderContext& renderer): m_renderer(renderer) {}
 	~CommandProcessor() = default;

 	KYTY_CLASS_NO_COPY(CommandProcessor);

 	void Reset();
-	void ApplyContextStateOperation(ContextStateOperation operation);

-	void            BufferInit();
-	void            BufferFlush();
-	void            BufferFlushAndWait();
-	void            BufferWait();
+	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;
+	}
+
 	HW::Context&    GetCtx() { return m_ctx; }
 	HW::UserConfig& GetUcfg() { return m_ucfg; }
 	HW::Shader&     GetShCtx() { return m_sh_ctx; }
@@ -74,9 +77,13 @@ public:
 	void SetDrawIndirectArgsBaseAddress(uint64_t draw_indirect_args_base_addr);
 	void SetDispatchIndirectArgsBaseAddress(uint64_t dispatch_indirect_args_base_addr);
 	void SetNumInstances(uint32_t num_instances);
-	void DrawIndex(DrawIndexArgs args);
-	void DrawIndexOffset(uint32_t index_offset, uint32_t index_count);
-	void DrawIndexAuto(DrawAutoArgs args);
+	void DrawIndex(uint32_t index_count, const void* index_addr, uint32_t flags, uint32_t type,
+	               uint32_t instance_count = 0, const void* object_ids = nullptr,
+	               uint32_t render_target_slice_offset = 0, int32_t vertex_offset_add = 0,
+	               uint32_t first_instance = 0);
+	void DrawIndexOffset(uint32_t index_offset, uint32_t index_count, uint32_t flags);
+	void DrawIndexAuto(uint32_t index_count, uint32_t flags,
+	                   uint32_t render_target_slice_offset = 0);
 	void DrawIndirect(uint32_t data_offset, uint32_t draw_initiator, bool indexed);
 	void DrawIndirectMulti(uint32_t data_offset, uint32_t max_count_or_count,
 	                       const volatile uint32_t* count_addr, uint32_t stride_in_bytes,
@@ -101,7 +108,7 @@ 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, uint64_t event_address = 0);
+	void TriggerEvent(uint32_t event_type, uint32_t event_index);

 	void SetUserDataMarker(HW::UserSgprType type) { m_user_data_marker = type; }
 	[[nodiscard]] HW::UserSgprType GetUserDataMarker() const { return m_user_data_marker; }
@@ -110,7 +117,6 @@ public:
 	void SetCeComplete(bool complete) { m_ce_complete = complete; }
 	void WaitCe();
 	void WaitDeDiff(uint32_t diff);
-	void WaitForRewind(bool valid);
 	void IncrementDe();
 	void IncrementCe();

@@ -129,14 +135,13 @@ public:
 	                    const volatile void* address, uint32_t count_in_dwords);
 	[[nodiscard]] bool ShouldSkipPredicatedPackets() const { return m_predicate_skip; }

-	Pm4ProcessResult Process(Pm4Execution& execution, std::span<const uint32_t> commands);
-	void             ProcessIndirectBuffer(std::span<const uint32_t> commands, bool chain);
+	Pm4ProcessResult Process(Pm4Execution& execution, uint32_t* buffer, uint32_t size_dw);
+	void             ProcessIndirectBuffer(uint32_t* buffer, uint32_t size_dw);

 	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>
@@ -144,16 +149,19 @@ private:
 	                      uint32_t cache_action, uint32_t event_index, uint32_t event_write_source,
 	                      void* dst_gpu_addr, T value, uint32_t interrupt_selector,
 	                      uint32_t interrupt_context_id);
-	void ProcessPm4(Pm4Execution& execution);
+	void ProcessPm4(Pm4Execution& execution, size_t stop_depth);
 	void SuspendPm4();
-	CommandScheduler&   GetScheduler() const { return m_renderer.GetCommandScheduler(); }
-	CommandBuffer&      CurrentBuffer() { return GetScheduler().Current(); }
-	void                CheckBuffer() const { GetScheduler().CheckActive(); }
+	void SubmitNonIndexedDraw(uint32_t vertex_count, uint32_t flags,
+	                          uint32_t render_target_slice_offset, uint32_t first_vertex,
+	                          uint32_t first_instance);
+
+	CommandScheduler&    GetScheduler() const { return m_renderer.GetCommandScheduler(); }
+	RenderCommandBuffer& CurrentBuffer() { return GetScheduler().Current(); }
+	void                 CheckBuffer() const { GetScheduler().CheckActive(); }
+	GpuResourceManager&  GetGpuResources() const { return m_renderer.GetGpuResources(); }

 	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;
@@ -165,17 +173,16 @@ 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;
+	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_const_ram[0x3000] = {0};

-	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;
+	FlipInfo m_flip;
+	uint64_t m_submit_id      = 0;
+	bool     m_predicate_skip = false;
 };

 } // namespace Libs::Graphics
@@ -23,7 +23,9 @@ constexpr auto MakeContextDispatchTable() {
 	g_hw_ctx_func[Pm4::DB_STENCIL_CLEAR]             = HwCtxSetStencilClear;
 	g_hw_ctx_func[Pm4::DB_DEPTH_CLEAR]               = HwCtxSetDepthClear;
 	g_hw_ctx_func[Pm4::PA_SC_SCREEN_SCISSOR_TL]      = HwCtxSetScreenScissor;
+	g_hw_ctx_func[Pm4::DB_Z_INFO]                    = HwCtxSetDepthRenderTarget;
 	g_hw_ctx_func[Pm4::DB_STENCIL_INFO]              = HwCtxSetStencilInfo;
+	g_hw_ctx_func[Pm4::DB_HTILE_SURFACE]             = HwCtxSetDepthHtileSurface;
 	g_hw_ctx_func[Pm4::PA_SU_HARDWARE_SCREEN_OFFSET] = HwCtxSetHardwareScreenOffset;
 	g_hw_ctx_func[Pm4::PA_SC_WINDOW_OFFSET]          = HwCtxSetWindowOffset;
 	g_hw_ctx_func[Pm4::PA_SC_WINDOW_SCISSOR_TL]      = HwCtxSetWindowScissor;
@@ -33,6 +35,7 @@ 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;
@@ -73,13 +76,16 @@ constexpr auto MakeContextDispatchTable() {
 	g_hw_ctx_func[Pm4::PA_SC_BINNER_CNTL_0]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::PA_SC_BINNER_CNTL_1]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::PA_SC_CONSERVATIVE_RASTERIZATION_CNTL] = HwCtxSetPaScExtendedControl;
+	g_hw_ctx_func[Pm4::PA_SC_NGG_MODE_CNTL]                   = HwCtxSetPaScExtendedControl;
 	g_hw_ctx_func[Pm4::DB_ALPHA_TO_MASK]                      = HwCtxSetAlphaToMask;
 	g_hw_ctx_func[Pm4::VGT_DRAW_PAYLOAD_CNTL]                 = HwCtxSetDrawPayloadControl;
 	g_hw_ctx_func[Pm4::VGT_PRIMITIVEID_RESET]                 = HwCtxSetPrimitiveIdReset;
 	g_hw_ctx_func[Pm4::PA_CL_OBJPRIM_ID_CNTL]                 = HwCtxSetObjprimIdControl;
 	g_hw_ctx_func[Pm4::VGT_SHADER_STAGES_EN]                  = HwCtxSetShaderStages;
+	g_hw_ctx_func[Pm4::PA_CL_GB_VERT_CLIP_ADJ]                = HwCtxSetGuardBands;

 	for (uint32_t slot = 0; slot < 8; slot++) {
+		g_hw_ctx_func[Pm4::CB_COLOR0_BASE + slot * 15] = HwCtxSetRenderTarget;
 		g_hw_ctx_func[Pm4::CB_COLOR0_INFO + slot * 15] = HwCtxSetColorInfo;

 		g_hw_ctx_func[Pm4::CB_BLEND0_CONTROL + slot * 1] = HwCtxSetBlendControl;
@@ -112,21 +118,25 @@ constexpr auto MakeShaderDispatchTable() {
 	}

 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_PS_0 + slot] = HwShIgnoreUserAccumulator;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_PS_0 + slot] = HwShSetPsUserAccumSgpr;
 	}

 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_ESGS_0 + slot] = HwShIgnoreUserAccumulator;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_ESGS_0 + slot] = HwShSetGsUserAccumSgpr;
 	}
 	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_LSHS_0 + slot] = HwShIgnoreUserAccumulator;
+		g_hw_sh_func[Pm4::SPI_SHADER_USER_ACCUM_LSHS_0 + slot] = HwShSetHsUserAccumSgpr;
 	}

-	for (uint32_t slot = 0; slot < 4; slot++) {
-		g_hw_sh_func[Pm4::COMPUTE_USER_ACCUM_0 + slot] = HwShIgnoreUserAccumulator;
+	for (uint32_t slot = 0; Pm4::COMPUTE_USER_ACCUM_0 + slot < Pm4::COMPUTE_USER_DATA_0; slot++) {
+		g_hw_sh_func[Pm4::COMPUTE_USER_ACCUM_0 + slot] = HwShSetCsUserAccumSgpr;
 	}
 	g_hw_sh_func[Pm4::COMPUTE_PGM_LO]                  = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_HI]                  = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_TBA_LO]                  = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_TBA_HI]                  = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_TMA_LO]                  = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_TMA_HI]                  = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC1]               = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC2]               = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_START_X]                 = HwShSetCsRegisters;
@@ -136,16 +146,30 @@ constexpr auto MakeShaderDispatchTable() {
 	g_hw_sh_func[Pm4::COMPUTE_NUM_THREAD_Y]            = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_NUM_THREAD_Z]            = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_RESOURCE_LIMITS]         = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE0]      = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE1]      = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_TMPRING_SIZE]            = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE2]      = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_DESTINATION_EN_SE3]      = HwShSetCsRegisters;
 	g_hw_sh_func[Pm4::COMPUTE_PGM_RSRC3]               = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::COMPUTE_PACE_ID]                 = HwShSetCsRegisters;
-	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_PS]  = HwShSetRegisters;
-	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_GS]  = HwShSetRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_GS] = HwShSetRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_GS] = HwShSetRegisters;
-	g_hw_sh_func[Pm4::SPI_GRAPHICS_SHADER_CONTROL_HS]  = HwShSetRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_HS] = HwShSetRegisters;
-	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_HS] = HwShSetRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_SHADER_CHKSUM]           = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::COMPUTE_DISPATCH_TUNNEL]         = HwShSetCsRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_TBA_LO_PS]            = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_PS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_PS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_TMA_LO_PS]            = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_TMA_HI_PS]            = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_PS]          = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_GS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_GS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_GS] = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_GS] = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_ESGS]        = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC3_HS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_PGM_RSRC4_HS]         = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_LO_HS] = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_USER_DATA_ADDR_HI_HS] = HwShIgnoreRegisters;
+	g_hw_sh_func[Pm4::SPI_SHADER_REQ_CTRL_LSHS]        = HwShIgnoreRegisters;

 	return g_hw_sh_func;
 }
@@ -157,7 +181,6 @@ constexpr auto MakeUconfigDispatchTable() {
 	g_hw_uc_func[Pm4::VGT_INDEX_TYPE]            = HwUcSetIndexType;
 	g_hw_uc_func[Pm4::VGT_OBJECT_ID]             = HwUcSetObjectId;
 	g_hw_uc_func[Pm4::TEXTURE_GRADIENT_FACTORS]  = HwUcSetTextureGradientFactors;
-	g_hw_uc_func[Pm4::TEXTURE_GRADIENT_CONTROL]  = HwUcSetTextureGradientFactors;
 	g_hw_uc_func[Pm4::IA_MULTI_VGT_PARAM]        = HwUcSetIaMultiVgtParam;
 	g_hw_uc_func[Pm4::GE_MULTI_PRIM_IB_RESET_EN] = HwUcSetMultiPrimIbReset;
 	g_hw_uc_func[Pm4::TA_CS_BC_BASE_ADDR]        = HwUcSetBorderColorTableAddr;
@@ -194,7 +217,6 @@ 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;
@@ -202,7 +224,6 @@ 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;
@@ -217,7 +238,6 @@ 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;
 }
@@ -226,13 +246,14 @@ 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;
 }
@@ -44,7 +44,11 @@ uint32_t HwCtxSetBorderColorTableAddr(CommandProcessor&, uint32_t, uint32_t, con
 uint32_t HwCtxSetStencilClear(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetDepthClear(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetScreenScissor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwCtxSetDepthRenderTarget(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
+                                   uint32_t);
 uint32_t HwCtxSetStencilInfo(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwCtxSetDepthHtileSurface(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
+                                   uint32_t);
 uint32_t HwCtxSetHardwareScreenOffset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
                                      uint32_t);
 uint32_t HwCtxSetWindowOffset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -52,6 +56,7 @@ 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);
@@ -85,6 +90,8 @@ uint32_t HwCtxSetDrawPayloadControl(CommandProcessor&, uint32_t, uint32_t, const
 uint32_t HwCtxSetPrimitiveIdReset(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetObjprimIdControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetShaderStages(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwCtxSetGuardBands(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwCtxSetRenderTarget(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetColorInfo(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetBlendControl(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwCtxSetViewportScissor(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -95,10 +102,12 @@ uint32_t HwShSetPsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t
 uint32_t HwShSetGsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShSetHsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShSetCsUserSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShIgnoreUserAccumulator(CommandProcessor&, uint32_t, uint32_t, const uint32_t*,
-                                   uint32_t);
+uint32_t HwShSetPsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwShSetGsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwShSetHsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwShSetCsUserAccumSgpr(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwShSetCsRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
-uint32_t HwShSetRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
+uint32_t HwShIgnoreRegisters(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwUcSetPrimitiveType(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwUcSetIndexType(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
 uint32_t HwUcSetObjectId(CommandProcessor&, uint32_t, uint32_t, const uint32_t*, uint32_t);
@@ -112,7 +121,6 @@ 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);
@@ -124,7 +132,6 @@ 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);
@@ -3,8 +3,15 @@

 #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,
@@ -53,34 +60,24 @@ 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_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,
+	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,
 };

 enum class ChannelType : uint32_t {
@@ -108,52 +105,32 @@ enum class ChannelOrder : uint32_t {
 	kAltReversed = 3,
 };

-// 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.
+// 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.
 struct ColorComponentMapping {
-	static constexpr uint8_t kIdentity = 0xe4u;
-
-	uint8_t packed = kIdentity;
+	uint8_t packed = 0xe4u; // RGBA: 0, 1, 2, 3

 	[[nodiscard]] constexpr uint32_t Map(uint32_t component) const {
 		return component < 4u ? (packed >> (component * 2u)) & 0x3u : component;
 	}

-	[[nodiscard]] constexpr uint32_t ApplyMask(uint32_t logical_mask) const {
+	[[nodiscard]] constexpr uint32_t ApplyMask(uint32_t mask) const {
 		uint32_t mapped = 0;
-		for (uint32_t physical_component = 0; physical_component < 4u; physical_component++) {
-			mapped |= ((logical_mask >> Map(physical_component)) & 1u) << physical_component;
+		for (uint32_t component = 0; component < 4u; component++) {
+			mapped |= ((mask >> component) & 1u) << Map(component);
 		}
 		return mapped;
 	}

-	// 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; }
+	[[nodiscard]] constexpr bool IsIdentity() const { return packed == 0xe4u; }

 	bool operator==(const ColorComponentMapping&) const = default;
 };

 static_assert(sizeof(ColorComponentMapping) == sizeof(uint8_t));

-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
+inline constexpr ColorComponentMapping ColorMappingBgra {0xc6u}; // 2, 1, 0, 3

 enum class DepthFormat : uint32_t {
 	kInvalid = 0,
@@ -401,25 +378,11 @@ 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_S2_F1       = 156,
-	kFmask8_S4_F1       = 157,
-	kFmask8_S8_F1       = 158,
-	kFmask8_S2_F2       = 159,
-	kFmask8_S4_F2       = 160,
 	kFmask8_S4_F4       = 161,
-	kFmask16_S16_F1     = 162,
-	kFmask16_S8_F2      = 163,
-	kFmask32_S16_F2     = 164,
-	kFmask32_S8_F4      = 165,
-	kFmask32_S8_F8      = 166,
-	kFmask64_S16_F4     = 167,
-	kFmask64_S16_F8     = 168,
 	kBc1UNorm           = 169,
 	kBc1Srgb            = 170,
 	kBc2UNorm           = 171,
@@ -3,235 +3,129 @@
 #include "graphics/guest_gpu/gpu_defs.h"

 #include <array>
-#include <bit>

 namespace Libs::Graphics::Prospero {
 namespace {

 struct FormatInfo {
-	BufferFormat       format;
-	uint32_t           bytes_per_element;
-	uint32_t           block_compressed_bytes_per_block;
-	uint32_t           render_target_bytes_per_element;
-	bool               sampled_texture;
-	bool               uint_texture;
-	bool               sint_texture = false;
+	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;
 };

 constexpr FormatInfo kFormatInfo[] = {
-	{BufferFormat::k8UNorm, 1, 0, 1, true, false},
-	{BufferFormat::k8SNorm, 0, 0, 1, false, false},
-	{BufferFormat::k8UInt, 1, 0, 1, true, true},
-	{BufferFormat::k16UNorm, 2, 0, 2, true, false},
-	{BufferFormat::k16SNorm, 2, 0, 2, true, false},
-	{BufferFormat::k16UInt, 2, 0, 2, true, true},
-	{BufferFormat::k16SInt, 2, 0, 2, true, false, true},
-	{BufferFormat::k16Float, 2, 0, 2, true, false},
-	{BufferFormat::k8_8UNorm, 2, 0, 2, true, false},
-	{BufferFormat::k8_8SNorm, 2, 0, 2, true, false},
-	{BufferFormat::k8_8UInt, 2, 0, 2, true, true},
-	{BufferFormat::k8_8SInt, 2, 0, 2, true, false, true},
-	{BufferFormat::k32UInt, 4, 0, 4, true, true},
-	{BufferFormat::k32SInt, 4, 0, 4, true, false, true},
-	{BufferFormat::k32Float, 4, 0, 4, true, false},
-	{BufferFormat::k16_16UNorm, 4, 0, 4, true, false},
-	{BufferFormat::k16_16SNorm, 4, 0, 4, true, false},
-	{BufferFormat::k16_16UInt, 4, 0, 4, true, true},
-	{BufferFormat::k16_16SInt, 4, 0, 4, true, false, true},
-	{BufferFormat::k16_16Float, 4, 0, 4, true, false},
-	{BufferFormat::k11_11_10UInt, 4, 0, 4, true, true},
-	{BufferFormat::k11_11_10Float, 4, 0, 4, true, false},
-	{BufferFormat::k10_10_10_2UNorm, 4, 0, 4, true, false},
-	{BufferFormat::k10_10_10_2UInt, 4, 0, 4, true, true},
-	{BufferFormat::k8_8_8_8UNorm, 4, 0, 4, true, false},
-	{BufferFormat::k8_8_8_8SNorm, 4, 0, 4, true, false},
-	{BufferFormat::k8_8_8_8UInt, 4, 0, 4, true, true},
-	{BufferFormat::k8_8_8_8SInt, 4, 0, 4, true, false, true},
-	{BufferFormat::k32_32UInt, 8, 0, 8, true, true},
-	{BufferFormat::k32_32SInt, 8, 0, 8, true, false, true},
-	{BufferFormat::k32_32Float, 8, 0, 8, true, false},
-	{BufferFormat::k16_16_16_16UNorm, 8, 0, 8, true, false},
-	{BufferFormat::k16_16_16_16SNorm, 8, 0, 8, true, false},
-	{BufferFormat::k16_16_16_16UInt, 8, 0, 8, true, true},
-	{BufferFormat::k16_16_16_16SInt, 8, 0, 8, true, false, true},
-	{BufferFormat::k16_16_16_16Float, 8, 0, 8, true, false},
-	{BufferFormat::k32_32_32UInt, 12, 0, 12, true, true},
-	{BufferFormat::k32_32_32SInt, 12, 0, 12, true, false, true},
-	{BufferFormat::k32_32_32Float, 12, 0, 12, true, false},
-	{BufferFormat::k32_32_32_32UInt, 16, 0, 16, true, true},
-	{BufferFormat::k32_32_32_32SInt, 16, 0, 16, true, false, true},
-	{BufferFormat::k32_32_32_32Float, 16, 0, 16, true, false},
-	{BufferFormat::k8Srgb, 1, 0, 0, true, false},
-	{BufferFormat::k8_8Srgb, 2, 0, 0, true, false},
-	{BufferFormat::k8_8_8_8Srgb, 4, 0, 4, true, false},
-	{BufferFormat::k9_9_9_5Float, 4, 0, 0, true, false},
-	{BufferFormat::k5_6_5UNorm, 2, 0, 2, true, false},
-	{BufferFormat::k5_5_5_1UNorm, 2, 0, 2, true, false},
-	{BufferFormat::k1_5_5_5UNorm, 0, 0, 2, false, false},
-	{BufferFormat::k4_4_4_4UNorm, 2, 0, 2, true, false},
-	{BufferFormat::kFmask8_S2_F1, 1, 0, 1, true, false},
-	{BufferFormat::kFmask8_S4_F1, 1, 0, 1, true, false},
-	{BufferFormat::kFmask8_S8_F1, 1, 0, 1, true, false},
-	{BufferFormat::kFmask8_S2_F2, 1, 0, 1, true, false},
-	{BufferFormat::kFmask8_S4_F2, 1, 0, 1, true, false},
-	{BufferFormat::kFmask8_S4_F4, 1, 0, 1, true, false},
-	{BufferFormat::kFmask16_S16_F1, 2, 0, 2, true, false},
-	{BufferFormat::kFmask16_S8_F2, 2, 0, 2, true, false},
-	{BufferFormat::kFmask32_S16_F2, 4, 0, 4, true, false},
-	{BufferFormat::kFmask32_S8_F4, 4, 0, 4, true, false},
-	{BufferFormat::kFmask32_S8_F8, 4, 0, 4, true, false},
-	{BufferFormat::kFmask64_S16_F4, 8, 0, 8, true, false},
-	{BufferFormat::kFmask64_S16_F8, 8, 0, 8, true, false},
-	{BufferFormat::kBc1UNorm, 0, 8, 0, true, false},
-	{BufferFormat::kBc1Srgb, 0, 8, 0, true, false},
-	{BufferFormat::kBc2UNorm, 0, 16, 0, true, false},
-	{BufferFormat::kBc2Srgb, 0, 16, 0, true, false},
-	{BufferFormat::kBc3UNorm, 0, 16, 0, true, false},
-	{BufferFormat::kBc3Srgb, 0, 16, 0, true, false},
-	{BufferFormat::kBc4UNorm, 0, 8, 0, true, false},
-	{BufferFormat::kBc4SNorm, 0, 8, 0, true, false},
-	{BufferFormat::kBc5UNorm, 0, 16, 0, true, false},
-	{BufferFormat::kBc5SNorm, 0, 16, 0, true, false},
-	{BufferFormat::kBc6UFloat, 0, 16, 0, true, false},
-	{BufferFormat::kBc6SFloat, 0, 16, 0, true, false},
-	{BufferFormat::kBc7UNorm, 0, 16, 0, true, false},
-	{BufferFormat::kBc7Srgb, 0, 16, 0, true, false},
+    {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},
 };

-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 size_t kMaxFormat = static_cast<size_t>(BufferFormat::kBc7Srgb);
+	constexpr uint32_t                            kMaxFormat = GpuEnumValue(BufferFormat::kBc7Srgb);
 	std::array<const FormatInfo*, kMaxFormat + 1> lookup {};
 	for (const auto& info: kFormatInfo) {
-		lookup[static_cast<size_t>(info.format)] = &info;
+		lookup[info.format] = &info;
 	}
 	return lookup;
 }

 constexpr auto kFormatInfoLookup = MakeFormatInfoLookup();

-const FormatInfo* FindFormatInfo(BufferFormat format) {
-	const auto index = static_cast<size_t>(format);
-	return index < kFormatInfoLookup.size() ? kFormatInfoLookup[index] : nullptr;
+const FormatInfo* FindFormatInfo(uint32_t format) {
+	return format < kFormatInfoLookup.size() ? kFormatInfoLookup[format] : nullptr;
 }

 } // namespace

-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) {
+uint32_t NumBytesPerElement(uint32_t format) {
 	const auto* info = FindFormatInfo(format);
 	return info != nullptr ? info->bytes_per_element : 0;
 }

-uint32_t BlockCompressedBytesPerBlock(BufferFormat format) {
+uint32_t BlockCompressedBytesPerBlock(uint32_t format) {
 	const auto* info = FindFormatInfo(format);
 	return info != nullptr ? info->block_compressed_bytes_per_block : 0;
 }

-uint32_t RenderTargetBytesPerElement(BufferFormat format) {
+uint32_t RenderTargetBytesPerElement(uint32_t format) {
 	const auto* info = FindFormatInfo(format);
 	return info != nullptr ? info->render_target_bytes_per_element : 0;
 }

-bool IsFmaskTextureFormat(BufferFormat format) {
-	return format >= BufferFormat::kFmask8_S2_F1 && format <= BufferFormat::kFmask64_S16_F8;
-}
-
-TextureNumericClass SampledTextureNumericClass(BufferFormat format) {
+bool IsSupportedTextureFormat(uint32_t format) {
 	const auto* info = FindFormatInfo(format);
-	if (info == nullptr || !info->sampled_texture) {
-		return TextureNumericClass::Unsupported;
-	}
-	if (info->sint_texture) {
-		return TextureNumericClass::Sint;
-	}
-	return info->uint_texture ? TextureNumericClass::Uint : TextureNumericClass::Float;
+	return info != nullptr && info->supported_sampled_texture_format;
 }

-BufferFormat RemapTextureFormat(BufferFormat format) {
-	return format == BufferFormat::k11_11_10UInt ? BufferFormat::k32UInt : format;
+bool IsUintTextureFormat(uint32_t format) {
+	const auto* info = FindFormatInfo(format);
+	return info != nullptr && info->unsigned_integer_texture_format;
+}
+
+bool IsFmaskTextureFormat(uint32_t format) {
+	return format == GpuEnumValue(BufferFormat::kFmask8_S4_F4);
 }

 } // namespace Libs::Graphics::Prospero
@@ -1,51 +1,16 @@
 #ifndef EMULATOR_INCLUDE_EMULATOR_GRAPHICS_GUEST_GPU_GPU_FORMAT_H_
 #define EMULATOR_INCLUDE_EMULATOR_GRAPHICS_GUEST_GPU_GPU_FORMAT_H_

-#include "graphics/guest_gpu/gpu_defs.h"
+#include "common/common.h"

 namespace Libs::Graphics::Prospero {

-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;
-		}
-	}
-};
-
-enum class TextureNumericClass { Unsupported, Float, Uint, Sint };
-
-RenderTargetFormatEncoding ResolveRenderTargetFormat(ChannelLayout layout, ChannelType type);
-uint32_t                   NumBytesPerElement(BufferFormat format);
-uint32_t                   BlockCompressedBytesPerBlock(BufferFormat format);
-uint32_t                   RenderTargetBytesPerElement(BufferFormat format);
-bool                       IsFmaskTextureFormat(BufferFormat format);
-TextureNumericClass        SampledTextureNumericClass(BufferFormat format);
-BufferFormat               RemapTextureFormat(BufferFormat format);
+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);

 } // namespace Libs::Graphics::Prospero

@@ -3,100 +3,41 @@

 #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 GuestGpu final {
+class Gpu final {
 public:
-	explicit GuestGpu(RenderContext& renderer);
-	~GuestGpu();
-	KYTY_CLASS_NO_COPY(GuestGpu);
+	explicit Gpu(RenderContext& renderer);
+	~Gpu();
+	KYTY_CLASS_NO_COPY(Gpu);

 	void               Shutdown();
 	[[nodiscard]] bool IsStopping();
 	void               SendCommand(Common::UniqueFunction<void>&& command);
 	void               SendCommandSync(Common::UniqueFunction<void>&& command);
+	void SendCommandSyncWithProcessor(Common::UniqueFunction<void, CommandProcessor&>&& command);

-	// 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();
+	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();
 	[[nodiscard]] int GetFrameNum() const;

-	[[nodiscard]] static bool IsGpuThread() noexcept;
+	[[nodiscard]] static bool              IsCommandProcessorThread() noexcept;
+	[[nodiscard]] static CommandProcessor* CurrentCommandProcessor() noexcept;

 private:
-	static constexpr uint32_t ComputePipeCount     = 7;
-	static constexpr uint32_t QueuesPerComputePipe = 8;
-	static constexpr uint32_t ComputeQueueCount    = ComputePipeCount * QueuesPerComputePipe;
-	static constexpr uint32_t ComputeQueueBase     = 0x20;
-	static constexpr uint32_t QueueCount           = 1 + ComputeQueueCount;
-
-	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;
+	std::unique_ptr<GpuState> m_state;
 };
 } // namespace Libs::Graphics

@@ -11,6 +11,15 @@ struct ColorBase {
 	uint64_t addr = 0;
 };

+struct ColorPitch {
+	uint32_t pitch_div8_minus1       = 0;
+	uint32_t fmask_pitch_div8_minus1 = 0;
+};
+
+struct ColorSlice {
+	uint32_t slice_div64_minus1 = 0;
+};
+
 struct ColorView {
 	uint32_t base_array_slice_index = 0;
 	uint32_t last_array_slice_index = 0;
@@ -20,20 +29,25 @@ 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;
-	Prospero::ChannelLayout format                         = Prospero::ChannelLayout::kInvalid;
-	Prospero::ChannelType   channel_type                   = Prospero::ChannelType::kUNorm;
-	Prospero::ChannelOrder  channel_order                  = Prospero::ChannelOrder::kStandard;
+	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;
 };

 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;
 };
@@ -45,33 +59,41 @@ struct ColorAttrib2 {
 };

 struct ColorAttrib3 {
-	uint32_t           depth                        = 0;
-	Prospero::TileMode tile_mode                    = Prospero::TileMode::kRenderTarget;
-	uint32_t           dimension                    = 0;
-	bool               metadata_pipe_aligned        = false;
-	bool               write_vrs_rate_hint_to_cmask = false;
+	uint32_t depth              = 0;
+	uint32_t tile_mode          = 0;
+	uint32_t dimension          = 0;
+	bool     cmask_pipe_aligned = false;
+	bool     dcc_pipe_aligned   = false;
 };

 struct ColorDccControl {
-	enum class IndependentBlockSize : uint8_t { Disabled, Bytes64, Bytes128 };
-
-	uint32_t             max_uncompressed_block_size    = 2;
-	uint32_t             max_compressed_block_size      = 2;
-	uint32_t             color_transform                = 0;
-	bool                 dcc_clear_key_enable           = false;
-	bool                 overwrite_combiner_disable     = false;
-	IndependentBlockSize independent_block_size         = IndependentBlockSize::Disabled;
-	bool                 data_write_on_dcc_clear_to_reg = false;
+	uint32_t max_uncompressed_block_size    = 0;
+	uint32_t max_compressed_block_size      = 0;
+	uint32_t min_compressed_block_size      = 0;
+	uint32_t color_transform                = 0;
+	bool     dcc_clear_key_enable           = false;
+	bool     overwrite_combiner_disable     = false;
+	bool     independent_64b_blocks         = false;
+	bool     independent_128b_blocks        = false;
+	bool     data_write_on_dcc_clear_to_reg = false;
 };

 struct ColorCmask {
 	uint64_t addr = 0;
 };

+struct ColorCmaskSlice {
+	uint32_t slice_minus1 = 0;
+};
+
 struct ColorFmask {
 	uint64_t addr = 0;
 };

+struct ColorFmaskSlice {
+	uint32_t slice_minus1 = 0;
+};
+
 struct ColorClearWord0 {
 	uint32_t word0 = 0;
 };
@@ -84,8 +106,15 @@ struct ColorDccAddr {
 	uint64_t addr = 0;
 };

+struct ColorSize {
+	uint32_t width  = 0;
+	uint32_t height = 0;
+};
+
 struct RenderTarget {
 	ColorBase       base;
+	ColorPitch      pitch;
+	ColorSlice      slice;
 	ColorView       view;
 	ColorInfo       info;
 	ColorAttrib     attrib;
@@ -93,18 +122,21 @@ struct RenderTarget {
 	ColorAttrib3    attrib3;
 	ColorDccControl dcc;
 	ColorCmask      cmask;
+	ColorCmaskSlice cmask_slice;
 	ColorFmask      fmask;
+	ColorFmaskSlice fmask_slice;
 	ColorClearWord0 clear_word0;
 	ColorClearWord1 clear_word1;
 	ColorDccAddr    dcc_addr;
+	ColorSize       size;
 };

 struct DepthZInfo {
-	Prospero::DepthFormat                       format      = Prospero::DepthFormat::kInvalid;
+	uint32_t                                    format      = 0;
 	uint32_t                                    num_samples = 0;
 	Prospero::TextureCompatiblePlaneCompression texture_compatibility =
 	    Prospero::TextureCompatiblePlaneCompression::kDisable;
-	Prospero::ZCompareBase z_compare_base     = Prospero::ZCompareBase::kZMax;
+	Prospero::ZCompareBase z_compare_base     = Prospero::ZCompareBase::kZMin;
 	bool                   htile_acceleration = false;
 	bool                   expclear_enabled   = false;
 	bool                   partially_resident = false;
@@ -112,16 +144,16 @@ struct DepthZInfo {

 	[[nodiscard]] static DepthZInfo Decode(uint32_t value) {
 		DepthZInfo info;
-		info.format                = static_cast<Prospero::DepthFormat>(value & 0x3u);
+		info.format                = value & 0x3u;
 		info.num_samples           = (value >> 2u) & 0x3u;
 		info.texture_compatibility = static_cast<Prospero::TextureCompatiblePlaneCompression>(
-		    value & static_cast<uint32_t>(Prospero::TextureCompatiblePlaneCompression::kBitMask));
+		    value & Prospero::GpuEnumValue(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 & static_cast<uint32_t>(Prospero::ZCompareBase::kBitMask));
+		    value & Prospero::GpuEnumValue(Prospero::ZCompareBase::kBitMask));
 		return info;
 	}

@@ -135,18 +167,18 @@ struct DepthZInfo {
 };

 struct DepthStencilInfo {
-	Prospero::StencilFormat            format = Prospero::StencilFormat::kInvalid;
+	uint32_t                           format = 0;
 	Prospero::TextureCompatibleStencil texture_compatibility =
 	    Prospero::TextureCompatibleStencil::kDisable;
 	bool expclear_enabled       = false;
-	bool htile_stencil_disabled = true;
+	bool htile_stencil_disabled = false;
 	bool partially_resident     = false;

 	[[nodiscard]] static DepthStencilInfo Decode(uint32_t value) {
 		DepthStencilInfo info;
-		info.format                = static_cast<Prospero::StencilFormat>(value & 0x1u);
+		info.format                = value & 0x1u;
 		info.texture_compatibility = static_cast<Prospero::TextureCompatibleStencil>(
-		    value & static_cast<uint32_t>(Prospero::TextureCompatibleStencil::kBitMask));
+		    value & Prospero::GpuEnumValue(Prospero::TextureCompatibleStencil::kBitMask));
 		info.partially_resident     = (value & 0x00001000u) != 0;
 		info.expclear_enabled       = (value & 0x08000000u) != 0;
 		info.htile_stencil_disabled = (value & 0x20000000u) != 0;
@@ -162,6 +194,16 @@ struct DepthStencilInfo {
 	}
 };

+struct DepthRenderTargetDepthInfo {
+	uint32_t addr5_swizzle_mask = 0;
+	uint32_t array_mode         = 0;
+	uint32_t pipe_config        = 0;
+	uint32_t bank_width         = 0;
+	uint32_t bank_height        = 0;
+	uint32_t macro_tile_aspect  = 0;
+	uint32_t num_banks          = 0;
+};
+
 struct DepthDepthView {
 	uint32_t slice_start           = 0;
 	uint32_t slice_max             = 0;
@@ -176,18 +218,36 @@ struct DepthDepthSizeXY {
 	bool     valid = false;
 };

+struct DepthRenderTargetHTileSurface {
+	uint32_t linear                  = 0;
+	uint32_t full_cache              = 0;
+	uint32_t htile_uses_preload_win  = 0;
+	uint32_t preload                 = 0;
+	uint32_t prefetch_width          = 0;
+	uint32_t prefetch_height         = 0;
+	uint32_t dst_outside_zero_to_one = 0;
+};
+
 struct DepthRenderTarget {
-	DepthZInfo       z_info;
-	DepthStencilInfo stencil_info;
-	DepthDepthView   depth_view;
-	DepthDepthSizeXY size;
+	DepthZInfo                    z_info;
+	DepthStencilInfo              stencil_info;
+	DepthDepthView                depth_view;
+	DepthDepthSizeXY              size;
+	DepthRenderTargetDepthInfo    depth_info;
+	DepthRenderTargetHTileSurface htile_surface;

 	uint64_t z_read_base_addr        = 0;
 	uint64_t stencil_read_base_addr  = 0;
 	uint64_t z_write_base_addr       = 0;
 	uint64_t stencil_write_base_addr = 0;
+	uint32_t pitch_div8_minus1       = 0;
+	uint32_t height_div8_minus1      = 0;
+	bool     pitch_height_valid      = false;
+	uint32_t slice_div64_minus1      = 0;
 	uint64_t htile_data_base_addr    = 0;
-	uint8_t  shading_rate_encoding   = 0;
+	uint32_t width                   = 0;
+	uint32_t height                  = 0;
+	bool     width_height_valid      = false;
 };

 struct RenderControl {
@@ -196,19 +256,10 @@ struct RenderControl {
 	bool    resummarize_enable       = false;
 	bool    stencil_compress_disable = false;
 	bool    depth_compress_disable   = false;
-	bool    copy_depth_to_color      = false;
-	bool    copy_stencil_to_color    = false;
 	bool    copy_centroid            = false;
 	uint8_t copy_sample              = 0;
 };

-struct DepthRenderOverride {
-	bool force_z_valid       = false;
-	bool force_z_dirty       = false;
-	bool force_stencil_valid = false;
-	bool force_stencil_dirty = false;
-};
-
 struct GdsOaCounter {
 	uint32_t counter = 0;
 	uint32_t address = 0;
@@ -244,18 +295,23 @@ struct ClipControl {
 	bool    linear_attribute_clip_enable        = false;
 	bool    force_viewport_index_from_vs_enable = false;

-	[[nodiscard]] bool IsZClipEnabled() const { return !min_z_clip_disable && !max_z_clip_disable; }
+	[[nodiscard]] bool IsZClipModeRepresentable() const {
+		return min_z_clip_disable == max_z_clip_disable;
+	}
+	[[nodiscard]] bool IsZClipEnabled() const { return !min_z_clip_disable; }
 };

 struct DepthControl {
-	bool    stencil_enable      = false;
-	bool    z_enable            = false;
-	bool    z_write_enable      = false;
-	bool    depth_bounds_enable = false;
-	uint8_t zfunc               = 0;
-	bool    backface_enable     = false;
-	uint8_t stencilfunc         = 0;
-	uint8_t stencilfunc_bf      = 0;
+	bool    stencil_enable                     = false;
+	bool    z_enable                           = false;
+	bool    z_write_enable                     = false;
+	bool    depth_bounds_enable                = false;
+	uint8_t zfunc                              = 0;
+	bool    backface_enable                    = false;
+	uint8_t stencilfunc                        = 0;
+	uint8_t stencilfunc_bf                     = 0;
+	bool    color_writes_on_depth_fail_enable  = false;
+	bool    color_writes_on_depth_pass_disable = false;
 };

 struct StencilControl {
@@ -292,16 +348,6 @@ 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;
@@ -369,16 +415,16 @@ struct AaConfig {
 struct Viewport {
 	float zmin                                  = 0.0f;
 	float zmax                                  = 0.0f;
-	float xscale                                = 1.0f;
+	float xscale                                = 0.0f;
 	float xoffset                               = 0.0f;
-	float yscale                                = 1.0f;
+	float yscale                                = 0.0f;
 	float yoffset                               = 0.0f;
-	float zscale                                = 1.0f;
+	float zscale                                = 0.0f;
 	float zoffset                               = 0.0f;
 	int   viewport_scissor_left                 = 0;
 	int   viewport_scissor_top                  = 0;
-	int   viewport_scissor_right                = 16384;
-	int   viewport_scissor_bottom               = 16384;
+	int   viewport_scissor_right                = 0;
+	int   viewport_scissor_bottom               = 0;
 	bool  viewport_scissor_window_offset_enable = false;
 };

@@ -387,26 +433,26 @@ struct ScreenViewport {
 	uint32_t transform_control                    = 1087;
 	int      screen_scissor_left                  = 0;
 	int      screen_scissor_top                   = 0;
-	int      screen_scissor_right                 = 16384;
-	int      screen_scissor_bottom                = 16384;
+	int      screen_scissor_right                 = 0;
+	int      screen_scissor_bottom                = 0;
 	int      window_scissor_left                  = 0;
 	int      window_scissor_top                   = 0;
-	int      window_scissor_right                 = 16384;
-	int      window_scissor_bottom                = 16384;
+	int      window_scissor_right                 = 0;
+	int      window_scissor_bottom                = 0;
 	bool     window_scissor_window_offset_enable  = false;
 	int      generic_scissor_left                 = 0;
 	int      generic_scissor_top                  = 0;
-	int      generic_scissor_right                = 16384;
-	int      generic_scissor_bottom               = 16384;
+	int      generic_scissor_right                = 0;
+	int      generic_scissor_bottom               = 0;
 	bool     generic_scissor_window_offset_enable = false;
 	int      window_offset_x                      = 0;
 	int      window_offset_y                      = 0;
 	uint32_t hw_offset_x                          = 0;
 	uint32_t hw_offset_y                          = 0;
-	float    guard_band_horz_clip                 = 1.0f;
-	float    guard_band_vert_clip                 = 1.0f;
-	float    guard_band_horz_discard              = 1.0f;
-	float    guard_band_vert_discard              = 1.0f;
+	float    guard_band_horz_clip                 = 0.0f;
+	float    guard_band_vert_clip                 = 0.0f;
+	float    guard_band_horz_discard              = 0.0f;
+	float    guard_band_vert_discard              = 0.0f;
 	uint16_t clip_rect_rule                       = 0xffffu;
 	int      clip_rect_left[4]                    = {};
 	int      clip_rect_top[4]                     = {};
@@ -415,28 +461,30 @@ struct ScreenViewport {
 	bool     clip_rect_window_offset_enable[4]    = {};
 };

-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 VsShaderResource1 {
+	uint8_t vgprs                    = 0;
+	uint8_t sgprs                    = 0;
+	uint8_t priority                 = 0;
+	uint8_t float_mode               = 0;
+	bool    dx10_clamp               = false;
+	bool    ieee_mode                = false;
+	uint8_t vgpr_component_count     = 0;
+	bool    cu_group_enable          = false;
+	bool    require_forward_progress = false;
+	bool    fp16_overflow            = false;
 };

-struct HsShaderResource2 {
-	bool     scratch_en   = false;
-	uint8_t  user_sgpr    = 0;
-	uint16_t lds_size     = 0;
-	uint8_t  shared_vgprs = 0;
+struct VsShaderResource2 {
+	bool    scratch_en        = false;
+	uint8_t user_sgpr         = 0;
+	bool    offchip_lds       = false;
+	bool    streamout_enabled = false;
+	uint8_t shared_vgprs      = 0;
 };

 struct PsShaderResource1 {
 	uint8_t vgprs                    = 0;
+	uint8_t sgprs                    = 0;
 	uint8_t priority                 = 0;
 	uint8_t float_mode               = 0;
 	bool    dx10_clamp               = false;
@@ -452,7 +500,7 @@ struct PsShaderResource2 {
 	uint8_t user_sgpr              = 0;
 	bool    wave_cnt_en            = false;
 	uint8_t extra_lds_size         = 0;
-	uint8_t raster_ordered_shading = 0;
+	bool    raster_ordered_shading = false;
 	uint8_t shared_vgprs           = 0;
 };

@@ -460,33 +508,27 @@ struct PsStageRegisters {
 	uint64_t          data_addr = 0;
 	PsShaderResource1 rsrc1;
 	PsShaderResource2 rsrc2;
+	uint64_t          chksum = 0;
 };

 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  priority                  = 0;
-	uint8_t  float_mode                = 0;
-	bool     dx10_clamp                = false;
-	bool     debug_mode                = false;
-	bool     ieee_mode                 = false;
-	bool     require_forward_progress  = false;
-	bool     fp16_overflow             = false;
-	bool     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;
-	uint8_t  shared_vgprs              = 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  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;
 };

 struct EsStageRegisters {
@@ -494,28 +536,31 @@ struct EsStageRegisters {
 };

 struct LsStageRegisters {
-	uint64_t data_addr = 0;
+	uint64_t          data_addr = 0;
+	VsShaderResource1 rsrc1;
+	VsShaderResource2 rsrc2;
 };

 struct HsStageRegisters {
-	uint64_t          data_addr      = 0;
-	uint64_t          user_data_addr = 0;
-	HsShaderResource1 rsrc1;
-	HsShaderResource2 rsrc2;
+	uint64_t          data_addr = 0;
+	VsShaderResource1 rsrc1;
+	VsShaderResource2 rsrc2;
+	uint64_t          chksum = 0;
 };

 struct GsShaderResource1 {
-	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    cu_group_enable           = false;
-	bool    require_forward_progress  = false;
-	bool    threadgroup_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    lds_configuration        = false;
+	uint8_t gs_vgpr_component_count  = 0;
+	bool    fp16_overflow            = false;
 };

 struct GsShaderResource2 {
@@ -528,10 +573,10 @@ struct GsShaderResource2 {
 };

 struct GsStageRegisters {
-	uint64_t          data_addr      = 0;
-	uint64_t          user_data_addr = 0;
+	uint64_t          data_addr = 0;
 	GsShaderResource1 rsrc1;
 	GsShaderResource2 rsrc2;
+	uint64_t          chksum = 0;
 };

 struct ShaderRegisters {
@@ -587,7 +632,8 @@ struct ShaderRegisters {
 enum class UserSgprType { Unknown, Region, Vsharp };

 struct UserSgprInfo {
-	static constexpr int SGPRS_MAX = 32;
+	// Changed from 16 to 64
+	static constexpr int SGPRS_MAX = 64;

 	uint32_t     value[SGPRS_MAX] = {0};
 	UserSgprType type[SGPRS_MAX]  = {};
@@ -634,6 +680,12 @@ public:
 	void Reset() { *this = Context(); }

 	void SetColorBase(uint32_t slot, const ColorBase& base) { m_render_targets[slot].base = base; }
+	void SetColorPitch(uint32_t slot, const ColorPitch& pitch) {
+		m_render_targets[slot].pitch = pitch;
+	}
+	void SetColorSlice(uint32_t slot, const ColorSlice& slice) {
+		m_render_targets[slot].slice = slice;
+	}
 	void SetColorView(uint32_t slot, const ColorView& view) { m_render_targets[slot].view = view; }
 	void SetColorInfo(uint32_t slot, const ColorInfo& info) { m_render_targets[slot].info = info; }
 	void SetColorAttrib(uint32_t slot, const ColorAttrib& attrib) {
@@ -651,9 +703,15 @@ public:
 	void SetColorCmask(uint32_t slot, const ColorCmask& cmask) {
 		m_render_targets[slot].cmask = cmask;
 	}
+	void SetColorCmaskSlice(uint32_t slot, const ColorCmaskSlice& cmask_slice) {
+		m_render_targets[slot].cmask_slice = cmask_slice;
+	}
 	void SetColorFmask(uint32_t slot, const ColorFmask& fmask) {
 		m_render_targets[slot].fmask = fmask;
 	}
+	void SetColorFmaskSlice(uint32_t slot, const ColorFmaskSlice& fmask_slice) {
+		m_render_targets[slot].fmask_slice = fmask_slice;
+	}
 	void SetColorClearWord0(uint32_t slot, const ColorClearWord0& clear_word0) {
 		m_render_targets[slot].clear_word0 = clear_word0;
 	}
@@ -663,6 +721,7 @@ public:
 	void SetColorDccAddr(uint32_t slot, const ColorDccAddr& dcc_addr) {
 		m_render_targets[slot].dcc_addr = dcc_addr;
 	}
+	void SetColorSize(uint32_t slot, const ColorSize& size) { m_render_targets[slot].size = size; }
 	[[nodiscard]] const RenderTarget& GetRenderTarget(uint32_t slot) const {
 		return m_render_targets[slot];
 	}
@@ -817,18 +876,10 @@ public:
 	void SetClipControl(const ClipControl& control) { m_clip_control = control; }
 	[[nodiscard]] const RenderControl& GetRenderControl() const { return m_render_control; }
 	void SetRenderControl(const RenderControl& control) { m_render_control = control; }
-	[[nodiscard]] const DepthRenderOverride& GetDepthRenderOverride() const {
-		return m_depth_render_override;
-	}
-	void SetDepthRenderOverride(const DepthRenderOverride& control) {
-		m_depth_render_override = control;
-	}
 	[[nodiscard]] const DepthControl& GetDepthControl() const { return m_depth_control; }
 	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; }
@@ -853,10 +904,8 @@ 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]] uint32_t GetPrimitiveResetIndex() const { return m_primitive_reset_index; }
-	void SetPrimitiveResetIndex(uint32_t index) { m_primitive_reset_index = index; }
+	[[nodiscard]] float GetLineWidth() const { return m_line_width; }
+	void                SetLineWidth(float width) { m_line_width = width; }

 	[[nodiscard]] const ShaderRegisters& GetShaderRegisters() const { return m_sh_regs; }

@@ -901,8 +950,7 @@ public:
 	void SetScShaderControl(uint32_t value) { m_sh_regs.m_paScShaderControl = value; }

 private:
-	float    m_line_width            = 1.0f;
-	uint32_t m_primitive_reset_index = 0xffffffffu;
+	float m_line_width = 1.0f;

 	BlendControl    m_blend_control[8];
 	BlendColor      m_blend_color;
@@ -918,30 +966,22 @@ private:

 	uint32_t m_shader_stages = 0;

-	DepthRenderTarget   m_depth_render_target;
-	RenderControl       m_render_control;
-	DepthRenderOverride m_depth_render_override;
-	DepthControl        m_depth_control;
-	StencilControl      m_stencil_control;
-	StencilMask         m_stencil_mask;
-	float               m_depth_clear_value   = 0.0f;
-	float               m_depth_bounds_min    = 0.0f;
-	float               m_depth_bounds_max    = 1.0f;
-	uint8_t             m_stencil_clear_value = 0;
+	DepthRenderTarget m_depth_render_target;
+	RenderControl     m_render_control;
+	DepthControl      m_depth_control;
+	StencilControl    m_stencil_control;
+	StencilMask       m_stencil_mask;
+	float             m_depth_clear_value   = 0.0f;
+	float             m_depth_bounds_min    = 0.0f;
+	float             m_depth_bounds_max    = 1.0f;
+	uint8_t           m_stencil_clear_value = 0;

 	ModeControl m_mode_control;
-	PolyOffset  m_poly_offset;
 	EqaaControl m_eqaa_control;

 	ShaderRegisters m_sh_regs;
 };

-struct FsrView {
-	uint32_t control_points[2][4] {};
-	uint32_t alphas[2][2] {};
-	uint32_t window[2] {};
-};
-
 class UserConfig {
 public:
 	UserConfig()  = default;
@@ -951,14 +991,12 @@ public:

 	void Reset() { *this = UserConfig(); }

-	void SetPrimitiveType(Prospero::PrimitiveType prim_type) { m_prim_type = prim_type; }
-	[[nodiscard]] Prospero::PrimitiveType GetPrimType() const { return m_prim_type; }
+	void                   SetPrimitiveType(uint32_t prim_type) { m_prim_type = prim_type; }
+	[[nodiscard]] uint32_t 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; }
@@ -971,21 +1009,11 @@ public:
 	void SetGdsOaCntl(uint32_t value) { m_gds_oa.cntl = value; }
 	void SetGdsOaCounter(uint32_t value) { m_gds_oa.counters[m_gds_oa.GetIndex()].counter = value; }
 	void SetGdsOaAddress(uint32_t value) { m_gds_oa.counters[m_gds_oa.GetIndex()].address = value; }
-	void SetFsrControlPoint(uint32_t axis, uint32_t index, uint32_t value) {
-		m_fsr_view.control_points[axis][index] = value;
-	}
-	void SetFsrAlpha(uint32_t axis, uint32_t index, uint32_t value) {
-		m_fsr_view.alphas[axis][index] = value;
-	}
-	void SetFsrWindow(uint32_t index, uint32_t value) { m_fsr_view.window[index] = value; }
-	[[nodiscard]] const FsrView& GetFsrView() const { return m_fsr_view; }

 private:
-	FsrView m_fsr_view;
-	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;
+	uint32_t m_prim_type    = 0;
+	uint32_t m_index_offset = 0;
+	uint32_t m_object_id    = 0;

 	GeControl    m_ge_cntl;
 	GeUserVgprEn m_ge_user_vgpr_en;
@@ -1003,25 +1031,23 @@ public:

 	void SetEsShaderBase(uint64_t addr) { m_vs.es_regs.data_addr = addr; }
 	void SetLsShaderBase(uint64_t addr) { m_vs.ls_regs.data_addr = addr; }
+	void SetLsShaderResource1(const VsShaderResource1& rsrc1) { m_vs.ls_regs.rsrc1 = rsrc1; }
+	void SetLsShaderResource2(const VsShaderResource2& rsrc2) { m_vs.ls_regs.rsrc2 = rsrc2; }
 	void SetHsShaderBase(uint64_t addr) { m_vs.hs_regs.data_addr = addr; }
-	void SetHsUserDataAddress(uint32_t word, uint32_t value) {
-		SetUserDataAddressWord(m_vs.hs_regs.user_data_addr, word, value);
-	}
-	void SetHsShaderResource1(const HsShaderResource1& rsrc1) { m_vs.hs_regs.rsrc1 = rsrc1; }
-	void SetHsShaderResource2(const HsShaderResource2& rsrc2) { m_vs.hs_regs.rsrc2 = rsrc2; }
+	void SetHsShaderResource1(const VsShaderResource1& rsrc1) { m_vs.hs_regs.rsrc1 = rsrc1; }
+	void SetHsShaderResource2(const VsShaderResource2& rsrc2) { m_vs.hs_regs.rsrc2 = rsrc2; }
+	void SetHsShaderChksum(uint32_t value) { m_vs.hs_regs.chksum = value; }
 	void SetGsShaderBase(uint64_t addr) { m_vs.gs_regs.data_addr = addr; }
-	void SetGsUserDataAddress(uint32_t word, uint32_t value) {
-		SetUserDataAddressWord(m_vs.gs_regs.user_data_addr, word, value);
-	}
 	void SetGsShaderResource1(const GsShaderResource1& rsrc1) { m_vs.gs_regs.rsrc1 = rsrc1; }
 	void SetGsShaderResource2(const GsShaderResource2& rsrc2) { m_vs.gs_regs.rsrc2 = rsrc2; }
+	void SetGsShaderChksum(uint32_t value) { m_vs.gs_regs.chksum = value; }

 	void SetPsShaderBase(uint64_t addr) { m_ps.ps_regs.data_addr = addr; }
 	void SetPsShaderResource1(const PsShaderResource1& rsrc1) { m_ps.ps_regs.rsrc1 = rsrc1; }
 	void SetPsShaderResource2(const PsShaderResource2& rsrc2) { m_ps.ps_regs.rsrc2 = rsrc2; }
+	void SetPsShaderChksum(uint32_t value) { m_ps.ps_regs.chksum = value; }

 	void SetCsShader(const CsStageRegisters& cs_regs) { m_cs.cs_regs = cs_regs; }
-	void SetCsWaveSize(uint8_t wave_size) { m_cs.cs_regs.wave_size = wave_size; }

 	void SetPsUserSgpr(uint32_t id, uint32_t value, UserSgprType type) {
 		m_ps.ps_user_sgpr.value[id] = value;
@@ -1053,12 +1079,6 @@ public:
 	[[nodiscard]] const ComputeShaderInfo& GetCs() const { return m_cs; }

 private:
-	static void SetUserDataAddressWord(uint64_t& address, uint32_t word, uint32_t value) {
-		const auto shift = word * 32u;
-		address = (address & ~(uint64_t {0xffffffffu} << shift)) |
-		          (static_cast<uint64_t>(value) << shift);
-	}
-
 	VertexShaderInfo  m_vs;
 	PixelShaderInfo   m_ps;
 	ComputeShaderInfo m_cs;
@@ -21,14 +21,15 @@ 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;
 }

@@ -60,7 +61,6 @@ 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";
@@ -90,7 +90,6 @@ constexpr auto MakeOpcodeNames() {
 	names[IT_DISPATCH_DRAW_PREAMBLE]    = "IT_DISPATCH_DRAW_PREAMBLE";
 	names[IT_DISPATCH_DRAW]             = "IT_DISPATCH_DRAW";
 	names[IT_GET_LOD_STATS]             = "IT_GET_LOD_STATS";
-	names[IT_WAIT_REG_MEM_64]           = "IT_WAIT_REG_MEM_64";
 	names[IT_SET_CONTEXT_REG_INDIRECT]  = "IT_SET_CONTEXT_REG_INDIRECT";
 	return names;
 }
@@ -100,7 +99,7 @@ constexpr auto g_opcode_names   = MakeOpcodeNames();

 } // namespace

-void DumpPm4PacketStream(Common::File* file, const uint32_t* cmd_buffer, uint32_t start_dw,
+void DumpPm4PacketStream(Common::File* file, uint32_t* cmd_buffer, uint32_t start_dw,
                         uint32_t num_dw) {
 	// db_dump();

@@ -143,9 +142,9 @@ void DumpPm4PacketStream(Common::File* file, const uint32_t* cmd_buffer, uint32_
 				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<const uint32_t*>(
-					    (static_cast<uint64_t>(cmd[0]) & 0xfffffffcu) |
-					    (static_cast<uint64_t>(cmd[1]) << 32u));
+					auto* indirect_buffer =
+					    reinterpret_cast<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;
@@ -42,7 +42,6 @@ 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;
@@ -71,7 +70,6 @@ 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 */
@@ -80,14 +78,15 @@ 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;

@@ -98,10 +97,6 @@ constexpr uint32_t DB_RENDER_CONTROL_DEPTH_CLEAR_ENABLE_SHIFT       = 0;
 constexpr uint32_t DB_RENDER_CONTROL_DEPTH_CLEAR_ENABLE_MASK        = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_CLEAR_ENABLE_SHIFT     = 1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_CLEAR_ENABLE_MASK      = 0x1;
-constexpr uint32_t DB_RENDER_CONTROL_COPY_DEPTH_TO_COLOR_SHIFT      = 2;
-constexpr uint32_t DB_RENDER_CONTROL_COPY_DEPTH_TO_COLOR_MASK       = 0x1;
-constexpr uint32_t DB_RENDER_CONTROL_COPY_STENCIL_TO_COLOR_SHIFT    = 3;
-constexpr uint32_t DB_RENDER_CONTROL_COPY_STENCIL_TO_COLOR_MASK     = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_RESUMMARIZE_ENABLE_SHIFT       = 4;
 constexpr uint32_t DB_RENDER_CONTROL_RESUMMARIZE_ENABLE_MASK        = 0x1;
 constexpr uint32_t DB_RENDER_CONTROL_STENCIL_COMPRESS_DISABLE_SHIFT = 5;
@@ -167,6 +162,22 @@ constexpr uint32_t PA_SC_SCREEN_SCISSOR_BR_BR_Y_MASK  = 0xFFFF;

 constexpr uint32_t DB_DFSM_CONTROL = 0xE;

+constexpr uint32_t DB_DEPTH_INFO                          = 0xF;
+constexpr uint32_t DB_DEPTH_INFO_ADDR5_SWIZZLE_MASK_SHIFT = 0;
+constexpr uint32_t DB_DEPTH_INFO_ADDR5_SWIZZLE_MASK_MASK  = 0xF;
+constexpr uint32_t DB_DEPTH_INFO_ARRAY_MODE_SHIFT         = 4;
+constexpr uint32_t DB_DEPTH_INFO_ARRAY_MODE_MASK          = 0xF;
+constexpr uint32_t DB_DEPTH_INFO_PIPE_CONFIG_SHIFT        = 8;
+constexpr uint32_t DB_DEPTH_INFO_PIPE_CONFIG_MASK         = 0x1F;
+constexpr uint32_t DB_DEPTH_INFO_BANK_WIDTH_SHIFT         = 13;
+constexpr uint32_t DB_DEPTH_INFO_BANK_WIDTH_MASK          = 0x3;
+constexpr uint32_t DB_DEPTH_INFO_BANK_HEIGHT_SHIFT        = 15;
+constexpr uint32_t DB_DEPTH_INFO_BANK_HEIGHT_MASK         = 0x3;
+constexpr uint32_t DB_DEPTH_INFO_MACRO_TILE_ASPECT_SHIFT  = 17;
+constexpr uint32_t DB_DEPTH_INFO_MACRO_TILE_ASPECT_MASK   = 0x3;
+constexpr uint32_t DB_DEPTH_INFO_NUM_BANKS_SHIFT          = 19;
+constexpr uint32_t DB_DEPTH_INFO_NUM_BANKS_MASK           = 0x3;
+
 constexpr uint32_t DB_Z_INFO                               = 0x10;
 constexpr uint32_t DB_Z_INFO_FORMAT_SHIFT                  = 0;
 constexpr uint32_t DB_Z_INFO_FORMAT_MASK                   = 0x3;
@@ -210,6 +221,16 @@ constexpr uint32_t DB_STENCIL_READ_BASE  = 0x13;
 constexpr uint32_t DB_Z_WRITE_BASE       = 0x14;
 constexpr uint32_t DB_STENCIL_WRITE_BASE = 0x15;

+constexpr uint32_t DB_DEPTH_SIZE                       = 0x16;
+constexpr uint32_t DB_DEPTH_SIZE_PITCH_TILE_MAX_SHIFT  = 0;
+constexpr uint32_t DB_DEPTH_SIZE_PITCH_TILE_MAX_MASK   = 0x7FF;
+constexpr uint32_t DB_DEPTH_SIZE_HEIGHT_TILE_MAX_SHIFT = 11;
+constexpr uint32_t DB_DEPTH_SIZE_HEIGHT_TILE_MAX_MASK  = 0x7FF;
+
+constexpr uint32_t DB_DEPTH_SLICE                      = 0x17;
+constexpr uint32_t DB_DEPTH_SLICE_SLICE_TILE_MAX_SHIFT = 0;
+constexpr uint32_t DB_DEPTH_SLICE_SLICE_TILE_MAX_MASK  = 0x3FFFFF;
+
 constexpr uint32_t DB_Z_READ_BASE_HI                  = 0x1A;
 constexpr uint32_t DB_STENCIL_READ_BASE_HI            = 0x1B;
 constexpr uint32_t DB_Z_WRITE_BASE_HI                 = 0x1C;
@@ -295,8 +316,8 @@ 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_BLUE                = 0x106;
-constexpr uint32_t CB_BLEND_GREEN               = 0x107;
+constexpr uint32_t CB_BLEND_GREEN               = 0x106;
+constexpr uint32_t CB_BLEND_BLUE                = 0x107;
 constexpr uint32_t CB_BLEND_ALPHA               = 0x108;
 constexpr uint32_t CB_DCC_CONTROL               = 0x109;

@@ -350,7 +371,6 @@ 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;
@@ -365,6 +385,9 @@ 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;
@@ -402,6 +425,11 @@ constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_SHIFT                        = 8
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_MASK                         = 0x7;
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_BF_SHIFT                     = 20;
 constexpr uint32_t DB_DEPTH_CONTROL_STENCILFUNC_BF_MASK                      = 0x7;
+constexpr uint32_t DB_DEPTH_CONTROL_ENABLE_COLOR_WRITES_ON_DEPTH_FAIL_SHIFT  = 30;
+constexpr uint32_t DB_DEPTH_CONTROL_ENABLE_COLOR_WRITES_ON_DEPTH_FAIL_MASK   = 0x1;
+constexpr uint32_t DB_DEPTH_CONTROL_DISABLE_COLOR_WRITES_ON_DEPTH_PASS_SHIFT = 31;
+constexpr uint32_t DB_DEPTH_CONTROL_DISABLE_COLOR_WRITES_ON_DEPTH_PASS_MASK  = 0x1;
+
 constexpr uint32_t DB_EQAA                                  = 0x201;
 constexpr uint32_t DB_EQAA_MAX_ANCHOR_SAMPLES_SHIFT         = 0;
 constexpr uint32_t DB_EQAA_MAX_ANCHOR_SAMPLES_MASK          = 0x7;
@@ -533,32 +561,40 @@ constexpr uint32_t VGT_DRAW_PAYLOAD_CNTL  = 0x2A6;
 constexpr uint32_t VGT_ESGS_RING_ITEMSIZE = 0x2AB;
 constexpr uint32_t VGT_REUSE_OFF          = 0x2AD;

-constexpr uint32_t DB_SHADING_RATE_ENCODING                             = 0x2AF;
-constexpr uint32_t DB_SHADING_RATE_ENCODING_SHADING_RATE_ENCODING_SHIFT = 19;
-constexpr uint32_t DB_SHADING_RATE_ENCODING_SHADING_RATE_ENCODING_MASK  = 0x3;
+constexpr uint32_t DB_HTILE_SURFACE                               = 0x2AF;
+constexpr uint32_t DB_HTILE_SURFACE_LINEAR_SHIFT                  = 0;
+constexpr uint32_t DB_HTILE_SURFACE_LINEAR_MASK                   = 0x1;
+constexpr uint32_t DB_HTILE_SURFACE_FULL_CACHE_SHIFT              = 1;
+constexpr uint32_t DB_HTILE_SURFACE_FULL_CACHE_MASK               = 0x1;
+constexpr uint32_t DB_HTILE_SURFACE_HTILE_USES_PRELOAD_WIN_SHIFT  = 2;
+constexpr uint32_t DB_HTILE_SURFACE_HTILE_USES_PRELOAD_WIN_MASK   = 0x1;
+constexpr uint32_t DB_HTILE_SURFACE_PRELOAD_SHIFT                 = 3;
+constexpr uint32_t DB_HTILE_SURFACE_PRELOAD_MASK                  = 0x1;
+constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_WIDTH_SHIFT          = 4;
+constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_WIDTH_MASK           = 0x3F;
+constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_HEIGHT_SHIFT         = 10;
+constexpr uint32_t DB_HTILE_SURFACE_PREFETCH_HEIGHT_MASK          = 0x3F;
+constexpr uint32_t DB_HTILE_SURFACE_DST_OUTSIDE_ZERO_TO_ONE_SHIFT = 16;
+constexpr uint32_t DB_HTILE_SURFACE_DST_OUTSIDE_ZERO_TO_ONE_MASK  = 0x1;

-constexpr uint32_t DB_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 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 PA_SC_AA_CONFIG                             = 0x2F8;
 constexpr uint32_t PA_SC_AA_CONFIG_MSAA_NUM_SAMPLES_SHIFT      = 0;
@@ -582,6 +618,7 @@ constexpr uint32_t PA_SC_SHADER_CONTROL                  = 0x310;
 constexpr uint32_t PA_SC_BINNER_CNTL_0                   = 0x311;
 constexpr uint32_t PA_SC_BINNER_CNTL_1                   = 0x312;
 constexpr uint32_t PA_SC_CONSERVATIVE_RASTERIZATION_CNTL = 0x313;
+constexpr uint32_t PA_SC_NGG_MODE_CNTL                   = 0x314;
 constexpr uint32_t CB_COLOR0_BASE                        = 0x318;

 constexpr uint32_t CB_COLOR0_VIEW                   = 0x31B;
@@ -609,13 +646,24 @@ constexpr uint32_t CB_COLOR0_INFO_BLEND_BYPASS_SHIFT              = 16;
 constexpr uint32_t CB_COLOR0_INFO_BLEND_BYPASS_MASK               = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_ROUND_MODE_SHIFT                = 18;
 constexpr uint32_t CB_COLOR0_INFO_ROUND_MODE_MASK                 = 0x1;
+constexpr uint32_t CB_COLOR0_INFO_CMASK_IS_LINEAR_SHIFT           = 19;
+constexpr uint32_t CB_COLOR0_INFO_CMASK_IS_LINEAR_MASK            = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESSION_DISABLE_SHIFT = 26;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESSION_DISABLE_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESS_1FRAG_ONLY_SHIFT = 27;
 constexpr uint32_t CB_COLOR0_INFO_FMASK_COMPRESS_1FRAG_ONLY_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_INFO_DCC_ENABLE_SHIFT                = 28;
 constexpr uint32_t CB_COLOR0_INFO_DCC_ENABLE_MASK                 = 0x1;
+constexpr uint32_t CB_COLOR0_INFO_CMASK_ADDR_TYPE_SHIFT           = 29;
+constexpr uint32_t CB_COLOR0_INFO_CMASK_ADDR_TYPE_MASK            = 0x3;
+constexpr uint32_t CB_COLOR0_INFO_ALT_TILE_MODE_SHIFT             = 31;
+constexpr uint32_t CB_COLOR0_INFO_ALT_TILE_MODE_MASK              = 0x1;
+
 constexpr uint32_t CB_COLOR0_ATTRIB                             = 0x31D;
+constexpr uint32_t CB_COLOR0_ATTRIB_TILE_MODE_INDEX_SHIFT       = 0;
+constexpr uint32_t CB_COLOR0_ATTRIB_TILE_MODE_INDEX_MASK        = 0x1F;
+constexpr uint32_t CB_COLOR0_ATTRIB_FMASK_TILE_MODE_INDEX_SHIFT = 5;
+constexpr uint32_t CB_COLOR0_ATTRIB_FMASK_TILE_MODE_INDEX_MASK  = 0x1F;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_SAMPLES_SHIFT           = 12;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_SAMPLES_MASK            = 0x7;
 constexpr uint32_t CB_COLOR0_ATTRIB_NUM_FRAGMENTS_SHIFT         = 15;
@@ -630,6 +678,8 @@ constexpr uint32_t CB_COLOR0_DCC_CONTROL_KEY_CLEAR_ENABLE_SHIFT
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_KEY_CLEAR_ENABLE_MASK                  = 0x1;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_UNCOMPRESSED_BLOCK_SIZE_SHIFT      = 2;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_UNCOMPRESSED_BLOCK_SIZE_MASK       = 0x3;
+constexpr uint32_t CB_COLOR0_DCC_CONTROL_MIN_COMPRESSED_BLOCK_SIZE_SHIFT        = 4;
+constexpr uint32_t CB_COLOR0_DCC_CONTROL_MIN_COMPRESSED_BLOCK_SIZE_MASK         = 0x1;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_COMPRESSED_BLOCK_SIZE_SHIFT        = 5;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_MAX_COMPRESSED_BLOCK_SIZE_MASK         = 0x3;
 constexpr uint32_t CB_COLOR0_DCC_CONTROL_COLOR_TRANSFORM_SHIFT                  = 7;
@@ -686,25 +736,31 @@ constexpr uint32_t CB_COLOR0_ATTRIB3_CMASK_PIPE_ALIGNED_SHIFT = 26;
 constexpr uint32_t CB_COLOR0_ATTRIB3_CMASK_PIPE_ALIGNED_MASK  = 0x1;
 constexpr uint32_t CB_COLOR0_ATTRIB3_DCC_PIPE_ALIGNED_SHIFT   = 30;
 constexpr uint32_t CB_COLOR0_ATTRIB3_DCC_PIPE_ALIGNED_MASK    = 0x1;
-constexpr uint32_t CB_COLOR0_ATTRIB3_WRITE_VRS_RATE_HINT_TO_CMASK_SHIFT = 31;
-constexpr uint32_t CB_COLOR0_ATTRIB3_WRITE_VRS_RATE_HINT_TO_CMASK_MASK  = 0x1;

 constexpr uint32_t CB_COLOR7_ATTRIB3 = 0x3BF;

-constexpr uint32_t CX_NOP = 0xDB;
+/* Fake codes. Don't exist on real HW */
+
+constexpr uint32_t CX_NOP = 0x800003FF;

 constexpr uint32_t CX_NUM = 0x3FF + 1;

 /* Shader registers */

-constexpr uint32_t SPI_SHADER_PACE_ID_PS            = 0x6;
-constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_PS    = 0x7;
-constexpr uint32_t SPI_SHADER_PGM_LO_PS              = 0x8;
-constexpr uint32_t SPI_SHADER_PGM_HI_PS              = 0x9;
+constexpr uint32_t SPI_SHADER_TBA_LO_PS     = 0x0;
+constexpr uint32_t SPI_SHADER_PGM_RSRC4_PS  = 0x1;
+constexpr uint32_t SPI_SHADER_TMA_LO_PS     = 0x2;
+constexpr uint32_t SPI_SHADER_TMA_HI_PS     = 0x3;
+constexpr uint32_t SPI_SHADER_PGM_CHKSUM_PS = 0x6;
+constexpr uint32_t SPI_SHADER_PGM_RSRC3_PS  = 0x7;
+constexpr uint32_t SPI_SHADER_PGM_LO_PS     = 0x8;
+constexpr uint32_t SPI_SHADER_PGM_HI_PS     = 0x9;

 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS                        = 0xA;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_VGPRS_SHIFT            = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_VGPRS_MASK             = 0x3F;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_SGPRS_SHIFT            = 6;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_SGPRS_MASK             = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_PRIORITY_SHIFT         = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_PRIORITY_MASK          = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_PS_FLOAT_MODE_SHIFT       = 12;
@@ -722,30 +778,33 @@ 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_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_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_USER_DATA_PS_0  = 0xC;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_15 = 0x1B;
 constexpr uint32_t SPI_SHADER_USER_DATA_PS_31 = 0x2B;
+constexpr uint32_t SPI_SHADER_REQ_CTRL_PS     = 0x30;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_PS_0 = 0x32;
 constexpr uint32_t SPI_SHADER_USER_ACCUM_PS_3 = 0x35;

 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_VGPRS_SHIFT           = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_VGPRS_MASK            = 0x3F;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_SGPRS_SHIFT           = 6;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_SGPRS_MASK            = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_PRIORITY_SHIFT        = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_PRIORITY_MASK         = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_VS_FLOAT_MODE_SHIFT      = 12;
@@ -776,20 +835,19 @@ constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_USER_SGPR_MSB_MASK    = 0x1;
 constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_SHARED_VGPR_CNT_SHIFT = 28;
 constexpr uint32_t SPI_SHADER_PGM_RSRC2_VS_SHARED_VGPR_CNT_MASK  = 0xF;

-constexpr uint32_t SPI_SHADER_PACE_ID_GS           = 0x80;
-// AGC compiler records use this numeric tag as opaque metadata. It is not a shader checksum
-// register and intentionally has no checksum-state consumer in HardwareContext.
-constexpr uint32_t SPI_SHADER_PGM_CHKSUM_GS        = SPI_SHADER_PACE_ID_GS;
+constexpr uint32_t SPI_SHADER_PGM_CHKSUM_GS        = 0x80;
 constexpr uint32_t SPI_SHADER_PGM_RSRC4_GS         = 0x81;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_LO_GS = 0x82;
 constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_HI_GS = 0x83;
-constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_GS  = 0x87;
+constexpr uint32_t SPI_SHADER_PGM_RSRC3_GS         = 0x87;
 constexpr uint32_t SPI_SHADER_PGM_LO_GS            = 0x88;
 constexpr uint32_t SPI_SHADER_PGM_HI_GS            = 0x89;

 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS                        = 0x8A;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_VGPRS_SHIFT            = 0;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_VGPRS_MASK             = 0x3F;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_SGPRS_SHIFT            = 6;
+constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_SGPRS_MASK             = 0xF;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_PRIORITY_SHIFT         = 10;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_PRIORITY_MASK          = 0x3;
 constexpr uint32_t SPI_SHADER_PGM_RSRC1_GS_FLOAT_MODE_SHIFT       = 12;
@@ -827,98 +885,61 @@ 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_USER_ACCUM_ESGS_0                   = 0xB2;
-constexpr uint32_t SPI_SHADER_USER_ACCUM_ESGS_3                   = 0xB5;
-constexpr uint32_t SPI_SHADER_PGM_LO_ES                           = 0xC8;
-constexpr uint32_t SPI_SHADER_PGM_HI_ES                           = 0xC9;
-// Private AGC registers used by shader binaries submitted through SET_SH_REG_INDIRECT.
-constexpr uint32_t SPI_SHADER_PGM_CHKSUM_HS                       = 0x100;
-constexpr uint32_t SPI_SHADER_PGM_RSRC4_HS                        = 0x101;
-constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_LO_HS                = 0x102;
-constexpr uint32_t SPI_SHADER_USER_DATA_ADDR_HI_HS                = 0x103;
-constexpr uint32_t SPI_GRAPHICS_SHADER_CONTROL_HS                 = 0x107;
-constexpr uint32_t SPI_SHADER_PGM_LO_HS                           = 0x108;
-constexpr uint32_t SPI_SHADER_PGM_HI_HS                           = 0x109;
-constexpr uint32_t SPI_SHADER_PGM_RSRC1_HS                        = 0x10A;
-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_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 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 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 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_PRIORITY_SHIFT   = 10;
-constexpr uint32_t COMPUTE_PGM_RSRC1_PRIORITY_MASK    = 0x3;
-constexpr uint32_t COMPUTE_PGM_RSRC1_FLOAT_MODE_SHIFT = 12;
-constexpr uint32_t COMPUTE_PGM_RSRC1_FLOAT_MODE_MASK  = 0xFF;
-constexpr uint32_t COMPUTE_PGM_RSRC1_DX10_CLAMP_SHIFT = 21;
-constexpr uint32_t COMPUTE_PGM_RSRC1_DX10_CLAMP_MASK  = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_DEBUG_MODE_SHIFT = 22;
-constexpr uint32_t COMPUTE_PGM_RSRC1_DEBUG_MODE_MASK  = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_IEEE_MODE_SHIFT  = 23;
-constexpr uint32_t COMPUTE_PGM_RSRC1_IEEE_MODE_MASK   = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_FP16_OVFL_SHIFT  = 26;
-constexpr uint32_t COMPUTE_PGM_RSRC1_FP16_OVFL_MASK   = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_WGP_MODE_SHIFT   = 29;
-constexpr uint32_t COMPUTE_PGM_RSRC1_WGP_MODE_MASK    = 0x1;
-constexpr uint32_t COMPUTE_PGM_RSRC1_FWD_PROGRESS_SHIFT = 31;
-constexpr uint32_t COMPUTE_PGM_RSRC1_FWD_PROGRESS_MASK  = 0x1;
-
-constexpr uint32_t COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_SHIFT = 15;
-constexpr uint32_t COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_MASK  = 0x1;
-
-[[nodiscard]] constexpr uint8_t ComputeWaveSize(uint32_t dispatch_initiator) {
-	return ((dispatch_initiator >> COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_SHIFT) &
-	        COMPUTE_DISPATCH_INITIATOR_CS_W32_EN_MASK) != 0u
-	           ? 32u
-	           : 64u;
-}
+constexpr uint32_t COMPUTE_PGM_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_RSRC2                      = 0x213;
 constexpr uint32_t COMPUTE_PGM_RSRC2_SCRATCH_EN_SHIFT     = 0;
@@ -939,16 +960,21 @@ constexpr uint32_t COMPUTE_PGM_RSRC2_LDS_SIZE_SHIFT       = 15;
 constexpr uint32_t COMPUTE_PGM_RSRC2_LDS_SIZE_MASK        = 0x1FF;

 constexpr uint32_t COMPUTE_RESOURCE_LIMITS    = 0x215;
+constexpr uint32_t COMPUTE_DESTINATION_EN_SE0 = 0x216;
+constexpr uint32_t COMPUTE_DESTINATION_EN_SE1 = 0x217;
 constexpr uint32_t COMPUTE_TMPRING_SIZE       = 0x218;
+constexpr uint32_t COMPUTE_DESTINATION_EN_SE2 = 0x219;
+constexpr uint32_t COMPUTE_DESTINATION_EN_SE3 = 0x21A;
 constexpr uint32_t COMPUTE_USER_ACCUM_0       = 0x224;
 constexpr uint32_t COMPUTE_PGM_RSRC3          = 0x228;
-constexpr uint32_t COMPUTE_PGM_RSRC3_SHARED_VGPRS_SHIFT = 0;
-constexpr uint32_t COMPUTE_PGM_RSRC3_SHARED_VGPRS_MASK  = 0xF;
-constexpr uint32_t COMPUTE_PACE_ID             = 0x22A;
+constexpr uint32_t COMPUTE_SHADER_CHKSUM      = 0x22A;
 constexpr uint32_t COMPUTE_USER_DATA_0        = 0x240;
 constexpr uint32_t COMPUTE_USER_DATA_15       = 0x24F;
+constexpr uint32_t COMPUTE_DISPATCH_TUNNEL    = 0x27D;

-constexpr uint32_t SH_NOP = 0x280;
+/* Fake codes. Don't exist on real HW */
+
+constexpr uint32_t SH_NOP = 0x800002FF;

 constexpr uint32_t SH_NUM = 0x2FF + 1;

@@ -965,7 +991,6 @@ constexpr uint32_t GE_INDX_OFFSET            = 0x24A;
 constexpr uint32_t GE_MULTI_PRIM_IB_RESET_EN = 0x24B;
 constexpr uint32_t VGT_HS_OFFCHIP_PARAM      = 0x24F;
 constexpr uint32_t VGT_TF_MEMORY_BASE        = 0x250;
-constexpr uint32_t IA_MULTI_VGT_PARAM        = 0x258;

 constexpr uint32_t GE_CNTL                     = 0x25B;
 constexpr uint32_t GE_CNTL_PRIM_GRP_SIZE_SHIFT = 0;
@@ -977,7 +1002,6 @@ 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 UC_PARAMETER_OVERSUBSCRIPTION = 0x260;

 constexpr uint32_t GE_USER_VGPR_EN                     = 0x262;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR1_SHIFT = 0;
@@ -987,25 +1011,32 @@ constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR2_MASK  = 0x1;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR3_SHIFT = 2;
 constexpr uint32_t GE_USER_VGPR_EN_EN_USER_VGPR3_MASK  = 0x1;

-constexpr uint32_t FSR_WINDOW_LEFT           = 0x2BC;
-constexpr uint32_t FSR_CONTROL_POINTS_LEFT_X = 0x2C4;
-constexpr uint32_t FSR_CONTROL_POINTS_LEFT_Y = 0x2CC;
-constexpr uint32_t FSR_ALPHA_LEFT_X          = 0x2E4;
-constexpr uint32_t FSR_ALPHA_LEFT_Y          = 0x2E8;
-
+constexpr uint32_t IA_MULTI_VGT_PARAM       = 0x258;
 constexpr uint32_t TA_CS_BC_BASE_ADDR       = 0x380;
 constexpr uint32_t TA_CS_BC_BASE_ADDR_HI    = 0x381;
-constexpr uint32_t TEXTURE_GRADIENT_FACTORS = 0x383;
-constexpr uint32_t TEXTURE_GRADIENT_CONTROL = 0x384;
+constexpr uint32_t TEXTURE_GRADIENT_FACTORS = 0x382;
 constexpr uint32_t GDS_OA_CNTL              = 0x41D;
 constexpr uint32_t GDS_OA_COUNTER           = 0x41E;
 constexpr uint32_t GDS_OA_ADDRESS           = 0x41F;

-constexpr uint32_t UC_NOP = 0xA2;
+/* Fake codes. Don't exist on real HW */
+
+constexpr uint32_t FSR_EXTEND_SUBPIXEL_ROUNDING = 0x80003FF4;
+constexpr uint32_t FSR_ALPHA_VALUE0             = 0x80003FF5;
+constexpr uint32_t FSR_ALPHA_VALUE1             = 0x80003FF6;
+constexpr uint32_t FSR_CONTROL_POINT0           = 0x80003FF7;
+constexpr uint32_t FSR_CONTROL_POINT1           = 0x80003FF8;
+constexpr uint32_t FSR_CONTROL_POINT2           = 0x80003FF9;
+constexpr uint32_t FSR_CONTROL_POINT3           = 0x80003FFA;
+constexpr uint32_t FSR_WINDOW0                  = 0x80003FFB;
+constexpr uint32_t FSR_WINDOW1                  = 0x80003FFC;
+constexpr uint32_t TEXTURE_GRADIENT_CONTROL     = 0x80003FFD;
+constexpr uint32_t MEMORY_MAPPING_MASK          = 0x80003FFE;
+constexpr uint32_t UC_NOP                       = 0x80003FFF;

 constexpr uint32_t UC_NUM = 0x3FFF + 1;

-void DumpPm4PacketStream(Common::File* file, const uint32_t* cmd_buffer, uint32_t start_dw,
+void DumpPm4PacketStream(Common::File* file, uint32_t* cmd_buffer, uint32_t start_dw,
                         uint32_t num_dw);

 } // namespace Libs::Graphics::Pm4
@@ -3,7 +3,6 @@

 #include "common/abi.h"
 #include "common/common.h"
-#include "graphics/guest_gpu/gpu_defs.h"

 namespace Libs::Graphics {

@@ -40,7 +39,7 @@ enum class TileBlockFamily : uint32_t {
 };

 struct TileBlockLayout {
-	TileBlockFamily family            = TileBlockFamily::Count;
+	TileBlockFamily family            = TileBlockFamily::Standard256B;
 	uint32_t        bytes_per_element = 0;
 	uint32_t        block_size        = 0;
 	uint32_t        block_width       = 0;
@@ -48,74 +47,42 @@ struct TileBlockLayout {
 	uint32_t        block_depth       = 0;
 };

-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;
-	bool     operator==(const TileMipLayout&) const = default;
-};
-
-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]         = {};
+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] = {};
 };

 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     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);
+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);
 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,
@@ -123,21 +90,17 @@ uint32_t TileGetDepthPitch(uint32_t width, uint32_t bytes_per_element,
 bool     TileGetRenderTargetSize(uint32_t width, uint32_t height, uint32_t pitch,
                                 uint32_t bytes_per_element, TileSizeAlign& total_size,
                                 uint32_t num_fragments_log2 = 0);
-bool     TileGetDccSize(uint32_t width, uint32_t height, uint32_t slices,
-                        uint32_t bytes_per_element, uint32_t levels, Prospero::TileMode tile,
-                        TileSizeAlign& total_size, uint32_t num_fragments_log2 = 0);
 bool     TileGetRenderTargetMipLayout(uint32_t width, uint32_t height, uint32_t pitch,
                                      uint32_t bytes_per_element, uint32_t levels,
                                      TileSizeAlign& total_size, TileSizeOffset* level_sizes,
                                      TilePaddedSize* padded_size);
-void     TileGetTextureSize(Prospero::BufferFormat format, uint32_t width, uint32_t height,
-                            uint32_t levels, Prospero::TileMode tile, TileSizeAlign* total_size,
+void     TileGetTextureSize(uint32_t format, uint32_t width, uint32_t height, uint32_t pitch,
+                            uint32_t levels, uint32_t tile, TileSizeAlign* total_size,
                            TileSizeOffset* level_sizes, TilePaddedSize* padded_size);
-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);
+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);

 } // namespace Libs::Graphics

@@ -3,113 +3,51 @@

 #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;

-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                               compute_subgroup_size_control_enabled = false;
-	bool                               sample_rate_shading_enabled           = false;
-	bool                               attachment_feedback_loop_enabled      = false;
-	bool                               provoking_vertex_last_enabled         = false;
-	bool                               supports_block_texel_view              = false;
-	bool                                      mesh_shader_enabled                   = false;
-	vk::PhysicalDeviceMeshShaderPropertiesEXT mesh_shader_properties                = {};
-	uint32_t                           subgroup_size                         = 0;
-	uint32_t                           min_subgroup_size                     = 0;
-	uint32_t                           max_subgroup_size                     = 0;
-	uint32_t                           max_push_descriptors                  = 0;
-	vk::ShaderStageFlags               required_subgroup_size_stages         = {};
-	Common::Mutex                      queue_mutex;
-	uint32_t                           queue_family = static_cast<uint32_t>(-1);
-	vk::Queue                          queue        = nullptr;
-
-	[[nodiscard]] const vk::PhysicalDeviceProperties& GetPhysicalDeviceProperties() const {
-		return physical_device_properties;
-	}
-
-	[[nodiscard]] const vk::PhysicalDeviceMemoryProperties&
-	GetPhysicalDeviceMemoryProperties() const {
-		return physical_device_memory_properties;
-	}
-
-	[[nodiscard]] vk::FormatProperties GetFormatProperties(vk::Format format) const {
-		std::scoped_lock lock(m_format_properties_mutex);
-		auto [it, inserted] = m_format_properties.try_emplace(format);
-		if (inserted) {
-			physical_device.getFormatProperties(format, &it->second);
-		}
-		return it->second;
-	}
-
-	[[nodiscard]] vk::Result GetImageFormatProperties(vk::Format format, vk::ImageType type,
-	                                                  vk::ImageTiling            tiling,
-	                                                  vk::ImageUsageFlags        usage,
-	                                                  vk::ImageCreateFlags       flags,
-	                                                  vk::ImageFormatProperties* properties) const {
-		using Key = std::tuple<vk::Format, vk::ImageType, vk::ImageTiling, vk::ImageUsageFlags,
-		                       vk::ImageCreateFlags>;
-		std::scoped_lock lock(m_image_format_properties_mutex);
-		auto [it, inserted] =
-		    m_image_format_properties.try_emplace(Key {format, type, tiling, usage, flags});
-		if (inserted) {
-			it->second.first = physical_device.getImageFormatProperties(format, type, tiling, usage,
-			                                                            flags, &it->second.second);
-		}
-		if (properties != nullptr) {
-			*properties = it->second.second;
-		}
-		return it->second.first;
-	}
-
-	[[nodiscard]] bool SupportsComputeWave64() const noexcept {
-		return subgroup_size == 64u || compute_subgroup_size_control_enabled;
-	}
-
-	[[nodiscard]] vk::DeviceSize StorageMinAlignment() const {
-		const auto alignment = physical_device_properties.limits.minStorageBufferOffsetAlignment;
-		return alignment != 0 ? alignment : 1;
-	}
-
+struct GraphicContext: public VulkanInstance {
 	[[nodiscard]] bool CreateAllocator();
 	void               DestroyAllocator();
 	void               LogMemoryBudget() const;
 	[[nodiscard]] bool CanReportMemoryUsage() const noexcept { return memory_budget_ext_enabled; }
 	[[nodiscard]] uint64_t GetDeviceMemoryUsage() const;
 	[[nodiscard]] uint64_t GetTotalMemoryBudget() const;
+	void                   CreateBuffer(uint64_t size, VulkanBuffer& buffer);
+	void                   DeleteBuffer(VulkanBuffer& buffer);
 	[[nodiscard]] bool     CreateImage(const vk::ImageCreateInfo& info, VulkanImage& image);
 	void                   DeleteImage(VulkanImage& image);
+	void                   MapMemory(VulkanMemory& memory, void*& data);
+	void                   UnmapMemory(VulkanMemory& memory);
+	void                   AppendHardwareRayTracingDeviceExtensions(
+	    const std::vector<vk::ExtensionProperties>& available_extensions,
+	    std::vector<const char*>&                   device_extensions);
+	void LoadHardwareRayTracingFunctions() const;

 	uint32_t screen_width  = 0;
 	uint32_t screen_height = 0;
+};

-private:
-	mutable std::mutex                                 m_format_properties_mutex;
-	mutable std::map<vk::Format, vk::FormatProperties> m_format_properties;
-	mutable std::mutex                                 m_image_format_properties_mutex;
-	mutable std::map<std::tuple<vk::Format, vk::ImageType, vk::ImageTiling, vk::ImageUsageFlags,
-	                            vk::ImageCreateFlags>,
-	                 std::pair<vk::Result, vk::ImageFormatProperties>>
-	    m_image_format_properties;
+struct VulkanMemory {
+	vk::MemoryRequirements  requirements       = {};
+	vk::MemoryPropertyFlags property           = {};
+	vk::MemoryPropertyFlags preferred_property = {};
+	vk::DeviceMemory        memory             = nullptr;
+	VmaAllocation           allocation         = nullptr;
+	VmaAllocationInfo       allocation_info    = {};
+	vk::DeviceSize          offset             = 0;
+	uint32_t                type               = 0;
+	uint64_t                unique_id          = 0;
 };

 struct VulkanImageState {
@@ -125,6 +63,7 @@ struct VulkanImage {
 	vk::Format                    format      = vk::Format::eUndefined;
 	vk::ImageType                 image_type  = vk::ImageType::e2D;
 	vk::Extent3D                  extent      = {1, 1, 1};
+	uint32_t                      guest_pitch = 0;
 	uint32_t                      layers      = 1;
 	uint32_t                      mip_levels  = 1;
 	uint32_t                      samples     = 1;
@@ -133,10 +72,15 @@ struct VulkanImage {
 	vk::Image                     image       = nullptr;
 	VulkanImageState              state;
 	std::vector<VulkanImageState> subresource_states;
-	VmaAllocation                allocation = nullptr;
+	Graphics::VulkanMemory        memory;
 };

-
+struct VulkanBuffer {
+	vk::Buffer           buffer = nullptr;
+	VulkanMemory         memory;
+	vk::BufferUsageFlags usage       = {};
+	uint64_t             buffer_size = 0;
+};

 } // namespace Libs::Graphics

@@ -1,6 +1,5 @@
 #include "graphics/host_gpu/memoryTracker.h"

-#include "common/alignment.h"
 #include "common/assert.h"

 namespace Libs::Graphics {
@@ -9,6 +8,9 @@ static_assert(std::atomic<void*>::is_always_lock_free);

 MemoryTracker::MemoryTracker(PageManager& page_manager): m_page_manager(page_manager) {
 	m_regions = std::make_unique<std::atomic<RegionManager*>[]>(REGION_COUNT);
+	for (size_t i = 0; i < REGION_COUNT; i++) {
+		m_regions[i].store(nullptr, std::memory_order_relaxed);
+	}
 }

 MemoryTracker::~MemoryTracker() = default;
@@ -16,12 +18,15 @@ MemoryTracker::~MemoryTracker() = default;
 #if KYTY_BUILD == KYTY_BUILD_DEBUG
 void MemoryTracker::ValidateGpuDirtyPages(const RangeSet& dirty, uint64_t vaddr, uint64_t size,
                                          const char* operation) const noexcept {
-	if (!GuestRange {vaddr, size}.Valid() || (vaddr & (TRACKER_PAGE_SIZE - 1)) != 0 ||
-	    (size & (TRACKER_PAGE_SIZE - 1)) != 0) {
+	if (vaddr == 0 || size == 0 || size > UINT64_MAX - vaddr ||
+	    (vaddr & (TRACKER_PAGE_SIZE - 1)) != 0 || (size & (TRACKER_PAGE_SIZE - 1)) != 0) {
 		EXIT("MemoryTracker: invalid dirty-page validation range\n");
 	}
 	for (auto page = vaddr; page < vaddr + size; page += TRACKER_PAGE_SIZE) {
-		if (!dirty.Intersects(page, TRACKER_PAGE_SIZE)) {
+		bool found = false;
+		dirty.ForEachIntersection(page, TRACKER_PAGE_SIZE,
+		                          [&found](RangeSet::Range) { found = true; });
+		if (!found) {
 			EXIT("MemoryTracker: GPU-dirty tracker page has no dirty bytes, operation=%s "
 			     "addr=0x%016" PRIx64 "\n",
 			     operation, page);
@@ -32,10 +37,15 @@ void MemoryTracker::ValidateGpuDirtyPages(const RangeSet& dirty, uint64_t vaddr,
 void MemoryTracker::ValidateGpuDirtyOwnership(const RangeSet& dirty, uint64_t vaddr, uint64_t size,
                                              const char* operation) {
 	ValidateRange(vaddr, size);
-	const auto begin = Common::AlignDown(vaddr, TRACKER_PAGE_SIZE);
-	const auto end   = Common::AlignUp(vaddr + size, TRACKER_PAGE_SIZE);
+	if (vaddr + size > UINT64_MAX - (TRACKER_PAGE_SIZE - 1)) {
+		EXIT("MemoryTracker: dirty ownership range alignment overflow\n");
+	}
+	const auto begin = vaddr & ~(TRACKER_PAGE_SIZE - 1);
+	const auto end   = (vaddr + size + TRACKER_PAGE_SIZE - 1) & ~(TRACKER_PAGE_SIZE - 1);
 	for (auto page = begin; page < end; page += TRACKER_PAGE_SIZE) {
-		const bool has_dirty_bytes = dirty.Intersects(page, TRACKER_PAGE_SIZE);
+		bool has_dirty_bytes = false;
+		dirty.ForEachIntersection(page, TRACKER_PAGE_SIZE,
+		                          [&has_dirty_bytes](RangeSet::Range) { has_dirty_bytes = true; });
 		if (IsRegionGpuModified(page, TRACKER_PAGE_SIZE) != has_dirty_bytes) {
 			EXIT("MemoryTracker: tracker and byte ownership disagree, operation=%s "
 			     "addr=0x%016" PRIx64 "\n",
@@ -46,7 +56,8 @@ void MemoryTracker::ValidateGpuDirtyOwnership(const RangeSet& dirty, uint64_t va
 #endif

 void MemoryTracker::ValidateRange(uint64_t vaddr, uint64_t size) {
-	if (!GuestRange {vaddr, size}.Valid()) {
+	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
+	    size > TRACKER_ADDRESS_SIZE - vaddr) {
 		EXIT("invalid memory tracker range\n");
 	}
 }
@@ -68,6 +79,7 @@ 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);
@@ -76,6 +88,7 @@ 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);
@@ -84,6 +97,7 @@ 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);
@@ -92,6 +106,7 @@ 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);
@@ -100,14 +115,14 @@ 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::UntrackMemory(uint64_t vaddr, uint64_t size) {
-	CheckNotInUploadCallback();
+void MemoryTracker::UntrackMemoryLocked(uint64_t vaddr, uint64_t size) {
 	std::vector<RegionManager*> managers;
 	managers.reserve((vaddr % TRACKER_REGION_SIZE + size + TRACKER_REGION_SIZE - 1) /
 	                 TRACKER_REGION_SIZE);
@@ -128,6 +143,13 @@ void MemoryTracker::UntrackMemory(uint64_t vaddr, uint64_t size) {
 	Iterate<false>(vaddr, size, [](RegionManager* manager, uint64_t offset, uint64_t bytes) {
 		manager->ChangeState<DirtySource::Cpu, true>(manager->GetCpuAddr() + offset, bytes);
 	});
+	locks.clear();
+}
+
+void MemoryTracker::UntrackMemory(uint64_t vaddr, uint64_t size) {
+	CheckNotInUploadCallback();
+	std::lock_guard access(m_access_mutex);
+	UntrackMemoryLocked(vaddr, size);
 }

 } // namespace Libs::Graphics
@@ -29,28 +29,45 @@ 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) noexcept {
+	void InvalidateRegion(uint64_t vaddr, uint64_t size, Flush&& on_flush) {
 		static_assert(std::is_invocable_v<Flush&>);
 		CheckNotInUploadCallback();
+		ValidateRange(vaddr, size);

-		Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t offset, uint64_t bytes) {
-			const bool should_flush = [&] {
-				// 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 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);
 			}
-		});
+			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,
@@ -62,18 +79,44 @@ public:
 	void ValidateGpuDirtyOwnership(const RangeSet&, uint64_t, uint64_t, const char*) {}
 #endif

-	template <bool clear, typename Func>
-	void ForEachDownloadRange(uint64_t vaddr, uint64_t size, Func&& func) {
+	template <bool clear, typename Preflight, typename Func>
+	void ForEachDownloadRange(uint64_t vaddr, uint64_t size, Preflight&& preflight, Func&& func) {
+		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();
-		Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t offset, uint64_t bytes) {
-			std::scoped_lock lock(manager->lock);
-			const auto       address = manager->GetCpuAddr() + offset;
-			manager->template ForEachModifiedRange<DirtySource::Gpu, false>(address, bytes, func);
-			if constexpr (clear) {
-				manager->template ChangeState<DirtySource::Gpu, false>(address, bytes);
-			}
+		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 auto address = manager->GetCpuAddr() + offset;
+			manager->template ForEachModifiedRange<DirtySource::Gpu, false>(address, bytes,
+			                                                                preflight);
+		});
+		Iterate<false>(vaddr, size, [&](RegionManager* manager, uint64_t offset, uint64_t bytes) {
+			manager->template ForEachModifiedRange<DirtySource::Gpu, false>(
+			    manager->GetCpuAddr() + offset, bytes, func);
+		});
+		if constexpr (clear) {
+			Iterate<false>(vaddr, size,
+			               [&](RegionManager* manager, uint64_t offset, uint64_t bytes) {
+				               const auto address = manager->GetCpuAddr() + offset;
+				               manager->template ForEachModifiedRange<DirtySource::Gpu, true>(
+				                   address, bytes, [](uint64_t, uint64_t) noexcept {});
+			               });
+		}
+	}
+
+	template <bool clear, typename Func>
+	void ForEachDownloadRange(uint64_t vaddr, uint64_t size, Func&& func) {
+		ForEachDownloadRange<clear>(
+		    vaddr, size, [](uint64_t, uint64_t) noexcept {}, std::forward<Func>(func));
 	}

 	template <typename RangeFunc, typename UploadFunc>
@@ -82,6 +125,7 @@ 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) {
@@ -145,11 +189,13 @@ private:
 	}

 	static void    ValidateRange(uint64_t vaddr, uint64_t size);
+	void           UntrackMemoryLocked(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;
 };

@@ -1,7 +1,5 @@
 #include "graphics/host_gpu/pageManager.h"

-#include "common/alignment.h"
-#include "common/virtualMemory.h"
 #include "graphics/host_gpu/regionDefinitions.h"
 #include "kernel/memory.h"

@@ -22,7 +20,10 @@
 #include <windows.h>
 #undef min
 #undef max
+#elif defined(__APPLE__)
+#include <unistd.h>
 #else
+#include <execinfo.h>
 #include <unistd.h>
 #endif

@@ -34,12 +35,38 @@ constexpr uint64_t REGION_SIZE  = TRACKER_REGION_SIZE;
 constexpr uint64_t ADDRESS_SIZE = TRACKER_ADDRESS_SIZE;
 constexpr uint64_t REGION_COUNT = ADDRESS_SIZE / REGION_SIZE;

+#if KYTY_PLATFORM != KYTY_PLATFORM_WINDOWS
+// The tracker reuses Win32 memory-protection tags as internal page-state values.
+// Mirror their canonical numeric values so the shared state-machine logic is identical.
+constexpr uint32_t PAGE_NOACCESS  = 0x01;
+constexpr uint32_t PAGE_READONLY  = 0x02;
+constexpr uint32_t PAGE_READWRITE = 0x04;
+#endif
 constexpr uint64_t REGION_PAGES = REGION_SIZE / PAGE_SIZE;

+constexpr uint32_t NO_ACCESS_PROTECTION  = PAGE_NOACCESS;
+constexpr uint32_t READ_ONLY_PROTECTION  = PAGE_READONLY;
+constexpr uint32_t READ_WRITE_PROTECTION = PAGE_READWRITE;
+
 [[noreturn]] void FailFast(const char* reason = nullptr) noexcept {
 	std::fputs("PageManager fail-fast: ", stderr);
 	std::fputs(reason != nullptr ? reason : "invalid page state", stderr);
 	std::fputc('\n', stderr);
+#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++) {
+		const auto address = reinterpret_cast<uintptr_t>(frames[i]);
+		std::fprintf(stderr, "  frame[%u]=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)));
+	::backtrace_symbols_fd(frames, frame_count, STDERR_FILENO);
+#endif
 	std::fflush(stderr);
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
 	TerminateProcess(GetCurrentProcess(), static_cast<UINT>(EXCEPTION_NONCONTINUABLE_EXCEPTION));
@@ -58,6 +85,15 @@ constexpr uint64_t REGION_PAGES = REGION_SIZE / PAGE_SIZE;
 	std::_Exit(322);
 }

+Common::VirtualMemory::Mode ToMemoryMode(uint32_t protection) {
+	switch (protection) {
+		case NO_ACCESS_PROTECTION: return Common::VirtualMemory::Mode::NoAccess;
+		case READ_ONLY_PROTECTION: return Common::VirtualMemory::Mode::Read;
+		case READ_WRITE_PROTECTION: return Common::VirtualMemory::Mode::ReadWrite;
+		default: Fatal("unmappable protection 0x%08" PRIx32, protection);
+	}
+}
+
 class SpinGuard final {
 public:
 	explicit SpinGuard(std::atomic_flag& lock): m_lock(lock) {
@@ -73,11 +109,20 @@ private:
 };

 void ValidateRange(uint64_t vaddr, uint64_t size) {
-	if (!GuestRange {vaddr, size}.Valid()) {
+	if (vaddr == 0 || size == 0 || vaddr >= ADDRESS_SIZE || size > ADDRESS_SIZE - vaddr) {
 		Fatal("invalid range vaddr=0x%016" PRIx64 ", size=0x%016" PRIx64, vaddr, size);
 	}
 }

+uint64_t PageStart(uint64_t vaddr) {
+	return vaddr & ~(PAGE_SIZE - 1);
+}
+
+uint64_t PageEnd(uint64_t vaddr, uint64_t size) {
+	ValidateRange(vaddr, size);
+	return PageStart(vaddr + size - 1) + PAGE_SIZE;
+}
+
 } // namespace

 struct PageManager::Impl {
@@ -85,14 +130,14 @@ struct PageManager::Impl {
 		uint8_t write_watchers  : 7 = 0;
 		uint8_t access_watchers : 1 = 0;

-		[[nodiscard]] Common::VirtualMemory::Mode Perms() const noexcept {
+		[[nodiscard]] uint32_t Perms() const noexcept {
 			if (access_watchers != 0) {
-				return Common::VirtualMemory::Mode::NoAccess;
+				return NO_ACCESS_PROTECTION;
 			}
 			if (write_watchers != 0) {
-				return Common::VirtualMemory::Mode::Read;
+				return READ_ONLY_PROTECTION;
 			}
-			return Common::VirtualMemory::Mode::ReadWrite;
+			return READ_WRITE_PROTECTION;
 		}

 		template <int delta, bool is_read>
@@ -156,6 +201,9 @@ struct PageManager::Impl {
 		}
 #endif
 		regions = std::make_unique<std::atomic<Region*>[]>(REGION_COUNT);
+		for (uint64_t i = 0; i < REGION_COUNT; i++) {
+			regions[i].store(nullptr, std::memory_order_relaxed);
+		}
 	}

 	~Impl() {
@@ -190,10 +238,11 @@ struct PageManager::Impl {
 		return ptr;
 	}

-	void Protect(uint64_t vaddr, uint64_t size, Common::VirtualMemory::Mode mode) noexcept {
-		if (!Libs::LibKernel::Memory::ProtectGuestHostMemory(vaddr, size, mode)) {
-			Fatal("address-space protection failed at 0x%016" PRIx64 ", mode=0x%08" PRIx32, vaddr,
-			      static_cast<uint32_t>(mode));
+	void Protect(uint64_t vaddr, uint64_t size, uint32_t protection) noexcept {
+		if (!Libs::LibKernel::Memory::ProtectGuestHostMemory(vaddr, size,
+		                                                     ToMemoryMode(protection))) {
+			Fatal("address-space protection failed at 0x%016" PRIx64 ", new=0x%08" PRIx32, vaddr,
+			      protection);
 		}
 	}

@@ -250,12 +299,11 @@ struct PageManager::Impl {

 	template <bool track, bool is_read>
 	void UpdatePageWatchers(uint64_t vaddr, uint64_t size) {
-		ValidateRange(vaddr, size);
-		const auto begin = Common::AlignDown(vaddr, PAGE_SIZE);
-		const auto end   = Common::AlignUp(vaddr + size, PAGE_SIZE);
+		const auto begin = PageStart(vaddr);
+		const auto end   = PageEnd(vaddr, size);
 		for (auto chunk_begin = begin; chunk_begin < end;) {
-			const auto chunk_end = std::min(end, Common::AlignUp(chunk_begin + 1, REGION_SIZE));
-			const auto region_base = Common::AlignDown(chunk_begin, REGION_SIZE);
+			const auto chunk_end   = std::min(end, (chunk_begin / REGION_SIZE + 1) * REGION_SIZE);
+			const auto region_base = chunk_begin / REGION_SIZE * REGION_SIZE;
 			auto*      region = track ? GetOrCreateRegion(chunk_begin) : FindRegion(chunk_begin);
 			if (region == nullptr) {
 				Fatal("untracking unknown page 0x%016" PRIx64, chunk_begin);
@@ -322,4 +370,8 @@ template void PageManager::UpdatePageWatchersForRegion<true, false>(uint64_t, Re
 template void PageManager::UpdatePageWatchersForRegion<false, true>(uint64_t, RegionBits&);
 template void PageManager::UpdatePageWatchersForRegion<false, false>(uint64_t, RegionBits&);

+void PageManager::OnGpuMap(uint64_t, uint64_t) {}
+
+void PageManager::OnGpuUnmap(uint64_t, uint64_t) {}
+
 } // namespace Libs::Graphics
@@ -24,6 +24,8 @@ public:
 	void UpdatePageWatchers(uint64_t vaddr, uint64_t size);
 	template <bool track, bool is_read = false>
 	void UpdatePageWatchersForRegion(uint64_t base_addr, RegionBits& mask);
+	void OnGpuMap(uint64_t vaddr, uint64_t size);
+	void OnGpuUnmap(uint64_t vaddr, uint64_t size);

 private:
 	struct Impl;
@@ -6,11 +6,17 @@
 #include <algorithm>
 #include <cstdint>
 #include <map>
+#include <vector>

 namespace Libs::Graphics {

 class RangeSet final {
 public:
+	struct Range {
+		uint64_t address = 0;
+		uint64_t size    = 0;
+	};
+
 	void Add(uint64_t address, uint64_t size) {
 		const auto end = End(address, size);
 		auto       it  = m_ranges.lower_bound(address);
@@ -47,22 +53,10 @@ 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]] 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]] 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 Contains(uint64_t address, uint64_t size) const {
@@ -76,7 +70,7 @@ public:
 	}

 	template <typename Func>
-	void ForEachInRange(uint64_t address, uint64_t size, Func&& func) const {
+	void ForEachIntersection(uint64_t address, uint64_t size, Func&& func) const {
 		const auto end = End(address, size);
 		auto       it  = m_ranges.upper_bound(address);
 		if (it != m_ranges.begin()) {
@@ -86,7 +80,7 @@ public:
 			const auto begin = std::max(address, it->first);
 			const auto last  = std::min(end, it->second);
 			if (begin < last) {
-				func(begin, last);
+				func(Range {begin, last - begin});
 			}
 		}
 	}
@@ -95,7 +89,7 @@ public:

 private:
 	static uint64_t End(uint64_t address, uint64_t size) {
-		if (size == 0 || size > UINT64_MAX - address) {
+		if (address == 0 || size == 0 || size > UINT64_MAX - address) {
 			EXIT("invalid range-set address or size\n");
 		}
 		return address + size;
@@ -4,8 +4,6 @@
 #include "common/bitArray.h"
 #include "common/common.h"

-#include <compare>
-
 namespace Libs::Graphics {

 constexpr uint64_t TRACKER_PAGE_SIZE    = 4ull * 1024ull;
@@ -13,20 +11,6 @@ constexpr uint64_t TRACKER_REGION_SIZE  = 4ull * 1024ull * 1024ull;
 constexpr uint64_t TRACKER_ADDRESS_SIZE = 1ull << 40u;
 constexpr size_t   TRACKER_REGION_PAGES = TRACKER_REGION_SIZE / TRACKER_PAGE_SIZE;

-struct GuestRange {
-	uint64_t address = 0;
-	uint64_t size    = 0;
-
-	[[nodiscard]] constexpr bool Empty() const noexcept { return address == 0 && size == 0; }
-	[[nodiscard]] constexpr bool Valid() const noexcept {
-		return address != 0 && size != 0 && address < TRACKER_ADDRESS_SIZE &&
-		       size <= TRACKER_ADDRESS_SIZE - address;
-	}
-	[[nodiscard]] constexpr bool     ValidOrEmpty() const noexcept { return Empty() || Valid(); }
-	[[nodiscard]] constexpr uint64_t End() const noexcept { return address + size; }
-	auto                             operator<=>(const GuestRange&) const = default;
-};
-
 enum class DirtySource { Cpu, Gpu };
 using RegionBits = Common::BitArray<TRACKER_REGION_PAGES>;
 static_assert(sizeof(RegionBits) == TRACKER_REGION_PAGES / 8);
@@ -111,43 +111,56 @@ public:
 			bits.UnsetRange(start, end);
 		}
 		if constexpr (source == DirtySource::Cpu) {
-			UpdateProtection<!enable, false>();
+			UpdateCpuProtection<!enable>();
 		} else {
-			UpdateProtection<enable, true>();
+			UpdateGpuProtection<enable>();
 		}
 	}

 	template <DirtySource source, bool clear, typename Func>
 	void ForEachModifiedRange(uint64_t vaddr, uint64_t size, Func&& func) {
 		const auto [start, end] = GetPageRange(vaddr, size);
-		auto&      bits         = GetBits<source>();
-		RegionBits mask(bits, start, end);
+		RegionBits mask(GetBits<source>(), start, end);
 		if constexpr (clear) {
-			bits.UnsetRange(start, end);
-			if constexpr (source == DirtySource::Cpu) {
-				UpdateProtection<true, false>();
-			} else {
-				UpdateProtection<false, true>();
-			}
+			GetBits<source>().UnsetRange(start, end);
 		}
-		for (const auto [first, last]: mask) {
-			func(m_cpu_addr + first * TRACKER_PAGE_SIZE, (last - first) * TRACKER_PAGE_SIZE);
+		if constexpr (source == DirtySource::Cpu && clear) {
+			UpdateCpuProtection<true>();
+			ForEachRange(mask, std::forward<Func>(func));
+			return;
 		}
+		if constexpr (source == DirtySource::Gpu && clear) {
+			UpdateGpuProtection<false>();
+		}
+		ForEachRange(mask, std::forward<Func>(func));
 	}

 	TrackingSpinLock lock;

 private:
-	template <bool track, bool is_read>
-	void UpdateProtection() {
-		const auto protection = is_read ? ~m_gpu_dirty : m_cpu_dirty;
-		auto&      previous   = is_read ? m_readable : m_writable;
-		auto       mask       = protection ^ previous;
+	template <bool track>
+	void UpdateCpuProtection() {
+		auto mask  = m_cpu_dirty ^ m_writable;
+		m_writable = m_cpu_dirty;
 		if (mask.None()) {
 			return;
 		}
-		previous = protection;
-		m_page_manager.UpdatePageWatchersForRegion<track, is_read>(m_cpu_addr, mask);
+		m_page_manager.UpdatePageWatchersForRegion<track>(m_cpu_addr, mask);
+	}
+
+	template <bool track>
+	void UpdateGpuProtection() {
+		auto readable = ~m_gpu_dirty;
+		auto mask     = readable ^ m_readable;
+		m_readable    = readable;
+		if (mask.None()) {
+			return;
+		}
+		if constexpr (track) {
+			m_page_manager.UpdatePageWatchersForRegion<true, true>(m_cpu_addr, mask);
+		} else {
+			m_page_manager.UpdatePageWatchersForRegion<false, true>(m_cpu_addr, mask);
+		}
 	}

 	template <DirtySource source>
@@ -178,6 +191,13 @@ private:
 		        static_cast<size_t>((offset + size + TRACKER_PAGE_SIZE - 1) / TRACKER_PAGE_SIZE)};
 	}

+	template <typename Func>
+	void ForEachRange(const RegionBits& bits, Func&& func) const {
+		for (const auto [start, end]: bits) {
+			func(m_cpu_addr + start * TRACKER_PAGE_SIZE, (end - start) * TRACKER_PAGE_SIZE);
+		}
+	}
+
 	PageManager& m_page_manager;
 	uint64_t     m_cpu_addr = 0;
 	RegionBits   m_cpu_dirty;
@@ -3,15 +3,13 @@

 #include "common/abi.h"
 #include "common/common.h"
-#include "common/lruCache.h"
-#include "common/slotVector.h"
+#include "common/threads.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>
@@ -19,104 +17,93 @@
 namespace Libs::Graphics {

 struct GraphicContext;
+class CommandBuffer;
 class CommandScheduler;
 class TextureCache;
+class ResourceMutex;

-using BufferId = Common::SlotId;
-inline constexpr BufferId NULL_BUFFER_ID {0};
+struct BufferBinding {
+	std::shared_ptr<void> owner;
+	vk::Buffer            buffer = nullptr;
+	uint64_t              offset = 0;
+};
+
+struct ImageBufferSource {
+	Buffer*  buffer = nullptr;
+	uint64_t offset = 0;
+};

 class BufferCache {
 public:
-	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);
+	static constexpr uint64_t CACHING_PAGE_SIZE = 16ull * 1024ull;
+	static constexpr uint64_t GetBufferOffset(uint64_t vaddr) {
+		return vaddr & (CACHING_PAGE_SIZE - 1);
+	}

 	BufferCache(GraphicContext& graphics, CommandScheduler& scheduler, PageManager& page_manager,
-	            TextureCache& texture_cache);
+	            TextureCache& texture_cache, ResourceMutex& resource_mutex);
 	~BufferCache();
 	KYTY_CLASS_NO_COPY(BufferCache);

-	void                   InvalidateMemory(uint64_t vaddr, uint64_t size);
-	void                   ReadMemory(uint64_t vaddr, uint64_t size, bool is_write = false);
-	[[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);
+	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);
 	[[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;

-	bool IsBufferInvalid(BufferId id) const {
-		const auto* buffer = m_slot_buffers.try_get(id);
-		return buffer == nullptr || buffer->is_deleted;
-	}
-
-	using BufferMap = std::map<uint64_t, BufferId>;
-	struct OverlapResult {
-		BufferMap::iterator first;
-		BufferMap::iterator last;
-		uint64_t            begin;
-		uint64_t            end;
-		bool                has_stream_leap;
+	struct CacheRange {
+		uint64_t address = 0;
+		uint64_t size    = 0;
 	};
-
-	using PageTable = MultiLevelPageTable<BufferId, CACHING_PAGEBITS, 40, 16>;
-	static_assert(CACHING_PAGESIZE == (uint64_t {1} << PageTable::kPageBits));
-	void WriteDataBuffer(Buffer& buffer, uint64_t address, const void* source, uint64_t size);
-	void TouchBuffer(const Buffer& buffer);
-	[[nodiscard]] OverlapResult ResolveOverlaps(uint64_t vaddr, uint64_t size);
-	void JoinOverlap(BufferId new_id, BufferId overlap_id, bool accumulate_stream_score);
-	[[nodiscard]] BufferId CreateBuffer(uint64_t vaddr, uint64_t size);
-	void                   Register(BufferId id);
-	void Unregister(BufferId id);
-	template <bool insert>
-	void ChangeRegister(BufferId id);
-	void DeleteBuffer(BufferId id);
-	[[nodiscard]] bool SynchronizeBuffer(Buffer& buffer, uint64_t vaddr, uint64_t size,
-	                                     bool is_written, bool is_texel_buffer);
-	[[nodiscard]] vk::Buffer UploadCopies(Buffer& buffer, std::span<vk::BufferCopy> copies,
-	                                      uint64_t total_size);
+	struct CachedBuffer;
+	struct DownloadCopy;
+	struct DownloadRange;
+	struct RetiredBuffer;
+	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);
 	[[nodiscard]] bool SynchronizeBufferFromImage(Buffer& buffer, uint64_t vaddr, uint64_t size);
-	// Queues backing publication; callers wait before clearing dirty pages or reusing their data.
-	[[nodiscard]] bool DownloadBufferMemory(Buffer& buffer, uint64_t vaddr, uint64_t size);
+	[[nodiscard]] std::vector<DownloadRange> RecordDownloads(std::span<const DownloadCopy> copies);
+	void PublishDownloads(std::span<const DownloadRange> downloads);
+	void QueueGarbageDownload(std::span<const DownloadCopy> copies, RetiredBuffer retire);
+	void WriteHostMemory(uint64_t vaddr, std::span<const uint8_t> data);

 	GraphicContext&                                   m_graphics;
 	CommandScheduler&                                 m_scheduler;
-	FaultManager                                      m_fault_manager;
 	Buffer                                            m_gds_buffer;
-	Buffer                                            m_bda_pagetable_buffer;
-	Common::SlotVector<Buffer>                        m_slot_buffers;
-	Common::LeastRecentlyUsedCache<BufferId, uint64_t> m_lru_cache;
-	BufferMap                                         m_buffers;
-	PageTable                                         m_page_table;
+	Common::Mutex                                     m_mutex;
+	std::shared_ptr<Buffer>                           m_null_buffer;
+	std::map<uint64_t, std::unique_ptr<CachedBuffer>> m_buffers;
 	RangeSet                                          m_gpu_modified_ranges;
 	MemoryTracker                                     m_memory_tracker;
 	StreamBuffer                                      m_staging_buffer;
@@ -124,7 +111,8 @@ private:
 	StreamBuffer                                      m_download_buffer;
 	StreamBuffer                                      m_device_buffer;
 	TextureCache&                                     m_texture_cache;
-	uint64_t                                          m_total_used_memory  = 0;
+	ResourceMutex&                                    m_resource_mutex;
+	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;
@@ -1,151 +0,0 @@
-#include "graphics/host_gpu/renderer/cache/faultManager.h"
-
-#include "common/assert.h"
-#include "common/logging/log.h"
-#include "gpu_tiler_shaders/fault_buffer_process_spv.h"
-#include "graphics/host_gpu/graphicContext.h"
-#include "graphics/host_gpu/renderer/cache/bufferCache.h"
-#include "graphics/host_gpu/renderer/commandScheduler.h"
-#include "graphics/host_gpu/vulkanCommon.h"
-
-#include <bit>
-#include <cinttypes>
-#include <cstring>
-#include <limits>
-
-namespace Libs::Graphics {
-
-namespace {
-
-constexpr size_t MaxPageFaults    = 1024;
-constexpr size_t PageFaultAreaSize = MaxPageFaults * sizeof(uint64_t);
-
-} // namespace
-
-FaultManager::FaultManager(GraphicContext& graphics, CommandScheduler& scheduler,
-                           BufferCache& buffer_cache)
-    : m_graphics(graphics), m_scheduler(scheduler), m_buffer_cache(buffer_cache),
-      m_fault_buffer(graphics, scheduler, MemoryUsage::DeviceLocal, 0, AllFlags,
-                     BufferCache::CACHING_NUMPAGES / 8),
-      m_download_buffer(graphics, scheduler, MemoryUsage::Download, 0, AllFlags,
-                        MaxPendingFaults * PageFaultAreaSize) {
-	SetVulkanObjectNameF(m_graphics.device, m_fault_buffer.Handle(), "Fault Buffer");
-
-	const vk::DescriptorSetLayoutBinding bindings[] {
-	    {0, vk::DescriptorType::eStorageBuffer, 1, vk::ShaderStageFlagBits::eCompute, nullptr},
-	    {1, vk::DescriptorType::eStorageBuffer, 1, vk::ShaderStageFlagBits::eCompute, nullptr},
-	};
-	vk::DescriptorSetLayoutCreateInfo layout_info {};
-	layout_info.flags        = vk::DescriptorSetLayoutCreateFlagBits::ePushDescriptorKHR;
-	layout_info.bindingCount = std::size(bindings);
-	layout_info.pBindings    = bindings;
-	RequireVulkanSuccess(
-	    m_graphics.device.createDescriptorSetLayout(&layout_info, nullptr,
-	                                                &m_fault_process_desc_layout),
-	    "create fault-buffer descriptor layout");
-
-	const auto module = CompileSPV(FAULT_BUFFER_PROCESS_SPV, m_graphics.device);
-
-	vk::PipelineLayoutCreateInfo pipeline_layout_info {};
-	pipeline_layout_info.setLayoutCount = 1;
-	pipeline_layout_info.pSetLayouts    = &m_fault_process_desc_layout;
-	RequireVulkanSuccess(
-	    m_graphics.device.createPipelineLayout(&pipeline_layout_info, nullptr,
-	                                           &m_fault_process_pipeline_layout),
-	    "create fault-buffer pipeline layout");
-
-	vk::PipelineShaderStageCreateInfo stage {};
-	stage.stage  = vk::ShaderStageFlagBits::eCompute;
-	stage.module = module;
-	stage.pName  = "main";
-	vk::ComputePipelineCreateInfo pipeline_info {};
-	pipeline_info.stage  = stage;
-	pipeline_info.layout = m_fault_process_pipeline_layout;
-	const auto result = m_graphics.device.createComputePipelines(
-	    nullptr, 1, &pipeline_info, nullptr, &m_fault_process_pipeline);
-	m_graphics.device.destroyShaderModule(module, nullptr);
-	RequireVulkanSuccess(result, "create fault-buffer pipeline");
-	SetVulkanObjectNameF(m_graphics.device, m_fault_process_pipeline, "Fault Buffer Parser");
-}
-
-FaultManager::~FaultManager() {
-	m_graphics.device.destroyPipeline(m_fault_process_pipeline, nullptr);
-	m_graphics.device.destroyPipelineLayout(m_fault_process_pipeline_layout, nullptr);
-	m_graphics.device.destroyDescriptorSetLayout(m_fault_process_desc_layout, nullptr);
-}
-
-void FaultManager::ProcessFaultBuffer() {
-	if (const auto wait_tick = m_fault_areas[m_current_area]; wait_tick != 0) {
-		m_scheduler.Wait(wait_tick);
-		m_scheduler.PopPendingOperations();
-	}
-
-	const auto offset = m_current_area * PageFaultAreaSize;
-	auto*      mapped = m_download_buffer.Mapped().data() + offset;
-	std::memset(mapped, 0, PageFaultAreaSize);
-	m_download_buffer.Flush(offset, PageFaultAreaSize);
-
-	vk::BufferMemoryBarrier2 pre_barrier {};
-	pre_barrier.srcStageMask  = vk::PipelineStageFlagBits2::eAllCommands;
-	pre_barrier.srcAccessMask = vk::AccessFlagBits2::eShaderWrite;
-	pre_barrier.dstStageMask  = vk::PipelineStageFlagBits2::eComputeShader;
-	pre_barrier.dstAccessMask = vk::AccessFlagBits2::eShaderRead;
-	pre_barrier.buffer        = m_fault_buffer.Handle();
-	pre_barrier.offset        = 0;
-	pre_barrier.size           = m_fault_buffer.Size();
-	auto post_barrier         = pre_barrier;
-	post_barrier.srcStageMask  = vk::PipelineStageFlagBits2::eComputeShader;
-	post_barrier.srcAccessMask = vk::AccessFlagBits2::eShaderWrite;
-	post_barrier.dstStageMask  = vk::PipelineStageFlagBits2::eAllCommands;
-	post_barrier.dstAccessMask = vk::AccessFlagBits2::eShaderWrite;
-
-	const vk::DescriptorBufferInfo infos[] {
-	    {m_fault_buffer.Handle(), 0, m_fault_buffer.Size()},
-	    {m_download_buffer.Handle(), offset, PageFaultAreaSize},
-	};
-	std::array<vk::WriteDescriptorSet, 2> writes {};
-	for (uint32_t index = 0; index < writes.size(); ++index) {
-		writes[index].dstBinding      = index;
-		writes[index].descriptorCount = 1;
-		writes[index].descriptorType  = vk::DescriptorType::eStorageBuffer;
-		writes[index].pBufferInfo     = &infos[index];
-	}
-
-	m_scheduler.EndRendering();
-	auto command = m_scheduler.Current().Handle();
-	vk::DependencyInfo dependency {};
-	dependency.dependencyFlags          = vk::DependencyFlagBits::eByRegion;
-	dependency.bufferMemoryBarrierCount = 1;
-	dependency.pBufferMemoryBarriers    = &pre_barrier;
-	command.pipelineBarrier2(dependency);
-	command.bindPipeline(vk::PipelineBindPoint::eCompute, m_fault_process_pipeline);
-	command.pushDescriptorSetKHR(vk::PipelineBindPoint::eCompute,
-	                             m_fault_process_pipeline_layout, 0, writes);
-	const auto num_threads    = BufferCache::CACHING_NUMPAGES / 32;
-	const auto num_workgroups = (num_threads + 63) / 64;
-	command.dispatch(static_cast<uint32_t>(num_workgroups), 1, 1);
-	dependency.pBufferMemoryBarriers = &post_barrier;
-	command.pipelineBarrier2(dependency);
-
-	const auto area = m_current_area;
-	m_scheduler.DeferOperation([this, mapped, offset, area] {
-		m_download_buffer.Invalidate(offset, PageFaultAreaSize);
-		RangeSet    fault_ranges;
-		const auto* faults = std::bit_cast<const uint64_t*>(mapped);
-		const auto  count  = static_cast<uint32_t>(faults[0]);
-		for (uint32_t index = 1; index <= count; ++index) {
-			fault_ranges.Add(faults[index], BufferCache::CACHING_PAGESIZE);
-			LOGF("Accessed non-GPU cached memory at 0x%016" PRIx64 "\n", faults[index]);
-		}
-		fault_ranges.ForEach([this](uint64_t start, uint64_t end) {
-			EXIT_IF(end - start > std::numeric_limits<uint32_t>::max());
-			(void)m_buffer_cache.FindBuffer(start, end - start);
-		});
-		m_fault_areas[area] = 0;
-	});
-
-	m_fault_areas[m_current_area++] = m_scheduler.CurrentTick();
-	m_current_area %= MaxPendingFaults;
-}
-
-} // namespace Libs::Graphics
@@ -1,40 +0,0 @@
-#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_CACHE_FAULTMANAGER_H_
-#define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_CACHE_FAULTMANAGER_H_
-
-#include "common/abi.h"
-#include "graphics/host_gpu/renderer/cache/streamBuffer.h"
-
-#include <array>
-#include <cstdint>
-
-namespace Libs::Graphics {
-
-class BufferCache;
-
-class FaultManager {
-	static constexpr size_t MaxPendingFaults = 8;
-
-public:
-	FaultManager(GraphicContext& graphics, CommandScheduler& scheduler, BufferCache& buffer_cache);
-	~FaultManager();
-	KYTY_CLASS_NO_COPY(FaultManager);
-
-	[[nodiscard]] Buffer* GetFaultBuffer() noexcept { return &m_fault_buffer; }
-	void                  ProcessFaultBuffer();
-
-private:
-	GraphicContext&                            m_graphics;
-	CommandScheduler&                          m_scheduler;
-	BufferCache&                               m_buffer_cache;
-	Buffer                                     m_fault_buffer;
-	Buffer                                     m_download_buffer;
-	std::array<uint64_t, MaxPendingFaults>      m_fault_areas {};
-	uint32_t                                   m_current_area = 0;
-	vk::DescriptorSetLayout                    m_fault_process_desc_layout = nullptr;
-	vk::Pipeline                               m_fault_process_pipeline = nullptr;
-	vk::PipelineLayout                         m_fault_process_pipeline_layout = nullptr;
-};
-
-} // namespace Libs::Graphics
-
-#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_CACHE_FAULTMANAGER_H_
@@ -0,0 +1,140 @@
+#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) {}
+
+GpuResourceManager::~GpuResourceManager() = default;
+
+bool GpuResourceManager::HandleFault(PageFaultAccess access, uint64_t fault_vaddr) noexcept {
+	constexpr uint64_t fault_size = 8;
+	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));
+	}
+	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;
+}
+
+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);
+	return true;
+}
+
+bool GpuResourceManager::IsMapped(uint64_t vaddr, uint64_t size) const noexcept {
+	if (vaddr == 0 || size == 0 || vaddr >= TRACKER_ADDRESS_SIZE ||
+	    size > TRACKER_ADDRESS_SIZE - vaddr) {
+		return false;
+	}
+	std::shared_lock lock(m_mapped_ranges_mutex);
+	return m_mapped_ranges.Contains(vaddr, size);
+}
+
+void GpuResourceManager::MapMemory(uint64_t vaddr, uint64_t size) {
+	{
+		std::lock_guard lock(m_mapped_ranges_mutex);
+		m_mapped_ranges.Add(vaddr, size);
+	}
+	m_page_manager.OnGpuMap(vaddr, size);
+}
+
+void GpuResourceManager::UnmapMemory(uint64_t vaddr, uint64_t size) {
+	if (CommandScheduler::InDeferredOperation()) {
+		EXIT("unsupported memory unmap from an asynchronous GPU completion, "
+		     "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.WaitPriorityOperations(tick);
+		}
+		m_buffer_cache.UnmapMemory(vaddr, size);
+		m_texture_cache.UnmapMemory(vaddr, size);
+		m_page_manager.OnGpuUnmap(vaddr, size);
+		std::lock_guard lock(m_mapped_ranges_mutex);
+		m_mapped_ranges.Subtract(vaddr, size);
+	};
+	if (m_gpu == nullptr) {
+		unmap();
+		return;
+	}
+	m_gpu->SendCommandSync(unmap);
+}
+
+void GpuResourceManager::RunGarbageCollector() {
+	m_texture_cache.ProcessDownloadImages();
+	m_texture_cache.RunGarbageCollector();
+	m_buffer_cache.RunGarbageCollector();
+}
+
+} // namespace Libs::Graphics
@@ -0,0 +1,49 @@
+#ifndef EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_GPURESOURCEMANAGER_H_
+#define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_GPURESOURCEMANAGER_H_
+
+#include "common/abi.h"
+#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>
+#include <shared_mutex>
+
+namespace Libs::Graphics {
+
+class CommandScheduler;
+class Gpu;
+
+class GpuResourceManager {
+public:
+	GpuResourceManager(GraphicContext& graphics, CommandScheduler& scheduler);
+	~GpuResourceManager();
+	KYTY_CLASS_NO_COPY(GpuResourceManager);
+
+	[[nodiscard]] BufferCache&  GetBufferCache() { return m_buffer_cache; }
+	[[nodiscard]] TextureCache& GetTextureCache() { return m_texture_cache; }
+	void                        SetGpu(Gpu* 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               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;
+};
+
+} // namespace Libs::Graphics
+
+#endif // EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_GPURESOURCEMANAGER_H_
@@ -7,9 +7,8 @@
 #include <array>
 #include <cstddef>
 #include <cstdint>
-#include <cstring>
+#include <list>
 #include <memory>
-#include <type_traits>
 #include <utility>
 #include <vector>

@@ -104,156 +103,256 @@ private:
 	size_t                               m_allocated_buckets = 0;
 };

-// Inline owner storage for page-table entries. Texture pages normally have only a
-// handful of owners, so avoid a heap allocation on the first 16 registrations.
-//
-// Owners are packed into live uint64_t objects instead of being constructed in
-// raw storage. This keeps sparse bucket construction cheap without relying on
-// subtle implicit-lifetime or pointer-provenance rules.
-template <typename OwnerT, size_t InlineCapacity = 16>
-class InlinePageOwnerList final {
+// Removes only the requested owner. Other owners in the same page entry remain intact.
+template <typename Container, typename Value>
+[[nodiscard]] bool EraseExact(Container& owners, const Value& owner) {
+	const auto it = std::find(owners.begin(), owners.end(), owner);
+	if (it == owners.end()) {
+		return false;
+	}
+	owners.erase(it);
+	return true;
+}
+
+// Multi-range ownership with 1 MiB candidate buckets and precise 4 KiB
+// lifetime accounting. OwnerT only needs equality.
+template <typename OwnerT>
+class MultiRangePageOwnerIndex final {
+private:
+	struct Registration;
+	using MembershipList = std::vector<const Registration*>;
+
 public:
-	using value_type = OwnerT;
-	using size_type  = size_t;
-
-	class const_iterator final {
-	public:
-		[[nodiscard]] OwnerT operator*() const noexcept { return m_owners->At(m_index); }
-		const_iterator&      operator++() noexcept {
-			++m_index;
-			return *this;
-		}
-		bool operator==(const const_iterator&) const = default;
-
-	private:
-		friend class InlinePageOwnerList;
-		const_iterator(const InlinePageOwnerList* owners, size_type index) noexcept
-		    : m_owners(owners), m_index(index) {}
-
-		const InlinePageOwnerList* m_owners = nullptr;
-		size_type                  m_index  = 0;
+	struct ByteRange final {
+		uint64_t address = 0;
+		uint64_t size    = 0;
 	};

-	static_assert(InlineCapacity > 0);
-	static_assert(std::is_default_constructible_v<OwnerT>);
-	static_assert(std::is_trivially_copyable_v<OwnerT>);
-	static_assert(std::has_unique_object_representations_v<OwnerT>);
-	static_assert(sizeof(OwnerT) <= sizeof(uint64_t));
+	using CoarseTable   = MultiLevelPageTable<MembershipList, 20, 40, 10>;
+	using TrackingTable = MultiLevelPageTable<MembershipList, 12, 40, 18>;

-	InlinePageOwnerList() noexcept {}
-	InlinePageOwnerList(const InlinePageOwnerList&)            = delete;
-	InlinePageOwnerList& operator=(const InlinePageOwnerList&) = delete;
-	InlinePageOwnerList(InlinePageOwnerList&& other) noexcept
-	    : m_overflow(std::move(other.m_overflow)),
-	      m_inline_size(std::exchange(other.m_inline_size, 0)) {
-		if (m_overflow == nullptr) {
-			std::copy_n(other.m_inline.begin(), m_inline_size, m_inline.begin());
-		}
-	}
-	InlinePageOwnerList& operator=(InlinePageOwnerList&& other) noexcept {
-		if (this != &other) {
-			m_overflow    = std::move(other.m_overflow);
-			m_inline_size = std::exchange(other.m_inline_size, 0);
-			if (m_overflow == nullptr) {
-				std::copy_n(other.m_inline.begin(), m_inline_size, m_inline.begin());
-			}
-		}
-		return *this;
-	}
-
-	[[nodiscard]] const_iterator begin() const noexcept { return {this, 0}; }
-	[[nodiscard]] const_iterator end() const noexcept { return {this, size()}; }
-
-	[[nodiscard]] bool      empty() const noexcept { return size() == 0; }
-	[[nodiscard]] size_type size() const noexcept {
-		return m_overflow == nullptr ? m_inline_size : m_overflow->size();
-	}
-	[[nodiscard]] OwnerT front() const noexcept { return At(0); }
-	[[nodiscard]] OwnerT operator[](size_type index) const noexcept { return At(index); }
-
-	void push_back(const OwnerT& owner) {
-		const uint64_t packed = Pack(owner);
-		if (m_overflow != nullptr) {
-			m_overflow->push_back(packed);
-			return;
-		}
-		if (m_inline_size < InlineCapacity) {
-			m_inline[m_inline_size] = packed;
-			++m_inline_size;
-			return;
-		}
-		auto overflow = std::make_unique<std::vector<uint64_t>>();
-		overflow->reserve(InlineCapacity * 2);
-		overflow->assign(m_inline.begin(), m_inline.end());
-		overflow->push_back(packed);
-		m_overflow = std::move(overflow);
-	}
-
-	[[nodiscard]] bool Contains(const OwnerT& owner) const noexcept {
-		const uint64_t packed = Pack(owner);
-		if (m_overflow != nullptr) {
-			return std::find(m_overflow->begin(), m_overflow->end(), packed) != m_overflow->end();
-		}
-		return std::find(m_inline.begin(), m_inline.begin() + static_cast<ptrdiff_t>(m_inline_size),
-		                 packed) != m_inline.begin() + static_cast<ptrdiff_t>(m_inline_size);
-	}
-
-	[[nodiscard]] bool Erase(const OwnerT& owner) noexcept {
-		const uint64_t packed = Pack(owner);
-		if (m_overflow != nullptr) {
-			const auto found = std::find(m_overflow->begin(), m_overflow->end(), packed);
-			if (found == m_overflow->end()) {
-				return false;
-			}
-			m_overflow->erase(found);
-			if (m_overflow->size() == InlineCapacity) {
-				std::copy(m_overflow->begin(), m_overflow->end(), m_inline.begin());
-				m_inline_size = InlineCapacity;
-				m_overflow.reset();
-			}
-			return true;
-		}
-		const auto end   = m_inline.begin() + static_cast<ptrdiff_t>(m_inline_size);
-		const auto found = std::find(m_inline.begin(), end, packed);
-		if (found == end) {
+	[[nodiscard]] bool Register(const OwnerT& owner, const std::vector<ByteRange>& ranges) {
+		if (FindRegistration(owner) != m_registrations.end()) {
 			return false;
 		}
-		std::move(found + 1, end, found);
-		--m_inline_size;
+		auto normalized = Normalize(ranges);
+		if (normalized.empty()) {
+			return false;
+		}
+		m_registrations.push_back({owner, std::move(normalized)});
+		const Registration* registration = &m_registrations.back();
+		for (const size_t page: CollectPages<20>(registration->ranges)) {
+			m_coarse_pages[page].push_back(registration);
+		}
+		for (const size_t page: CollectPages<12>(registration->ranges)) {
+			m_tracking_pages[page].push_back(registration);
+		}
 		return true;
 	}

-	template <typename Func>
-	void ForEach(Func&& func) const {
-		if (m_overflow != nullptr) {
-			for (const uint64_t owner: *m_overflow) {
-				func(Unpack(owner));
+	// No state changes if an expected membership is absent. Final releases are
+	// sorted and coalesced 4 KiB spans whose final owner disappeared.
+	[[nodiscard]] bool Unregister(const OwnerT& owner, std::vector<ByteRange>& final_releases) {
+		final_releases.clear();
+		auto registration = FindRegistration(owner);
+		if (registration == m_registrations.end()) {
+			return false;
+		}
+		const auto          coarse_pages     = CollectPages<20>(registration->ranges);
+		const auto          tracking_pages   = CollectPages<12>(registration->ranges);
+		const Registration* registration_ptr = &*registration;
+		if (!HasAllMemberships(m_coarse_pages, coarse_pages, registration_ptr) ||
+		    !HasAllMemberships(m_tracking_pages, tracking_pages, registration_ptr)) {
+			return false;
+		}
+		std::vector<size_t> final_pages;
+		for (const size_t page: coarse_pages) {
+			(void)EraseExact(*m_coarse_pages.Find(page), registration_ptr);
+		}
+		for (const size_t page: tracking_pages) {
+			auto& owners = *m_tracking_pages.Find(page);
+			if (owners.size() == 1) {
+				final_pages.push_back(page);
 			}
-			return;
-		}
-		for (size_type index = 0; index < m_inline_size; ++index) {
-			func(Unpack(m_inline[index]));
+			(void)EraseExact(owners, registration_ptr);
 		}
+		m_registrations.erase(registration);
+		final_releases = CoalesceTrackingPages(final_pages);
+		return true;
+	}
+
+	[[nodiscard]] std::vector<OwnerT> Query(uint64_t address, uint64_t size) const {
+		return Query(address, size, [](const OwnerT&) { return true; });
+	}
+
+	template <typename Predicate>
+	[[nodiscard]] std::vector<OwnerT> Query(uint64_t address, uint64_t size,
+	                                        Predicate&& predicate) const {
+		return QueryImpl(address, size, true, std::forward<Predicate>(predicate));
+	}
+
+	// Fault paths use page candidates: exact byte-disjoint owners sharing a
+	// touched 4 KiB page are intentionally retained.
+	[[nodiscard]] std::vector<OwnerT> QueryCandidates(uint64_t address, uint64_t size) const {
+		return QueryCandidates(address, size, [](const OwnerT&) { return true; });
+	}
+
+	template <typename Predicate>
+	[[nodiscard]] std::vector<OwnerT> QueryCandidates(uint64_t address, uint64_t size,
+	                                                  Predicate&& predicate) const {
+		return QueryImpl(address, size, false, std::forward<Predicate>(predicate));
+	}
+
+	[[nodiscard]] size_t CoarseMembershipCount(size_t page) const {
+		const auto* owners = m_coarse_pages.Find(page);
+		return owners == nullptr ? 0 : owners->size();
+	}
+	[[nodiscard]] size_t TrackingMembershipCount(size_t page) const {
+		const auto* owners = m_tracking_pages.Find(page);
+		return owners == nullptr ? 0 : owners->size();
 	}

 private:
-	[[nodiscard]] OwnerT At(size_type index) const noexcept {
-		return Unpack(m_overflow == nullptr ? m_inline[index] : (*m_overflow)[index]);
-	}
-	[[nodiscard]] static uint64_t Pack(const OwnerT& owner) noexcept {
-		uint64_t packed = 0;
-		std::memcpy(&packed, &owner, sizeof(owner));
-		return packed;
-	}
-	[[nodiscard]] static OwnerT Unpack(uint64_t packed) noexcept {
-		OwnerT owner {};
-		std::memcpy(&owner, &packed, sizeof(owner));
-		return owner;
+	template <typename Predicate>
+	[[nodiscard]] std::vector<OwnerT> QueryImpl(uint64_t address, uint64_t size, bool strict_bytes,
+	                                            Predicate&& predicate) const {
+		typename CoarseTable::PageRange   coarse_range {};
+		typename TrackingTable::PageRange tracking_range {};
+		if (!CoarseTable::TryGetPageRange(address, size, coarse_range) ||
+		    (!strict_bytes && !TrackingTable::TryGetPageRange(address, size, tracking_range))) {
+			return {};
+		}
+		MembershipList candidates;
+		for (size_t page = coarse_range.first; page < coarse_range.last_exclusive; ++page) {
+			if (const auto* owners = m_coarse_pages.Find(page); owners != nullptr) {
+				AppendUnique(candidates, *owners);
+			}
+		}
+		std::vector<OwnerT> result;
+		for (const Registration* registration: candidates) {
+			if ((strict_bytes ? Overlaps(registration->ranges, address, size)
+			                  : HasTrackingMembership(registration, tracking_range)) &&
+			    predicate(registration->owner)) {
+				result.push_back(registration->owner);
+			}
+		}
+		return result;
 	}

-	std::array<uint64_t, InlineCapacity>   m_inline;
-	std::unique_ptr<std::vector<uint64_t>> m_overflow;
-	size_type                              m_inline_size = 0;
+	struct Registration final {
+		OwnerT                 owner;
+		std::vector<ByteRange> ranges;
+	};
+	using RegistrationIterator      = typename std::list<Registration>::iterator;
+	using ConstRegistrationIterator = typename std::list<Registration>::const_iterator;
+
+	[[nodiscard]] RegistrationIterator FindRegistration(const OwnerT& owner) {
+		return std::find_if(m_registrations.begin(), m_registrations.end(),
+		                    [&](const Registration& item) { return item.owner == owner; });
+	}
+	[[nodiscard]] ConstRegistrationIterator FindRegistration(const OwnerT& owner) const {
+		return std::find_if(m_registrations.begin(), m_registrations.end(),
+		                    [&](const Registration& item) { return item.owner == owner; });
+	}
+
+	[[nodiscard]] static std::vector<ByteRange> Normalize(const std::vector<ByteRange>& ranges) {
+		std::vector<ByteRange> sorted;
+		for (const auto& range: ranges) {
+			typename CoarseTable::PageRange ignored {};
+			if (!CoarseTable::TryGetPageRange(range.address, range.size, ignored)) {
+				return {};
+			}
+			sorted.push_back(range);
+		}
+		std::sort(sorted.begin(), sorted.end(), [](const ByteRange& lhs, const ByteRange& rhs) {
+			return lhs.address < rhs.address;
+		});
+		std::vector<ByteRange> merged;
+		for (const auto& range: sorted) {
+			if (merged.empty() || range.address > merged.back().address + merged.back().size) {
+				merged.push_back(range);
+			} else {
+				const uint64_t end = std::max(merged.back().address + merged.back().size,
+				                              range.address + range.size);
+				merged.back().size = end - merged.back().address;
+			}
+		}
+		return merged;
+	}
+
+	template <size_t Bits>
+	[[nodiscard]] static std::vector<size_t> CollectPages(const std::vector<ByteRange>& ranges) {
+		std::vector<size_t> pages;
+		for (const auto& range: ranges) {
+			const size_t first = static_cast<size_t>(range.address >> Bits);
+			const size_t last  = static_cast<size_t>((range.address + range.size - 1) >> Bits);
+			for (size_t page = first; page <= last; ++page) {
+				pages.push_back(page);
+			}
+		}
+		std::sort(pages.begin(), pages.end());
+		pages.erase(std::unique(pages.begin(), pages.end()), pages.end());
+		return pages;
+	}
+
+	template <typename Table>
+	[[nodiscard]] static bool HasAllMemberships(const Table&               table,
+	                                            const std::vector<size_t>& pages,
+	                                            const Registration*        registration) {
+		for (const size_t page: pages) {
+			const auto* owners = table.Find(page);
+			if (owners == nullptr ||
+			    std::find(owners->begin(), owners->end(), registration) == owners->end()) {
+				return false;
+			}
+		}
+		return true;
+	}
+
+	[[nodiscard]] bool HasTrackingMembership(const Registration*                      registration,
+	                                         const typename TrackingTable::PageRange& range) const {
+		for (size_t page = range.first; page < range.last_exclusive; ++page) {
+			const auto* owners = m_tracking_pages.Find(page);
+			if (owners != nullptr &&
+			    std::find(owners->begin(), owners->end(), registration) != owners->end()) {
+				return true;
+			}
+		}
+		return false;
+	}
+
+	[[nodiscard]] static bool Overlaps(const std::vector<ByteRange>& ranges, uint64_t address,
+	                                   uint64_t size) {
+		const uint64_t end = address + size;
+		return std::any_of(ranges.begin(), ranges.end(), [&](const ByteRange& range) {
+			return range.address < end && address < range.address + range.size;
+		});
+	}
+	static void AppendUnique(MembershipList& destination, const MembershipList& source) {
+		for (const Registration* registration: source) {
+			if (std::find(destination.begin(), destination.end(), registration) ==
+			    destination.end()) {
+				destination.push_back(registration);
+			}
+		}
+	}
+	[[nodiscard]] static std::vector<ByteRange>
+	CoalesceTrackingPages(const std::vector<size_t>& pages) {
+		std::vector<ByteRange> result;
+		for (const size_t page: pages) {
+			const uint64_t address = static_cast<uint64_t>(page) << 12;
+			if (!result.empty() && result.back().address + result.back().size == address) {
+				result.back().size += uint64_t {1} << 12;
+			} else {
+				result.push_back({address, uint64_t {1} << 12});
+			}
+		}
+		return result;
+	}
+
+	CoarseTable             m_coarse_pages;
+	TrackingTable           m_tracking_pages;
+	std::list<Registration> m_registrations;
 };

 } // namespace Libs::Graphics
@@ -0,0 +1,93 @@
+#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
@@ -0,0 +1,48 @@
+#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_
@@ -116,6 +116,9 @@ vk::Sampler SamplerCache::GetSampler(const ShaderSamplerResource& r) {
 		default: EXIT("unknown border color: %d", static_cast<int>(r.BorderColorType()));
 	}

+	sampler_info.sType     = vk::StructureType::eSamplerCreateInfo;
+	sampler_info.pNext     = nullptr;
+	sampler_info.flags     = {};
 	sampler_info.magFilter = to_vk_filter(mag_filter);
 	sampler_info.minFilter = to_vk_filter(min_filter);
 	sampler_info.mipmapMode =
@@ -1,12 +1,13 @@
 #include "graphics/host_gpu/renderer/cache/streamBuffer.h"

-#include "common/alignment.h"
 #include "common/assert.h"
 #include "common/profiler.h"
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/host_gpu/renderer/commandScheduler.h"
+#include "graphics/host_gpu/vma.h"

 #include <cstring>
+#include <limits>
 #include <numeric>
 #include <vk_mem_alloc.h>

@@ -45,11 +46,16 @@ constexpr size_t WATCHES_RESERVE_CHUNK   = 0x1000;
 		result = value;
 		return true;
 	}
-	const auto aligned = Common::AlignUp(value, alignment);
-	if (aligned < value) {
+	const auto remainder = value % alignment;
+	if (remainder == 0) {
+		result = value;
+		return true;
+	}
+	const auto increment = alignment - remainder;
+	if (value > std::numeric_limits<uint64_t>::max() - increment) {
 		return false;
 	}
-	result = aligned;
+	result = value + increment;
 	return true;
 }

@@ -58,20 +64,17 @@ constexpr size_t WATCHES_RESERVE_CHUNK   = 0x1000;
 Buffer::Buffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsage usage,
               uint64_t cpu_address, vk::BufferUsageFlags flags, uint64_t size)
    : m_graphics(&graphics), m_scheduler(&scheduler), m_usage(usage), m_cpu_address(cpu_address),
-      m_size(size) {
+      m_size(size), m_buffer(std::make_unique<VulkanBuffer>()) {
 	KYTY_PROFILER_FUNCTION();
 	EXIT_IF(graphics.allocator == nullptr || size == 0);

 	vk::BufferCreateInfo buffer_info {};
 	buffer_info.size        = size;
 	buffer_info.usage       = flags;
+	buffer_info.sharingMode = vk::SharingMode::eExclusive;

-	const bool with_bda = bool(flags & vk::BufferUsageFlagBits::eShaderDeviceAddress);
-	const VmaAllocationCreateFlags bda_flag =
-	    with_bda ? VMA_ALLOCATION_CREATE_DEDICATED_MEMORY_BIT : 0;
 	VmaAllocationCreateInfo allocation_info {};
-	allocation_info.flags =
-	    VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | bda_flag | AllocationFlags(usage);
+	allocation_info.flags = VMA_ALLOCATION_CREATE_WITHIN_BUDGET_BIT | AllocationFlags(usage);
 	allocation_info.usage = AllocationUsage(usage);
 	allocation_info.preferredFlags = usage == MemoryUsage::DeviceLocal
 	                                     ? VkMemoryPropertyFlags {}
@@ -81,70 +84,72 @@ Buffer::Buffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsag
 	VkBuffer          native_buffer = VK_NULL_HANDLE;
 	const auto        result        = static_cast<vk::Result>(vmaCreateBuffer(
 	    graphics.allocator, static_cast<const VkBufferCreateInfo*>(buffer_info), &allocation_info,
-	    &native_buffer, &m_allocation, &allocation_result));
+	    &native_buffer, &m_buffer->memory.allocation, &allocation_result));
 	if (result != vk::Result::eSuccess) {
 		graphics.LogMemoryBudget();
 	}
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);

-	m_buffer = native_buffer;
-	if (with_bda) {
-		vk::BufferDeviceAddressInfo address_info {};
-		address_info.buffer = m_buffer;
-		m_device_address    = graphics.device.getBufferAddress(address_info);
-		EXIT_IF(m_device_address == 0);
-	}
+	m_buffer->buffer                 = native_buffer;
+	m_buffer->usage                  = flags;
+	m_buffer->buffer_size            = size;
+	m_buffer->memory.allocation_info = allocation_result;
+	m_buffer->memory.memory          = allocation_result.deviceMemory;
+	m_buffer->memory.offset          = allocation_result.offset;
+	m_buffer->memory.type            = allocation_result.memoryType;
+	m_buffer->memory.unique_id       = VulkanNextMemoryUniqueId();
+	graphics.device.getBufferMemoryRequirements(m_buffer->buffer, &m_buffer->memory.requirements);

 	VkMemoryPropertyFlags properties = 0;
-	vmaGetAllocationMemoryProperties(graphics.allocator, m_allocation, &properties);
-	m_coherent = (properties & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) != 0;
+	vmaGetAllocationMemoryProperties(graphics.allocator, m_buffer->memory.allocation, &properties);
+	m_buffer->memory.property = vk::MemoryPropertyFlags(properties);
+	m_is_coherent             = (properties & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) != 0;
 	if (allocation_result.pMappedData != nullptr) {
 		m_mapped = {static_cast<uint8_t*>(allocation_result.pMappedData),
 		            static_cast<size_t>(size)};
 	}
+	VulkanTrackAllocation(m_buffer->memory);
 }

 Buffer::~Buffer() {
-	if (m_buffer != nullptr) {
-		vmaDestroyBuffer(m_graphics->allocator, m_buffer, m_allocation);
+	if (m_buffer->buffer != nullptr) {
+		VulkanUntrackAllocation(m_buffer->memory);
+		vmaDestroyBuffer(m_graphics->allocator, m_buffer->buffer, m_buffer->memory.allocation);
 	}
 }

-vk::DeviceAddress Buffer::BufferDeviceAddress() const noexcept {
-	EXIT_IF(m_device_address == 0);
-	return m_device_address;
+vk::Buffer Buffer::Handle() const noexcept {
+	return m_buffer->buffer;
 }

 bool Buffer::IsInBounds(uint64_t address, uint64_t size) const noexcept {
-	return address >= m_cpu_address && size <= Size() && address - m_cpu_address <= Size() - size;
+	return address >= m_cpu_address && size <= m_size && address - m_cpu_address <= m_size - size;
+}
+
+void Buffer::Write(uint64_t offset, const void* source, uint64_t size) {
+	EXIT_IF(source == nullptr || m_mapped.empty() || offset > m_size || size > m_size - offset);
+	std::memcpy(m_mapped.data() + offset, source, static_cast<size_t>(size));
+	Flush(offset, size);
 }

 void Buffer::Flush(uint64_t offset, uint64_t size) {
-	EXIT_IF(m_mapped.empty() || offset > Size() || size > Size() - offset);
-	if (!IsCoherent() && size != 0) {
+	EXIT_IF(m_mapped.empty() || offset > m_size || size > m_size - offset);
+	if (!m_is_coherent && size != 0) {
 		const auto result =
-		    vmaFlushAllocation(m_graphics->allocator, m_allocation, offset, size);
-		EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
-	}
-}
-
-void Buffer::Invalidate(uint64_t offset, uint64_t size) {
-	EXIT_IF(m_usage != MemoryUsage::Download || offset > Size() || size > Size() - offset);
-	if (!IsCoherent() && size != 0) {
-		const auto result =
-		    vmaInvalidateAllocation(m_graphics->allocator, m_allocation, offset, size);
+		    vmaFlushAllocation(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 > Size() || size > Size() - offset) {
+	if (Handle() == nullptr || size == 0 || offset > m_size || size > m_size - offset) {
 		EXIT("Buffer: invalid DMA barrier, handle=%p offset=0x%016" PRIx64 " size=0x%016" PRIx64
 		     " capacity=0x%016" PRIx64 "\n",
-		     static_cast<const void*>(Handle()), offset, size, Size());
+		     static_cast<const void*>(Handle()), offset, size, m_size);
 	}
 	vk::BufferMemoryBarrier barrier {};
+	barrier.sType               = vk::StructureType::eBufferMemoryBarrier;
 	barrier.srcAccessMask       = source;
 	barrier.dstAccessMask       = destination;
 	barrier.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
@@ -159,8 +164,8 @@ void Buffer::CopyFrom(CommandBuffer& command, const Buffer& source, uint64_t sou
                      uint64_t destination_offset, uint64_t size, vk::AccessFlags source_before,
                      vk::AccessFlags destination_before, vk::AccessFlags source_after,
                      vk::AccessFlags destination_after) {
-	if (size == 0 || source_offset > source.Size() || size > source.Size() - source_offset ||
-	    destination_offset > Size() || size > Size() - destination_offset) {
+	if (size == 0 || source_offset > source.m_size || size > source.m_size - source_offset ||
+	    destination_offset > m_size || size > m_size - destination_offset) {
 		EXIT("Buffer: invalid copy range\n");
 	}
 	if (source.Handle() == Handle() && source_offset < destination_offset + size &&
@@ -217,8 +222,10 @@ void Buffer::Fill(uint64_t offset, uint64_t size, uint32_t value) {

 StreamBuffer::StreamBuffer(GraphicContext& graphics, CommandScheduler& scheduler, MemoryUsage usage,
                           uint64_t size)
-    : Buffer(graphics, scheduler, usage, 0, AllFlags, size),
-      m_current_watches(WATCHES_INITIAL_RESERVE), m_previous_watches(WATCHES_INITIAL_RESERVE) {}
+    : Buffer(graphics, scheduler, usage, 0, AllFlags, size) {
+	ReserveWatches(m_current_watches, WATCHES_INITIAL_RESERVE);
+	ReserveWatches(m_previous_watches, WATCHES_INITIAL_RESERVE);
+}

 bool StreamBuffer::NormalizeReservation(bool coherent, uint64_t atom, uint64_t& size,
                                        uint64_t& alignment) {
@@ -283,8 +290,10 @@ std::pair<uint8_t*, uint64_t> StreamBuffer::Map(uint64_t size, uint64_t alignmen
 }

 void StreamBuffer::Commit() {
-	if (Usage() != MemoryUsage::Download && m_mapped_size != 0) {
-		Flush(m_offset, m_mapped_size);
+	if (!IsCoherent() && Usage() != MemoryUsage::Download && m_mapped_size != 0) {
+		const auto result = vmaFlushAllocation(
+		    Graphics().allocator, NativeBuffer().memory.allocation, m_offset, m_mapped_size);
+		EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
 	}

 	m_offset += m_mapped_size;
@@ -294,13 +303,23 @@ void StreamBuffer::Commit() {
 		return;
 	}
 	if (m_current_watch_cursor + 1 >= m_current_watches.size()) {
-		m_current_watches.resize(m_current_watches.size() + WATCHES_RESERVE_CHUNK);
+		ReserveWatches(m_current_watches, WATCHES_RESERVE_CHUNK);
 	}
 	auto& watch       = m_current_watches[m_current_watch_cursor++];
 	watch.upper_bound = m_offset;
 	watch.tick        = tick;
 }

+void StreamBuffer::Invalidate(uint64_t offset, uint64_t size) {
+	EXIT_IF(Usage() != MemoryUsage::Download || offset > Size() || size > Size() - offset);
+	if (IsCoherent() || size == 0) {
+		return;
+	}
+	const auto result = vmaInvalidateAllocation(Graphics().allocator,
+	                                            NativeBuffer().memory.allocation, offset, size);
+	EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
+}
+
 uint64_t StreamBuffer::Copy(const void* source, uint64_t size, uint64_t alignment) {
 	EXIT_IF(source == nullptr);
 	const auto [data, offset] = Map(size, alignment);
@@ -310,6 +329,10 @@ uint64_t StreamBuffer::Copy(const void* source, uint64_t size, uint64_t alignmen
 	return offset;
 }

+void StreamBuffer::ReserveWatches(std::vector<Watch>& watches, size_t grow_size) {
+	watches.resize(watches.size() + grow_size);
+}
+
 bool StreamBuffer::WaitPendingOperations(const std::vector<Watch>& watches,
                                         std::optional<size_t>     invalidation_mark,
                                         uint64_t requested_upper_bound, bool allow_wait,
@@ -6,19 +6,19 @@
 #include "graphics/host_gpu/vulkanCommon.h"

 #include <cstdint>
+#include <memory>
 #include <optional>
 #include <span>
 #include <utility>
 #include <vector>

-VK_DEFINE_HANDLE(VmaAllocation)
-
 namespace Libs::Graphics {

 class CommandBuffer;
 class CommandScheduler;
 struct StreamBufferTestAccess;
 struct GraphicContext;
+struct VulkanBuffer;

 enum class MemoryUsage : uint8_t {
 	DeviceLocal,
@@ -42,21 +42,18 @@ public:
 	~Buffer();
 	KYTY_CLASS_NO_COPY(Buffer);

-	[[nodiscard]] vk::Buffer         Handle() const noexcept { return m_buffer; }
+	[[nodiscard]] vk::Buffer         Handle() const noexcept;
 	[[nodiscard]] uint64_t           Size() const noexcept { return m_size; }
 	[[nodiscard]] std::span<uint8_t> Mapped() const noexcept { return m_mapped; }
-	[[nodiscard]] bool               IsCoherent() const noexcept { return m_coherent; }
+	[[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               IncreaseStreamScore(int score) noexcept { stream_score += score; }
-	[[nodiscard]] int  StreamScore() const noexcept { return stream_score; }
+	void               Write(uint64_t offset, const void* source, uint64_t size);
 	void               Flush(uint64_t offset, uint64_t size);
-	void               Invalidate(uint64_t offset, uint64_t size);
 	void CopyFrom(CommandBuffer& command, const Buffer& source, uint64_t source_offset,
 	              uint64_t destination_offset, uint64_t size,
 	              vk::AccessFlags source_before      = vk::AccessFlagBits::eMemoryWrite,
@@ -68,14 +65,10 @@ 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;
-	int    stream_score = 0;
-	size_t lru_id       = 0;
-
 protected:
 	[[nodiscard]] GraphicContext&   Graphics() const noexcept { return *m_graphics; }
 	[[nodiscard]] CommandScheduler& Scheduler() const noexcept { return *m_scheduler; }
+	[[nodiscard]] VulkanBuffer&     NativeBuffer() noexcept { return *m_buffer; }

 private:
 	[[nodiscard]] vk::BufferMemoryBarrier Barrier(uint64_t offset, uint64_t size,
@@ -86,12 +79,10 @@ private:
 	CommandScheduler*             m_scheduler   = nullptr;
 	MemoryUsage                   m_usage       = MemoryUsage::DeviceLocal;
 	uint64_t                      m_cpu_address = 0;
-	vk::DeviceAddress             m_device_address = 0;
-	vk::Buffer                    m_buffer     = nullptr;
-	VmaAllocation                 m_allocation = nullptr;
-	uint64_t                      m_size;
-	bool                          m_coherent = false;
+	uint64_t                      m_size        = 0;
+	std::unique_ptr<VulkanBuffer> m_buffer;
 	std::span<uint8_t>            m_mapped;
+	bool                          m_is_coherent = false;
 };

 class StreamBuffer final: public Buffer {
@@ -102,6 +93,9 @@ 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:
@@ -112,6 +106,7 @@ private:
 		uint64_t upper_bound = 0;
 	};

+	void                      ReserveWatches(std::vector<Watch>& watches, size_t grow_size);
 	[[nodiscard]] static bool NormalizeReservation(bool coherent, uint64_t atom, uint64_t& size,
 	                                               uint64_t& alignment);
 	[[nodiscard]] bool        WaitPendingOperations(const std::vector<Watch>& watches,
@@ -4,18 +4,19 @@
 #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 <type_traits>
-#include <unordered_map>
-#include <unordered_set>
+#include <memory>
+#include <optional>
+#include <set>
+#include <span>
+#include <utility>
 #include <vector>

 namespace Libs::Graphics {
@@ -25,7 +26,10 @@ class Buffer;
 class BufferCache;
 class CommandBuffer;
 class CommandScheduler;
+class ResourceMutex;
 class RenderExecutor;
+class StreamBuffer;
+class TileManager;
 struct TextureCacheTestAccess;

 class TextureCache {
@@ -38,30 +42,33 @@ public:
 		BindingType   type = BindingType::Texture;
 	};

+	struct RegionInfo {
+		bool image_pages     = false;
+		bool image_bytes     = false;
+		bool gpu_image_bytes = false;
+	};
+
 	TextureCache(GraphicContext& graphics, CommandScheduler& scheduler, PageManager& page_manager,
-	             BufferCache& buffer_cache);
+	             BufferCache& buffer_cache, ResourceMutex& resource_mutex);
 	~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) {
-		auto& image = m_slot_images[id];
-		TouchImage(image);
-		return image;
-	}
-	void MarkGpuWritten(ImageId id);
+	[[nodiscard]] Image&        GetImage(ImageId id);
+	[[nodiscard]] const Image&  GetImage(ImageId id) const;
+	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);
-	void               InvalidateMemoryFromGPU(uint64_t address, uint64_t size);
-	[[nodiscard]] bool IsRegionGpuModified(uint64_t address, uint64_t size);
+	[[nodiscard]] bool InvalidateMemoryFromGPU(uint64_t address, uint64_t size,
+	                                           bool formatted_buffer_write = false);
+	[[nodiscard]] RegionInfo QueryRegion(uint64_t address, uint64_t size);

 	[[nodiscard]] bool IsMeta(uint64_t address);
 	[[nodiscard]] bool IsMetaCleared(uint64_t address, uint32_t slice);
@@ -74,14 +81,16 @@ public:

 private:
 	enum class TransferDirection { Upload, Download };
-	struct TextureTransfer;
-	struct ImageDownload;
+	struct ColorTransferPlan;
+	struct DownloadPlan;
+
+	struct Slot {
+		std::shared_ptr<Image> image;
+		uint32_t               generation = 1;
+	};

 	struct MetaDataInfo {
-		enum class Type : uint8_t { CMask, FMask, HTile };
-
-		Type     type;
-		uint32_t clear_mask = UINT32_MAX;
+		uint32_t clear_mask = 0;
 	};

 	struct OverlapResult {
@@ -90,100 +99,87 @@ private:
 		int32_t layer = -1;
 	};

-	using ImageIds       = InlinePageOwnerList<ImageId, 16>;
-	using ImagePageTable = MultiLevelPageTable<ImageIds, 20, 40, 10>;
+	using ImageOwnerIndex = MultiRangePageOwnerIndex<ImageId>;

-	// Callers have validated the nonempty 40-bit range with TryGetPageRange.
-	template <typename Func>
-	static void ForEachPage(uint64_t address, size_t size, Func&& func) {
-		using FuncReturn = typename std::invoke_result<Func, uint64_t>::type;
-		static constexpr bool RETURNS_BOOL = std::is_same_v<FuncReturn, bool>;
-		const uint64_t page_end = (address + size - 1) >> ImagePageTable::kPageBits;
-		for (uint64_t page = address >> ImagePageTable::kPageBits; page <= page_end; ++page) {
-			if constexpr (RETURNS_BOOL) {
-				if (func(page)) {
-					break;
-				}
-			} else {
-				func(page);
-			}
-		}
-	}
-
-	[[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]] 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]] static bool SameBacking(const ImageInfo& cached, const ImageInfo& requested,
 	                                      bool exact_format);
 	[[nodiscard]] static BindingType UploadBinding(const Image& image);
 	[[nodiscard]] bool               SafeToDownload(const Image& image);

-	// Caller holds m_lock; it also serializes the per-image query epoch.
-	[[nodiscard]] ImageIds      FindImagesInRegion(uint64_t address, uint64_t size,
-	                                               bool page_overlap) const;
+	[[nodiscard]] std::vector<ImageId> FindImagesInRegion(uint64_t address, uint64_t size,
+	                                                      bool page_overlap) const;
 	[[nodiscard]] OverlapResult ResolveOverlap(const ImageInfo& requested, BindingType binding,
 	                                           ImageId cached, ImageId merged);
 	[[nodiscard]] ImageId       ResolveDepthOverlap(const ImageInfo& requested, BindingType binding,
 	                                                ImageId cached);
 	[[nodiscard]] ImageId       ExpandImage(const ImageInfo& info, ImageId source);
-	void                        RefreshImage(ImageId id);
-	void                        MaterializeDccClear(ImageId id, const ImageDesc& desc,
-	                                                uint32_t metadata_base_layer);
-	void                        InitializeImage(ImageId id);
-	[[nodiscard]] TextureTransfer
-	BuildTextureTransfer(const Image& image, BindingType binding, TransferDirection direction) const;
-	[[nodiscard]] ImageDownload BuildDownload(const Image& image) const;
-	void UploadImage(Image& image, Buffer& source, uint64_t source_offset);
-	void DownloadImage(Image& image, Buffer& destination, uint64_t destination_offset,
-	                       uint64_t destination_size, ImageDownload transfer);
+	void                        RefreshImage(ImageId id, const ImageDesc& desc);
+	void                        InitializeImage(ImageId id, const ImageDesc& desc);
+	[[nodiscard]] ColorTransferPlan BuildColorTransfer(const Image& image, BindingType binding,
+	                                                   TransferDirection direction) const;
+	[[nodiscard]] DownloadPlan      BuildDownload(const Image& image) const;
+	void UploadImage(Image& image, const ImageDesc& desc, Buffer& source, uint64_t source_offset);
+	void DownloadImageData(Image& image, Buffer& destination, uint64_t destination_offset,
+	                       uint64_t destination_size, DownloadPlan plan);
 	void DownloadDepth(Image& image, Buffer& destination, uint64_t destination_offset);
 	void CommitGpuWrite(Image& image);
-	// Caller holds m_lock. Volume layer ranges select depth slices.
-	void ClearImage(CommandBuffer& command, ImageId id, vk::Format format,
-	                const vk::ImageSubresourceRange& range, const vk::ClearValue& clear);
 	void PrepareImageCopy(Image& image);
 	void RefreshCopySource(ImageId id);
 	[[nodiscard]] bool CopyD16(Image& destination, Image& source);
 	void               CopyImage(ImageId destination, ImageId source);
 	void               AssociateStencil(ImageId depth, GuestRange stencil);
+	void               AssociateStencilLocked(ImageId depth, GuestRange stencil);
 	void CopyImageMip(ImageId destination, ImageId source, uint32_t mip, uint32_t layer);
 	void ValidateImageDesc(const ImageDesc& desc) const;

-	void               InvalidateCpuAliases(uint64_t address, uint64_t size);
-	[[nodiscard]] bool DownloadImageMemory(ImageId id);
+	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);

 	GraphicContext&                                   m_graphics;
 	CommandScheduler&                                 m_scheduler;
 	TrackingSpinLock                                  m_lock;
 	PageManager&                                      m_page_manager;
 	BlitHelper                                        m_blit_helper;
-	TileManager                                       m_tiler;
+	std::unique_ptr<TileManager>                      m_tiler;
 	BufferCache&                                      m_buffer_cache;
-	Common::SlotVector<Image>                         m_slot_images;
-	ImagePageTable                                    m_image_page_table;
-	std::unordered_map<vk::Format, ImageId>           m_null_images;
+	ResourceMutex&                                    m_resource_mutex;
+	std::vector<Slot>                                 m_slots;
+	std::vector<uint32_t>                             m_free_slots;
+	ImageOwnerIndex                                   m_image_owner_index;
+	std::map<vk::Format, ImageId>                     m_null_images;
 	Common::LeastRecentlyUsedCache<ImageId, uint64_t> m_lru_cache;
-	std::unordered_set<ImageId>                       m_download_images;
+	std::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;
 	uint64_t                                          m_pressure_gc_memory = 1536ull * 1024 * 1024;
-	uint64_t         m_critical_gc_memory     = 3ull * 1024 * 1024 * 1024;
-	uint64_t         m_gc_tick                = 0;
-	mutable uint32_t m_image_query_epoch      = 0;
-	bool             m_readback_linear_images = false;
+	uint64_t m_critical_gc_memory     = 3ull * 1024 * 1024 * 1024;
+	uint64_t m_gc_tick                = 0;
+	bool     m_readback_linear_images = false;

 	friend struct TextureCacheTestAccess;
 	friend class BufferCache;
@@ -4,58 +4,42 @@
 #include "common/logging/log.h"
 #include "common/profiler.h"
 #include "graphics/guest_gpu/gpu_defs.h"
-#include "graphics/guest_gpu/gpu_format.h"
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/guest_gpu/tile.h"
 #include "graphics/host_gpu/graphicContext.h"
 #include "graphics/host_gpu/renderer/debug.h"
 #include "graphics/host_gpu/renderer/image/textureCommon.h"
+#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vulkanCommon.h"

 #include <algorithm>
-#include <array>
 #include <atomic>

 namespace Libs::Graphics {

 static std::atomic<uint32_t> g_render_color_log_count = 0;

-static bool DccAlphaOnMsb(const HW::ColorInfo& info) {
-	switch (info.format) {
-		case Prospero::ChannelLayout::k10_10_10_2:
-		case Prospero::ChannelLayout::k10_10_10_2Float:
-		case Prospero::ChannelLayout::k5_5_5_1: return true;
-		case Prospero::ChannelLayout::k2_10_10_10:
-		case Prospero::ChannelLayout::k1_5_5_5: return false;
-		default: break;
-	}
-	const auto components =
-	    Prospero::ResolveRenderTargetFormat(info.format, info.channel_type).components;
-	if (components == 1) {
-		return info.channel_order != Prospero::ChannelOrder::kStandard;
-	}
-	return components == 3 || info.channel_order == Prospero::ChannelOrder::kStandard ||
-	       info.channel_order == Prospero::ChannelOrder::kAlt;
-}
-
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
-void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColorInfo& r,
+void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandBuffer& buffer,
+                                              RenderColorInfo& r,
                                              uint32_t         render_target_slice_offset,
-                                              uint32_t rt_slot, bool ignore_target_mask,
+                                              uint32_t render_target_slot, bool ignore_target_mask,
                                              bool exact_format) {
 	KYTY_PROFILER_FUNCTION();
 	const auto& hw = buffer.GetRegisters();

+	const auto  rt_slot = (render_target_slot == UINT32_MAX ? render_target_first_bound_slot(buffer)
+	                                                        : render_target_slot);
 	const auto& rt      = hw.GetRenderTarget(rt_slot);
 	auto        mask    = render_target_mask_slot(hw.GetRenderTargetMask(), rt_slot);
 	if (ignore_target_mask && rt.base.addr != 0 && mask == 0) {
 		mask = 0x0f;
 	}

-	r             = {};
-	r.target_slot = rt_slot;
+	r.target_slot    = rt_slot;
+	r.export_mapping = {};

 	if (rt.base.addr == 0 || mask == 0) {
 		if (graphics_debug_dump_enabled()) {
@@ -70,6 +54,21 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 			}
 		}

+		// No color output
+		r.type               = RenderColorType::NoColorOutput;
+		r.desc               = {};
+		r.base_addr          = 0;
+		r.image_id           = {};
+		r.image_view         = nullptr;
+		r.format             = vk::Format::eUndefined;
+		r.extent             = {};
+		r.base_mip_level     = 0;
+		r.base_array_layer   = 0;
+		r.buffer_size        = 0;
+		r.samples            = 1;
+		r.export_mapping     = {};
+		r.color_clear_enable = false;
+		r.color_clear_value  = {};
 		return;
 	}
 	const auto samples = render_sample_count(rt.attrib.num_fragments);
@@ -77,41 +76,8 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 		EXIT("unsupported render-target sample configuration: samples=%u fragments=%u\n",
 		     rt.attrib.num_samples, rt.attrib.num_fragments);
 	}
-	const uint32_t levels = rt.attrib2.num_mip_levels + 1u;
-	if (levels == 0 || levels > 16 || rt.view.current_mip_level >= levels) {
-		EXIT("unsupported render-target mip range: current=%u levels=%u\n",
-		     rt.view.current_mip_level, levels);
-	}
-	static constexpr std::array image_types {Prospero::ImageType::kColor1D,
-	                                         Prospero::ImageType::kColor2D,
-	                                         Prospero::ImageType::kColor3D};
-	if (rt.attrib3.dimension >= image_types.size()) {
-		EXIT("unsupported render-target dimension: %u\n", rt.attrib3.dimension);
-	}
-	const auto image_type = image_types[rt.attrib3.dimension];
-	const bool is_1d      = image_type == Prospero::ImageType::kColor1D;
-	const bool volume     = image_type == Prospero::ImageType::kColor3D;
-	if (is_1d && rt.attrib2.height != 0) {
-		EXIT("1D render target has nonzero height: %u\n", rt.attrib2.height);
-	}
-	if (!volume && rt.attrib3.depth != 0) {
-		EXIT("non-3D render target has nonzero depth: %u\n", rt.attrib3.depth);
-	}
-	if (is_1d && samples != 1) {
-		EXIT("multisampled 1D render targets are unsupported\n");
-	}
-	if (volume && samples != 1) {
-		EXIT("multisampled 3D render targets are unsupported\n");
-	}
-	const uint32_t depth = volume ? rt.attrib3.depth + 1u : 1u;
-	// For volumes, CB_COLOR_VIEW bounds exported slices; ATTRIB3 defines storage depth.
-	// The host attachment contains only the selected slices that exist in this mip.
-	const uint32_t last_layer = volume
-	                                ? std::min(rt.view.last_array_slice_index,
-	                                           std::max(depth >> rt.view.current_mip_level, 1u) - 1u)
-	                                : rt.view.last_array_slice_index;
 	const auto view = ResolveTargetViewInfo(
-	    rt.view.base_array_slice_index, last_layer, render_target_slice_offset);
+	    rt.view.base_array_slice_index, rt.view.last_array_slice_index, render_target_slice_offset);
 	switch (view.type) {
 		case TargetViewType::Image2D:
 		case TargetViewType::Image2DArray: break;
@@ -120,6 +86,12 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 			     rt.view.base_array_slice_index, rt.view.last_array_slice_index,
 			     render_target_slice_offset);
 	}
+	r.base_array_layer    = view.base_layer;
+	const uint32_t levels = rt.attrib2.num_mip_levels + 1u;
+	if (levels == 0 || levels > 16 || rt.view.current_mip_level >= levels) {
+		EXIT("unsupported render-target mip range: current=%u levels=%u\n",
+		     rt.view.current_mip_level, levels);
+	}
 	if (graphics_debug_dump_enabled()) {
 		static std::atomic_uint log_count = 0;
 		const auto              log_id    = log_count.fetch_add(1, std::memory_order_relaxed);
@@ -129,36 +101,45 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 			     " 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,
-			     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));
+			     rt.attrib3.tile_mode, rt.attrib3.dimension, rt.info.format, rt.info.channel_type,
+			     rt.info.channel_order);
 		}
 	}

-	// 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.
+	// 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.
 	// Fast color clears are metadata driven and must be handled explicitly when
 	// that metadata path is implemented; render-pass load must preserve contents.
+	r.color_clear_enable = false;
+	r.color_clear_value  = {};
+
 	uint32_t   width  = 0;
 	uint32_t   height = 0;
 	uint32_t   pitch  = 0;
 	uint64_t   size   = 0;
 	bool       tile   = false;
-	const bool     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;
+	const bool volume = rt.attrib3.dimension == 2;
+	if (rt.attrib3.dimension != 1 && !volume) {
+		EXIT("unsupported render-target dimension: %u\n", rt.attrib3.dimension);
+	}
+	if (!volume && rt.attrib3.depth != 0) {
+		EXIT("2D render target has nonzero depth: %u\n", rt.attrib3.depth);
+	}
+	if (volume && samples != 1) {
+		EXIT("multisampled 3D render targets are unsupported\n");
+	}
+	const uint32_t depth = volume ? rt.attrib3.depth + 1u : 1u;
+	const bool     standard64 =
+	    rt.attrib3.tile_mode == Prospero::GpuEnumValue(Prospero::TileMode::kStandard64KB);

 	switch (rt.attrib3.tile_mode) {
-		case Prospero::TileMode::kLinear:
-		case Prospero::TileMode::kStandard4KB:
-		case Prospero::TileMode::kStandard64KB:
-		case Prospero::TileMode::kDepth:
-		case Prospero::TileMode::kRenderTarget:
+		case Prospero::GpuEnumValue(Prospero::TileMode::kLinear):
+		case Prospero::GpuEnumValue(Prospero::TileMode::kStandard64KB):
+		case Prospero::GpuEnumValue(Prospero::TileMode::kRenderTarget):
 			tile = !RenderIsColorTileModeLinear(rt.attrib3.tile_mode);
 			break;
-		default: EXIT("unknown tile mode: %u\n", static_cast<uint32_t>(rt.attrib3.tile_mode));
+		default: EXIT("unknown tile mode: %u\n", rt.attrib3.tile_mode);
 	}
 	if (!tile && levels > 1) {
 		EXIT("linear mipmapped render targets are unsupported\n");
@@ -166,9 +147,6 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 	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;
@@ -179,66 +157,61 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 		EXIT("render-target format has no valid element size\n");
 	}
 	const auto transfer_format = ImageOps::RenderTargetTransferFormat(bytes_per_element);
-	TileTextureBlockLayout texture_tile_layout {};
-	if (texture_tile &&
-	    (!TileGetTextureBlockLayout(transfer_format, rt.attrib3.tile_mode, volume,
-	                                texture_tile_layout) ||
-	     (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.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"
+	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
 		     " 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, 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);
+		     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);
 	}
-	if ((standard64 || depth_tile) &&
-	    (rt.attrib3.dimension != 1 || rt.attrib3.depth != 0 ||
-	     (depth_tile && (view.base_layer != 0 || view.image_layers != 1)))) {
-		EXIT("unsupported 64KB texture-tiled render-target view: dimension=%u depth=%u"
-		     " layer=%u/%u\n",
-		     rt.attrib3.dimension, rt.attrib3.depth, view.base_layer, view.image_layers);
-	}
-	// PPSA28068: the linear EULA target and its sampled view must share the padded pitch.
-	if (!tile || volume || texture_tile) {
-		pitch = TileGetTexturePitch(transfer_format, width, rt.attrib3.tile_mode);
+	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);
+		} else {
+			pitch = TileGetRenderTargetPitch(width, bytes_per_element, rt.attrib.num_fragments);
+		}
+		if (pitch == 0) {
+			EXIT("unsupported render-target pitch: width=%u bytes=%u\n", width, bytes_per_element);
+		}
 	} else {
-		pitch = TileGetRenderTargetPitch(width, bytes_per_element, rt.attrib.num_fragments);
-	}
-	if (pitch == 0) {
-		EXIT("unsupported render-target pitch: width=%u bytes=%u\n", width, bytes_per_element);
+		pitch = width;
 	}

-	TileSizeOffset    mip_sizes[16] {};
-	TilePaddedSize    mip_padded[16] {};
-	TileSurfaceLayout volume_layout {};
-	uint64_t          backing_size = 0;
+	TileSizeOffset   mip_sizes[16] {};
+	TilePaddedSize   mip_padded[16] {};
+	TileVolumeLayout volume_layout {};
+	uint64_t         backing_size = 0;
 	if (volume) {
-		const TileSurfaceDescription description {transfer_format,
-		                                          rt.attrib3.tile_mode,
-		                                          TileSurfaceDimension::Dim3D,
-		                                          width,
-		                                          height,
-		                                          depth,
-		                                          levels,
-		                                          1};
-		if (!tile || !TileGetTiledTextureLayout(description, volume_layout)) {
+		if (!tile || !TileGetTextureVolumeLayout(transfer_format, width, height, depth, levels,
+		                                         rt.attrib3.tile_mode, volume_layout)) {
 			EXIT("unsupported 3D render-target layout: %ux%ux%u levels=%u tile=%u\n", width, height,
-			     depth, levels, static_cast<uint32_t>(rt.attrib3.tile_mode));
+			     depth, levels, 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 (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;
+		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;
 		} else {
 			valid_layout =
 			    levels == 1 ? TileGetRenderTargetSize(width, height, pitch, bytes_per_element,
@@ -264,6 +237,12 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 		mip_sizes[0]  = {static_cast<uint32_t>(size), 0, 0, 0, 0, 0};
 		mip_padded[0] = {pitch, height};
 	}
+	if (rt.slice.slice_div64_minus1 != 0 &&
+	    (static_cast<uint64_t>(rt.slice.slice_div64_minus1) + 1u) * 64u != size) {
+		EXIT("render-target slice span mismatch: encoded=0x%016" PRIx64 " derived=0x%016" PRIx64
+		     "\n",
+		     (static_cast<uint64_t>(rt.slice.slice_div64_minus1) + 1u) * 64u, size);
+	}
 	if (size == 0 || (!volume && size > UINT64_MAX / view.image_layers)) {
 		EXIT("render-target memory footprint is invalid\n");
 	}
@@ -273,12 +252,18 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 	if (backing_size == 0) {
 		EXIT("render-target backing is empty\n");
 	}
-	if (!GuestRange {rt.base.addr, backing_size}.Valid()) {
+	if (backing_size > TRACKER_ADDRESS_SIZE - rt.base.addr) {
 		EXIT("render-target backing range is invalid\n");
 	}

 	const vk::Extent2D view_extent = {std::max(width >> rt.view.current_mip_level, 1u),
 	                                  std::max(height >> rt.view.current_mip_level, 1u)};
+	const uint32_t     view_depth  = std::max(depth >> rt.view.current_mip_level, 1u);
+	if (volume &&
+	    (view.base_layer >= view_depth || view.layer_count > view_depth - view.base_layer)) {
+		EXIT("3D render-target view exceeds mip depth: base=%u count=%u depth=%u mip=%u\n",
+		     view.base_layer, view.layer_count, view_depth, rt.view.current_mip_level);
+	}

 	auto decision_log_id = g_render_color_log_count.fetch_add(1);
 	if (decision_log_id < 128) {
@@ -286,9 +271,8 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 		     " 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,
-		     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");
+		     view_extent.width, view_extent.height, levels, pitch, rt.info.format,
+		     rt.info.channel_type, rt.info.channel_order, samples, tile ? "tiled" : "linear");
 	}

 	TextureCache::ImageDesc desc {};
@@ -296,32 +280,20 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 	desc.info.data         = {rt.base.addr, backing_size};
 	desc.info.pixel_format = target_format.format;
 	desc.info.guest_format = transfer_format;
-	desc.info.type         = image_type;
+	desc.info.type         = volume ? Prospero::ImageType::kColor3D : Prospero::ImageType::kColor2D;
 	desc.info.extent       = {width, height, depth};
 	desc.info.resources    = {levels, volume ? 1u : view.image_layers};
 	desc.info.pitch        = pitch;
 	desc.info.bytes_per_block = bytes_per_element;
 	desc.info.samples         = samples;
 	desc.info.tile_mode       = rt.attrib3.tile_mode;
-	const bool has_dcc        = rt.info.dcc_compression_enable && rt.dcc_addr.addr != 0;
-	if (has_dcc) {
-		TileSizeAlign metadata_size {};
-		(void)TileGetDccSize(width, height, volume ? depth : view.image_layers, bytes_per_element,
-		                     levels, rt.attrib3.tile_mode, metadata_size, rt.attrib.num_fragments);
-		desc.info.metadata.kind                     = ImageMetadataKind::Dcc;
-		desc.info.metadata.range                    = {rt.dcc_addr.addr, metadata_size.size};
-		desc.info.metadata.dcc_clear_word           = rt.clear_word0.word0;
-		desc.info.metadata.dcc_clear_register_valid = true;
-		desc.info.metadata.dcc_alpha_msb            = DccAlphaOnMsb(rt.info);
-	}
 	for (uint32_t level = 0; level < levels; level++) {
 		if (volume) {
-			const auto& mip             = volume_layout.mips[level];
 			desc.info.mip_layout[level] = {
-			    mip.offset,
-			    mip.size,
-			    mip.padded_width,
-			    mip.padded_height,
+			    volume_layout.level_offsets[level],
+			    volume_layout.level_sizes[level],
+			    volume_layout.level_widths[level],
+			    volume_layout.level_heights[level],
 			};
 			continue;
 		}
@@ -339,13 +311,8 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 		};
 	}
 	desc.view_info.format = target_format.format;
-	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.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;
@@ -354,10 +321,18 @@ void RenderExecutor::ResolveRenderColorTarget(CommandBuffer& buffer, RenderColor
 	desc.view_info.usage       = vk::ImageUsageFlagBits::eColorAttachment;
 	auto& texture_cache        = m_context.GetTextureCache();
 	r.desc                     = std::move(desc);
-	r.guest_mip_level          = rt.view.current_mip_level;
-	r.guest_array_layer        = view.base_layer;
 	r.image_id                 = texture_cache.FindImage(r.desc, exact_format);
+	r.type                     = RenderColorType::RenderTexture;
+	r.base_addr                = rt.base.addr;
+	r.image_view               = nullptr;
+	r.format                   = r.desc.view_info.format;
+	r.extent                   = view_extent;
+	r.base_mip_level           = rt.view.current_mip_level;
+	r.buffer_size              = backing_size;
+	r.samples                  = samples;
 	r.export_mapping           = target_format.export_mapping;
+	r.color_clear_enable       = false;
+	r.color_clear_value        = {};
 	BindRenderTarget(r.image_id);
 }

@@ -6,24 +6,33 @@
 #include "graphics/host_gpu/renderer/renderTarget.h"
 #include "graphics/host_gpu/vulkanCommon.h"

-#include <algorithm>
 #include <cstdint>

 namespace Libs::Graphics {

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

 } // namespace Libs::Graphics
@@ -1,92 +1,96 @@
 #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;

-namespace {
-
-void ReportVulkanFatal(const char* what, vk::Result result, uint64_t tick, uint32_t debug_op,
-                       uint64_t debug_submit, uint32_t arg0, uint32_t arg1, uint32_t arg2,
-                       uint32_t arg3, uint64_t arg4) {
-	LOGF("%s failed: %s (%d), tick=%" PRIu64 " debug_op=%u debug_submit=%" PRIu64
-	     " args=%u,%u,%u,%u,0x%016" PRIx64 "\n",
-	     what, vk::to_string(result).c_str(), static_cast<int>(result), tick, debug_op,
-	     debug_submit, arg0, arg1, arg2, arg3, arg4);
-	std::printf("%s failed: %s (%d), tick=%" PRIu64 " debug_op=%u debug_submit=%" PRIu64
-	            " args=%u,%u,%u,%u,0x%016" PRIx64 "\n",
-	            what, vk::to_string(result).c_str(), static_cast<int>(result), tick, debug_op,
-	            debug_submit, arg0, arg1, arg2, arg3, arg4);
-	std::fflush(stdout);
+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
+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;

-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::eTransient |
-	                          vk::CommandPoolCreateFlagBits::eResetCommandBuffer;
+	create.flags            = vk::CommandPoolCreateFlagBits::eResetCommandBuffer;
 	const auto result       = graphics.device.createCommandPool(&create, nullptr, &m_pool);
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess || m_pool == nullptr);
 }

-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);
+CommandSlot* CommandScheduler::CommandPool::CreateSlot() {
+	EXIT_IF(m_graphics == nullptr);
+	auto& graphics = *m_graphics;

 	vk::CommandBufferAllocateInfo allocate {};
+	allocate.sType              = vk::StructureType::eCommandBufferAllocateInfo;
 	allocate.commandPool        = m_pool;
 	allocate.level              = vk::CommandBufferLevel::ePrimary;
-	allocate.commandBufferCount = static_cast<uint32_t>(GrowStep);
-	EXIT_IF(m_graphics.device.allocateCommandBuffers(&allocate, m_buffers.data() + first) !=
-	        vk::Result::eSuccess);
-	return first;
+	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;
 }

-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;
-	};
+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;
+}

-	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());
+void CommandScheduler::CommandPool::Destroy() {
+	Common::LockGuard lock(m_mutex);
+	if (m_pool == nullptr) {
+		return;
 	}
-	if (!found) {
-		found = search(0, m_hint);
+	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           = Grow();
-		m_ticks[*found] = m_master.CurrentTick();
-	}
-
-	m_hint = (*found + 1) % m_ticks.size();
-	return m_buffers[*found];
+	m_graphics->device.destroyCommandPool(m_pool, nullptr);
+	m_slots.clear();
+	m_pool     = nullptr;
+	m_graphics = nullptr;
 }

 bool CommandScheduler::InDeferredOperation() noexcept {
@@ -95,7 +99,6 @@ 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() {
@@ -122,11 +125,9 @@ void CommandScheduler::Shutdown() {
 		}
 		m_operation_state = OperationState::Draining;
 	}
-	if (!m_command.IsInvalid()) {
-		Submit();
+	if (Active() && m_recording) {
+		Finish();
 	}
-	m_master.Wait(CurrentTick() - 1);
-	PopPendingOperations();
 	DrainPriorityOperations();
 	m_priority_thread.request_stop();
 	m_operation_available.notify_all();
@@ -148,10 +149,21 @@ void CommandScheduler::Begin(HW::Context& registers, HW::UserConfig& user_config
 		std::lock_guard lock(m_operation_mutex);
 		EXIT_IF(m_operation_state != OperationState::Open);
 	}
-	m_command.Bind(registers, user_config, shaders);
+	m_registers   = &registers;
+	m_user_config = &user_config;
+	m_shaders     = &shaders;

-	if (m_command.IsInvalid()) {
-		BeginNext();
+	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;
 	}
 }

@@ -160,7 +172,7 @@ void CommandScheduler::BeginRendering(const RenderState& state) {
 }

 void CommandScheduler::EndRendering() {
-	if (Active() && !m_command.IsInvalid()) {
+	if (Active() && m_recording) {
 		Current().EndRendering();
 	}
 }
@@ -171,57 +183,71 @@ void CommandScheduler::Flush() {
 }

 void CommandScheduler::Flush(SubmitInfo& submit) {
-	Submit(submit);
+	SubmitCurrent(submit);
 	BeginNext();
 }

-void CommandScheduler::FlushAndWait() {
-	const auto tick = Submit();
-	m_master.Wait(tick);
+CommandBuffer& CommandScheduler::FlushAndGetSubmitted() {
+	SubmitInfo submit;
+	auto&      submitted = SubmitCurrent(submit);
 	BeginNext();
+	return submitted;
 }

 void CommandScheduler::Finish() {
 	CheckActive();
-	if (!m_command.IsInvalid()) {
-		Submit();
+	const auto tick = CurrentTick();
+	if (m_recording) {
+		SubmitInfo submit;
+		SubmitCurrent(submit);
 	}
-	m_master.Wait(CurrentTick() - 1);
-	BeginNext();
+	for (auto& buffer: m_buffers) {
+		buffer->WaitForFenceAndReset();
+	}
+	m_master.Wait(tick);
 	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()) {
-		CheckActive();
+	if (tick >= CurrentTick()) {
 		// A stream-buffer wrap can wait while a draw is being prepared through a reference to
-		// 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);
+		// Current(). Recycle the same command object so that reference remains valid.
+		FinishCurrent();
+		return;
 	}
+	m_master.Wait(tick);
+	PopPendingOperations();
 }

 void CommandScheduler::PopPendingOperations() {
 	m_master.Refresh();
 	for (;;) {
-		PendingOperation operation;
+		Common::UniqueFunction<void> callback;
 		{
 			std::lock_guard lock(m_operation_mutex);
 			if (m_pending_operations.empty() ||
 			    !m_master.IsFree(m_pending_operations.front().tick)) {
 				return;
 			}
-			operation = std::move(m_pending_operations.front());
+			callback = std::move(m_pending_operations.front().callback);
 			m_pending_operations.pop();
 		}
-		WaitPriorityOperations(operation.tick);
-		RunOperation(std::move(operation.callback));
+		RunOperation(std::move(callback));
 	}
 }

@@ -327,73 +353,49 @@ 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());
+	EXIT_IF(!Active() || m_current >= BufferCount);
 }

-CommandBuffer& CommandScheduler::Current() {
+RenderCommandBuffer& CommandScheduler::Current() const {
 	CheckActive();
-	return m_command;
+	EXIT_IF(m_buffers[m_current] == nullptr);
+	return *m_buffers[m_current];
 }

-CommandBuffer& CommandScheduler::BeginCommand() {
-	EXIT_IF(!m_command.IsInvalid());
-	m_command.m_buffer = m_command_pool.Commit();
-	m_command.Begin();
-	return m_command;
+void CommandScheduler::BindCurrent() const {
+	EXIT_IF(m_registers == nullptr || m_user_config == nullptr || m_shaders == nullptr);
+	Current().Bind(*m_registers, *m_user_config, *m_shaders);
 }

-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.waitSemaphoreValueCount   = submit.num_wait_semaphores;
-		timeline_info.pWaitSemaphoreValues      = submit.wait_ticks.data();
-		timeline_info.signalSemaphoreValueCount = submit.num_signal_semaphores;
-		timeline_info.pSignalSemaphoreValues    = submit.signal_ticks.data();
-
-		vk::SubmitInfo submit_info {};
-		submit_info.pNext                = &timeline_info;
-		submit_info.waitSemaphoreCount   = submit.num_wait_semaphores;
-		submit_info.pWaitSemaphores      = submit.wait_semaphores.data();
-		submit_info.pWaitDstStageMask    = submit.wait_stages.data();
-		submit_info.commandBufferCount   = 1;
-		submit_info.pCommandBuffers      = &buffer;
-		submit_info.signalSemaphoreCount = submit.num_signal_semaphores;
-		submit_info.pSignalSemaphores    = submit.signal_semaphores.data();
-
-		result = graphics.queue.submit(1, &submit_info, nullptr);
-	}
-
-	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;
+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;
 }

 void CommandScheduler::BeginNext() {
-	CheckActive();
-	BeginCommand();
+	EXIT_IF(m_recording);
+	m_current = (m_current + 1) % BufferCount;
+	Current().WaitForFenceAndReset();
+	PopPendingOperations();
+	BindCurrent();
+	Current().Begin();
+	m_recording = true;
 }

 } // namespace Libs::Graphics
@@ -6,18 +6,33 @@
 #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);
@@ -27,10 +42,9 @@ public:
 	void           EndRendering();
 	void           Flush();
 	void           Flush(SubmitInfo& submit);
-	void           FlushAndWait();
+	CommandBuffer& FlushAndGetSubmitted();
 	void           Finish();
-	CommandBuffer& BeginCommand();
-	uint64_t       Submit(SubmitInfo submit = {});
+	void           FinishCurrent();
 	// Deferred callbacks can observe an externally owned drain, but cannot initiate shutdown:
 	// the priority runner cannot join itself.
 	void                      Shutdown();
@@ -42,35 +56,32 @@ public:
 	void                      DeferPriorityOperation(Common::UniqueFunction<void>&& operation);
 	[[nodiscard]] static bool InDeferredOperation() noexcept;

-	[[nodiscard]] bool Active() const noexcept { return m_command.m_registers != nullptr; }
-	void                           CheckActive() const;
-	CommandBuffer&                 Current();
-	[[nodiscard]] uint64_t         CurrentTick() const noexcept { return m_master.CurrentTick(); }
-	[[nodiscard]] bool             IsFree(uint64_t tick);
-	[[nodiscard]] MasterSemaphore& GetMasterSemaphore() noexcept { return m_master; }
-	[[nodiscard]] RenderContext&   Context() const noexcept { return m_context; }
-	[[nodiscard]] GraphicContext&  Graphics() const noexcept { return m_graphics; }
+	[[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; }

 private:
 	class CommandPool {
 	public:
-		CommandPool(GraphicContext& graphics, MasterSemaphore& master);
+		CommandPool() = default;
 		~CommandPool();
 		KYTY_CLASS_NO_COPY(CommandPool);

-		vk::CommandBuffer Commit();
+		CommandSlot* Allocate(GraphicContext& graphics);

 	private:
-		static constexpr size_t GrowStep = 4;
+		void         Create(GraphicContext& graphics);
+		CommandSlot* CreateSlot();
+		void         Destroy();

-		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;
+		GraphicContext*         m_graphics = nullptr;
+		Common::Mutex           m_mutex;
+		vk::CommandPool         m_pool = nullptr;
+		std::deque<CommandSlot> m_slots;
 	};

 	enum class OperationState { Open, Draining, Closed };
@@ -80,23 +91,35 @@ private:
 		uint64_t                     tick = 0;
 	};

-	void BeginNext();
-	void PriorityOperationsThread(std::stop_token stop);
-	void RunOperation(Common::UniqueFunction<void>&& operation);
+	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;

-	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;
+	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;
 };

 } // namespace Libs::Graphics
@@ -1,5 +1,7 @@
 #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"
@@ -7,33 +9,108 @@
 #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"
 #include "graphics/host_gpu/vulkanCommon.h"

 #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_graphics(scheduler.Graphics()) {}
+    : m_context(scheduler.Context()), m_scheduler(scheduler), m_graphics(scheduler.Graphics()),
+      m_slot(scheduler.AllocateCommandBuffer()) {}
+
+CommandBuffer::~CommandBuffer() {
+	Release();
+}

 bool CommandBuffer::IsInvalid() const {
-	return m_buffer == nullptr;
+	return m_slot == nullptr;
 }

 vk::CommandBuffer CommandBuffer::Handle() const {
 	EXIT_IF(IsInvalid());
-	return m_buffer;
+
+	const auto handle = m_slot->buffer;
+	EXIT_IF(handle == nullptr);
+	return handle;
 }

-void CommandBuffer::Begin() {
-	EXIT_IF(m_rendering || IsInvalid());
+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);
 	auto buffer = Handle();

 	vk::CommandBufferBeginInfo begin_info {};
-	begin_info.flags = vk::CommandBufferUsageFlagBits::eOneTimeSubmit;
+	begin_info.sType            = vk::StructureType::eCommandBufferBeginInfo;
+	begin_info.pNext            = nullptr;
+	begin_info.flags            = {};
+	begin_info.pInheritanceInfo = nullptr;

 	auto result = buffer.begin(&begin_info);

@@ -60,17 +137,137 @@ 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);
 	if (m_rendering && m_render_state == state) {
 		return;
 	}
-	EXIT_IF(state.width == 0 || state.height == 0 || state.num_layers == 0 ||
-	        state.num_color_attachments > RENDER_COLOR_ATTACHMENTS_MAX);
 	EndRendering();

 	std::array<vk::RenderingAttachmentInfo, RENDER_COLOR_ATTACHMENTS_MAX> colors {};
 	for (uint32_t i = 0; i < state.num_color_attachments; i++) {
 		const auto& attachment = state.color_attachments[i];
+		colors[i].sType        = vk::StructureType::eRenderingAttachmentInfo;
 		colors[i].imageView    = attachment.image_view;
 		colors[i].imageLayout  = attachment.image_layout;
 		colors[i].loadOp =
@@ -81,6 +278,7 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {

 	const auto&                 depth_stencil = state.depth_stencil_attachment;
 	vk::RenderingAttachmentInfo depth {};
+	depth.sType       = vk::StructureType::eRenderingAttachmentInfo;
 	depth.imageView   = depth_stencil.image_view;
 	depth.imageLayout = depth_stencil.image_layout;
 	depth.loadOp =
@@ -89,6 +287,7 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {
 	depth.clearValue.depthStencil.depth = std::bit_cast<float>(depth_stencil.clear_value[0]);

 	vk::RenderingAttachmentInfo stencil {};
+	stencil.sType       = vk::StructureType::eRenderingAttachmentInfo;
 	stencil.imageView   = depth_stencil.image_view;
 	stencil.imageLayout = depth_stencil.image_layout;
 	stencil.loadOp =
@@ -97,6 +296,7 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {
 	stencil.clearValue.depthStencil.stencil = depth_stencil.clear_value[1];

 	vk::RenderingInfo rendering {};
+	rendering.sType                = vk::StructureType::eRenderingInfo;
 	rendering.renderArea.extent    = {state.width, state.height};
 	rendering.layerCount           = state.num_layers;
 	rendering.colorAttachmentCount = state.num_color_attachments;
@@ -4,18 +4,20 @@
 #include "common/common.h"
 #include "common/emulatorConfig.h"
 #include "common/logging/log.h"
+#include "common/stringUtils.h"
 #include "graphics/guest_gpu/gpu_defs.h"
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/host_gpu/renderer/render.h"

 #include <algorithm>
-#include <array>
 #include <atomic>
 #include <cmath>
 #include <fmt/format.h>

 namespace Libs::Graphics {

+static std::atomic<uint32_t> g_scissor_default_log_count = 0;
+
 uint32_t render_target_mask_slot(uint32_t mask, uint32_t slot) {
 	return (mask >> (slot * 4u)) & 0x0fu;
 }
@@ -24,7 +26,7 @@ static bool RenderTargetMaskHasMrt(uint32_t mask) {
 	return (mask & ~0x0fu) != 0;
 }

-static bool RenderTargetMaskHasBoundMrt(const CommandBuffer& buffer) {
+static bool RenderTargetMaskHasBoundMrt(const RenderCommandBuffer& buffer) {
 	const auto& hw   = buffer.GetRegisters();
 	const auto  mask = hw.GetRenderTargetMask();

@@ -42,7 +44,7 @@ static bool RenderTargetMaskHasBoundMrt(const CommandBuffer& buffer) {
 	return bound_targets > 1;
 }

-uint32_t render_target_first_bound_slot(const CommandBuffer& buffer) {
+uint32_t render_target_first_bound_slot(const RenderCommandBuffer& buffer) {
 	const auto& hw   = buffer.GetRegisters();
 	const auto  mask = hw.GetRenderTargetMask();
 	for (uint32_t i = 0; i < 8; i++) {
@@ -56,7 +58,7 @@ uint32_t render_target_first_bound_slot(const CommandBuffer& buffer) {

 bool graphics_debug_dump_enabled() {
 	return Config::GraphicsDebugDumpEnabled() &&
-	       Config::GetPrintfDirection() != Config::LogDirection::Silent;
+	       Config::GetPrintfDirection() != Config::OutputDirection::Silent;
 }

 void uc_print(const char* func, const HW::UserConfig& uc) {
@@ -68,115 +70,163 @@ 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",
-	     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");
+	     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");
 }

 void uc_check(const HW::UserConfig& uc) {
+	const auto& ge_cntl = uc.GetGeControl();
 	const auto& user_en = uc.GetGeUserVgprEn();

+	if (ge_cntl.primitive_group_size > 0x0040) {
+		LOGF("\t warning: unsupported GE_CNTL primitive_group_size = 0x%04" PRIx16 "\n",
+		     ge_cntl.primitive_group_size);
+	}
+	if (ge_cntl.vertex_group_size > 0x0040) {
+		LOGF("\t warning: unsupported GE_CNTL vertex_group_size = 0x%04" PRIx16 "\n",
+		     ge_cntl.vertex_group_size);
+	}
 	EXIT_NOT_IMPLEMENTED(user_en.vgpr1 != false);
 	EXIT_NOT_IMPLEMENTED(user_en.vgpr2 != false);
 	EXIT_NOT_IMPLEMENTED(user_en.vgpr3 != false);
 }

-std::string rt_print(const char* func, const HW::RenderTarget& rt) {
-	std::string dst;
-	dst.reserve(4096);
+void sh_print(const char* func, const HW::Shader& /*uc*/) {
+	LOGF("%s\n", func);
+}

-	dst += fmt::format("{}\n", func);
+std::vector<std::string> rt_print(const char* func, const HW::RenderTarget& rt) {
+	std::vector<std::string> dst;
+	dst.reserve(53);

-	dst += fmt::format("\t base.addr                       = 0x{:016x}\n", rt.base.addr);
-	dst += fmt::format("\t view.base_array_slice_index     = 0x{:08x}\n",
-	                  rt.view.base_array_slice_index);
-	dst += fmt::format("\t view.last_array_slice_index     = 0x{:08x}\n",
-	                  rt.view.last_array_slice_index);
-	dst += fmt::format("\t view.current_mip_level          = 0x{:08x}\n", rt.view.current_mip_level);
-	dst += fmt::format("\t info.fmask_compression_enable   = {}\n",
-	                  rt.info.fmask_compression_enable ? "true" : "false");
+	dst.push_back(fmt::format("{}\n", func));

-	dst += fmt::format("\t info.fmask_data_compression_disable = {}\n",
-	                  rt.info.fmask_data_compression_disable ? "true" : "false");
-	dst += fmt::format("\t info.fmask_one_frag_mode        = {}\n",
-	                  rt.info.fmask_one_frag_mode ? "true" : "false");
+	dst.push_back(fmt::format("\t base.addr                       = 0x{:016x}\n", rt.base.addr));
+	dst.push_back(
+	    fmt::format("\t pitch.pitch_div8_minus1         = 0x{:08x}\n", rt.pitch.pitch_div8_minus1));
+	dst.push_back(fmt::format("\t pitch.fmask_pitch_div8_minus1   = 0x{:08x}\n",
+	                          rt.pitch.fmask_pitch_div8_minus1));
+	dst.push_back(fmt::format("\t slice.slice_div64_minus1        = 0x{:08x}\n",
+	                          rt.slice.slice_div64_minus1));
+	dst.push_back(fmt::format("\t view.base_array_slice_index     = 0x{:08x}\n",
+	                          rt.view.base_array_slice_index));
+	dst.push_back(fmt::format("\t view.last_array_slice_index     = 0x{:08x}\n",
+	                          rt.view.last_array_slice_index));
+	dst.push_back(
+	    fmt::format("\t view.current_mip_level          = 0x{:08x}\n", rt.view.current_mip_level));
+	dst.push_back(fmt::format("\t info.fmask_compression_enable   = {}\n",
+	                          rt.info.fmask_compression_enable ? "true" : "false"));

-	dst += fmt::format("\t info.cmask_fast_clear_enable    = {}\n",
-	                  rt.info.cmask_fast_clear_enable ? "true" : "false");
-	dst += fmt::format("\t info.dcc_compression_enable     = {}\n",
-	                  rt.info.dcc_compression_enable ? "true" : "false");
-	dst += fmt::format("\t info.format                     = 0x{:08x}\n",
-	                  static_cast<uint32_t>(rt.info.format));
-	dst += fmt::format("\t info.channel_type               = 0x{:08x}\n",
-	                  static_cast<uint32_t>(rt.info.channel_type));
-	dst += fmt::format("\t info.channel_order              = 0x{:08x}\n",
-	                  static_cast<uint32_t>(rt.info.channel_order));
-	dst += fmt::format("\t info.blend_bypa                 = {}\n",
-	                  rt.info.blend_bypass ? "true" : "false");
-	dst += fmt::format("\t info.blend_clamp                = {}\n",
-	                  rt.info.blend_clamp ? "true" : "false");
-	dst += fmt::format("\t info.round_mode                 = {}\n",
-	                  rt.info.round_mode ? "true" : "false");
-	dst += fmt::format("\t attrib.force_dest_alpha_to_one  = {}\n",
-	                  rt.attrib.force_dest_alpha_to_one ? "true" : "false");
-	dst += fmt::format("\t attrib.num_samples              = 0x{:08x}\n", rt.attrib.num_samples);
-	dst += fmt::format("\t attrib.num_fragments            = 0x{:08x}\n", rt.attrib.num_fragments);
-	dst += fmt::format("\t attrib2.width                   = 0x{:08x}\n", rt.attrib2.width);
-	dst += fmt::format("\t attrib2.height                  = 0x{:08x}\n", rt.attrib2.height);
-	dst += fmt::format("\t attrib2.num_mip_levels          = 0x{:08x}\n", rt.attrib2.num_mip_levels);
-	dst += fmt::format("\t attrib3.depth                   = 0x{:08x}\n", rt.attrib3.depth);
-	dst += fmt::format("\t attrib3.tile_mode               = 0x{:08x}\n",
-	                  static_cast<uint32_t>(rt.attrib3.tile_mode));
-	dst += fmt::format("\t attrib3.dimension               = 0x{:08x}\n", rt.attrib3.dimension);
-	dst += fmt::format("\t attrib3.metadata_pipe_aligned   = {}\n",
-	                  rt.attrib3.metadata_pipe_aligned ? "true" : "false");
-	dst += fmt::format("\t attrib3.write_vrs_rate_hint_to_cmask = {}\n",
-	                  rt.attrib3.write_vrs_rate_hint_to_cmask ? "true" : "false");
-	dst += fmt::format("\t dcc.max_uncompressed_block_size = 0x{:08x}\n",
-	                  rt.dcc.max_uncompressed_block_size);
-	dst += fmt::format("\t dcc.max_compressed_block_size   = 0x{:08x}\n",
-	                  rt.dcc.max_compressed_block_size);
-	dst += fmt::format("\t dcc.color_transform             = 0x{:08x}\n", rt.dcc.color_transform);
-	dst += fmt::format("\t dcc.overwrite_combiner_disable  = {}\n",
-	                  rt.dcc.overwrite_combiner_disable ? "true" : "false");
-	dst += fmt::format("\t dcc.independent_block_size      = 0x{:02x}\n",
-	                  static_cast<uint8_t>(rt.dcc.independent_block_size));
-	dst += fmt::format("\t data_write_on_dcc_clear_to_reg  = {}\n",
-	                  rt.dcc.data_write_on_dcc_clear_to_reg ? "true" : "false");
-	dst += fmt::format("\t dcc.dcc_clear_key_enable        = {}\n",
-	                  rt.dcc.dcc_clear_key_enable ? "true" : "false");
-	dst += fmt::format("\t cmask.addr                      = 0x{:016x}\n", rt.cmask.addr);
-	dst += fmt::format("\t fmask.addr                      = 0x{:016x}\n", rt.fmask.addr);
-	dst += fmt::format("\t clear_word0.word0               = 0x{:08x}\n", rt.clear_word0.word0);
-	dst += fmt::format("\t clear_word1.word1               = 0x{:08x}\n", rt.clear_word1.word1);
-	dst += fmt::format("\t dcc_addr.addr                   = 0x{:016x}\n", rt.dcc_addr.addr);
+	// dst.push_back(fmt::format("\t info.fmask_compression_mode     = 0x{:08x}\n", //
+	// rt.info.fmask_compression_mode));
+	dst.push_back(fmt::format("\t info.fmask_data_compression_disable = {}\n",
+	                          rt.info.fmask_data_compression_disable ? "true" : "false"));
+	dst.push_back(fmt::format("\t info.fmask_one_frag_mode        = {}\n",
+	                          rt.info.fmask_one_frag_mode ? "true" : "false"));
+
+	dst.push_back(fmt::format("\t info.cmask_fast_clear_enable    = {}\n",
+	                          rt.info.cmask_fast_clear_enable ? "true" : "false"));
+	dst.push_back(fmt::format("\t info.dcc_compression_enable     = {}\n",
+	                          rt.info.dcc_compression_enable ? "true" : "false"));
+	dst.push_back(
+	    fmt::format("\t info.cmask_is_linear            = 0x{:08x}\n", rt.info.cmask_is_linear));
+	dst.push_back(
+	    fmt::format("\t info.cmask_addr_type            = 0x{:08x}\n", rt.info.cmask_addr_type));
+	dst.push_back(fmt::format("\t info.alt_tile_mode              = {}\n",
+	                          rt.info.alt_tile_mode ? "true" : "false"));
+	dst.push_back(fmt::format("\t info.format                     = 0x{:08x}\n", 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.blend_bypa                 = {}\n",
+	                          rt.info.blend_bypass ? "true" : "false"));
+	dst.push_back(fmt::format("\t info.blend_clamp                = {}\n",
+	                          rt.info.blend_clamp ? "true" : "false"));
+	dst.push_back(fmt::format("\t info.round_mode                 = {}\n",
+	                          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.num_samples              = 0x{:08x}\n", rt.attrib.num_samples));
+	dst.push_back(
+	    fmt::format("\t attrib.num_fragments            = 0x{:08x}\n", rt.attrib.num_fragments));
+	dst.push_back(fmt::format("\t attrib2.width                   = 0x{:08x}\n", rt.attrib2.width));
+	dst.push_back(
+	    fmt::format("\t attrib2.height                  = 0x{:08x}\n", rt.attrib2.height));
+	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.dimension               = 0x{:08x}\n", rt.attrib3.dimension));
+	dst.push_back(fmt::format("\t attrib3.cmask_pipe_aligned      = {}\n",
+	                          rt.attrib3.cmask_pipe_aligned ? "true" : "false"));
+	dst.push_back(fmt::format("\t attrib3.dcc_pipe_aligned        = {}\n",
+	                          rt.attrib3.dcc_pipe_aligned ? "true" : "false"));
+	dst.push_back(fmt::format("\t dcc.max_uncompressed_block_size = 0x{:08x}\n",
+	                          rt.dcc.max_uncompressed_block_size));
+	dst.push_back(fmt::format("\t dcc.max_compressed_block_size   = 0x{:08x}\n",
+	                          rt.dcc.max_compressed_block_size));
+	dst.push_back(fmt::format("\t dcc.min_compressed_block_size   = 0x{:08x}\n",
+	                          rt.dcc.min_compressed_block_size));
+	dst.push_back(
+	    fmt::format("\t dcc.color_transform             = 0x{:08x}\n", rt.dcc.color_transform));
+	dst.push_back(fmt::format("\t dcc.overwrite_combiner_disable  = {}\n",
+	                          rt.dcc.overwrite_combiner_disable ? "true" : "false"));
+	dst.push_back(fmt::format("\t dcc.independent_64b_blocks      = {}\n",
+	                          rt.dcc.independent_64b_blocks ? "true" : "false"));
+	dst.push_back(fmt::format("\t dcc.independent_128b_blocks     = {}\n",
+	                          rt.dcc.independent_128b_blocks ? "true" : "false"));
+	dst.push_back(fmt::format("\t data_write_on_dcc_clear_to_reg  = {}\n",
+	                          rt.dcc.data_write_on_dcc_clear_to_reg ? "true" : "false"));
+	dst.push_back(fmt::format("\t dcc.dcc_clear_key_enable        = {}\n",
+	                          rt.dcc.dcc_clear_key_enable ? "true" : "false"));
+	dst.push_back(fmt::format("\t cmask.addr                      = 0x{:016x}\n", rt.cmask.addr));
+	dst.push_back(fmt::format("\t cmask_slice.slice_minus1        = 0x{:08x}\n",
+	                          rt.cmask_slice.slice_minus1));
+	dst.push_back(fmt::format("\t fmask.addr                      = 0x{:016x}\n", rt.fmask.addr));
+	dst.push_back(fmt::format("\t fmask_slice.slice_minus1        = 0x{:08x}\n",
+	                          rt.fmask_slice.slice_minus1));
+	dst.push_back(
+	    fmt::format("\t clear_word0.word0               = 0x{:08x}\n", rt.clear_word0.word0));
+	dst.push_back(
+	    fmt::format("\t clear_word1.word1               = 0x{:08x}\n", rt.clear_word1.word1));
+	dst.push_back(
+	    fmt::format("\t dcc_addr.addr                   = 0x{:016x}\n", rt.dcc_addr.addr));
+	dst.push_back(fmt::format("\t size.width                      = 0x{:08x}\n", rt.size.width));
+	dst.push_back(fmt::format("\t size.height                     = 0x{:08x}\n", rt.size.height));

 	return dst;
 }

-static bool RenderIsColorTileMode(Prospero::TileMode tile_mode) {
+static bool RenderIsColorTileMode(uint32_t tile_mode) {
 	// AGC CxRenderTarget::TileMode shifted by CB_COLOR*_ATTRIB3.COLOR_SW_MODE.
 	switch (tile_mode) {
-		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;
+		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;
 		default: return false;
 	}
 }

-bool RenderIsColorTileModeLinear(Prospero::TileMode tile_mode) {
-	return tile_mode == Prospero::TileMode::kLinear;
+bool RenderIsColorTileModeLinear(uint32_t tile_mode) {
+	return tile_mode == Prospero::GpuEnumValue(Prospero::TileMode::kLinear);
 }

 static bool RenderIsColorDimension(uint32_t dimension) {
@@ -186,8 +236,13 @@ static bool RenderIsColorDimension(uint32_t dimension) {
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
 static void RtCheck(const HW::RenderTarget& rt) {
 	if (rt.base.addr != 0) {
+		// bool render_to_texture = (rt.attrib.tile_mode == 0x0d);
 		//  EXIT_NOT_IMPLEMENTED(rt.base_addr == 0);

+		EXIT_NOT_IMPLEMENTED(rt.pitch.pitch_div8_minus1 != 0);
+		EXIT_NOT_IMPLEMENTED(rt.pitch.fmask_pitch_div8_minus1 != 0);
+		EXIT_NOT_IMPLEMENTED(rt.slice.slice_div64_minus1 != 0);
+
 		EXIT_NOT_IMPLEMENTED(rt.view.base_array_slice_index > rt.view.last_array_slice_index);
 		if (rt.view.base_array_slice_index != 0x00000000 ||
 		    rt.view.last_array_slice_index != 0x00000000) {
@@ -211,14 +266,16 @@ static void RtCheck(const HW::RenderTarget& rt) {
 			EXIT_NOT_IMPLEMENTED(rt.attrib.num_samples == 0 && rt.attrib.num_fragments == 0);
 			static bool logged = false;
 			if (!logged) {
-				LOGF("RenderTarget: using native Vulkan MSAA without guest FMASK metadata, "
-				     "fmask=0x%016" PRIx64 "\n",
+				LOGF("RenderTarget: temporary: ignoring PS5 FMASK metadata for single-sample MSAA "
+				     "fallback, 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;
@@ -231,6 +288,22 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
+		if (rt.info.cmask_is_linear != 0x00000000 || rt.info.cmask_addr_type != 0x00000000) {
+			static bool logged = false;
+			if (!logged) {
+				LOGF("RenderTarget: temporary: ignoring PS5 CMASK metadata fields "
+				     "is_linear=%u addr_type=%u\n",
+				     rt.info.cmask_is_linear, rt.info.cmask_addr_type);
+				logged = true;
+			}
+		}
+		if (rt.info.alt_tile_mode) {
+			static bool logged = false;
+			if (!logged) {
+				LOGF("RenderTarget: temporary: ignoring alternate tile mode flag\n");
+				logged = true;
+			}
+		}
 		if (rt.info.blend_bypass) {
 			static bool logged = false;
 			if (!logged) {
@@ -257,6 +330,8 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
+		// EXIT_NOT_IMPLEMENTED(rt.tile_mode != 0x0000000a);
+		// EXIT_NOT_IMPLEMENTED(rt.fmask_tile_mode != 0x0000000a);
 		if (rt.attrib.num_samples != 0x00000000 || rt.attrib.num_fragments != 0x00000000) {
 			static bool logged = false;
 			if (!logged) {
@@ -285,28 +360,39 @@ static void RtCheck(const HW::RenderTarget& rt) {
 			}
 		}
 		if (!RenderIsColorTileMode(rt.attrib3.tile_mode)) {
-			EXIT("unknown PS5 render-target tile mode: 0x%08" PRIx32 "\n",
-			     static_cast<uint32_t>(rt.attrib3.tile_mode));
+			EXIT("unknown PS5 render-target tile mode: 0x%08" PRIx32 "\n", rt.attrib3.tile_mode);
 		}
 		if (!RenderIsColorDimension(rt.attrib3.dimension)) {
 			EXIT("unknown PS5 render-target dimension: 0x%08" PRIx32 "\n", rt.attrib3.dimension);
 		}
-		if (!rt.attrib3.metadata_pipe_aligned) {
+		if (!rt.attrib3.cmask_pipe_aligned) {
 			static bool logged = false;
 			if (!logged) {
-				LOGF("RenderTarget: temporary: accepting unaligned PS5 metadata pipe flag\n");
+				LOGF("RenderTarget: temporary: accepting unaligned PS5 CMASK pipe flag\n");
+				logged = true;
+			}
+		}
+		if (!rt.attrib3.dcc_pipe_aligned) {
+			static bool logged = false;
+			if (!logged) {
+				LOGF("RenderTarget: temporary: accepting unaligned PS5 DCC pipe flag\n");
 				logged = true;
 			}
 		}
-		EXIT_NOT_IMPLEMENTED(rt.attrib3.write_vrs_rate_hint_to_cmask);

 		// EXIT_NOT_IMPLEMENTED(rt.dcc_max_uncompressed_block_size != 0x00000002);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.max_compressed_block_size != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(rt.dcc.min_compressed_block_size != 0x00000000);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.color_transform != 0x00000000);
 		EXIT_NOT_IMPLEMENTED(rt.dcc.overwrite_combiner_disable != false);
+		// EXIT_NOT_IMPLEMENTED(rt.dcc.force_independent_blocks != false);
+		// EXIT_NOT_IMPLEMENTED(rt.dcc.independent_128b_blocks != false);
 		// EXIT_NOT_IMPLEMENTED(rt.dcc.data_write_on_dcc_clear_to_reg != false);
 		EXIT_NOT_IMPLEMENTED(rt.dcc.dcc_clear_key_enable != false);
-		if (rt.cmask.addr != 0x0000000000000000 || rt.fmask.addr != 0x0000000000000000 ||
+		if (rt.cmask.addr != 0x0000000000000000 || rt.cmask_slice.slice_minus1 != 0x00000000 ||
+		    rt.fmask.addr != 0x0000000000000000 ||
+		    (rt.fmask_slice.slice_minus1 != 0x00000000 &&
+		     rt.fmask_slice.slice_minus1 != rt.slice.slice_div64_minus1) ||
 		    rt.dcc_addr.addr != 0x0000000000000000) {
 			static bool logged = false;
 			if (!logged) {
@@ -316,6 +402,9 @@ static void RtCheck(const HW::RenderTarget& rt) {
 				logged = true;
 			}
 		}
+
+		EXIT_NOT_IMPLEMENTED(rt.size.width != 0);
+		EXIT_NOT_IMPLEMENTED(rt.size.height != 0);
 	}
 }

@@ -335,37 +424,160 @@ static void ZPrint(const char* func, const HW::DepthRenderTarget& z) {
 	     "\t stencil_info.htile_stencil_disabled   = %s\n"
 	     "\t stencil_info.expclear_enabled         = %s\n"
 	     "\t stencil_info.partially_resident       = %s\n"
+	     "\t depth_info.addr5_swizzle_mask         = 0x%08" PRIx32 "\n"
+	     "\t depth_info.array_mode                 = 0x%08" PRIx32 "\n"
+	     "\t depth_info.pipe_config                = 0x%08" PRIx32 "\n"
+	     "\t depth_info.bank_width                 = 0x%08" PRIx32 "\n"
+	     "\t depth_info.bank_height                = 0x%08" PRIx32 "\n"
+	     "\t depth_info.macro_tile_aspect          = 0x%08" PRIx32 "\n"
+	     "\t depth_info.num_banks                  = 0x%08" PRIx32 "\n"
 	     "\t depth_view.slice_start                = 0x%08" PRIx32 "\n"
 	     "\t depth_view.slice_max                  = 0x%08" PRIx32 "\n"
 	     "\t depth_view.current_mip_level          = 0x%02" PRIx8 "\n"
 	     "\t depth_view.depth_write_disable        = %s\n"
 	     "\t depth_view.stencil_write_disable      = %s\n"
+	     "\t htile_surface.linear                  = 0x%08" PRIx32 "\n"
+	     "\t htile_surface.full_cache              = 0x%08" PRIx32 "\n"
+	     "\t htile_surface.htile_uses_preload_win  = 0x%08" PRIx32 "\n"
+	     "\t htile_surface.preload                 = 0x%08" PRIx32 "\n"
+	     "\t htile_surface.prefetch_width          = 0x%08" PRIx32 "\n"
+	     "\t htile_surface.prefetch_height         = 0x%08" PRIx32 "\n"
+	     "\t htile_surface.dst_outside_zero_to_one = 0x%08" PRIx32 "\n"
 	     "\t z_read_base_addr                      = 0x%016" PRIx64 "\n"
 	     "\t stencil_read_base_addr                = 0x%016" PRIx64 "\n"
 	     "\t z_write_base_addr                     = 0x%016" PRIx64 "\n"
 	     "\t stencil_write_base_addr               = 0x%016" PRIx64 "\n"
+	     "\t pitch_div8_minus1                     = 0x%08" PRIx32 "\n"
+	     "\t height_div8_minus1                    = 0x%08" PRIx32 "\n"
+	     "\t slice_div64_minus1                    = 0x%08" PRIx32 "\n"
 	     "\t htile_data_base_addr                  = 0x%016" PRIx64 "\n"
-	     "\t shading_rate_encoding                 = 0x%02" PRIx8 "\n"
+	     "\t width                                 = 0x%08" PRIx32 "\n"
+	     "\t height                                = 0x%08" PRIx32 "\n"
 	     "\t size.x_max                            = 0x%04" PRIx16 "\n"
-	     "\t size.y_max                            = 0x%04" PRIx16 "\n"
-	     "\t size.valid                            = %s\n",
-	     static_cast<uint32_t>(z.z_info.format), z.z_info.num_samples,
-	     static_cast<uint32_t>(z.z_info.texture_compatibility),
+	     "\t size.y_max                            = 0x%04" PRIx16 "\n",
+	     z.z_info.format, z.z_info.num_samples,
+	     Prospero::GpuEnumValue(z.z_info.texture_compatibility),
 	     z.z_info.htile_acceleration ? "true" : "false",
 	     z.z_info.expclear_enabled ? "true" : "false",
-	     static_cast<uint32_t>(z.z_info.z_compare_base),
+	     Prospero::GpuEnumValue(z.z_info.z_compare_base),
 	     z.z_info.partially_resident ? "true" : "false", z.z_info.max_mip_level,
-	     static_cast<uint32_t>(z.stencil_info.format),
-	     static_cast<uint32_t>(z.stencil_info.texture_compatibility),
+	     z.stencil_info.format, Prospero::GpuEnumValue(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_view.slice_start,
-	     z.depth_view.slice_max, z.depth_view.current_mip_level,
+	     z.stencil_info.partially_resident ? "true" : "false", z.depth_info.addr5_swizzle_mask,
+	     z.depth_info.array_mode, z.depth_info.pipe_config, z.depth_info.bank_width,
+	     z.depth_info.bank_height, z.depth_info.macro_tile_aspect, z.depth_info.num_banks,
+	     z.depth_view.slice_start, z.depth_view.slice_max, z.depth_view.current_mip_level,
 	     z.depth_view.depth_write_disable ? "true" : "false",
-	     z.depth_view.stencil_write_disable ? "true" : "false", z.z_read_base_addr,
-	     z.stencil_read_base_addr, z.z_write_base_addr, z.stencil_write_base_addr,
-	     z.htile_data_base_addr, z.shading_rate_encoding, z.size.x_max, z.size.y_max,
-	     z.size.valid ? "true" : "false");
+	     z.depth_view.stencil_write_disable ? "true" : "false", z.htile_surface.linear,
+	     z.htile_surface.full_cache, z.htile_surface.htile_uses_preload_win,
+	     z.htile_surface.preload, z.htile_surface.prefetch_width, z.htile_surface.prefetch_height,
+	     z.htile_surface.dst_outside_zero_to_one, z.z_read_base_addr, z.stencil_read_base_addr,
+	     z.z_write_base_addr, z.stencil_write_base_addr, z.pitch_div8_minus1, z.height_div8_minus1,
+	     z.slice_div64_minus1, z.htile_data_base_addr, z.width, z.height, z.size.x_max,
+	     z.size.y_max);
+}
+
+// NOLINTNEXTLINE(readability-function-cognitive-complexity)
+static void ZCheck(const HW::DepthRenderTarget& z, const HW::DepthControl& dc,
+                   const HW::RenderControl& rc) {
+	const bool depth_active =
+	    dc.z_enable || dc.z_write_enable || dc.depth_bounds_enable || rc.depth_clear_enable;
+	const bool stencil_active = dc.stencil_enable || rc.stencil_clear_enable;
+	if (!depth_active && !stencil_active) {
+		return;
+	}
+
+	EXIT_NOT_IMPLEMENTED(!z.z_info.HasValidTextureCompatibility());
+	EXIT_NOT_IMPLEMENTED(!z.stencil_info.HasValidTextureCompatibility());
+	if (z.z_info.format == 0) {
+		EXIT_NOT_IMPLEMENTED(z.z_info.format != 0);
+		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);
+		if (z.z_info.num_samples != 0x00000000) {
+			static bool logged = false;
+			if (!logged) {
+				LOGF("DepthTarget: using native num_samples=0x%08" PRIx32 "\n",
+				     z.z_info.num_samples);
+				logged = true;
+			}
+		}
+		EXIT_NOT_IMPLEMENTED(z.z_info.expclear_enabled != false);
+		EXIT_NOT_IMPLEMENTED(z.z_info.partially_resident != false);
+		EXIT_NOT_IMPLEMENTED(z.z_info.max_mip_level != 0);
+	}
+
+	if (z.stencil_info.format == 0) {
+		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.expclear_enabled != false);
+		EXIT_NOT_IMPLEMENTED(z.stencil_info.partially_resident != false);
+	}
+
+	if (z.z_info.format != 0 || z.stencil_info.format != 0) {
+
+		EXIT_NOT_IMPLEMENTED(z.depth_info.addr5_swizzle_mask != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.depth_info.array_mode != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.depth_info.pipe_config != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.depth_info.bank_width != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.depth_info.bank_height != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.depth_info.macro_tile_aspect != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.depth_info.num_banks != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.htile_surface.linear != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.htile_surface.full_cache != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.htile_surface.htile_uses_preload_win != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.htile_surface.preload != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.htile_surface.prefetch_width != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.htile_surface.prefetch_height != 0x00000000);
+		EXIT_NOT_IMPLEMENTED(z.htile_surface.dst_outside_zero_to_one != 0x00000000);
+
+		if (z.depth_view.current_mip_level != 0x00000000) {
+			static std::atomic<uint32_t> log_count {0};
+			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {
+				LOGF("DepthTarget: temporary: ignoring PS5 current mip level=0x%08" PRIx32 "\n",
+				     z.depth_view.current_mip_level);
+			}
+		}
+		if (z.depth_view.depth_write_disable || z.depth_view.stencil_write_disable) {
+			static std::atomic<uint32_t> log_count {0};
+			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {
+				LOGF("DepthTarget: honoring write disable depth=%s, stencil=%s\n",
+				     z.depth_view.depth_write_disable ? "true" : "false",
+				     z.depth_view.stencil_write_disable ? "true" : "false");
+			}
+		}
+		EXIT_NOT_IMPLEMENTED(z.z_read_base_addr != z.z_write_base_addr);
+		EXIT_NOT_IMPLEMENTED(z.stencil_read_base_addr != z.stencil_write_base_addr);
+		EXIT_NOT_IMPLEMENTED(z.z_write_base_addr == 0);
+		if (z.stencil_info.format != 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.pitch_div8_minus1 != 0x000000ff);
+		// EXIT_NOT_IMPLEMENTED(z.height_div8_minus1 != 0x0000008f);
+		// EXIT_NOT_IMPLEMENTED(z.slice_div64_minus1 != 0x00008fff);
+		// EXIT_NOT_IMPLEMENTED(z.htile_data_base_addr == 0);
+		// EXIT_NOT_IMPLEMENTED(z.width != 0x00000780);
+		// EXIT_NOT_IMPLEMENTED(z.height != 0x00000438);
+
+		EXIT_NOT_IMPLEMENTED(z.width != 0);
+		EXIT_NOT_IMPLEMENTED(z.height != 0);
+		if (z.size.x_max == 0 || z.size.y_max == 0) {
+			static std::atomic<uint32_t> log_count {0};
+			if (log_count.fetch_add(1) < 16) {
+				LOGF("DepthTarget: temporary: inferring PS5 default depth extent x_max=0x%04" PRIx16
+				     ", y_max=0x%04" PRIx16 "\n",
+				     z.size.x_max, z.size.y_max);
+			}
+		}
+	}
 }

 static void ClipPrint(const char* func, const HW::ClipControl& c) {
@@ -396,7 +608,8 @@ static void ClipCheck(const HW::ClipControl& c) {
 	// dx_linear_attr_clip_enable preserves linear (noperspective) attributes at clip-generated
 	// vertices, which Vulkan provides as part of clipping and interpolation.
 	EXIT_NOT_IMPLEMENTED(c.user_clip_planes != 0 || c.user_clip_plane_mode != 0 ||
-	                     c.vertex_kill_any || c.user_clip_plane_negate_y ||
+	                     c.vertex_kill_any || !c.IsZClipModeRepresentable() ||
+	                     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);
 }
@@ -420,6 +633,17 @@ 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);
@@ -452,6 +676,24 @@ static void McCheck(const HW::ModeControl& c) {
 	// EXIT_NOT_IMPLEMENTED(c.cull_front != false);
 	// EXIT_NOT_IMPLEMENTED(c.cull_back != false);
 	// EXIT_NOT_IMPLEMENTED(c.face != false);
+	if (c.poly_mode != 0) {
+		static bool logged = false;
+		if (!logged) {
+			LOGF("\t temporary: PA_SU_SC_MODE_CNTL.POLY_MODE is not fully implemented; continuing "
+			     "with filled polygons\n");
+			logged = true;
+		}
+	}
+	EXIT_NOT_IMPLEMENTED(c.polymode_front_ptype != 0 && c.polymode_front_ptype != 2);
+	EXIT_NOT_IMPLEMENTED(c.polymode_back_ptype != 0 && c.polymode_back_ptype != 2);
+	if (c.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) {
@@ -461,6 +703,7 @@ static void McCheck(const HW::ModeControl& c) {
 			logged = true;
 		}
 	}
+	EXIT_NOT_IMPLEMENTED(c.provoking_vtx_last != false);
 	EXIT_NOT_IMPLEMENTED(c.persp_corr_dis != false);
 }

@@ -487,7 +730,16 @@ static void BcPrint(const char* func, const HW::BlendControl& c, const HW::Blend
 	     color.red, color.green, color.blue, color.alpha, cc.mode, cc.op);
 }

-static void BcCheck(const HW::BlendColor& color, const HW::ColorControl& cc) {
+static void BcCheck(const HW::BlendControl& /*c*/, const HW::BlendColor& color,
+                    const HW::ColorControl& cc) {
+	// EXIT_NOT_IMPLEMENTED(c.color_srcblend != 0);
+	// EXIT_NOT_IMPLEMENTED(c.color_comb_fcn != 0);
+	// EXIT_NOT_IMPLEMENTED(c.color_destblend != 0);
+	// EXIT_NOT_IMPLEMENTED(c.alpha_srcblend != 0);
+	// EXIT_NOT_IMPLEMENTED(c.alpha_comb_fcn != 0);
+	// EXIT_NOT_IMPLEMENTED(c.alpha_destblend != 0);
+	// EXIT_NOT_IMPLEMENTED(c.separate_alpha_blend != false);
+	// EXIT_NOT_IMPLEMENTED(c.enable != false);
 	if (color.red != 0.0f || color.green != 0.0f || color.blue != 0.0f || color.alpha != 0.0f) {
 		static bool logged = false;
 		if (!logged) {
@@ -527,6 +779,8 @@ static void DPrint(const char* func, const HW::DepthControl& c, const HW::Stenci
 	     "\t backface_enable      = %s\n"
 	     "\t stencilfunc          = %" PRIu8 "\n"
 	     "\t stencilfunc_bf       = %" PRIu8 "\n"
+	     "\t color_writes_on_depth_fail_enable  = %s\n"
+	     "\t color_writes_on_depth_pass_disable = %s\n"
 	     "\t stencil_fail         = %" PRIu8 "\n"
 	     "\t stencil_zpass        = %" PRIu8 "\n"
 	     "\t stencil_zfail        = %" PRIu8 "\n"
@@ -543,11 +797,37 @@ static void DPrint(const char* func, const HW::DepthControl& c, const HW::Stenci
 	     "\t stencil_opval_bf     = %" PRIu8 "\n",
 	     c.stencil_enable ? "true" : "false", c.z_enable ? "true" : "false",
 	     c.z_write_enable ? "true" : "false", c.depth_bounds_enable ? "true" : "false", c.zfunc,
-	     c.backface_enable ? "true" : "false", c.stencilfunc, c.stencilfunc_bf, s.stencil_fail,
-	     s.stencil_zpass, s.stencil_zfail, s.stencil_fail_bf, s.stencil_zpass_bf,
-	     s.stencil_zfail_bf, sm.stencil_testval, sm.stencil_mask, sm.stencil_writemask,
-	     sm.stencil_opval, sm.stencil_testval_bf, sm.stencil_mask_bf, sm.stencil_writemask_bf,
-	     sm.stencil_opval_bf);
+	     c.backface_enable ? "true" : "false", c.stencilfunc, c.stencilfunc_bf,
+	     c.color_writes_on_depth_fail_enable ? "true" : "false",
+	     c.color_writes_on_depth_pass_disable ? "true" : "false", s.stencil_fail, s.stencil_zpass,
+	     s.stencil_zfail, s.stencil_fail_bf, s.stencil_zpass_bf, s.stencil_zfail_bf,
+	     sm.stencil_testval, sm.stencil_mask, sm.stencil_writemask, sm.stencil_opval,
+	     sm.stencil_testval_bf, sm.stencil_mask_bf, sm.stencil_writemask_bf, sm.stencil_opval_bf);
+}
+
+static void DCheck(const HW::DepthControl& c, const HW::StencilControl& s,
+                   const HW::StencilMask& /*sm*/) {
+	// EXIT_NOT_IMPLEMENTED(c.stencil_enable != false);
+	// EXIT_NOT_IMPLEMENTED(c.z_enable != false);
+	// EXIT_NOT_IMPLEMENTED(c.z_write_enable != false);
+	// EXIT_NOT_IMPLEMENTED(c.zfunc != 0);
+	// Back-face stencil state is handled separately when enabled.
+	// EXIT_NOT_IMPLEMENTED(c.stencilfunc != 0);
+	// EXIT_NOT_IMPLEMENTED(c.stencilfunc_bf != 0);
+	EXIT_NOT_IMPLEMENTED(c.color_writes_on_depth_fail_enable != false);
+	EXIT_NOT_IMPLEMENTED(c.color_writes_on_depth_pass_disable != false);
+	// EXIT_NOT_IMPLEMENTED(s.stencil_fail != 0);
+	// EXIT_NOT_IMPLEMENTED(s.stencil_zpass != 0);
+	// EXIT_NOT_IMPLEMENTED(s.stencil_zfail != 0);
+	// Back-face stencil ops may legitimately differ from the front-face ops.
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_testval != 0);
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_mask != 0);
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_writemask != 0);
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_opval != 0);
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_testval_bf != 0);
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_mask_bf != 0);
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_writemask_bf != 0);
+	// EXIT_NOT_IMPLEMENTED(sm.stencil_opval_bf != 0);
 }

 static void EqaaPrint(const char* func, const HW::EqaaControl& c) {
@@ -567,12 +847,7 @@ static void EqaaPrint(const char* func, const HW::EqaaControl& c) {
 	     c.static_anchor_associations ? "true" : "false");
 }

-static void EqaaCheck(const HW::EqaaControl& c, const HW::AaConfig& cf) {
-	// The EQAA controls only take effect while multisampling is on, so a single-sample
-	// target leaves them inert and has nothing to warn about.
-	if (cf.msaa_num_samples == 0) {
-		return;
-	}
+static void EqaaCheck(const HW::EqaaControl& c) {
 	if (c.max_anchor_samples != 0 || c.ps_iter_samples != 0 || c.mask_export_num_samples != 0 ||
 	    c.alpha_to_mask_num_samples != 0 || c.high_quality_intersections ||
 	    c.incoherent_eqaa_reads || c.interpolate_comp_z || c.static_anchor_associations) {
@@ -624,10 +899,12 @@ void LogDrawPhase(const char* draw_name, const char* phase) {
 	}
 }

-void LogPipelineTrace(const char* phase, uint64_t vertex_program_id, uint64_t pixel_program_id) {
+void LogPipelineTrace(const char* phase, uint32_t vs_hash0, uint32_t vs_crc32, uint32_t ps_hash0,
+                      uint32_t ps_crc32) {
 	if (graphics_debug_dump_enabled()) {
-		LOGF("PipelineTrace: %s VS=%" PRIu64 " PS=%" PRIu64 "\n", phase, vertex_program_id,
-		     pixel_program_id);
+		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);
 	}
 }

@@ -708,6 +985,15 @@ static void VpCheck(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc
 	// EXIT_NOT_IMPLEMENTED(smc.vport_scissor_enable);
 	EXIT_NOT_IMPLEMENTED(smc.line_stipple_enable);

+	if (vp.viewports[0].zmin > 0.000000 || vp.viewports[0].zmax != 1.000000) {
+		static bool logged = false;
+		if (!logged) {
+			LOGF("\t warning: non-default viewport depth clamp zmin = %f, zmax = %f; using "
+			     "viewport scale/offset for Vulkan depth range\n",
+			     vp.viewports[0].zmin, vp.viewports[0].zmax);
+			logged = true;
+		}
+	}
 	// EXIT_NOT_IMPLEMENTED(vp.viewports[0].xscale != 960.000000);
 	// EXIT_NOT_IMPLEMENTED(vp.viewports[0].xoffset != 960.000000);
 	// EXIT_NOT_IMPLEMENTED(vp.viewports[0].yscale != -540.000000);
@@ -727,7 +1013,7 @@ static void VpCheck(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc
 	// EXIT_NOT_IMPLEMENTED(fabsf(vp.guard_band_horz_clip - 33.133327f) > 0.001f);
 	// EXIT_NOT_IMPLEMENTED(fabsf(vp.guard_band_vert_clip - 59.629623f) > 0.001f);

-	if (vp.guard_band_horz_discard != 1.0f || vp.guard_band_vert_discard != 1.0f) {
+	if (vp.guard_band_horz_discard != 0.0f || vp.guard_band_vert_discard != 0.0f) {
 		static std::atomic<uint32_t> log_count {0};
 		if (log_count.fetch_add(1) < 16) {
 			LOGF("\t warning: unsupported PS5 guard band discard = %f, %f, continuing\n",
@@ -747,6 +1033,10 @@ static bool ScissorRectValid(const ScissorRect& r) {
 	return r.right > r.left && r.bottom > r.top;
 }

+static bool ScissorRectSet(const ScissorRect& r) {
+	return r.left != 0 || r.top != 0 || r.right != 0 || r.bottom != 0;
+}
+
 static ScissorRect ScissorRectOffset(ScissorRect r, int x, int y) {
 	r.left += x;
 	r.right += x;
@@ -756,8 +1046,8 @@ static ScissorRect ScissorRectOffset(ScissorRect r, int x, int y) {
 }

 static ScissorRect ScissorRectIntersect(const ScissorRect& a, const ScissorRect& b) {
-	return {std::max(a.left, b.left), std::max(a.top, b.top),
-	        std::min(a.right, b.right), std::min(a.bottom, b.bottom)};
+	return {a.left > b.left ? a.left : b.left, a.top > b.top ? a.top : b.top,
+	        a.right < b.right ? a.right : b.right, a.bottom < b.bottom ? a.bottom : b.bottom};
 }

 static ScissorRect ScissorRectClamp(ScissorRect r, uint32_t width, uint32_t height) {
@@ -777,24 +1067,18 @@ 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);

-	static constexpr auto rules = MakeScissorIntersectionRules();
-	for (uint32_t candidate = 0; candidate < rules.size(); candidate++) {
-		if (rules[candidate] == rule) {
+	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) {
 			*mask = static_cast<uint8_t>(candidate);
 			return true;
 		}
@@ -804,26 +1088,49 @@ static bool ScissorClipRuleToIntersectionMask(uint16_t rule, uint8_t* mask) {
 }

 ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc,
-                               vk::Extent2D extent, uint32_t viewport_index) {
-	EXIT_IF(viewport_index >= std::size(vp.viewports));
-	ScissorRect final {vp.screen_scissor_left, vp.screen_scissor_top, vp.screen_scissor_right,
-	                   vp.screen_scissor_bottom};
-	const auto intersect = [&](ScissorRect rect, bool window_offset) {
-		if (window_offset) {
-			rect = ScissorRectOffset(rect, vp.window_offset_x, vp.window_offset_y);
-		}
-		final = ScissorRectIntersect(final, rect);
-	};
-	intersect({vp.window_scissor_left, vp.window_scissor_top, vp.window_scissor_right,
-	           vp.window_scissor_bottom}, vp.window_scissor_window_offset_enable);
-	intersect({vp.generic_scissor_left, vp.generic_scissor_top, vp.generic_scissor_right,
-	           vp.generic_scissor_bottom}, vp.generic_scissor_window_offset_enable);
+                               vk::Extent2D extent) {
+	ScissorRect screen {vp.screen_scissor_left, vp.screen_scissor_top, vp.screen_scissor_right,
+	                    vp.screen_scissor_bottom};
+	ScissorRect final = screen;

-	const auto& viewport = vp.viewports[viewport_index];
-	if (smc.vport_scissor_enable) {
-		intersect({viewport.viewport_scissor_left, viewport.viewport_scissor_top,
-		           viewport.viewport_scissor_right, viewport.viewport_scissor_bottom},
-		          viewport.viewport_scissor_window_offset_enable);
+	if (!ScissorRectSet(screen)) {
+		final = {0, 0, static_cast<int>(extent.width), static_cast<int>(extent.height)};
+
+		auto log_id = g_scissor_default_log_count.fetch_add(1);
+		if (log_id < 32) {
+			LOGF("temporary: default unset screen scissor to framebuffer extent %ux%u\n",
+			     extent.width, extent.height);
+		}
+	}
+
+	ScissorRect window {vp.window_scissor_left, vp.window_scissor_top, vp.window_scissor_right,
+	                    vp.window_scissor_bottom};
+	if (ScissorRectSet(window)) {
+		if (vp.window_scissor_window_offset_enable) {
+			window = ScissorRectOffset(window, vp.window_offset_x, vp.window_offset_y);
+		}
+		final = ScissorRectIntersect(final, window);
+	}
+
+	ScissorRect generic {vp.generic_scissor_left, vp.generic_scissor_top, vp.generic_scissor_right,
+	                     vp.generic_scissor_bottom};
+	if (ScissorRectSet(generic)) {
+		if (vp.generic_scissor_window_offset_enable) {
+			generic = ScissorRectOffset(generic, vp.window_offset_x, vp.window_offset_y);
+		}
+		final = ScissorRectIntersect(final, generic);
+	}
+
+	const auto& viewport = vp.viewports[0];
+	ScissorRect viewport_scissor {viewport.viewport_scissor_left, viewport.viewport_scissor_top,
+	                              viewport.viewport_scissor_right,
+	                              viewport.viewport_scissor_bottom};
+	if (smc.vport_scissor_enable && ScissorRectSet(viewport_scissor)) {
+		if (viewport.viewport_scissor_window_offset_enable) {
+			viewport_scissor =
+			    ScissorRectOffset(viewport_scissor, vp.window_offset_x, vp.window_offset_y);
+		}
+		final = ScissorRectIntersect(final, viewport_scissor);
 	}

 	if (vp.clip_rect_rule == 0) {
@@ -836,8 +1143,12 @@ ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeC
 					continue;
 				}

-				intersect({vp.clip_rect_left[i], vp.clip_rect_top[i], vp.clip_rect_right[i],
-				           vp.clip_rect_bottom[i]}, vp.clip_rect_window_offset_enable[i]);
+				ScissorRect clip {vp.clip_rect_left[i], vp.clip_rect_top[i], vp.clip_rect_right[i],
+				                  vp.clip_rect_bottom[i]};
+				if (vp.clip_rect_window_offset_enable[i]) {
+					clip = ScissorRectOffset(clip, vp.window_offset_x, vp.window_offset_y);
+				}
+				final = ScissorRectIntersect(final, clip);
 			}
 		} else {
 			static std::atomic<uint32_t> log_count {0};
@@ -851,14 +1162,19 @@ ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeC
 	return ScissorRectClamp(final, extent.width, extent.height);
 }

-void hw_check(const CommandBuffer& buffer) {
+void hw_check(const RenderCommandBuffer& buffer) {
 	const auto& hw      = buffer.GetRegisters();
 	const auto  rt_slot = render_target_first_bound_slot(buffer);
 	const auto& rt      = hw.GetRenderTarget(rt_slot);
+	const auto& bc      = hw.GetBlendControl(rt_slot);
 	const auto& bclr    = hw.GetBlendColor();
 	const auto& vp      = hw.GetScreenViewport();
+	const auto& z       = hw.GetDepthRenderTarget();
 	const auto& c       = hw.GetClipControl();
 	const auto& rc      = hw.GetRenderControl();
+	const auto& d       = hw.GetDepthControl();
+	const auto& s       = hw.GetStencilControl();
+	const auto& sm      = hw.GetStencilMask();
 	const auto& mc      = hw.GetModeControl();
 	const auto& eqaa    = hw.GetEqaaControl();
 	const auto& cc      = hw.GetColorControl();
@@ -879,30 +1195,32 @@ void hw_check(const CommandBuffer& buffer) {
 	RtCheck(rt);
 	log_phase("vp");
 	VpCheck(vp, smc);
+	log_phase("z");
+	ZCheck(z, d, rc);
 	log_phase("clip");
 	ClipCheck(c);
 	log_phase("rc");
 	RcCheck(rc);
 	log_phase("depth");
+	DCheck(d, s, sm);
 	log_phase("mode");
 	McCheck(mc);
 	log_phase("blend");
-	BcCheck(bclr, cc);
+	BcCheck(bc, bclr, cc);
 	log_phase("eqaa");
-	EqaaCheck(eqaa, ac);
+	EqaaCheck(eqaa);
 	log_phase("aa");
 	AaCheck(aa, ac);
 	log_phase("done");

-	if (graphics_debug_dump_enabled() && RenderTargetMaskHasBoundMrt(buffer)) {
+	if (RenderTargetMaskHasBoundMrt(buffer)) {
 		LOGF("MRT render target mask: 0x%08" PRIx32 "\n", hw.GetRenderTargetMask());
 		for (uint32_t i = 0; i < 8; i++) {
 			const auto& mrt = hw.GetRenderTarget(i);
 			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,
-			     static_cast<uint32_t>(mrt.info.format), mrt.attrib2.width + 1,
-			     mrt.attrib2.height + 1, static_cast<uint32_t>(mrt.attrib3.tile_mode));
+			     i, mrt.base.addr, (hw.GetRenderTargetMask() >> (i * 4u)) & 0x0fu, mrt.info.format,
+			     mrt.attrib2.width + 1, mrt.attrib2.height + 1, mrt.attrib3.tile_mode);
 		}
 	}
 	if (rc.depth_clear_enable && hw.GetDepthClearValue() != 0.0f &&
@@ -916,7 +1234,7 @@ void hw_check(const CommandBuffer& buffer) {
 	// EXIT_NOT_IMPLEMENTED(hw.GetStencilClearValue() != 0);
 }

-void hw_print(const CommandBuffer& buffer) {
+void hw_print(const RenderCommandBuffer& buffer) {
 	const auto& hw      = buffer.GetRegisters();
 	const auto  rt_slot = render_target_first_bound_slot(buffer);
 	const auto& rt      = hw.GetRenderTarget(rt_slot);
@@ -941,12 +1259,11 @@ void hw_print(const CommandBuffer& buffer) {
 		     "\t GetRenderTargetMask()   = 0x%08" PRIx32 "\n"
 		     "\t GetDepthClearValue()    = %f\n"
 		     "\t GetStencilClearValue()  = %" PRIu8 "\n"
-		     "\t GetLineWidth()          = %f\n"
-		     "\t primitive_reset_index   = 0x%08" PRIx32 "\n",
+		     "\t GetLineWidth()          = %f\n",
 		     hw.GetRenderTargetMask(), hw.GetDepthClearValue(), hw.GetStencilClearValue(),
-		     hw.GetLineWidth(), hw.GetPrimitiveResetIndex());
+		     hw.GetLineWidth());

-		LOGF("%s", rt_print("RenderTraget:", rt).c_str());
+		LOGF("%s", Common::Concat(rt_print("RenderTraget:", rt), "").c_str());

 		ZPrint("DepthRenderTraget:", z);
 		VpPrint("ScreenViewport:", vp, smc);
@@ -5,13 +5,15 @@

 #include <cstdint>
 #include <string>
+#include <vector>

 namespace Libs::Graphics {

-class CommandBuffer;
+class RenderCommandBuffer;

 namespace HW {
 class Context;
+class Shader;
 class UserConfig;
 struct RenderTarget;
 struct ScanModeControl;
@@ -26,17 +28,18 @@ struct ScissorRect {
 };

 uint32_t                 render_target_mask_slot(uint32_t mask, uint32_t slot);
-uint32_t                 render_target_first_bound_slot(const CommandBuffer& buffer);
+uint32_t                 render_target_first_bound_slot(const RenderCommandBuffer& buffer);
 bool                     graphics_debug_dump_enabled();
 void                     uc_print(const char* func, const HW::UserConfig& uc);
 void                     uc_check(const HW::UserConfig& uc);
-std::string              rt_print(const char* func, const HW::RenderTarget& rt);
-bool                     RenderIsColorTileModeLinear(Prospero::TileMode tile_mode);
-void                     hw_print(const CommandBuffer& buffer);
-void                     hw_check(const CommandBuffer& buffer);
+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);
 void                     LogDrawPhase(const char* draw_name, const char* phase);
 ScissorRect calc_final_scissor(const HW::ScreenViewport& vp, const HW::ScanModeControl& smc,
-                               vk::Extent2D extent, uint32_t viewport_index);
+                               vk::Extent2D extent);

 } // namespace Libs::Graphics

@@ -10,16 +10,15 @@
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/guest_gpu/tile.h"
 #include "graphics/host_gpu/graphicContext.h"
-#include "graphics/host_gpu/renderer/commandScheduler.h"
 #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"

 #include <algorithm>
-#include <array>
 #include <atomic>
 #include <cmath>
 #include <cstdarg>
@@ -39,74 +38,29 @@ namespace Libs::Graphics {
 	EXIT("unsupported render state; details were printed above\n");
 }

-static vk::StencilOp ConvertStencilOp(uint8_t value, uint8_t write_mask, uint8_t op_value) {
-	if (write_mask == 0) {
-		return vk::StencilOp::eKeep;
-	}
+static vk::StencilOp ConvertStencilOp(uint8_t value) {
 	switch (static_cast<Prospero::StencilOp>(value)) {
 		case Prospero::StencilOp::kKeep: return vk::StencilOp::eKeep;
 		case Prospero::StencilOp::kZero: return vk::StencilOp::eZero;
-		case Prospero::StencilOp::kReplaceTest: return vk::StencilOp::eReplace;
-		case Prospero::StencilOp::kReplaceOp:
-			if ((op_value & write_mask) == 0) {
-				return vk::StencilOp::eZero;
-			}
-			return vk::StencilOp::eReplace;
+		case Prospero::StencilOp::kReplaceTest:
+		case Prospero::StencilOp::kReplaceOp: return vk::StencilOp::eReplace;
 		case Prospero::StencilOp::kAddClamp: return vk::StencilOp::eIncrementAndClamp;
 		case Prospero::StencilOp::kSubClamp: return vk::StencilOp::eDecrementAndClamp;
 		case Prospero::StencilOp::kInvert: return vk::StencilOp::eInvert;
 		case Prospero::StencilOp::kAddWrap: return vk::StencilOp::eIncrementAndWrap;
 		case Prospero::StencilOp::kSubWrap: return vk::StencilOp::eDecrementAndWrap;
-		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);
+		default: DepthFatal("unsupported stencil operation");
 	}
 }

-static vk::StencilOpState ConvertStencilState(
-    uint8_t compare, const std::array<uint8_t, 3>& operations, uint8_t op_value,
-    const vk::StencilOpState& state) {
-	const auto test_value = state.reference;
-	auto reference       = test_value;
-	auto required_bits   = state.compareMask;
-	if (compare == static_cast<uint8_t>(vk::CompareOp::eAlways) ||
-	    compare == static_cast<uint8_t>(vk::CompareOp::eNever)) {
-		required_bits = 0;
-	}
-	std::array<vk::StencilOp, 3> converted {};
-	for (size_t i = 0; i < operations.size(); i++) {
-		converted[i] = ConvertStencilOp(operations[i], state.writeMask, op_value);
-		if (converted[i] != vk::StencilOp::eReplace) {
-			continue;
-		}
-		auto replacement = test_value;
-		if (static_cast<Prospero::StencilOp>(operations[i]) == Prospero::StencilOp::kReplaceOp) {
-			replacement = op_value;
-		}
-		if (((reference ^ replacement) & required_bits & state.writeMask) != 0) {
-			DepthFatal("unsupported stencil replacement: compare=%u, compare mask=0x%02" PRIx32
-			           ", write mask=0x%02" PRIx32 ", operation value=0x%02" PRIx8
-			           ", test value=0x%02" PRIx32,
-			           compare, state.compareMask, state.writeMask, op_value, test_value);
-		}
-		// Vulkan shares one reference between comparison and every replacement on this face.
-		reference = (reference & ~state.writeMask) | (replacement & state.writeMask);
-		required_bits |= state.writeMask;
-	}
-	return {converted[0], converted[1], converted[2], static_cast<vk::CompareOp>(compare),
-	        state.compareMask, state.writeMask, reference};
+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);
 }

-[[nodiscard]] static vk::Format ResolveHostDepthAttachmentFormat(const CommandBuffer&     buffer,
-                                                                 const DepthFormatPolicy& policy,
+[[nodiscard]] static vk::Format ResolveHostDepthAttachmentFormat(const RenderCommandBuffer& buffer,
+                                                                 const DepthFormatPolicy&   policy,
                                                                 bool     has_stencil,
                                                                 uint32_t samples) {
 	auto&      graphics         = buffer.GetGraphics();
@@ -131,20 +85,63 @@ static vk::StencilOpState ConvertStencilState(
 	return vk::Format::eUndefined;
 }

-static TextureCache::ImageDesc MakeDepthTargetDesc(const CommandBuffer& buffer,
-                                                  const HW::DepthRenderTarget& z,
-                                                  bool write_buffer = false) {
-	const bool has_stencil = z.stencil_info.format != Prospero::StencilFormat::kInvalid;
-	const auto depth_address = write_buffer ? z.z_write_base_addr : z.z_read_base_addr;
-	const auto stencil_address =
-	    write_buffer ? z.stencil_write_base_addr : z.stencil_read_base_addr;
+// NOLINTNEXTLINE(readability-function-cognitive-complexity)
+void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandBuffer& 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 =
+	    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) {
+		return;
+	}
 	if (!z.z_info.HasValidTextureCompatibility() ||
 	    !z.stencil_info.HasValidTextureCompatibility()) {
 		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.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 &&
+	    !z.stencil_info.expclear_enabled && !z.stencil_info.partially_resident &&
+	    z.depth_view.slice_start == 0 && z.depth_view.slice_max == 0 &&
+	    z.depth_view.current_mip_level == 0 && !z.depth_view.depth_write_disable &&
+	    !z.depth_view.stencil_write_disable && z.depth_info.addr5_swizzle_mask == 0 &&
+	    z.depth_info.array_mode == 0 && z.depth_info.pipe_config == 0 &&
+	    z.depth_info.bank_width == 0 && z.depth_info.bank_height == 0 &&
+	    z.depth_info.macro_tile_aspect == 0 && z.depth_info.num_banks == 0 &&
+	    z.htile_surface.linear == 0 && z.htile_surface.full_cache == 0 &&
+	    z.htile_surface.htile_uses_preload_win == 0 && z.htile_surface.preload == 0 &&
+	    z.htile_surface.prefetch_width == 0 && z.htile_surface.prefetch_height == 0 &&
+	    z.htile_surface.dst_outside_zero_to_one == 0 && z.z_read_base_addr == 0 &&
+	    z.z_write_base_addr == 0 && z.stencil_read_base_addr == 0 &&
+	    z.stencil_write_base_addr == 0 && z.htile_data_base_addr == 0 &&
+	    // DB_DEPTH_SIZE_XY is independent state and may remain programmed after the attachment
+	    // formats and addresses are unbound. A zero encoding is the valid 1x1 value, so its
+	    // presence alone must not manufacture a depth attachment.
+	    !z.z_info.htile_acceleration && !z.width_height_valid && !z.pitch_height_valid &&
+	    z.size.x_max == 0 && z.size.y_max == 0 && z.pitch_div8_minus1 == 0 &&
+	    z.height_div8_minus1 == 0 && z.slice_div64_minus1 == 0 && z.width == 0 && z.height == 0;
+	if (attachment_unbound) {
+		static std::atomic_bool logged = false;
+		if (!logged.exchange(true, std::memory_order_relaxed)) {
+			LOGF("DepthTarget: ignoring enabled depth/stencil state without a bound attachment\n");
+		}
+		return;
+	}
 	const bool has_htile = z.z_info.htile_acceleration;
-	const bool unsupported_shading_rate_encoding =
-	    z.shading_rate_encoding > 1 || (z.shading_rate_encoding != 0 && !has_htile);
 	const auto samples   = render_sample_count(z.z_info.num_samples);
 	if (samples == 0) {
 		DepthFatal("unsupported depth fragment count: %u", z.z_info.num_samples);
@@ -159,19 +156,33 @@ static TextureCache::ImageDesc MakeDepthTargetDesc(const CommandBuffer& buffer,
 			DepthFatal("invalid depth view: base=%u last=%u", z.depth_view.slice_start,
 			           z.depth_view.slice_max);
 	}
-	// EXPCLEAR permits an HTile acceleration state; the host attachment is already expanded.
-	if (z.z_info.partially_resident ||
-	    z.stencil_info.partially_resident || z.z_info.max_mip_level != 0 ||
-	    z.depth_view.current_mip_level != 0 || unsupported_shading_rate_encoding ||
-	    depth_address == 0 || (depth_address & 0xffffu) != 0) {
+	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 ||
+	    (z.z_read_base_addr & 0xffffu) != 0 ||
+	    dc.zfunc > static_cast<uint8_t>(vk::CompareOp::eAlways)) {
 		DepthFatal("unsupported depth register state");
 	}
 	if (has_stencil) {
-		if (z.stencil_info.format != Prospero::StencilFormat::k8UInt || !htile_stencil_compat ||
-		    stencil_address == 0 || (stencil_address & 0xffffu) != 0) {
+		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) {
 			DepthFatal("unsupported stencil attachment state");
 		}
-	} else if (z.stencil_read_base_addr != 0 || z.stencil_write_base_addr != 0) {
+	} else if (z.stencil_read_base_addr != 0 || z.stencil_write_base_addr != 0 ||
+	           !htile_stencil_compat ||
+	           z.stencil_info.texture_compatibility !=
+	               Prospero::TextureCompatibleStencil::kDisable) {
 		DepthFatal("stencil state without an active stencil attachment");
 	}
 	if (has_htile) {
@@ -181,28 +192,40 @@ static TextureCache::ImageDesc MakeDepthTargetDesc(const CommandBuffer& buffer,
 		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");
 	}
-	if (!z.size.valid) {
+	const bool size_xy_valid = z.size.valid;
+	const bool wh_valid      = z.width_height_valid && z.width != 0 && z.height != 0;
+	if (!size_xy_valid && !wh_valid) {
 		DepthFatal("missing depth extent");
 	}
-	const uint32_t width  = static_cast<uint32_t>(z.size.x_max) + 1u;
-	const uint32_t height = static_cast<uint32_t>(z.size.y_max) + 1u;
-	if (width > 16384 || height > 16384) {
-		DepthFatal("invalid depth extent");
+	const uint32_t width  = size_xy_valid ? static_cast<uint32_t>(z.size.x_max) + 1u : z.width;
+	const uint32_t height = size_xy_valid ? static_cast<uint32_t>(z.size.y_max) + 1u : z.height;
+	if (width > 16384 || height > 16384 ||
+	    (size_xy_valid && wh_valid && (width != z.width || height != z.height)) ||
+	    (!z.pitch_height_valid &&
+	     (z.pitch_div8_minus1 != 0 || z.height_div8_minus1 != 0 || z.slice_div64_minus1 != 0))) {
+		DepthFatal("inconsistent depth extent or encoded layout");
 	}
 	const auto* policy = FindDepthFormatPolicy(z.z_info.format);
 	if (policy == nullptr) {
 		DepthFatal("unsupported depth/stencil format pair");
 	}
 	const auto ideal_format = DepthAttachmentFormat(*policy, has_stencil);
-	const auto format = ResolveHostDepthAttachmentFormat(buffer, *policy, has_stencil, samples);
-	if (format == vk::Format::eUndefined) {
+	r.format = ResolveHostDepthAttachmentFormat(buffer, *policy, has_stencil, samples);
+	if (r.format == vk::Format::eUndefined) {
 		DepthFatal("no host depth/stencil format supports required usage for %s",
-		           vk::to_string(ideal_format).c_str());
+		           VulkanToString(ideal_format).c_str());
 	}
-	const auto     guest_format = policy->guest_format;
+	const uint32_t guest_format = Prospero::GpuEnumValue(policy->guest_format);
 	const uint32_t bytes        = policy->bytes_per_element;
 	const auto     pitch        = TileGetDepthPitch(width, bytes, z.z_info.num_samples);
+	if (z.pitch_height_valid && ((static_cast<uint64_t>(z.pitch_div8_minus1) + 1u) * 8u != pitch ||
+	                             (static_cast<uint64_t>(z.height_div8_minus1) + 1u) * 8u !=
+	                                 ((static_cast<uint64_t>(height) + 7u) & ~7ull))) {
+		DepthFatal("encoded depth pitch or height mismatch");
+	}
 	TileSizeAlign depth_size {};
 	TileSizeAlign stencil_size {};
 	TileSizeAlign htile_size {};
@@ -213,6 +236,14 @@ static TextureCache::ImageDesc MakeDepthTargetDesc(const CommandBuffer& buffer,
 	    (has_htile != (htile_size.align == 32768 && htile_size.size != 0))) {
 		DepthFatal("unsupported depth/stencil/HTile footprint");
 	}
+	if (z.pitch_height_valid &&
+	    (static_cast<uint64_t>(z.slice_div64_minus1) + 1u) * 64u != depth_size.size) {
+		DepthFatal("depth footprint mismatch: extent=%ux%u pitch=%u expected=0x%016" PRIx64
+		           " align=0x%016" PRIx64 " encoded_valid=%u encoded=0x%016" PRIx64,
+		           width, height, pitch, depth_size.size, depth_size.align,
+		           z.pitch_height_valid ? 1u : 0u,
+		           (static_cast<uint64_t>(z.slice_div64_minus1) + 1u) * 64u);
+	}
 	if (depth_size.size > UINT64_MAX / view.image_layers ||
 	    stencil_size.size > UINT64_MAX / view.image_layers ||
 	    htile_size.size > UINT64_MAX / view.image_layers) {
@@ -221,17 +252,78 @@ static TextureCache::ImageDesc MakeDepthTargetDesc(const CommandBuffer& buffer,
 	const auto depth_backing_size   = depth_size.size * view.image_layers;
 	const auto stencil_backing_size = stencil_size.size * view.image_layers;
 	const auto htile_backing_size   = htile_size.size * view.image_layers;
-	if (!GuestRange {depth_address, depth_backing_size}.Valid() ||
-	    (has_stencil && !GuestRange {stencil_address, stencil_backing_size}.Valid()) ||
-	    (has_htile && !GuestRange {z.htile_data_base_addr, htile_backing_size}.Valid())) {
+	if (depth_backing_size > TRACKER_ADDRESS_SIZE - z.z_read_base_addr ||
+	    (has_stencil && stencil_backing_size > TRACKER_ADDRESS_SIZE - z.stencil_read_base_addr) ||
+	    (has_htile && htile_backing_size > TRACKER_ADDRESS_SIZE - z.htile_data_base_addr)) {
 		DepthFatal("layered depth backing range is invalid");
 	}
+	r.htile                   = has_htile;
+	r.width                   = width;
+	r.height                  = height;
+	r.samples                 = samples;
+	r.depth_buffer_size       = depth_backing_size;
+	r.depth_buffer_vaddr      = z.z_read_base_addr;
+	r.stencil_buffer_size     = has_stencil ? stencil_backing_size : 0;
+	r.stencil_buffer_vaddr    = has_stencil ? z.stencil_read_base_addr : 0;
+	r.htile_buffer_size       = has_htile ? htile_backing_size : 0;
+	r.htile_buffer_vaddr      = has_htile ? z.htile_data_base_addr : 0;
+	r.depth_clear_enable      = rc.depth_clear_enable;
+	r.depth_meta_clear_enable = false;
+	r.depth_load_clear_enable = r.depth_clear_enable;
+	r.depth_clear_value       = hw.GetDepthClearValue();
+	r.depth_test_enable       = dc.z_enable;
+	r.depth_write_enable      = dc.z_write_enable && !z.depth_view.depth_write_disable;
+	r.depth_compare_op        = static_cast<vk::CompareOp>(dc.zfunc);
+
+	r.depth_bounds_test_enable = dc.depth_bounds_enable;
+	r.depth_min_bounds         = hw.GetDepthBoundsMin();
+	r.depth_max_bounds         = hw.GetDepthBoundsMax();
+
+	r.stencil_clear_enable = has_stencil && rc.stencil_clear_enable;
+	r.stencil_clear_value  = hw.GetStencilClearValue();
+	r.stencil_test_enable  = has_stencil && dc.stencil_enable;
+	if (r.stencil_test_enable) {
+		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) &&
+		     sm.stencil_opval != sm.stencil_testval) ||
+		    (dc.backface_enable &&
+		     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};
+		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};
+		} else {
+			r.stencil_static_back  = r.stencil_static_front;
+			r.stencil_dynamic_back = r.stencil_dynamic_front;
+		}
+	}
+	r.vaddr_num = has_stencil ? 2 : 1;
+	r.vaddr[0]  = r.depth_buffer_vaddr;
+	r.size[0]   = r.depth_buffer_size;
+	if (has_stencil) {
+		r.vaddr[1] = r.stencil_buffer_vaddr;
+		r.size[1]  = r.stencil_buffer_size;
+	}
 	TextureCache::ImageDesc desc {};
 	desc.type                 = TextureCache::BindingType::DepthTarget;
-	desc.info.data            = {depth_address, depth_backing_size};
-	desc.info.stencil =
-	    has_stencil ? GuestRange {stencil_address, stencil_backing_size} : GuestRange {};
-	desc.info.pixel_format    = format;
+	desc.info.data            = {r.depth_buffer_vaddr, r.depth_buffer_size};
+	desc.info.stencil         = {r.stencil_buffer_vaddr, r.stencil_buffer_size};
+	desc.info.pixel_format    = r.format;
 	desc.info.guest_format    = guest_format;
 	desc.info.type            = Prospero::ImageType::kColor2D;
 	desc.info.extent          = {width, height, 1};
@@ -239,182 +331,29 @@ static TextureCache::ImageDesc MakeDepthTargetDesc(const CommandBuffer& buffer,
 	desc.info.pitch           = pitch;
 	desc.info.bytes_per_block = bytes;
 	desc.info.samples         = samples;
-	desc.info.tile_mode       = Prospero::TileMode::kDepth;
-	desc.info.mip_layout[0]   = {0, depth_backing_size, pitch, height};
-	desc.info.metadata.range =
-	    has_htile ? GuestRange {z.htile_data_base_addr, htile_backing_size} : GuestRange {};
+	desc.info.tile_mode       = Prospero::GpuEnumValue(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;
 	desc.info.metadata.stencil_compressed =
 	    has_stencil && has_htile && !z.stencil_info.htile_stencil_disabled;
-	desc.view_info.format = format;
+	desc.view_info.format = r.format;
 	desc.view_info.type =
 	    view.layer_count == 1 ? vk::ImageViewType::e2D : vk::ImageViewType::e2DArray;
-	desc.view_info.aspect      = ImageViewOps::DepthAspectMask(format);
+	desc.view_info.aspect      = ImageViewOps::DepthAspectMask(r.format);
 	desc.view_info.base_level  = 0;
 	desc.view_info.level_count = 1;
 	desc.view_info.base_layer  = view.base_layer;
 	desc.view_info.layer_count = view.layer_count;
 	desc.view_info.usage       = vk::ImageUsageFlagBits::eDepthStencilAttachment;
-	return desc;
-}
-
-// NOLINTNEXTLINE(readability-function-cognitive-complexity)
-void RenderExecutor::ResolveRenderDepthTarget(CommandBuffer& buffer, RenderDepthInfo& r) {
-	KYTY_PROFILER_FUNCTION();
-	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.depth_bounds_enable ||
-	                          rc.depth_clear_enable || rc.copy_depth_to_color;
-	const bool stencil_active =
-	    has_stencil && (dc.stencil_enable || rc.stencil_clear_enable || rc.copy_stencil_to_color);
-	if (!depth_active && !stencil_active) {
-		return;
-	}
-	const bool attachment_unbound =
-	    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 &&
-	    !z.stencil_info.expclear_enabled && !z.stencil_info.partially_resident &&
-	    z.depth_view.slice_start == 0 && z.depth_view.slice_max == 0 &&
-	    z.depth_view.current_mip_level == 0 && !z.depth_view.depth_write_disable &&
-	    !z.depth_view.stencil_write_disable && z.z_read_base_addr == 0 && z.z_write_base_addr == 0 &&
-	    z.stencil_read_base_addr == 0 && z.stencil_write_base_addr == 0 &&
-	    z.htile_data_base_addr == 0 &&
-	    // DB_DEPTH_SIZE_XY is independent state and may remain programmed after the attachment
-	    // formats and addresses are unbound. A zero encoding is the valid 1x1 value, so its
-	    // presence alone must not manufacture a depth attachment.
-	    !z.z_info.htile_acceleration && z.shading_rate_encoding == 0 && z.size.x_max == 0 &&
-	    z.size.y_max == 0;
-	if (attachment_unbound) {
-		static std::atomic_bool logged = false;
-		if (!logged.exchange(true, std::memory_order_relaxed)) {
-			LOGF("DepthTarget: ignoring enabled depth/stencil state without a bound attachment\n");
-		}
-		return;
-	}
-	if (rc.copy_depth_to_color || rc.copy_stencil_to_color || rc.copy_centroid ||
-	    rc.copy_sample != 0 || dc.zfunc > static_cast<uint8_t>(vk::CompareOp::eAlways) ||
-	    (!z.depth_view.depth_write_disable && z.z_write_base_addr != z.z_read_base_addr) ||
-	    (has_stencil && !z.depth_view.stencil_write_disable &&
-	     z.stencil_write_base_addr != z.stencil_read_base_addr)) {
-		DepthFatal("unsupported depth register state");
-	}
-	r.desc = MakeDepthTargetDesc(buffer, z);
-	r.depth_clear_enable      = rc.depth_clear_enable;
-	r.depth_load_clear_enable = r.depth_clear_enable;
-	r.depth_clear_value       = hw.GetDepthClearValue();
-	r.depth_test_enable       = dc.z_enable;
-	r.depth_write_enable      = r.depth_test_enable && dc.z_write_enable &&
-	                            !z.depth_view.depth_write_disable && !r.depth_clear_enable;
-	r.depth_compare_op        = static_cast<vk::CompareOp>(dc.zfunc);
-
-	r.depth_bounds_test_enable = dc.depth_bounds_enable;
-	r.depth_min_bounds         = hw.GetDepthBoundsMin();
-	r.depth_max_bounds         = hw.GetDepthBoundsMax();
-
-	r.stencil_clear_enable =
-	    has_stencil && rc.stencil_clear_enable && !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))) {
-			DepthFatal("unsupported stencil compare state");
-		}
-		r.stencil_front = ConvertStencilState(
-		    dc.stencilfunc, {sc.stencil_fail, sc.stencil_zpass, sc.stencil_zfail},
-		    sm.stencil_opval, {.compareMask = sm.stencil_mask, .writeMask = front_write_mask,
-		                       .reference = sm.stencil_testval});
-		if (dc.backface_enable) {
-			r.stencil_back = ConvertStencilState(
-			    dc.stencilfunc_bf, {sc.stencil_fail_bf, sc.stencil_zpass_bf, sc.stencil_zfail_bf},
-			    sm.stencil_opval_bf, {.compareMask = sm.stencil_mask_bf, .writeMask = back_write_mask,
-			                          .reference = sm.stencil_testval_bf});
-		} else {
-			r.stencil_back = r.stencil_front;
-		}
-	}
-	auto& cache = m_context.GetTextureCache();
-	r.image_id = cache.FindImage(r.desc);
+	r.desc                     = std::move(desc);
+	auto& cache                = m_context.GetTextureCache();
+	r.image_id                 = cache.FindImage(r.desc);
+	r.image_view               = nullptr;
 	BindRenderTarget(r.image_id);
 }

-bool RenderExecutor::DepthStencilCopy(CommandBuffer& buffer) {
-	const auto& hw       = buffer.GetRegisters();
-	const auto& z        = hw.GetDepthRenderTarget();
-	const auto& override = hw.GetDepthRenderOverride();
-	if (hw.GetColorControl().mode != 0) {
-		return false;
-	}
-	const bool depth_copy = override.force_z_dirty && override.force_z_valid &&
-	                        z.z_info.format != Prospero::DepthFormat::kInvalid &&
-	                        z.z_read_base_addr != 0 && z.z_write_base_addr != 0 &&
-	                        z.z_read_base_addr != z.z_write_base_addr;
-	const bool stencil_copy = override.force_stencil_dirty && override.force_stencil_valid &&
-	                          z.stencil_info.format != Prospero::StencilFormat::kInvalid &&
-	                          z.stencil_read_base_addr != 0 && z.stencil_write_base_addr != 0 &&
-	                          z.stencil_read_base_addr != z.stencil_write_base_addr;
-	if (!depth_copy && !stencil_copy) {
-		return false;
-	}
-
-	auto  read_desc  = MakeDepthTargetDesc(buffer, z);
-	auto  write_desc = MakeDepthTargetDesc(buffer, z, true);
-	auto& cache      = m_context.GetTextureCache();
-	const auto read_id = cache.FindImage(read_desc);
-	BindRenderTarget(read_id);
-	const auto write_id = cache.FindImage(write_desc);
-	BindRenderTarget(write_id);
-	cache.UpdateImage(read_id);
-	cache.UpdateImage(write_id);
-	cache.MarkGpuWritten(write_id);
-	auto& source      = cache.GetImage(read_id);
-	auto& destination = cache.GetImage(write_id);
-	EXIT_IF(read_id == write_id || source.backing.format != destination.backing.format);
-
-	auto& scheduler = m_context.GetCommandScheduler();
-	scheduler.EndRendering();
-	const auto command = scheduler.Current().Handle();
-	const ImageSubresourceRange range {read_desc.view_info.base_level, 1,
-	                                  read_desc.view_info.base_layer,
-	                                  read_desc.view_info.layer_count};
-	source.Transit(vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead,
-	               range, command);
-	destination.Transit(vk::ImageLayout::eTransferDstOptimal, vk::AccessFlagBits2::eTransferWrite,
-	                    range, command);
-	std::array<vk::ImageCopy, 2> regions {};
-	uint32_t                   count = 0;
-	for (const auto aspect: {vk::ImageAspectFlagBits::eDepth, vk::ImageAspectFlagBits::eStencil}) {
-		if ((aspect == vk::ImageAspectFlagBits::eDepth && !depth_copy) ||
-		    (aspect == vk::ImageAspectFlagBits::eStencil && !stencil_copy)) {
-			continue;
-		}
-		auto& region = regions[count++];
-		region.srcSubresource = {aspect, range.base_level, range.base_layer,
-		                         range.layer_count};
-		region.dstSubresource = region.srcSubresource;
-		region.extent = write_desc.info.extent;
-	}
-	command.copyImage(source.backing.image, vk::ImageLayout::eTransferSrcOptimal,
-	                  destination.backing.image, vk::ImageLayout::eTransferDstOptimal,
-	                  count, regions.data());
-	return true;
-}
-
 vk::ImageAspectFlags RenderDepthInfo::AttachmentWriteAspects() const {
-	const auto format = desc.view_info.format;
 	if (format == vk::Format::eUndefined) {
 		return {};
 	}
@@ -422,20 +361,21 @@ vk::ImageAspectFlags RenderDepthInfo::AttachmentWriteAspects() const {
 	const auto           available = ImageViewOps::DepthAspectMask(format);
 	vk::ImageAspectFlags writes {};
 	if ((available & vk::ImageAspectFlagBits::eDepth) &&
-	    (depth_load_clear_enable || depth_write_enable)) {
+	    (depth_load_clear_enable || (depth_test_enable && depth_write_enable))) {
 		writes |= vk::ImageAspectFlagBits::eDepth;
 	}
 	if (!(available & vk::ImageAspectFlagBits::eStencil)) {
 		return writes;
 	}

-	const auto face_writes = [&](const vk::StencilOpState& state) {
-		if (state.writeMask == 0) {
+	const auto face_writes = [&](const PipelineStencilStaticState&  state,
+	                             const PipelineStencilDynamicState& dynamic) {
+		if (dynamic.writeMask == 0) {
 			return false;
 		}
 		bool can_pass = state.compareOp != vk::CompareOp::eNever;
 		bool can_fail = state.compareOp != vk::CompareOp::eAlways;
-		if (state.compareMask == 0) {
+		if (dynamic.compareMask == 0) {
 			switch (state.compareOp) {
 				case vk::CompareOp::eEqual:
 				case vk::CompareOp::eLessOrEqual:
@@ -461,7 +401,8 @@ vk::ImageAspectFlags RenderDepthInfo::AttachmentWriteAspects() const {
 		       (can_pass && depth_fail && state.depthFailOp != vk::StencilOp::eKeep);
 	};
 	if (stencil_clear_enable ||
-	    (stencil_test_enable && (face_writes(stencil_front) || face_writes(stencil_back)))) {
+	    (stencil_test_enable && (face_writes(stencil_static_front, stencil_dynamic_front) ||
+	                             face_writes(stencil_static_back, stencil_dynamic_back)))) {
 		writes |= vk::ImageAspectFlagBits::eStencil;
 	}
 	return writes;
@@ -11,21 +11,34 @@

 namespace Libs::Graphics {

-// Without a stencil plane, Hi-Stencil fields are inactive. An active plane is compatible when
-// Hi-Stencil is disabled or HTile backing is present.
+class RenderCommandBuffer;
+
 inline constexpr bool depth_htile_stencil_acceleration_compatible(bool has_stencil, bool has_htile,
                                                                  bool htile_stencil_disabled) {
-	return !has_stencil || htile_stencil_disabled || has_htile;
+	return htile_stencil_disabled || (has_stencil && has_htile);
 }

 struct RenderDepthInfo {
-	// Discovery keeps guest image information but can remap the view into a larger cache image.
 	TextureCache::ImageDesc     desc;
+	vk::Format                  format                   = vk::Format::eUndefined;
+	uint32_t                    width                    = 0;
+	uint32_t                    height                   = 0;
+	uint32_t                    samples                  = 1;
+	bool                        htile                    = false;
+	uint64_t                    depth_buffer_size        = 0;
+	uint64_t                    depth_buffer_vaddr       = 0;
+	uint64_t                    depth_tile_swizzle       = 0;
+	uint64_t                    stencil_buffer_size      = 0;
+	uint64_t                    stencil_buffer_vaddr     = 0;
+	uint64_t                    stencil_tile_swizzle     = 0;
+	uint64_t                    htile_buffer_size        = 0;
+	uint64_t                    htile_buffer_vaddr       = 0;
+	uint64_t                    htile_tile_swizzle       = 0;
 	bool                        depth_clear_enable       = false;
 	bool                        depth_load_clear_enable  = false;
+	bool                        depth_meta_clear_enable  = false;
 	float                       depth_clear_value        = 0.0f;
 	bool                        depth_test_enable        = false;
-	// Effective draw writes; discovery applies test, target-write and clear controls.
 	bool                        depth_write_enable       = false;
 	vk::CompareOp               depth_compare_op         = vk::CompareOp::eNever;
 	bool                        depth_bounds_test_enable = false;
@@ -34,20 +47,29 @@ struct RenderDepthInfo {
 	bool                        stencil_clear_enable     = false;
 	uint8_t                     stencil_clear_value      = 0;
 	bool                        stencil_test_enable      = false;
-	vk::StencilOpState          stencil_front;
-	vk::StencilOpState          stencil_back;
+	PipelineStencilStaticState  stencil_static_front;
+	PipelineStencilStaticState  stencil_static_back;
+	PipelineStencilDynamicState stencil_dynamic_front;
+	PipelineStencilDynamicState stencil_dynamic_back;
 	ImageId                     image_id;
+	vk::ImageView               image_view = nullptr;
+	uint64_t                    vaddr[3]   = {};
+	uint64_t                    size[3]    = {};
+	int                         vaddr_num  = 0;

 	[[nodiscard]] vk::ImageAspectFlags AttachmentWriteAspects() const;
 };

+inline bool depth_attachment_read_only(const RenderDepthInfo& depth) {
+	return !depth.AttachmentWriteAspects();
+}
+
 inline vk::ImageLayout depth_attachment_layout(const RenderDepthInfo& depth) {
-	const auto available     = ImageViewOps::DepthAspectMask(depth.desc.view_info.format);
+	const auto available     = ImageViewOps::DepthAspectMask(depth.format);
 	const auto writes        = depth.AttachmentWriteAspects();
 	const bool has_depth     = static_cast<bool>(available & vk::ImageAspectFlagBits::eDepth);
 	const bool has_stencil   = static_cast<bool>(available & vk::ImageAspectFlagBits::eStencil);
-	// LoadOp clears are separate from guest depth writes.
-	const bool depth_write   = depth.depth_write_enable;
+	const bool depth_write   = static_cast<bool>(writes & vk::ImageAspectFlagBits::eDepth);
 	const bool stencil_write = static_cast<bool>(writes & vk::ImageAspectFlagBits::eStencil);
 	if (!has_stencil) {
 		return depth_write ? vk::ImageLayout::eDepthAttachmentOptimal