Qt: Clean up Setup Wizard's labels

Fixes issues with AA1 with flat shading.

.github/workflows/cron_update_base_translation.yml

@@ -11,7 +11,7 @@ jobs:
     name: "Update Base Translation"
     runs-on: ubuntu-22.04
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
 
       - name: Update Base Translation
         run: ./.github/workflows/scripts/common/update_base_translation.sh

.github/workflows/cron_update_controller_db.yml

@@ -9,7 +9,7 @@ jobs:
     if: github.repository == 'PCSX2/pcsx2'
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
 
       - name: Get Latest DB and Prepare DB File
         run: |

.github/workflows/lint_gamedb.yml

@@ -18,7 +18,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout Repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Install Packages
         run: |

.github/workflows/linux_build_flatpak.yml

@@ -51,7 +51,7 @@ jobs:
 
     steps:
       - name: Checkout Repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
           set-safe-directory: ${{ env.GITHUB_WORKSPACE }}
           # 10 here, since the odds of having 10 untagged commits in a row should be slim to none
@@ -157,4 +157,21 @@ jobs:
         with:
           name: ${{ steps.artifact-metadata.outputs.artifact-name }}
           path: ci-artifacts
+  report-error:
+    runs-on: ubuntu-latest
+    needs: build_linux
+    if: always() && needs.build_linux.result == 'failure' && inputs.publish == true
 
+    steps:
+      - uses: actions/checkout@v7
+
+      - name: Push Discord Webhook
+        env:
+          OWNER: PCSX2
+          REPO: pcsx2
+          DISCORD_BUILD_WEBHOOK: ${{ secrets.DISCORD_BUILD_WEBHOOK }}
+          GITHUB_TOKEN: ${{ github.token }}
+        run: |
+          cd ./.github/workflows/scripts/releases/announce-release
+          npm ci
+          FLATHUB_FAILURE=1 node index.js

.github/workflows/linux_build_matrix.yml

@@ -20,6 +20,18 @@ jobs:
       cmakeflags: ""
       buildAppImage: true
     secrets: inherit
+  build_linux_debug:
+    name: "Debug"
+    if: github.repository != 'PCSX2/pcsx2' || github.event_name == 'pull_request'
+    uses: ./.github/workflows/linux_build_qt.yml
+    with:
+      jobName: "Debug Build"
+      artifactPrefixName: "PCSX2-linux-Qt-x64-debug"
+      compiler: clang
+      cmakeflags: ""
+      buildAppImage: true
+      cmakeBuildType: Debug
+    secrets: inherit
   build_linux_flatpak:
     name: "Flatpak"
     if: github.repository != 'PCSX2/pcsx2' || github.event_name == 'pull_request'

.github/workflows/linux_build_qt.yml

@@ -39,6 +39,10 @@ on:
         required: false
         type: boolean
         default: false
+      cmakeBuildType:
+        required: false
+        type: string
+        default: Release
 
 jobs:
   build_linux:
@@ -55,7 +59,7 @@ jobs:
 
     steps:
       - name: Checkout Repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
         with:
            fetch-depth: 0
 
@@ -129,7 +133,7 @@ jobs:
           ADDITIONAL_CMAKE_ARGS: ${{ inputs.cmakeflags }}
         run: |
           cmake -B build -G Ninja \
-            -DCMAKE_BUILD_TYPE=Release \
+            -DCMAKE_BUILD_TYPE="${{ inputs.cmakeBuildType }}" \
             -DCMAKE_INTERPROCEDURAL_OPTIMIZATION=ON \
             -DCMAKE_PREFIX_PATH="$HOME/deps" \
             -DCMAKE_C_COMPILER=clang-17 \

.github/workflows/macos_build.yml

@@ -52,7 +52,7 @@ jobs:
 
     steps:
       - name: Checkout Repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       # actions/checkout elides tags, fetch them primarily for releases
       - name: Fetch Tags

.github/workflows/release_cut_new.yml

@@ -35,7 +35,7 @@ jobs:
     outputs:
       new_tag: ${{ steps.tag_version.outputs.new_tag }}
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
 
       # Docs - https://github.com/mathieudutour/github-tag-action
       - name: Bump Version and Push Tag
@@ -176,7 +176,7 @@ jobs:
     name: "Upload Artifacts"
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
 
       - name: Prepare Artifact Folder
         run: mkdir ./ci-artifacts/

.github/workflows/scripts/linux/build-dependencies-qt.sh

@@ -23,11 +23,11 @@ QTAPNG=1.3.0
 
 FFMPEG=8.1
 LIBBACKTRACE=ad106d5fdd5d960bd33fae1c48a351af567fd075
-LIBJPEGTURBO=3.1.4.1
+LIBJPEGTURBO=3.2.0
 LIBPNG=1.6.58
 LIBWEBP=1.6.0
 NVENC=13.0.19.0
-SDL=SDL3-3.4.8
+SDL=SDL3-3.4.12
 LZ4=1.10.0
 VULKAN=1.4.328.1
 ZSTD=1.5.7
@@ -57,10 +57,10 @@ f1d3be3489f758efe1a8f12118a212febbe611aa670af32e0159fa3c1feab2a6  QtApng-$QTAPNG
 
 b072aed6871998cce9b36e7774033105ca29e33632be5b6347f3206898e0756a  ffmpeg-$FFMPEG.tar.xz
 96e5c2d7f2c482a60d5804da48a2eb9a0db0719b2c65dcc169fbfdcf37f3a45d  libbacktrace-$LIBBACKTRACE.tar.gz
-ecae8008e2cc9ade2f2c1bb9d5e6d4fb73e7c433866a056bd82980741571a022  libjpeg-turbo-$LIBJPEGTURBO.tar.gz
+6f30092cef9fb839779646608f4ee14ae3cbac989c47fa05e841b0841f09878e  libjpeg-turbo-$LIBJPEGTURBO.tar.gz
 28eb403f51f0f7405249132cecfe82ea5c0ef97f1b32c5a65828814ae0d34775  libpng-$LIBPNG.tar.xz
 e4ab7009bf0629fd11982d4c2aa83964cf244cffba7347ecd39019a9e38c4564  libwebp-$LIBWEBP.tar.gz
-e9fff7467fb60f037e6708da18b25560649e4c63edc2a69bb871b960d9cbfbba  $SDL.tar.gz
+f07b958a9ac5020fb7a44cadb957f658b2149c3c8abb4f63145fac9303249db7  $SDL.tar.gz
 eee7dea22ed502868017971c86c63c4ed1e6085de0baebfdcc3d3322f00f3eb0  libpng-$LIBPNG-apng.patch.gz
 537512904744b35e232912055ccf8ec66d768639ff3abe5788d90d792ec5f48b  lz4-$LZ4.tar.gz
 13da39edb3a40ed9713ae390ca89faa2f1202c9dda869ef306a8d4383e242bee  nv-codec-headers-$NVENC.tar.gz

.github/workflows/scripts/linux/flatpak/modules/20-sdl3.json

@@ -14,8 +14,8 @@
   "sources": [
     {
       "type": "archive",
-      "url": "https://libsdl.org/release/SDL3-3.4.8.tar.gz",
-      "sha256": "e9fff7467fb60f037e6708da18b25560649e4c63edc2a69bb871b960d9cbfbba"
+      "url": "https://libsdl.org/release/SDL3-3.4.12.tar.gz",
+      "sha256": "f07b958a9ac5020fb7a44cadb957f658b2149c3c8abb4f63145fac9303249db7"
     }
   ],
   "cleanup": [

.github/workflows/scripts/linux/pcsx2-qt.desktop

@@ -3,7 +3,7 @@ Version=1.0
 Terminal=false
 Type=Application
 Name=PCSX2
-StartupWMClass=PCSX2
+StartupWMClass=pcsx2-qt
 GenericName=PlayStation 2 Emulator
 Comment=Sony PlayStation 2 emulator
 Exec=pcsx2-qt

.github/workflows/scripts/macos/build-dependencies-universal.sh

@@ -47,12 +47,12 @@ QT=6.11.1
 QTAPNG=1.3.0
 
 FREETYPE=2.14.3
-SDL=SDL3-3.4.8
+SDL=SDL3-3.4.12
 HARFBUZZ=14.2.0
 ZSTD=1.5.7
 LZ4=1.10.0
 LIBPNG=1.6.58
-LIBJPEGTURBO=3.1.4.1
+LIBJPEGTURBO=3.2.0
 LIBWEBP=1.6.0
 FFMPEG=8.1
 MOLTENVK=1.4.1
@@ -93,14 +93,14 @@ b2bf6c6845ac175ed7f819145483ba4676f617aaa6a5012c8efee63c8bbac413  qtimageformats
 f1d3be3489f758efe1a8f12118a212febbe611aa670af32e0159fa3c1feab2a6  QtApng-$QTAPNG.tar.gz
 
 36bc4f1cc413335368ee656c42afca65c5a3987e8768cc28cf11ba775e785a5f  freetype-$FREETYPE.tar.xz
-e9fff7467fb60f037e6708da18b25560649e4c63edc2a69bb871b960d9cbfbba  $SDL.tar.gz
+f07b958a9ac5020fb7a44cadb957f658b2149c3c8abb4f63145fac9303249db7  $SDL.tar.gz
 c652d5d94971031654ab3989891a490a895d3e3f2b71171c62692b28e94b1b93  harfbuzz-$HARFBUZZ.tar.gz
 eb33e51f49a15e023950cd7825ca74a4a2b43db8354825ac24fc1b7ee09e6fa3  zstd-$ZSTD.tar.gz
 537512904744b35e232912055ccf8ec66d768639ff3abe5788d90d792ec5f48b  lz4-$LZ4.tar.gz
 28eb403f51f0f7405249132cecfe82ea5c0ef97f1b32c5a65828814ae0d34775  libpng-$LIBPNG.tar.xz
 e4ab7009bf0629fd11982d4c2aa83964cf244cffba7347ecd39019a9e38c4564  libwebp-$LIBWEBP.tar.gz
 eee7dea22ed502868017971c86c63c4ed1e6085de0baebfdcc3d3322f00f3eb0  libpng-$LIBPNG-apng.patch.gz
-ecae8008e2cc9ade2f2c1bb9d5e6d4fb73e7c433866a056bd82980741571a022  libjpeg-turbo-$LIBJPEGTURBO.tar.gz
+6f30092cef9fb839779646608f4ee14ae3cbac989c47fa05e841b0841f09878e  libjpeg-turbo-$LIBJPEGTURBO.tar.gz
 b072aed6871998cce9b36e7774033105ca29e33632be5b6347f3206898e0756a  ffmpeg-$FFMPEG.tar.xz
 9985f141902a17de818e264d17c1ce334b748e499ee02fcb4703e4dc0038f89c  MoltenVK-$MOLTENVK.tar.gz
 51dbf24fe72e43dd7cb9a289d3cab47112010f1a2ed69b6fc8ac0dff31991ed2  KDDockWidgets-$KDDOCKWIDGETS.tar.gz

.github/workflows/scripts/macos/build-dependencies.sh

@@ -24,12 +24,12 @@ QT=6.11.1
 QTAPNG=1.3.0
 
 FREETYPE=2.14.3
-SDL=SDL3-3.4.8
+SDL=SDL3-3.4.12
 HARFBUZZ=14.2.0
 ZSTD=1.5.7
 LZ4=1.10.0
 LIBPNG=1.6.58
-LIBJPEGTURBO=3.1.4.1
+LIBJPEGTURBO=3.2.0
 LIBWEBP=1.6.0
 FFMPEG=8.1
 MOLTENVK=1.4.1
@@ -69,14 +69,14 @@ b2bf6c6845ac175ed7f819145483ba4676f617aaa6a5012c8efee63c8bbac413  qtimageformats
 f1d3be3489f758efe1a8f12118a212febbe611aa670af32e0159fa3c1feab2a6  QtApng-$QTAPNG.tar.gz
 
 36bc4f1cc413335368ee656c42afca65c5a3987e8768cc28cf11ba775e785a5f  freetype-$FREETYPE.tar.xz
-e9fff7467fb60f037e6708da18b25560649e4c63edc2a69bb871b960d9cbfbba  $SDL.tar.gz
+f07b958a9ac5020fb7a44cadb957f658b2149c3c8abb4f63145fac9303249db7  $SDL.tar.gz
 c652d5d94971031654ab3989891a490a895d3e3f2b71171c62692b28e94b1b93  harfbuzz-$HARFBUZZ.tar.gz
 eb33e51f49a15e023950cd7825ca74a4a2b43db8354825ac24fc1b7ee09e6fa3  zstd-$ZSTD.tar.gz
 537512904744b35e232912055ccf8ec66d768639ff3abe5788d90d792ec5f48b  lz4-$LZ4.tar.gz
 28eb403f51f0f7405249132cecfe82ea5c0ef97f1b32c5a65828814ae0d34775  libpng-$LIBPNG.tar.xz
 e4ab7009bf0629fd11982d4c2aa83964cf244cffba7347ecd39019a9e38c4564  libwebp-$LIBWEBP.tar.gz
 eee7dea22ed502868017971c86c63c4ed1e6085de0baebfdcc3d3322f00f3eb0  libpng-$LIBPNG-apng.patch.gz
-ecae8008e2cc9ade2f2c1bb9d5e6d4fb73e7c433866a056bd82980741571a022  libjpeg-turbo-$LIBJPEGTURBO.tar.gz
+6f30092cef9fb839779646608f4ee14ae3cbac989c47fa05e841b0841f09878e  libjpeg-turbo-$LIBJPEGTURBO.tar.gz
 b072aed6871998cce9b36e7774033105ca29e33632be5b6347f3206898e0756a  ffmpeg-$FFMPEG.tar.xz
 9985f141902a17de818e264d17c1ce334b748e499ee02fcb4703e4dc0038f89c  v$MOLTENVK.tar.gz
 51dbf24fe72e43dd7cb9a289d3cab47112010f1a2ed69b6fc8ac0dff31991ed2  KDDockWidgets-$KDDOCKWIDGETS.tar.gz

.github/workflows/scripts/releases/announce-release/index.js

@@ -38,39 +38,54 @@ const octokit = new Octokit({
   }
 });
 
-if (process.env.TAG_VAL === undefined || process.env.TAG_VAL === "") {
-  console.log(`Not announcing - TAG_VAL not defined`);
-  process.exit(1);
+let embed;
+
+if (process.env.FLATHUB_FAILURE === undefined)
+{
+  if (process.env.TAG_VAL === undefined || process.env.TAG_VAL === "") {
+    console.log(`Not announcing - TAG_VAL not defined`);
+    process.exit(1);
+  }
+
+  const { data: releaseInfo } = await octokit.rest.repos.getReleaseByTag({
+    owner: owner,
+    repo: repo,
+    tag: process.env.TAG_VAL,
+  });
+
+  if (releaseInfo === undefined) {
+    console.log(`Not announcing - could not locate release with tag ${process.env.TAG_VAL}`);
+    process.exit(1);
+  }
+
+  if (!releaseInfo.prerelease) {
+    console.log("Not announcing - release was not a pre-release (aka a Nightly)");
+    process.exit(0);
+  }
+
+  // Publish Webhook
+  embed = new MessageEmbed()
+    .setColor('#FF8000')
+    .setTitle('New PCSX2 Nightly Build Available!')
+    .setDescription("To download the latest or previous builds, [visit the official downloads page](https://pcsx2.net/downloads/).")
+    .addFields(
+      { name: 'Version', value: releaseInfo.tag_name, inline: true },
+      { name: 'Installation Steps', value: '[See Here](https://pcsx2.net/docs/category/setup)', inline: true },
+      { name: 'Included Changes', value: releaseInfo.body, inline: false }
+    );
+  console.log(embed);
 }
+else
+{
+  // Flathub upload failure
 
-const { data: releaseInfo } = await octokit.rest.repos.getReleaseByTag({
-  owner: owner,
-  repo: repo,
-  tag: process.env.TAG_VAL,
-});
-
-if (releaseInfo === undefined) {
-  console.log(`Not announcing - could not locate release with tag ${process.env.TAG_VAL}`);
-  process.exit(1);
+  embed = new MessageEmbed()
+    .setColor('#FF0000')
+    .setTitle('PCSX2 Failed to Build/Upload to FlatHub')
+    .setDescription("Please check the latest Flathub build job to determine the root cause.")
+  console.log(embed);
 }
 
-if (!releaseInfo.prerelease) {
-  console.log("Not announcing - release was not a pre-release (aka a Nightly)");
-  process.exit(0);
-}
-
-// Publish Webhook
-const embed = new MessageEmbed()
-  .setColor('#FF8000')
-  .setTitle('New PCSX2 Nightly Build Available!')
-  .setDescription("To download the latest or previous builds, [visit the official downloads page](https://pcsx2.net/downloads/).")
-  .addFields(
-    { name: 'Version', value: releaseInfo.tag_name, inline: true },
-    { name: 'Installation Steps', value: '[See Here](https://pcsx2.net/docs/category/setup)', inline: true },
-    { name: 'Included Changes', value: releaseInfo.body, inline: false }
-  );
-console.log(embed);
-
 // Get all webhooks, simple comma-sep string
 const webhookUrls = process.env.DISCORD_BUILD_WEBHOOK.split(",");
 

.github/workflows/scripts/windows/build-dependencies-arm64.bat

@@ -60,8 +60,8 @@ set QTAPNG=1.3.0
 
 set FREETYPE=2.14.3
 set HARFBUZZ=14.2.0
-set SDL=SDL3-3.4.8
-set LIBJPEGTURBO=3.1.4.1
+set SDL=SDL3-3.4.12
+set LIBJPEGTURBO=3.2.0
 set LIBPNG=1658
 set LIBPNGLONG=1.6.58
 set LZ4=1.10.0
@@ -92,9 +92,9 @@ call :downloadfile "freetype-%FREETYPE%.tar.gz" https://sourceforge.net/projects
 call :downloadfile "harfbuzz-%HARFBUZZ%.zip" https://github.com/harfbuzz/harfbuzz/archive/refs/tags/%HARFBUZZ%.zip bb2f83255706b1c92d731541c7cefaf98bb5b93e8f76d16f6deda05225ff20ee || goto error
 call :downloadfile "lpng%LIBPNG%.zip" https://download.sourceforge.net/libpng/lpng1658.zip b32f170855dbbe3e6d9e645af40b538137041773672c3ba3e02db5816c82d376 || goto error
 call :downloadfile "lpng%LIBPNG%-apng.patch.gz" https://download.sourceforge.net/libpng-apng/libpng-%LIBPNGLONG%-apng.patch.gz eee7dea22ed502868017971c86c63c4ed1e6085de0baebfdcc3d3322f00f3eb0 || goto error
-call :downloadfile "libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" "https://github.com/libjpeg-turbo/libjpeg-turbo/releases/download/%LIBJPEGTURBO%/libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" ecae8008e2cc9ade2f2c1bb9d5e6d4fb73e7c433866a056bd82980741571a022 || goto error
+call :downloadfile "libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" "https://github.com/libjpeg-turbo/libjpeg-turbo/releases/download/%LIBJPEGTURBO%/libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" 6f30092cef9fb839779646608f4ee14ae3cbac989c47fa05e841b0841f09878e || goto error
 call :downloadfile "libwebp-%WEBP%.tar.gz" "https://storage.googleapis.com/downloads.webmproject.org/releases/webp/libwebp-%WEBP%.tar.gz" e4ab7009bf0629fd11982d4c2aa83964cf244cffba7347ecd39019a9e38c4564 || goto error
-call :downloadfile "%SDL%.zip" "https://libsdl.org/release/%SDL%.zip" 506206e02f90c1f37e048eacaf9e8a3a7dc682fd27783eb0ff15a7d2dcc9c2af || goto error
+call :downloadfile "%SDL%.zip" "https://libsdl.org/release/%SDL%.zip" 3d4de8967a49c0451e775a0c1e9022092c19fdef41ba38a83fcf031c5a6496e2 || goto error
 call :downloadfile "lz4-%LZ4%.zip" "https://github.com/lz4/lz4/archive/refs/tags/v%LZ4%.zip" 3224b4c80f351f194984526ef396f6079bd6332dd9825c72ac0d7a37b3cdc565 || goto error
 call :downloadfile "zlib%ZLIBSHORT%.zip" "https://github.com/madler/zlib/releases/download/v%ZLIB%/zlib%ZLIBSHORT%.zip" e8bf55f3017aa181690990cb58a994e77885da140609fc8f94abe9b65d2cae28 || goto error
 call :downloadfile "zstd-%ZSTD%.zip" "https://github.com/facebook/zstd/archive/refs/tags/v%ZSTD%.zip" 7897bc5d620580d9b7cd3539c44b59d78f3657d33663fe97a145e07b4ebd69a4 || goto error

.github/workflows/scripts/windows/build-dependencies.bat

@@ -77,8 +77,8 @@ set LIBX264=b35605ace3ddf7c1a5d67a2eb553f034aef41d55
 
 set FREETYPE=2.14.3
 set HARFBUZZ=14.2.0
-set SDL=SDL3-3.4.8
-set LIBJPEGTURBO=3.1.4.1
+set SDL=SDL3-3.4.12
+set LIBJPEGTURBO=3.2.0
 set LIBPNG=1658
 set LIBPNGLONG=1.6.58
 set LZ4=1.10.0
@@ -97,6 +97,7 @@ set SHADERC_SPIRVHEADERS=58006c901d1d5c37dece6b6610e9af87fa951375
 set SHADERC_SPIRVTOOLS=6337eb62cadd7d124ac6789bf39c0f71148f0a73
 
 set AGILITYSDK=1.619.2
+set DXHEADERS=1.619.1
 
 call :downloadfile "qtbase-everywhere-src-%QT%.zip" "https://download.qt.io/official_releases/qt/%QTMINOR%/%QT%/submodules/qtbase-everywhere-src-%QT%.zip" 3529cc37297a5a7aae4486843b9fd41c30df1d79a770f85e240b537dcc327ca5 || goto error
 call :downloadfile "qtimageformats-everywhere-src-%QT%.zip" "https://download.qt.io/official_releases/qt/%QTMINOR%/%QT%/submodules/qtimageformats-everywhere-src-%QT%.zip" 37fba768f2780580dfae535ad6654cb9dc0bf2272e71b9b9781988de9ed0dac0 || goto error
@@ -119,9 +120,9 @@ call :downloadfile "freetype-%FREETYPE%.tar.gz" https://sourceforge.net/projects
 call :downloadfile "harfbuzz-%HARFBUZZ%.zip" https://github.com/harfbuzz/harfbuzz/archive/refs/tags/%HARFBUZZ%.zip bb2f83255706b1c92d731541c7cefaf98bb5b93e8f76d16f6deda05225ff20ee || goto error
 call :downloadfile "lpng%LIBPNG%.zip" https://download.sourceforge.net/libpng/lpng1658.zip b32f170855dbbe3e6d9e645af40b538137041773672c3ba3e02db5816c82d376 || goto error
 call :downloadfile "lpng%LIBPNG%-apng.patch.gz" https://download.sourceforge.net/libpng-apng/libpng-%LIBPNGLONG%-apng.patch.gz eee7dea22ed502868017971c86c63c4ed1e6085de0baebfdcc3d3322f00f3eb0 || goto error
-call :downloadfile "libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" "https://github.com/libjpeg-turbo/libjpeg-turbo/releases/download/%LIBJPEGTURBO%/libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" ecae8008e2cc9ade2f2c1bb9d5e6d4fb73e7c433866a056bd82980741571a022 || goto error
+call :downloadfile "libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" "https://github.com/libjpeg-turbo/libjpeg-turbo/releases/download/%LIBJPEGTURBO%/libjpeg-turbo-%LIBJPEGTURBO%.tar.gz" 6f30092cef9fb839779646608f4ee14ae3cbac989c47fa05e841b0841f09878e || goto error
 call :downloadfile "libwebp-%WEBP%.tar.gz" "https://storage.googleapis.com/downloads.webmproject.org/releases/webp/libwebp-%WEBP%.tar.gz" e4ab7009bf0629fd11982d4c2aa83964cf244cffba7347ecd39019a9e38c4564 || goto error
-call :downloadfile "%SDL%.zip" "https://libsdl.org/release/%SDL%.zip" 506206e02f90c1f37e048eacaf9e8a3a7dc682fd27783eb0ff15a7d2dcc9c2af || goto error
+call :downloadfile "%SDL%.zip" "https://libsdl.org/release/%SDL%.zip" 3d4de8967a49c0451e775a0c1e9022092c19fdef41ba38a83fcf031c5a6496e2 || goto error
 call :downloadfile "lz4-%LZ4%.zip" "https://github.com/lz4/lz4/archive/refs/tags/v%LZ4%.zip" 3224b4c80f351f194984526ef396f6079bd6332dd9825c72ac0d7a37b3cdc565 || goto error
 call :downloadfile "zlib%ZLIBSHORT%.zip" "https://github.com/madler/zlib/releases/download/v%ZLIB%/zlib%ZLIBSHORT%.zip" e8bf55f3017aa181690990cb58a994e77885da140609fc8f94abe9b65d2cae28 || goto error
 call :downloadfile "zstd-%ZSTD%.zip" "https://github.com/facebook/zstd/archive/refs/tags/v%ZSTD%.zip" 7897bc5d620580d9b7cd3539c44b59d78f3657d33663fe97a145e07b4ebd69a4 || goto error
@@ -129,6 +130,7 @@ call :downloadfile "KDDockWidgets-%KDDOCKWIDGETS%.zip" "https://github.com/KDAB/
 call :downloadfile "plutovg-%PLUTOVG%.zip" "https://github.com/sammycage/plutovg/archive/v%PLUTOVG%.zip" 4fe4e48f28aa80171b2166d45c0976ab0f21eecedb52cd4c3ef73b5afb48fac9 || goto error
 call :downloadfile "plutosvg-%PLUTOSVG%.zip" "https://github.com/sammycage/plutosvg/archive/v%PLUTOSVG%.zip" 82dee2c57ad712bdd6d6d81d3e76249d89caa4b5a4214353660fd5adff12201a || goto error
 call :downloadfile "agility-sdk-%AGILITYSDK%.nupkg" "https://www.nuget.org/api/v2/package/Microsoft.Direct3D.D3D12/%AGILITYSDK%" eb92d90bb23b2ec23410c41d791e41dbdbec942ab946924d1fdcb31eac6f0735 || goto error
+call :downloadfile "DirectX-Headers-%DXHEADERS%.zip" "https://github.com/microsoft/DirectX-Headers/archive/v%DXHEADERS%.zip" 9eb8b102a90a42e4ea72a825f7d249d55ec90d164f030966c9b7784b93374927 || goto error
 call :downloadfile "rapidyaml-%RAPIDYAML%-src.zip" "https://github.com/biojppm/rapidyaml/releases/download/v%RAPIDYAML%/rapidyaml-%RAPIDYAML%-src.zip" 96276f55b9fa7837ac8f3f72fd52965879cbb5d5d2e6af548c69a177fb078304 || goto error
 
 call :downloadfile "shaderc-%SHADERC%.zip" "https://github.com/google/shaderc/archive/refs/tags/v%SHADERC%.zip" f9401cc5cb36c276cd1e072b6595dbd728148e8dba389e50f7339e2d388dbc08 || goto error
@@ -469,7 +471,6 @@ echo "Building PlutoVG..."
 rmdir /S /Q "plutovg-%PLUTOVG%"
 %SEVENZIP% x "plutovg-%PLUTOVG%.zip" || goto error
 cd "plutovg-%PLUTOVG%" || goto error
-%PATCH% -p1 < "%SCRIPTDIR%\plutovg-workaround-msvc-ice.patch" || goto error
 cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_PREFIX_PATH="%INSTALLDIR%" -DCMAKE_INSTALL_PREFIX="%INSTALLDIR%" -DBUILD_SHARED_LIBS=ON -DPLUTOVG_BUILD_EXAMPLES=OFF -B build -G Ninja || goto error
 cmake --build build --parallel || goto error
 ninja -C build install || goto error
@@ -507,6 +508,16 @@ if %DEBUG%==1 (
 )
 cd .. || goto error
 
+rem DirectX Headers include a CMakeList file, which is absent in the Nuget package
+echo Unpacking DirectX Headers
+rmdir /S /Q "DirectX-Headers-%DXHEADERS%"
+%SEVENZIP% x "DirectX-Headers-%DXHEADERS%.zip" || goto error
+cd "DirectX-Headers-%DXHEADERS%" || goto error
+cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_PREFIX_PATH="%INSTALLDIR%" -DCMAKE_INSTALL_PREFIX="%INSTALLDIR%" -DDXHEADERS_BUILD_TEST=OFF -DDXHEADERS_BUILD_GOOGLE_TEST=OFF -B build -G Ninja || goto error
+cmake --build build --parallel || goto error
+ninja -C build install || goto error
+cd .. || goto error
+
 echo Building shaderc...
 rmdir /S /Q "shaderc-%SHADERC%"
 %SEVENZIP% x "shaderc-%SHADERC%.zip" || goto error

.github/workflows/scripts/windows/plutovg-workaround-msvc-ice.patch

@@ -1,52 +0,0 @@
-diff --git a/source/plutovg-font.c b/source/plutovg-font.c
-index 0ad89b1..110de7f 100644
---- a/source/plutovg-font.c
-+++ b/source/plutovg-font.c
-@@ -836,7 +836,8 @@ int plutovg_font_face_cache_load_file(plutovg_font_face_cache_t* cache, const ch
-         size_t family_index = 0;
-         if(unicode_family_name) {
-             const stbtt_uint8* family_name = unicode_family_name;
--            while(family_length) {
-+            //while(family_length) {
-+            for (int i = family_length; i > 0; i -= 2) {
-                 stbtt_uint16 ch = family_name[0] * 256 + family_name[1];
-                 if(ch < 0x80) {
-                     entry->family[family_index++] = ch;
-@@ -853,7 +854,8 @@ int plutovg_font_face_cache_load_file(plutovg_font_face_cache_t* cache, const ch
-                     entry->family[family_index++] = (0x80 + ((c) & 0x3f));
- 
-                     family_name += 2;
--                    family_length -= 2;
-+                    //family_length -= 2;
-+                    i -= 2;
-                 } else {
-                     entry->family[family_index++] = (0xe0 + (ch >> 12));
-                     entry->family[family_index++] = (0x80 + ((ch >> 6) & 0x3f));
-@@ -861,7 +863,7 @@ int plutovg_font_face_cache_load_file(plutovg_font_face_cache_t* cache, const ch
-                 }
- 
-                 family_name += 2;
--                family_length -= 2;
-+                //family_length -= 2;
-             }
- 
-             entry->family[family_index] = '\0';
-@@ -902,7 +904,8 @@ int plutovg_font_face_cache_load_file(plutovg_font_face_cache_t* cache, const ch
-             };
- 
-             const stbtt_uint8* family_name = roman_family_name;
--            while(family_length) {
-+            //while(family_length) {
-+            for (int i2 = family_length; i2 > 0; i2--) {
-                 stbtt_uint16 ch = MAC_ROMAN_TABLE[family_name[0]];
-                 if(ch < 0x80) {
-                     entry->family[family_index++] = ch;
-@@ -916,7 +919,7 @@ int plutovg_font_face_cache_load_file(plutovg_font_face_cache_t* cache, const ch
-                 }
- 
-                 family_name += 1;
--                family_length -= 1;
-+                //family_length -= 1;
-             }
- 
-             entry->family[family_index] = '\0';

.github/workflows/windows_build_matrix.yml

@@ -16,7 +16,7 @@ jobs:
     runs-on: windows-2025-vs2026
     steps:
       - name: Checkout Repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
       - name: Verify VS Project Files
         run: .github\workflows\scripts\windows\validate-vs-filters.ps1
 

.github/workflows/windows_build_qt.yml

@@ -56,7 +56,7 @@ jobs:
 
     steps:
       - name: Checkout Repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
 
       - name: Configure MSBuild Clang Version
         if: inputs.configuration != 'CMake'

3rdparty/cpuinfo/include/cpuinfo.h

@@ -361,6 +361,8 @@ enum cpuinfo_uarch {
 	cpuinfo_uarch_raptor_cove = 0x0010020F,
 	/** Intel Redwood Cove microarchitecture (Granite Rapids). */
 	cpuinfo_uarch_redwood_cove = 0x00100210,
+	/** Intel Coyote Cove microarchitecture. */
+	cpuinfo_uarch_coyote_cove = 0x00100211,
 
 	/** Pentium 4 with Willamette, Northwood, or Foster cores. */
 	cpuinfo_uarch_willamette = 0x00100300,

3rdparty/cpuinfo/src/api.c

@@ -1,3 +1,6 @@
+/* for syscall() */
+#define _DEFAULT_SOURCE
+
 #include <stdbool.h>
 #include <stddef.h>
 

3rdparty/cpuinfo/src/arm/api.h

@@ -36,6 +36,7 @@ enum cpuinfo_arm_chipset_vendor {
 	cpuinfo_arm_chipset_vendor_texas_instruments,
 	cpuinfo_arm_chipset_vendor_unisoc,
 	cpuinfo_arm_chipset_vendor_wondermedia,
+	cpuinfo_arm_chipset_vendor_google,
 	cpuinfo_arm_chipset_vendor_max,
 };
 
@@ -73,6 +74,7 @@ enum cpuinfo_arm_chipset_series {
 	cpuinfo_arm_chipset_series_unisoc_t,
 	cpuinfo_arm_chipset_series_unisoc_ums,
 	cpuinfo_arm_chipset_series_wondermedia_wm,
+	cpuinfo_arm_chipset_series_google_tensor,
 	cpuinfo_arm_chipset_series_max,
 };
 

3rdparty/cpuinfo/src/arm/linux/api.h

@@ -366,7 +366,7 @@ CPUINFO_INTERNAL struct cpuinfo_arm_chipset cpuinfo_arm_android_decode_chipset_f
 CPUINFO_INTERNAL struct cpuinfo_arm_chipset cpuinfo_arm_android_decode_chipset_from_ro_hardware_chipname(
 	const char ro_hardware_chipname[restrict static CPUINFO_BUILD_PROP_VALUE_MAX]);
 CPUINFO_INTERNAL struct cpuinfo_arm_chipset cpuinfo_arm_android_decode_chipset_from_ro_soc_model(
-	const char ro_soc_model[restrict static CPUINFO_BUILD_PROP_VALUE_MAX]);
+	const char soc_model[restrict static CPUINFO_BUILD_PROP_VALUE_MAX]);
 #else
 CPUINFO_INTERNAL struct cpuinfo_arm_chipset cpuinfo_arm_linux_decode_chipset_from_proc_cpuinfo_revision(
 	const char proc_cpuinfo_revision[restrict static CPUINFO_REVISION_VALUE_MAX]);

3rdparty/cpuinfo/src/arm/linux/chipset.c

@@ -3492,9 +3492,20 @@ struct cpuinfo_arm_chipset cpuinfo_arm_android_decode_chipset_from_ro_chipname(
 	};
 }
 
+/*
+ * Decodes chipset name from ro.soc.model Android system property.
+ *
+ * @param[in] soc_model - ro.soc.model value.
+ *
+ * @returns Decoded chipset name. If chipset could not be decoded, the resulting
+ * structure would use `unknown` vendor and series identifiers.
+ */
 struct cpuinfo_arm_chipset cpuinfo_arm_android_decode_chipset_from_ro_soc_model(
 	const char soc_model[restrict static CPUINFO_BUILD_PROP_VALUE_MAX]) {
-	struct cpuinfo_arm_chipset chipset;
+	struct cpuinfo_arm_chipset chipset = {
+		.vendor = cpuinfo_arm_chipset_vendor_unknown,
+		.series = cpuinfo_arm_chipset_series_unknown,
+	};
 	const size_t soc_model_length = strnlen(soc_model, CPUINFO_BUILD_PROP_VALUE_MAX);
 	const char* soc_model_end = soc_model + soc_model_length;
 
@@ -3516,10 +3527,28 @@ struct cpuinfo_arm_chipset cpuinfo_arm_android_decode_chipset_from_ro_soc_model(
 		return chipset;
 	}
 
-	return (struct cpuinfo_arm_chipset){
-		.vendor = cpuinfo_arm_chipset_vendor_unknown,
-		.series = cpuinfo_arm_chipset_series_unknown,
-	};
+	if (soc_model[0] != '\0') {
+		if (strncmp(soc_model, "Tensor", 6) == 0) {
+			chipset.vendor = cpuinfo_arm_chipset_vendor_google;
+			chipset.series = cpuinfo_arm_chipset_series_google_tensor;
+			const char* suffix_start = soc_model + 6;
+			while (*suffix_start == ' ') {
+				suffix_start++;
+			}
+			const size_t suffix_length = strnlen(suffix_start, CPUINFO_ARM_CHIPSET_SUFFIX_MAX);
+			if (suffix_length > 0) {
+				strncpy(chipset.suffix, suffix_start, suffix_length);
+			}
+			return chipset;
+		} else if (strncmp(soc_model, "GS201", 5) == 0) {
+			chipset.vendor = cpuinfo_arm_chipset_vendor_google;
+			chipset.series = cpuinfo_arm_chipset_series_google_tensor;
+			strncpy(chipset.suffix, "G2", 2);
+			return chipset;
+		}
+	}
+
+	return chipset;
 }
 #endif /* __ANDROID__ */
 
@@ -3858,6 +3887,7 @@ static const char* chipset_vendor_string[cpuinfo_arm_chipset_vendor_max] = {
 	[cpuinfo_arm_chipset_vendor_texas_instruments] = "Texas Instruments",
 	[cpuinfo_arm_chipset_vendor_unisoc] = "Unisoc",
 	[cpuinfo_arm_chipset_vendor_wondermedia] = "WonderMedia",
+	[cpuinfo_arm_chipset_vendor_google] = "Google",
 };
 
 /* Map from ARM chipset series ID to its string representation */
@@ -3895,6 +3925,7 @@ static const char* chipset_series_string[cpuinfo_arm_chipset_series_max] = {
 	[cpuinfo_arm_chipset_series_unisoc_t] = "T",
 	[cpuinfo_arm_chipset_series_unisoc_ums] = "UMS",
 	[cpuinfo_arm_chipset_series_wondermedia_wm] = "WM",
+	[cpuinfo_arm_chipset_series_google_tensor] = "Tensor",
 };
 
 /* Convert chipset name represented by cpuinfo_arm_chipset structure to a string
@@ -3913,14 +3944,35 @@ void cpuinfo_arm_chipset_to_string(
 	const char* vendor_string = chipset_vendor_string[vendor];
 	const char* series_string = chipset_series_string[series];
 	const uint32_t model = chipset->model;
+	const size_t suffix_length = strnlen(chipset->suffix, CPUINFO_ARM_CHIPSET_SUFFIX_MAX);
 	if (model == 0) {
-		if (series == cpuinfo_arm_chipset_series_unknown) {
-			strncpy(name, vendor_string, CPUINFO_ARM_CHIPSET_NAME_MAX);
+		if (suffix_length > 0) {
+			if (series == cpuinfo_arm_chipset_series_unknown) {
+				snprintf(
+					name,
+					CPUINFO_ARM_CHIPSET_NAME_MAX,
+					"%s %.*s",
+					vendor_string,
+					(int)suffix_length,
+					chipset->suffix);
+			} else {
+				snprintf(
+					name,
+					CPUINFO_ARM_CHIPSET_NAME_MAX,
+					"%s %s %.*s",
+					vendor_string,
+					series_string,
+					(int)suffix_length,
+					chipset->suffix);
+			}
 		} else {
-			snprintf(name, CPUINFO_ARM_CHIPSET_NAME_MAX, "%s %s", vendor_string, series_string);
+			if (series == cpuinfo_arm_chipset_series_unknown) {
+				strncpy(name, vendor_string, CPUINFO_ARM_CHIPSET_NAME_MAX);
+			} else {
+				snprintf(name, CPUINFO_ARM_CHIPSET_NAME_MAX, "%s %s", vendor_string, series_string);
+			}
 		}
 	} else {
-		const size_t suffix_length = strnlen(chipset->suffix, CPUINFO_ARM_CHIPSET_SUFFIX_MAX);
 		snprintf(
 			name,
 			CPUINFO_ARM_CHIPSET_NAME_MAX,
@@ -4051,7 +4103,6 @@ static enum cpuinfo_arm_chipset_vendor disambiguate_chipset_vendor(
 	    (vendor_a == cpuinfo_arm_chipset_vendor_spreadtrum && vendor_b == cpuinfo_arm_chipset_vendor_unisoc)) {
 		return cpuinfo_arm_chipset_vendor_unisoc;
 	}
-
 	return cpuinfo_arm_chipset_vendor_unknown;
 }
 

3rdparty/cpuinfo/src/arm/linux/init.c

@@ -256,12 +256,12 @@ void cpuinfo_arm_linux_init(void) {
 	}
 
 #if defined(__ANDROID__)
-	struct cpuinfo_android_properties android_properties;
+	struct cpuinfo_android_properties android_properties = {0};
 	cpuinfo_arm_android_parse_properties(&android_properties);
 #else
-	char proc_cpuinfo_hardware[CPUINFO_HARDWARE_VALUE_MAX];
+	char proc_cpuinfo_hardware[CPUINFO_HARDWARE_VALUE_MAX] = {0};
 #endif
-	char proc_cpuinfo_revision[CPUINFO_REVISION_VALUE_MAX];
+	char proc_cpuinfo_revision[CPUINFO_REVISION_VALUE_MAX] = {0};
 
 	if (!cpuinfo_arm_linux_parse_proc_cpuinfo(
 #if defined(__ANDROID__)

3rdparty/cpuinfo/src/arm/uarch.c

@@ -11,6 +11,11 @@ void cpuinfo_arm_decode_vendor_uarch(
 #endif /* CPUINFO_ARCH_ARM */
 	enum cpuinfo_vendor vendor[RESTRICT_STATIC 1],
 	enum cpuinfo_uarch uarch[RESTRICT_STATIC 1]) {
+	/* Ensure the out-parameters are always initialized, including for
+	 * implementers that are not handled in the switch below. */
+	*vendor = cpuinfo_vendor_unknown;
+	*uarch = cpuinfo_uarch_unknown;
+
 	switch (midr_get_implementer(midr)) {
 		case 'A':
 			*vendor = cpuinfo_vendor_arm;
@@ -438,6 +443,50 @@ void cpuinfo_arm_decode_vendor_uarch(
 						midr_get_part(midr));
 			}
 			break;
+		case 'a':
+			*vendor = cpuinfo_vendor_apple;
+			switch (midr_get_part(midr)) {
+				case 0x022:
+				case 0x024:
+				case 0x028:
+					*uarch = cpuinfo_uarch_icestorm;
+					break;
+				case 0x023:
+				case 0x025:
+				case 0x029:
+					*uarch = cpuinfo_uarch_firestorm;
+					break;
+				case 0x032:
+				case 0x034:
+				case 0x038:
+					*uarch = cpuinfo_uarch_blizzard;
+					break;
+				case 0x033:
+				case 0x035:
+				case 0x039:
+					*uarch = cpuinfo_uarch_avalanche;
+					break;
+				case 0x042:
+				case 0x044:
+				case 0x048:
+					*uarch = cpuinfo_uarch_coll_sawtooth;
+					break;
+				case 0x043:
+				case 0x045:
+				case 0x049:
+					*uarch = cpuinfo_uarch_coll_everest;
+					break;
+				case 0x052:
+					*uarch = cpuinfo_uarch_donan_sawtooth;
+					break;
+				case 0x053:
+					*uarch = cpuinfo_uarch_donan_everest;
+					break;
+				default:
+					cpuinfo_log_warning(
+						"unknown Apple CPU part 0x%03" PRIx32 " ignored", midr_get_part(midr));
+			}
+			break;
 #if CPUINFO_ARCH_ARM
 		case 'V':
 			*vendor = cpuinfo_vendor_marvell;

3rdparty/cpuinfo/src/x86/name.c

@@ -307,6 +307,15 @@ static bool transform_token(char* token_start, char* token_end, struct parser_st
 			}
 			break;
 		case 4:
+			/*
+			 * Erase everything starting with "with" on AMD
+			 * processors, e.g. "AMD Ryzen 5 PRO 6650U with Radeon
+			 * Graphics"
+			 */
+			if (erase_matching(token_start, token_length, "with")) {
+				return false;
+			}
+
 			/* Remember to erase "Dual Core" in "AMD Athlon(tm) 64
 			 * X2 Dual Core Processor 3800+" */
 			if (memcmp(token_start, "Dual", token_length) == 0) {

3rdparty/cpuinfo/src/x86/uarch.c

@@ -270,8 +270,15 @@ enum cpuinfo_uarch cpuinfo_x86_decode_uarch(
 							return cpuinfo_uarch_prescott;
 					}
 					break;
+				case 0x12:
+					switch (model_info->model) {
+						case 0x01: // Nova Lake P-core (Coyote Cove)
+							return cpuinfo_uarch_coyote_cove;
+					}
+					break;
 			}
 			break;
+
 		case cpuinfo_vendor_amd:
 			switch (model_info->family) {
 #if CPUINFO_ARCH_X86

3rdparty/cpuinfo/src/x86/windows/init.c

@@ -265,8 +265,6 @@ BOOL CALLBACK cpuinfo_x86_windows_init(PINIT_ONCE init_once, PVOID parameter, PV
 		 * APIC order */
 		const uint32_t core_id = cores_count++;
 		uint32_t smt_id = 0;
-		/* Reconstruct core part of APIC ID */
-		const uint32_t core_apic_id = (core_id & core_bits_mask) << x86_processor.topology.core_bits_offset;
 		/* Iterate processor groups and set the core & SMT parts of APIC
 		 * ID */
 		for (uint32_t i = 0; i < core_info->Processor.GroupCount; i++) {

3rdparty/d3d12memalloc/CMakeLists.txt

@@ -3,5 +3,7 @@ add_library(D3D12MemAlloc
 	src/D3D12MemAlloc.cpp
 )
 
+target_link_libraries(D3D12MemAlloc PRIVATE Microsoft::DirectX-Headers)
+
 target_include_directories(D3D12MemAlloc PUBLIC "${CMAKE_CURRENT_SOURCE_DIR}/include")
 disable_compiler_warnings_for_target(D3D12MemAlloc)

3rdparty/d3d12memalloc/d3d12memalloc.vcxproj

@@ -20,6 +20,7 @@
   <ImportGroup Label="PropertySheets">
     <Import Project="..\DefaultProjectRootDir.props" />
     <Import Project="..\3rdparty.props" />
+    <Import Project="..\..\common\vsprops\DepsDir.props" />
     <Import Condition="$(Configuration.Contains(Debug))" Project="..\..\common\vsprops\CodeGen_Debug.props" />
     <Import Condition="$(Configuration.Contains(Devel))" Project="..\..\common\vsprops\CodeGen_Devel.props" />
     <Import Condition="$(Configuration.Contains(Release))" Project="..\..\common\vsprops\CodeGen_Release.props" />
@@ -37,9 +38,9 @@
   </ItemGroup>
   <ItemDefinitionGroup>
     <ClCompile>
-      <PreprocessorDefinitions>%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>D3D12MA_USING_DIRECTX_HEADERS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <WarningLevel>TurnOffAllWarnings</WarningLevel>
-      <AdditionalIncludeDirectories>$(ProjectDir)include;$(ProjectDir)src;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <AdditionalIncludeDirectories>$(DepsIncludeDir);$(ProjectDir)include;$(ProjectDir)src;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
     </ClCompile>
   </ItemDefinitionGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />

3rdparty/d3d12memalloc/include/D3D12MemAlloc.h

@@ -24,7 +24,7 @@
 
 /** \mainpage D3D12 Memory Allocator
 
-<b>Version 3.1.0</b> (2026-02-23)
+<b>Version 3.2.0</b> (2026-06-05)
 
 Copyright (c) 2019-2026 Advanced Micro Devices, Inc. All rights reserved. \n
 License: MIT
@@ -889,10 +889,17 @@ enum POOL_FLAGS
     */
     POOL_FLAG_MSAA_TEXTURES_ALWAYS_COMMITTED = 0x2,
     /** Every allocation made in this pool will be created as a committed resource - will have its own memory block.
-    
+     
     There is also an equivalent flag for the entire allocator: D3D12MA::ALLOCATOR_FLAG_ALWAYS_COMMITTED.
     */
     POOL_FLAG_ALWAYS_COMMITTED = 0x4,
+    /** Disables tight resource alignment for all resources created from this pool.
+    
+    When tight alignment is supported and enabled for the allocator, resources created from custom pools
+    normally use it automatically. Specify this flag to restore pre-tight-alignment behavior for this pool:
+    tight alignment won't be requested and small buffers may again be preferred as committed resources.
+    */
+    POOL_FLAG_DONT_USE_TIGHT_ALIGNMENT = 0x8,
 
     // Bit mask to extract only `ALGORITHM` bits from entire set of flags.
     POOL_FLAG_ALGORITHM_MASK = POOL_FLAG_ALGORITHM_LINEAR
@@ -1114,7 +1121,8 @@ enum ALLOCATOR_FLAGS
     It can also be disabled for a single allocation by using #ALLOCATION_FLAG_STRATEGY_MIN_TIME.
 
     If the tight resource alignment feature is used by the library (which happens automatically whenever supported,
-    unless you use flag #ALLOCATOR_FLAG_DONT_USE_TIGHT_ALIGNMENT), then small buffers are not preferred as committed.
+    unless you use flag #ALLOCATOR_FLAG_DONT_USE_TIGHT_ALIGNMENT or
+    #POOL_FLAG_DONT_USE_TIGHT_ALIGNMENT on a custom pool), then small buffers are not preferred as committed.
     Long story short, you don't need to specify any of these flags.
     The library chooses the most optimal method automatically.
     */
@@ -2892,6 +2900,7 @@ as it is no longer needed.
 
 You can check if the tight alignment it is available in the current system by calling D3D12MA::Allocator::IsTightAlignmentSupported().
 You can tell the library to not use it by specifying D3D12MA::ALLOCATOR_FLAG_DONT_USE_TIGHT_ALIGNMENT.
+Custom pools can opt out independently by specifying D3D12MA::POOL_FLAG_DONT_USE_TIGHT_ALIGNMENT.
 Typically, you don't need to do any of those.
 
 The library automatically aligns all buffers to at least 256 B, even when the system supports smaller alignment.

3rdparty/d3d12memalloc/src/D3D12MemAlloc.cpp

@@ -163,6 +163,14 @@ especially to test compatibility with D3D12_RESOURCE_HEAP_TIER_1 on modern GPUs.
     #endif
 #endif
 
+#ifndef D3D12MA_OPTIONS4_SUPPORTED
+    #ifdef __ID3D12Device5_INTERFACE_DEFINED__
+        #define D3D12MA_OPTIONS4_SUPPORTED 1
+    #else
+        #define D3D12MA_OPTIONS4_SUPPORTED 0
+    #endif
+#endif
+
 #ifndef D3D12MA_DEBUG_LOG
    #define D3D12MA_DEBUG_LOG(format, ...)
    /*
@@ -703,6 +711,13 @@ static UINT GetBitsPerPixel(DXGI_FORMAT format)
     case DXGI_FORMAT_R24_UNORM_X8_TYPELESS:
     case DXGI_FORMAT_X24_TYPELESS_G8_UINT:
         return 32;
+    case DXGI_FORMAT_B8G8R8A8_UNORM:
+    case DXGI_FORMAT_B8G8R8X8_UNORM:
+    case DXGI_FORMAT_B8G8R8A8_TYPELESS:
+    case DXGI_FORMAT_B8G8R8A8_UNORM_SRGB:
+    case DXGI_FORMAT_B8G8R8X8_TYPELESS:
+    case DXGI_FORMAT_B8G8R8X8_UNORM_SRGB:
+        return 32;
     case DXGI_FORMAT_R8G8_TYPELESS:
     case DXGI_FORMAT_R8G8_UNORM:
     case DXGI_FORMAT_R8G8_UINT:
@@ -717,6 +732,9 @@ static UINT GetBitsPerPixel(DXGI_FORMAT format)
     case DXGI_FORMAT_R16_SNORM:
     case DXGI_FORMAT_R16_SINT:
         return 16;
+    case DXGI_FORMAT_B5G6R5_UNORM:
+    case DXGI_FORMAT_B5G5R5A1_UNORM:
+        return 16;
     case DXGI_FORMAT_R8_TYPELESS:
     case DXGI_FORMAT_R8_UNORM:
     case DXGI_FORMAT_R8_UINT:
@@ -767,46 +785,175 @@ static ResourceClass ResourceDescToResourceClass(const D3D12_RESOURCE_DESC_T& re
         (resDesc.Flags & (D3D12_RESOURCE_FLAG_ALLOW_RENDER_TARGET | D3D12_RESOURCE_FLAG_ALLOW_DEPTH_STENCIL)) != 0;
     return isRenderTargetOrDepthStencil ? ResourceClass::RT_DS_Texture : ResourceClass::Non_RT_DS_Texture;
 }
+
+static bool ResourceDimensionIsTexture(D3D12_RESOURCE_DIMENSION dimension)
+{
+    return dimension == D3D12_RESOURCE_DIMENSION_TEXTURE1D ||
+        dimension == D3D12_RESOURCE_DIMENSION_TEXTURE2D ||
+        dimension == D3D12_RESOURCE_DIMENSION_TEXTURE3D;
+}
+
+static bool ResourceFlagsContainRenderTargetOrDepthStencil(D3D12_RESOURCE_FLAGS flags)
+{
+    return (flags & (D3D12_RESOURCE_FLAG_ALLOW_RENDER_TARGET | D3D12_RESOURCE_FLAG_ALLOW_DEPTH_STENCIL)) != 0;
+}
     
-// This algorithm is overly conservative.
+struct SmallAlignmentTileShape
+{
+    UINT Width;
+    UINT Height;
+    UINT Depth;
+};
+
+static bool GetSmallAlignmentTileShape(
+    D3D12_RESOURCE_DIMENSION dimension,
+    bool isMsaa,
+    UINT bitsPerUnit,
+    SmallAlignmentTileShape& outTileShape)
+{
+    if (isMsaa)
+    {
+        if (dimension != D3D12_RESOURCE_DIMENSION_TEXTURE2D)
+            return false;
+
+        switch (bitsPerUnit)
+        {
+        case    8: outTileShape = { 256, 256, 1 }; return true;
+        case   16: outTileShape = { 256, 128, 1 }; return true;
+        case   32: outTileShape = { 128, 128, 1 }; return true;
+        case   64: outTileShape = { 128,  64, 1 }; return true;
+        case  128: outTileShape = {  64,  64, 1 }; return true;
+        case  256: outTileShape = {  64,  32, 1 }; return true;
+        case  512: outTileShape = {  32,  32, 1 }; return true;
+        case 1024: outTileShape = {  32,  16, 1 }; return true;
+        case 2048: outTileShape = {  16,  16, 1 }; return true;
+        default: return false;
+        }
+    }
+
+    switch (dimension)
+    {
+    case D3D12_RESOURCE_DIMENSION_TEXTURE1D:
+        switch (bitsPerUnit)
+        {
+        case   8: outTileShape = { 4096, 1, 1 }; return true;
+        case  16: outTileShape = { 2048, 1, 1 }; return true;
+        case  32: outTileShape = { 1024, 1, 1 }; return true;
+        case  64: outTileShape = {  512, 1, 1 }; return true;
+        case 128: outTileShape = {  256, 1, 1 }; return true;
+        default: return false;
+        }
+
+    case D3D12_RESOURCE_DIMENSION_TEXTURE2D:
+        switch (bitsPerUnit)
+        {
+        case   8: outTileShape = { 64, 64, 1 }; return true;
+        case  16: outTileShape = { 64, 32, 1 }; return true;
+        case  32: outTileShape = { 32, 32, 1 }; return true;
+        case  64: outTileShape = { 32, 16, 1 }; return true;
+        case 128: outTileShape = { 16, 16, 1 }; return true;
+        default: return false;
+        }
+
+    case D3D12_RESOURCE_DIMENSION_TEXTURE3D:
+        // 4 KB counterparts of the documented 64 KB volume-tile shapes.
+        switch (bitsPerUnit)
+        {
+        case   8: outTileShape = { 16, 16, 16 }; return true;
+        case  16: outTileShape = { 16, 16,  8 }; return true;
+        case  32: outTileShape = { 16,  8,  8 }; return true;
+        case  64: outTileShape = {  8,  8,  8 }; return true;
+        case 128: outTileShape = {  8,  8,  4 }; return true;
+        default: return false;
+        }
+
+    default:
+        return false;
+    }
+}
+
+// This algorithm is overly conservative and applies only to the legacy small-alignment heuristic path.
 template<typename D3D12_RESOURCE_DESC_T>
 static bool CanUseSmallAlignment(const D3D12_RESOURCE_DESC_T& resourceDesc)
 {
-    if (resourceDesc.Dimension != D3D12_RESOURCE_DIMENSION_TEXTURE2D)
+    if (resourceDesc.SampleDesc.Count == 0 ||
+        resourceDesc.Layout != D3D12_TEXTURE_LAYOUT_UNKNOWN ||
+        resourceDesc.Width > UINT_MAX)
+    {
         return false;
-    if ((resourceDesc.Flags & (D3D12_RESOURCE_FLAG_ALLOW_RENDER_TARGET | D3D12_RESOURCE_FLAG_ALLOW_DEPTH_STENCIL)) != 0)
+    }
+
+    const bool isMsaa = resourceDesc.SampleDesc.Count > 1;
+    const bool isRtDs = ResourceFlagsContainRenderTargetOrDepthStencil(resourceDesc.Flags);
+    if (isMsaa && !isRtDs)
+    {
         return false;
-    if (resourceDesc.SampleDesc.Count > 1)
-        return false;
-    if (resourceDesc.DepthOrArraySize != 1)
+    }
+    if (!isMsaa &&
+        isRtDs)
+    {
         return false;
+    }
 
     UINT sizeX = (UINT)resourceDesc.Width;
-    UINT sizeY = resourceDesc.Height;
-    UINT bitsPerPixel = GetBitsPerPixel(resourceDesc.Format);
-    if (bitsPerPixel == 0)
+    UINT sizeY = 1;
+    UINT sizeZ = 1;
+    UINT bitsPerUnit = GetBitsPerPixel(resourceDesc.Format);
+    if (bitsPerUnit == 0)
         return false;
 
-    if (IsFormatCompressed(resourceDesc.Format))
+    switch (resourceDesc.Dimension)
     {
-        sizeX = DivideRoundingUp(sizeX, 4u);
-        sizeY = DivideRoundingUp(sizeY, 4u);
-        bitsPerPixel *= 16;
+    case D3D12_RESOURCE_DIMENSION_TEXTURE1D:
+        if (isMsaa)
+            return false;
+        // For 1D arrays, small-alignment eligibility depends on the mip0 slice size, not array size.
+        if (IsFormatCompressed(resourceDesc.Format))
+            return false;
+        break;
+
+    case D3D12_RESOURCE_DIMENSION_TEXTURE2D:
+        // For 2D arrays, small-alignment eligibility depends on mip0 width/height, not array size.
+        sizeY = resourceDesc.Height;
+        if (isMsaa && IsFormatCompressed(resourceDesc.Format))
+            return false;
+        if (IsFormatCompressed(resourceDesc.Format))
+        {
+            sizeX = DivideRoundingUp(sizeX, 4u);
+            sizeY = DivideRoundingUp(sizeY, 4u);
+            bitsPerUnit *= 16;
+        }
+        break;
+
+    case D3D12_RESOURCE_DIMENSION_TEXTURE3D:
+        if (isMsaa)
+            return false;
+        sizeY = resourceDesc.Height;
+        sizeZ = resourceDesc.DepthOrArraySize;
+        if (IsFormatCompressed(resourceDesc.Format))
+        {
+            sizeX = DivideRoundingUp(sizeX, 4u);
+            sizeY = DivideRoundingUp(sizeY, 4u);
+            bitsPerUnit *= 16;
+        }
+        break;
+
+    default:
+        return false;
     }
 
-    UINT tileSizeX = 0, tileSizeY = 0;
-    switch (bitsPerPixel)
-    {
-    case   8: tileSizeX = 64; tileSizeY = 64; break;
-    case  16: tileSizeX = 64; tileSizeY = 32; break;
-    case  32: tileSizeX = 32; tileSizeY = 32; break;
-    case  64: tileSizeX = 32; tileSizeY = 16; break;
-    case 128: tileSizeX = 16; tileSizeY = 16; break;
-    default: return false;
-    }
+    bitsPerUnit *= resourceDesc.SampleDesc.Count;
 
-    const UINT tileCount = DivideRoundingUp(sizeX, tileSizeX) * DivideRoundingUp(sizeY, tileSizeY);
-    return tileCount <= 16;
+    SmallAlignmentTileShape tileShape = {};
+    if (!GetSmallAlignmentTileShape(resourceDesc.Dimension, isMsaa, bitsPerUnit, tileShape))
+        return false;
+
+    const UINT64 tileCount =
+        UINT64(DivideRoundingUp(sizeX, tileShape.Width)) *
+        UINT64(DivideRoundingUp(sizeY, tileShape.Height)) *
+        UINT64(DivideRoundingUp(sizeZ, tileShape.Depth));
+    // Non-MSAA compares 64 KB against 4 KB tiles => 16 pages. MSAA compares 4 MB against 64 KB tiles => 64 pages.
+    return tileCount <= (isMsaa ? 64 : 16);
 }
     
 static bool ValidateAllocateMemoryParameters(
@@ -5988,6 +6135,7 @@ public:
     BOOL IsCacheCoherentUMA() const { return m_D3D12Architecture.CacheCoherentUMA; }
     bool SupportsResourceHeapTier2() const { return m_D3D12Options.ResourceHeapTier >= D3D12_RESOURCE_HEAP_TIER_2; }
     bool IsGPUUploadHeapSupported() const { return m_GPUUploadHeapSupported != FALSE; }
+    bool IsMsaa64KBAlignedTextureSupported() const { return m_MSAA64KBAlignedTextureSupported != FALSE; }
     bool IsTightAlignmentSupported() const { return m_TightAlignmentSupported != FALSE; }
     bool IsTightAlignmentEnabled() const { return IsTightAlignmentSupported() && m_UseTightAlignment; }
     bool UseMutex() const { return m_UseMutex; }
@@ -6076,7 +6224,7 @@ private:
     const bool m_UseMutex;
     const bool m_AlwaysCommitted;
     const bool m_MsaaAlwaysCommitted;
-    bool m_PreferSmallBuffersCommitted;
+    const bool m_PreferSmallBuffersCommitted;
     const bool m_UseTightAlignment;
     bool m_DefaultPoolsNotZeroed = false;
     ID3D12Device* m_Device; // AddRef
@@ -6105,6 +6253,7 @@ private:
     DXGI_ADAPTER_DESC m_AdapterDesc;
     D3D12_FEATURE_DATA_D3D12_OPTIONS m_D3D12Options;
     BOOL m_GPUUploadHeapSupported = FALSE;
+    BOOL m_MSAA64KBAlignedTextureSupported = FALSE;
     BOOL m_TightAlignmentSupported = FALSE;
     D3D12_FEATURE_DATA_ARCHITECTURE m_D3D12Architecture;
     AllocationObjectAllocator m_AllocationObjectAllocator;
@@ -6121,7 +6270,7 @@ private:
     */
     template<typename D3D12_RESOURCE_DESC_T>
     bool PrefersCommittedAllocation(const D3D12_RESOURCE_DESC_T& resourceDesc,
-        ALLOCATION_FLAGS strategy);
+        ALLOCATION_FLAGS strategy, bool useTightAlignment) const;
 
     // Allocates and registers new committed resource with implicit heap, as dedicated allocation.
     // Creates and returns Allocation object and optionally D3D12 resource.
@@ -6141,6 +6290,7 @@ private:
     template<typename D3D12_RESOURCE_DESC_T>
     HRESULT CalcAllocationParams(const ALLOCATION_DESC& allocDesc, UINT64 allocSize,
         const D3D12_RESOURCE_DESC_T* resDesc, // Optional
+        bool useTightAlignment,
         BlockVector*& outBlockVector, CommittedAllocationParameters& outCommittedAllocationParams, bool& outPreferCommitted);
 
     // Returns UINT32_MAX if index cannot be calculcated.
@@ -6174,7 +6324,10 @@ private:
     template<typename D3D12_RESOURCE_DESC_T>
     HRESULT GetResourceAllocationInfo(D3D12_RESOURCE_DESC_T& inOutResourceDesc,
         UINT32 NumCastableFormats, const DXGI_FORMAT* pCastableFormats,
-        D3D12_RESOURCE_ALLOCATION_INFO& outAllocInfo) const;
+        D3D12_RESOURCE_ALLOCATION_INFO& outAllocInfo,
+        bool useTightAlignment) const;
+
+    bool IsTightAlignmentEnabled(const ALLOCATION_DESC& allocDesc) const;
 
     bool NewAllocationWithinBudget(D3D12_HEAP_TYPE heapType, UINT64 size);
 
@@ -6277,6 +6430,17 @@ HRESULT AllocatorPimpl::Init(const ALLOCATOR_DESC& desc)
     }
 #endif // #if D3D12MA_OPTIONS16_SUPPORTED
 
+#if D3D12MA_OPTIONS4_SUPPORTED
+    {
+        D3D12_FEATURE_DATA_D3D12_OPTIONS4 options4 = {};
+        hr = m_Device->CheckFeatureSupport(D3D12_FEATURE_D3D12_OPTIONS4, &options4, sizeof(options4));
+        if (SUCCEEDED(hr))
+        {
+            m_MSAA64KBAlignedTextureSupported = options4.MSAA64KBAlignedTextureSupported;
+        }
+    }
+#endif // #if D3D12MA_OPTIONS4_SUPPORTED
+
 #if D3D12MA_TIGHT_ALIGNMENT_SUPPORTED
     {
         D3D12_FEATURE_DATA_TIGHT_ALIGNMENT tightAlignment = {};
@@ -6285,13 +6449,6 @@ HRESULT AllocatorPimpl::Init(const ALLOCATOR_DESC& desc)
         {
             m_TightAlignmentSupported = tightAlignment.SupportTier >= D3D12_TIGHT_ALIGNMENT_TIER_1;
             
-            // If tight alignment is supported (checked by the code above) and wasn't disabled by the developer
-            // (with ALLOCATOR_FLAG_DONT_USE_TIGHT_ALIGNMENT), disable the preference for creating small buffers as committed,
-            // as if ALLOCATOR_FLAG_DONT_PREFER_SMALL_BUFFERS_COMMITTED was specified.
-            if (IsTightAlignmentEnabled())
-            {
-                m_PreferSmallBuffersCommitted = false;
-            }
         }
     }
 #endif // #if D3D12MA_TIGHT_ALIGNMENT_SUPPORTED
@@ -6490,6 +6647,7 @@ HRESULT AllocatorPimpl::CreateResource(
     }
 
     HRESULT hr = E_NOINTERFACE;
+    const bool useTightAlignment = IsTightAlignmentEnabled(*pAllocDesc);
     CREATE_RESOURCE_PARAMS finalCreateParams = createParams;
     D3D12_RESOURCE_DESC finalResourceDesc;
 #ifdef __ID3D12Device8_INTERFACE_DEFINED__
@@ -6500,7 +6658,7 @@ HRESULT AllocatorPimpl::CreateResource(
     {
         finalResourceDesc = *createParams.GetResourceDesc();
         finalCreateParams.AccessResourceDesc() = &finalResourceDesc;
-        hr = GetResourceAllocationInfo(finalResourceDesc, 0, NULL, resAllocInfo);
+        hr = GetResourceAllocationInfo(finalResourceDesc, 0, NULL, resAllocInfo, useTightAlignment);
     }
 #ifdef __ID3D12Device8_INTERFACE_DEFINED__
     else if (createParams.Variant == CREATE_RESOURCE_PARAMS::VARIANT_WITH_STATE_AND_DESC1)
@@ -6509,7 +6667,7 @@ HRESULT AllocatorPimpl::CreateResource(
         {
             finalResourceDesc1 = *createParams.GetResourceDesc1();
             finalCreateParams.AccessResourceDesc1() = &finalResourceDesc1;
-            hr = GetResourceAllocationInfo(finalResourceDesc1, 0, NULL, resAllocInfo);
+            hr = GetResourceAllocationInfo(finalResourceDesc1, 0, NULL, resAllocInfo, useTightAlignment);
         }
     }
 #endif
@@ -6521,7 +6679,7 @@ HRESULT AllocatorPimpl::CreateResource(
             finalResourceDesc1 = *createParams.GetResourceDesc1();
             finalCreateParams.AccessResourceDesc1() = &finalResourceDesc1;
             hr = GetResourceAllocationInfo(finalResourceDesc1,
-                createParams.GetNumCastableFormats(), createParams.GetCastableFormats(), resAllocInfo);
+                createParams.GetNumCastableFormats(), createParams.GetCastableFormats(), resAllocInfo, useTightAlignment);
         }
     }
 #endif
@@ -6547,14 +6705,14 @@ HRESULT AllocatorPimpl::CreateResource(
     if (createParams.Variant >= CREATE_RESOURCE_PARAMS::VARIANT_WITH_STATE_AND_DESC1)
     {
         hr = CalcAllocationParams<D3D12_RESOURCE_DESC1>(*pAllocDesc, resAllocInfo.SizeInBytes,
-            createParams.GetResourceDesc1(),
+            createParams.GetResourceDesc1(), useTightAlignment,
             blockVector, committedAllocationParams, preferCommitted);
     }
     else
 #endif
     {
         hr = CalcAllocationParams<D3D12_RESOURCE_DESC>(*pAllocDesc, resAllocInfo.SizeInBytes,
-            createParams.GetResourceDesc(),
+            createParams.GetResourceDesc(), useTightAlignment,
             blockVector, committedAllocationParams, preferCommitted);
     }
     if (FAILED(hr))
@@ -6601,6 +6759,7 @@ HRESULT AllocatorPimpl::AllocateMemory(
     bool preferCommitted = false;
     HRESULT hr = CalcAllocationParams<D3D12_RESOURCE_DESC>(*pAllocDesc, pAllocInfo->SizeInBytes,
         NULL, // pResDesc
+        false, // useTightAlignment - irrelevant for AllocateMemory
         blockVector, committedAllocationParams, preferCommitted);
     if (FAILED(hr))
         return hr;
@@ -6649,7 +6808,7 @@ HRESULT AllocatorPimpl::CreateAliasingResource(
     {
         finalResourceDesc = *createParams.GetResourceDesc();
         finalCreateParams.AccessResourceDesc() = &finalResourceDesc;
-        hr = GetResourceAllocationInfo(finalResourceDesc, 0, NULL, resAllocInfo);
+        hr = GetResourceAllocationInfo(finalResourceDesc, 0, NULL, resAllocInfo, IsTightAlignmentEnabled());
     }
 #ifdef __ID3D12Device8_INTERFACE_DEFINED__
     else if (createParams.Variant == CREATE_RESOURCE_PARAMS::VARIANT_WITH_STATE_AND_DESC1)
@@ -6658,7 +6817,7 @@ HRESULT AllocatorPimpl::CreateAliasingResource(
         {
             finalResourceDesc1 = *createParams.GetResourceDesc1();
             finalCreateParams.AccessResourceDesc1() = &finalResourceDesc1;
-            hr = GetResourceAllocationInfo(finalResourceDesc1, 0, NULL, resAllocInfo);
+            hr = GetResourceAllocationInfo(finalResourceDesc1, 0, NULL, resAllocInfo, IsTightAlignmentEnabled());
         }
     }
 #endif
@@ -6670,7 +6829,7 @@ HRESULT AllocatorPimpl::CreateAliasingResource(
             finalResourceDesc1 = *createParams.GetResourceDesc1();
             finalCreateParams.AccessResourceDesc1() = &finalResourceDesc1;
             hr = GetResourceAllocationInfo(finalResourceDesc1,
-                createParams.GetNumCastableFormats(), createParams.GetCastableFormats(), resAllocInfo);
+                createParams.GetNumCastableFormats(), createParams.GetCastableFormats(), resAllocInfo, IsTightAlignmentEnabled());
         }
     }
 #endif
@@ -7305,13 +7464,14 @@ void AllocatorPimpl::FreeStatsString(WCHAR* pStatsString)
 
 template<typename D3D12_RESOURCE_DESC_T>
 bool AllocatorPimpl::PrefersCommittedAllocation(const D3D12_RESOURCE_DESC_T& resourceDesc,
-    ALLOCATION_FLAGS strategy)
+    ALLOCATION_FLAGS strategy, bool useTightAlignment) const
 {
     // Prefer creating small buffers <= 32 KB as committed, because drivers pack them better,
     // while placed buffers require 64 KB alignment.
     if(resourceDesc.Dimension == D3D12_RESOURCE_DIMENSION_BUFFER &&
         resourceDesc.Width <= D3D12_DEFAULT_RESOURCE_PLACEMENT_ALIGNMENT / 2 &&
         strategy != ALLOCATION_FLAG_STRATEGY_MIN_TIME &&  // Creating as committed would be slower.
+        !useTightAlignment &&
         m_PreferSmallBuffersCommitted)
     {
         return true;
@@ -7526,7 +7686,7 @@ HRESULT AllocatorPimpl::AllocateHeap(
 
 template<typename D3D12_RESOURCE_DESC_T>
 HRESULT AllocatorPimpl::CalcAllocationParams(const ALLOCATION_DESC& allocDesc, UINT64 allocSize,
-    const D3D12_RESOURCE_DESC_T* resDesc,
+    const D3D12_RESOURCE_DESC_T* resDesc, bool useTightAlignment,
     BlockVector*& outBlockVector, CommittedAllocationParameters& outCommittedAllocationParams, bool& outPreferCommitted)
 {
     outBlockVector = NULL;
@@ -7599,7 +7759,7 @@ HRESULT AllocatorPimpl::CalcAllocationParams(const ALLOCATION_DESC& allocDesc, U
     {
         if (resDesc->SampleDesc.Count > 1 && msaaAlwaysCommitted)
             outBlockVector = NULL;
-        if (!outPreferCommitted && PrefersCommittedAllocation(*resDesc, allocDesc.Flags & ALLOCATION_FLAG_STRATEGY_MASK))
+        if (!outPreferCommitted && PrefersCommittedAllocation(*resDesc, allocDesc.Flags & ALLOCATION_FLAG_STRATEGY_MASK, useTightAlignment))
             outPreferCommitted = true;
     }
 
@@ -7810,12 +7970,13 @@ template<typename D3D12_RESOURCE_DESC_T>
 HRESULT AllocatorPimpl::GetResourceAllocationInfo(
     D3D12_RESOURCE_DESC_T& inOutResourceDesc,
     UINT32 NumCastableFormats, const DXGI_FORMAT* pCastableFormats,
-    D3D12_RESOURCE_ALLOCATION_INFO& outAllocInfo) const
+    D3D12_RESOURCE_ALLOCATION_INFO& outAllocInfo,
+    bool useTightAlignment) const
 {
 #ifdef __ID3D12Device1_INTERFACE_DEFINED__
 
 #if D3D12MA_TIGHT_ALIGNMENT_SUPPORTED
-    if (IsTightAlignmentEnabled())
+    if (useTightAlignment)
     {
         // Don't allow USE_TIGHT_ALIGNMENT together with ALLOW_CROSS_ADAPTER as there is a D3D Debug Layer error:
         // D3D12 ERROR: ID3D12Device::GetResourceAllocationInfo: D3D12_RESOURCE_DESC::Flag D3D12_RESOURCE_FLAG_USE_TIGHT_ALIGNMENT will be ignored since D3D12_RESOURCE_FLAG_ALLOW_CROSS_ADAPTER is set. [ STATE_CREATION ERROR #599: CREATERESOURCE_INVALIDMISCFLAGS]
@@ -7839,7 +8000,7 @@ HRESULT AllocatorPimpl::GetResourceAllocationInfo(
     */
     if (inOutResourceDesc.Alignment == 0 &&
         inOutResourceDesc.Dimension == D3D12_RESOURCE_DIMENSION_BUFFER &&
-        !IsTightAlignmentEnabled())
+        !useTightAlignment)
     {
         outAllocInfo = {
             AlignUp<UINT64>(inOutResourceDesc.Width, D3D12_DEFAULT_RESOURCE_PLACEMENT_ALIGNMENT), // SizeInBytes
@@ -7852,16 +8013,33 @@ HRESULT AllocatorPimpl::GetResourceAllocationInfo(
     HRESULT hr = S_OK;
 
 #if D3D12MA_USE_SMALL_RESOURCE_PLACEMENT_ALIGNMENT
-    if (inOutResourceDesc.Alignment == 0 &&
-        (inOutResourceDesc.Flags & D3D12_RESOURCE_FLAG_USE_TIGHT_ALIGNMENT_COPY) == 0 &&
-        (inOutResourceDesc.Dimension == D3D12_RESOURCE_DIMENSION_TEXTURE1D ||
-            inOutResourceDesc.Dimension == D3D12_RESOURCE_DIMENSION_TEXTURE2D ||
-            inOutResourceDesc.Dimension == D3D12_RESOURCE_DIMENSION_TEXTURE3D) &&
-        (inOutResourceDesc.Flags & (D3D12_RESOURCE_FLAG_ALLOW_RENDER_TARGET | D3D12_RESOURCE_FLAG_ALLOW_DEPTH_STENCIL)) == 0
+    bool trySmallAlignment = false;
+
+    if (inOutResourceDesc.Alignment == 0
+        && (inOutResourceDesc.Flags & D3D12_RESOURCE_FLAG_USE_TIGHT_ALIGNMENT_COPY) == 0
+        && inOutResourceDesc.Layout == D3D12_TEXTURE_LAYOUT_UNKNOWN
+        && ResourceDimensionIsTexture(inOutResourceDesc.Dimension))
+    {
+        // MSAA texture.
+        if(inOutResourceDesc.SampleDesc.Count > 1)
+        {
+            trySmallAlignment = IsMsaa64KBAlignedTextureSupported()
+                && ResourceFlagsContainRenderTargetOrDepthStencil(inOutResourceDesc.Flags)
+                && inOutResourceDesc.Dimension == D3D12_RESOURCE_DIMENSION_TEXTURE2D;
+        }
+        // Non-MSAA texture.
+        else
+        {
+            trySmallAlignment = !ResourceFlagsContainRenderTargetOrDepthStencil(inOutResourceDesc.Flags);
+        }
+    }
+
 #if D3D12MA_USE_SMALL_RESOURCE_PLACEMENT_ALIGNMENT == 1
-        && CanUseSmallAlignment(inOutResourceDesc)
+    if(trySmallAlignment)
+        trySmallAlignment = CanUseSmallAlignment(inOutResourceDesc);
 #endif
-        )
+    
+    if (trySmallAlignment)
     {
         /*
         The algorithm here is based on Microsoft sample: "Small Resources Sample"
@@ -7886,6 +8064,20 @@ HRESULT AllocatorPimpl::GetResourceAllocationInfo(
         inOutResourceDesc, NumCastableFormats, pCastableFormats, outAllocInfo);
 }
 
+bool AllocatorPimpl::IsTightAlignmentEnabled(const ALLOCATION_DESC& allocDesc) const
+{
+    if (!IsTightAlignmentEnabled())
+        return false;
+
+    if (allocDesc.CustomPool != NULL &&
+        (allocDesc.CustomPool->m_Pimpl->GetDesc().Flags & POOL_FLAG_DONT_USE_TIGHT_ALIGNMENT) != 0)
+    {
+        return false;
+    }
+
+    return true;
+}
+
 bool AllocatorPimpl::NewAllocationWithinBudget(D3D12_HEAP_TYPE heapType, UINT64 size)
 {
     Budget budget = {};

3rdparty/vulkan/LICENSE.md

@@ -1,12 +1,13 @@
-Copyright 2015-2023 The Khronos Group Inc.
+Copyright 2015-2026 The Khronos Group Inc.
 
 Files in this repository fall under one of these licenses:
 
 - `Apache-2.0`
 - `MIT`
 
-Note: With the exception of `parse_dependency.py` the files using `MIT` license
-also fall under `Apache-2.0`. Example:
+Note: With the exception of `parse_dependency.py`, which is based on an
+external file, files using the `MIT` license also fall under `Apache-2.0`.
+Example:
 
 ```
 SPDX-License-Identifier: Apache-2.0 OR MIT

3rdparty/vulkan/include/vk_video/vulkan_video_codec_av1std.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_av1std_decode.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_av1std_encode.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_h264std.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_h264std_decode.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_h264std_encode.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_h265std.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_h265std_decode.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_h265std_encode.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_vp9std.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codec_vp9std_decode.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vk_video/vulkan_video_codecs_common.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vk_icd.h

@@ -2,8 +2,7 @@
  * Copyright 2015-2023 The Khronos Group Inc.
  * Copyright 2015-2023 Valve Corporation
  * Copyright 2015-2023 LunarG, Inc.
- *
- * SPDX-License-Identifier: Apache-2.0
+ * SPDX-License-Identifier: Apache-2.0 OR MIT
  */
 #pragma once
 

3rdparty/vulkan/include/vulkan/vk_layer.h

@@ -2,8 +2,7 @@
  * Copyright 2015-2023 The Khronos Group Inc.
  * Copyright 2015-2023 Valve Corporation
  * Copyright 2015-2023 LunarG, Inc.
- *
- * SPDX-License-Identifier: Apache-2.0
+ * SPDX-License-Identifier: Apache-2.0 OR MIT
  */
 #pragma once
 

3rdparty/vulkan/include/vulkan/vk_platform.h

@@ -3,8 +3,7 @@
 //
 /*
 ** Copyright 2014-2026 The Khronos Group Inc.
-**
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 

3rdparty/vulkan/include/vulkan/vulkan.h

@@ -3,8 +3,7 @@
 
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
-**
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 #include "vk_platform.h"

3rdparty/vulkan/include/vulkan/vulkan_android.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_beta.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_directfb.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_fuchsia.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_ggp.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_ios.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_macos.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_metal.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_ohos.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_screen.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_vi.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_wayland.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_win32.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_xcb.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_xlib.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

3rdparty/vulkan/include/vulkan/vulkan_xlib_xrandr.h

@@ -4,7 +4,7 @@
 /*
 ** Copyright 2015-2026 The Khronos Group Inc.
 **
-** SPDX-License-Identifier: Apache-2.0
+** SPDX-License-Identifier: Apache-2.0 OR MIT
 */
 
 /*

bin/resources/game_controller_db.txt

@@ -683,7 +683,7 @@
 03000000a306000020f6000000000000,Saitek PS2700 PlayStation Controller,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a4,start:b9,x:b0,y:b3,platform:Windows,
 03000000300f00001101000000000000,Saitek Rumble,a:b2,b:b3,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b10,lefttrigger:b5,leftx:a0,lefty:a1,rightshoulder:b6,rightstick:b11,righttrigger:b7,rightx:a3,righty:a2,start:b9,x:b0,y:b1,platform:Windows,
 03000000e804000000a0000000000000,Samsung EIGP20,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftx:a0,lefty:a1,rightshoulder:b7,rightx:a3,righty:a4,start:b11,x:b3,y:b4,platform:Windows,
-03000000bd12000010d1000000000000,Sanwa 4Button,a:b0,b:b1,x:b2,y:b3,dpup:-a1,dpdown:+a1,dpleft:-a0,dpright:+a0,platform:Windows,
+03000000bd12000010d1000000000000,Sanwa 4Button,a:b0,b:b1,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,x:b2,y:b3,platform:Windows,
 03000000c01100000252000000000000,Sanwa Easy Grip,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,lefttrigger:b7,leftx:a0,lefty:a1,rightshoulder:b5,righttrigger:b2,start:b9,x:b3,y:b4,platform:Windows,
 03000000c01100004350000000000000,Sanwa Micro Grip P3,a:b1,b:b0,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b5,leftx:a0,lefty:a1,rightshoulder:b6,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,x:b3,y:b2,platform:Windows,
 03000000411200004550000000000000,Sanwa Micro Grip Pro,a:b0,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b5,leftx:a0,lefty:a1,rightshoulder:b6,rightstick:b11,righttrigger:b7,rightx:a1,righty:a2,start:b9,x:b1,y:b3,platform:Windows,
@@ -697,14 +697,14 @@
 030000004f04000028b3000000000000,Score A,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a5,start:b9,x:b0,y:b3,platform:Windows,
 03000000952e00002577000000000000,Scuf PS4 Controller,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Windows,
 03000000a30c00002500000000000000,Sega Genesis Mini 3B Controller,a:b2,b:b1,dpdown:+a4,dpleft:-a3,dpright:+a3,dpup:-a4,righttrigger:b5,start:b9,platform:Windows,
-03000000a30c00002400000000000000,Sega Mega Drive Mini 6B Controller,a:b2,b:b1,dpdown:+a4,dpleft:-a3,dpright:+a3,dpup:-a4,rightshoulder:b4,righttrigger:b5,leftshoulder:b6,lefttrigger:b7,back:b8,start:b9,x:b3,y:b0,platform:Windows,
+03000000a30c00002400000000000000,Sega Mega Drive Mini 6B Controller,a:b2,b:b1,back:b8,dpdown:+a4,dpleft:-a3,dpright:+a3,dpup:-a4,leftshoulder:b6,lefttrigger:b7,rightshoulder:b4,righttrigger:b5,start:b9,x:b3,y:b0,platform:Windows,
 03000000d804000086e6000000000000,Sega Multi Controller,a:b0,b:b1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:a2,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,righttrigger:b2,start:b8,x:b3,y:b4,platform:Windows,
 0300000000050000289b000000000000,Sega Saturn Adapter,a:b1,b:b2,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,leftshoulder:b6,lefttrigger:b4,rightshoulder:b7,righttrigger:b5,start:b9,x:b0,y:b3,platform:Windows,
 0300000000f000000800000000000000,Sega Saturn Controller,a:b1,b:b2,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,rightshoulder:b7,righttrigger:b3,start:b0,x:b5,y:b6,platform:Windows,
 03000000730700000601000000000000,Sega Saturn Controller,a:b0,b:b1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,lefttrigger:b7,leftx:a0,lefty:a1,rightshoulder:b5,righttrigger:b2,start:b9,x:b3,y:b4,platform:Windows,
 03000000b40400000a01000000000000,Sega Saturn Controller,a:b0,b:b1,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,leftshoulder:b6,lefttrigger:b7,rightshoulder:b5,righttrigger:b2,start:b8,x:b3,y:b4,platform:Windows,
 030000003b07000004a1000000000000,SFX,a:b0,b:b2,back:b7,leftshoulder:b6,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b9,righttrigger:b5,start:b8,x:b1,y:b3,platform:Windows,
-03000000632500002705000000000000,ShanWan Q36,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,lefttrigger:b8,rightshoulder:b7,righttrigger:b9,start:b11,x:b3,y:b4,platform:Windows,
+03000000632500002705000000000000,ShanWan Gamepad,+lefty:+a2,-lefty:-a1,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftstick:b13,lefttrigger:b8,leftx:a0,rightshoulder:b7,rightstick:b14,righttrigger:b9,start:b11,x:b3,y:b4,platform:Windows,
 03000000f82100001900000000000000,Shogun Bros Chameleon X1,a:b2,b:b1,back:b9,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b7,righttrigger:b5,rightx:a2,righty:a3,start:b8,x:b3,y:b0,platform:Windows,
 03000000120c00001c1e000000000000,SnakeByte 4S PS4 Controller,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Windows,
 03000000140300000918000000000000,SNES Controller,a:b0,b:b1,back:b6,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,leftshoulder:b4,rightshoulder:b5,start:b7,x:b2,y:b3,platform:Windows,
@@ -733,6 +733,7 @@
 03000000120c0000160e000000000000,Steel Play Metaltech PS4 Controller,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Windows,
 03000000110100001914000000000000,SteelSeries,a:b0,b:b1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftstick:b13,lefttrigger:b6,leftx:a0,lefty:a1,rightstick:b14,righttrigger:b7,rightx:a3,righty:a4,start:b11,x:b3,y:b4,platform:Windows,
 03000000381000001214000000000000,SteelSeries Free,a:b0,b:b1,back:b12,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftx:a0,lefty:a1,rightshoulder:b7,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Windows,
+03000000381000004114000000000000,SteelSeries Nimbus Cloud,a:b0,b:b1,x:b3,y:b4,back:b10,guide:b15,start:b11,leftstick:b13,rightstick:b14,leftshoulder:b6,rightshoulder:b7,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,paddle1:b5,paddle2:b2,leftx:a0,lefty:a1,rightx:a3,righty:a4,lefttrigger:b8,righttrigger:b9,platform:Windows,
 03000000110100003114000000000000,SteelSeries Stratus Duo,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftstick:b13,lefttrigger:b8,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:b9,rightx:a3,righty:a4,start:b11,x:b3,y:b4,platform:Windows,
 03000000381000003014000000000000,SteelSeries Stratus Duo,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b10,leftshoulder:b4,leftstick:b8,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b9,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Windows,
 03000000381000003114000000000000,SteelSeries Stratus Duo,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b10,leftshoulder:b4,leftstick:b8,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b9,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Windows,
@@ -776,6 +777,7 @@
 030000005f140000c501000000000000,Trust Gamepad,a:b2,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b3,y:b0,platform:Windows,
 03000000b80500000210000000000000,Trust Gamepad,a:b2,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b3,y:b0,platform:Windows,
 030000005f1400002a01000000000000,Trust Predator GM1200,a:b0,b:b1,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,leftshoulder:b6,lefttrigger:b7,rightshoulder:b5,righttrigger:b2,x:b3,y:b4,platform:Windows,
+03000000580400000a10000000000000,Trust Sight Fighter,a:b0,b:b1,back:b5,leftshoulder:b6,lefttrigger:b8,leftx:a0,lefty:a1,rightshoulder:b7,righttrigger:b9,start:b2,x:b3,y:b4,platform:Windows,
 030000004f04000087b6000000000000,TWCS Throttle,dpdown:b8,dpleft:b9,dpright:b7,dpup:b6,leftstick:b5,lefttrigger:-a5,leftx:a0,lefty:a1,righttrigger:+a5,platform:Windows,
 03000000411200000450000000000000,Twin Shock,a:b2,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftstick:b9,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b10,righttrigger:b5,rightx:a3,righty:a4,start:b11,x:b3,y:b0,platform:Windows,
 03000000d90400000200000000000000,TwinShock PS2 Controller,a:b2,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftstick:b10,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b11,righttrigger:b5,rightx:a3,righty:a2,start:b9,x:b3,y:b0,platform:Windows,
@@ -939,6 +941,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000503200000110000045010000,Atari VCS Classic,a:b0,b:b1,back:b4,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b3,start:b2,platform:Mac OS X,
 03000000503200000110000047010000,Atari VCS Classic Controller,a:b0,b:b1,back:b4,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b3,start:b2,platform:Mac OS X,
 03000000503200000210000047010000,Atari VCS Modern Controller,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b9,leftshoulder:b4,leftstick:b6,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b7,righttrigger:a4,rightx:a2,righty:a3,start:b8,x:b2,y:b3,platform:Mac OS X,
+03000000e30500008106000000010000,Austgame GameCube Adapter,a:b0,b:b2,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,lefttrigger:+a4,leftx:a0,lefty:a1,misc3:b5,misc4:b6,rightshoulder:b4,righttrigger:-a4,rightx:a2,righty:a3,start:b7,x:b1,y:b3,platform:Mac OS X,
 030000008a3500000102000000010000,Backbone One,a:b0,b:b1,back:b16,dpdown:b11,dpleft:b13,dpright:b12,dpup:b10,guide:b17,leftshoulder:b4,leftstick:b8,lefttrigger:b6,leftx:a0,lefty:a1~,rightshoulder:b5,rightstick:b9,righttrigger:b7,rightx:a2,righty:a3~,start:b15,x:b2,y:b3,platform:Mac OS X,
 030000008a3500000201000000010000,Backbone One,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Mac OS X,
 030000008a3500000202000000010000,Backbone One,a:b0,b:b1,back:b16,dpdown:b11,dpleft:b13,dpright:b12,dpup:b10,guide:b17,leftshoulder:b4,leftstick:b8,lefttrigger:b6,leftx:a0,lefty:a1~,rightshoulder:b5,rightstick:b9,righttrigger:b7,rightx:a2,righty:a3~,start:b15,x:b2,y:b3,platform:Mac OS X,
@@ -1116,6 +1119,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000110100002014000000000000,SteelSeries Nimbus,a:b0,b:b1,dpdown:b9,dpleft:b11,dpright:b10,dpup:b8,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1~,rightshoulder:b5,righttrigger:b7,rightx:a2,righty:a3~,x:b2,y:b3,platform:Mac OS X,
 03000000110100002014000001000000,SteelSeries Nimbus,a:b0,b:b1,dpdown:b9,dpleft:b11,dpright:b10,dpup:b8,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1~,rightshoulder:b5,righttrigger:b7,rightx:a2,righty:a3~,x:b2,y:b3,platform:Mac OS X,
 03000000381000002014000001000000,SteelSeries Nimbus,a:b0,b:b1,dpdown:b9,dpleft:b11,dpright:b10,dpup:b8,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1~,rightshoulder:b5,righttrigger:b7,rightx:a2,righty:a3~,x:b2,y:b3,platform:Mac OS X,
+03000000381000004114000000010000,SteelSeries Nimbus Cloud,a:b0,b:b1,x:b3,y:b4,back:b10,guide:b15,start:b11,leftstick:b13,rightstick:b14,leftshoulder:b6,rightshoulder:b7,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,paddle1:b5,paddle2:b2,leftx:a0,lefty:a1,rightx:a2,righty:a3,lefttrigger:a5,righttrigger:a4,platform:Mac OS X,
 05000000484944204465766963650000,SteelSeries Nimbus Plus,a:b0,b:b1,back:b15,dpdown:b11,dpleft:b13,dpright:b12,dpup:b10,guide:b16,leftshoulder:b4,leftstick:b8,lefttrigger:b6,leftx:a0,lefty:a1~,rightshoulder:b5,rightstick:b9,righttrigger:b7,rightx:a2,righty:a3~,start:b14,x:b2,y:b3,platform:Mac OS X,
 050000004e696d6275732b0000000000,SteelSeries Nimbus Plus,a:b0,b:b1,back:b15,dpdown:b11,dpleft:b13,dpright:b12,dpup:b10,guide:b16,leftshoulder:b4,leftstick:b8,lefttrigger:b6,leftx:a0,lefty:a1~,rightshoulder:b5,rightstick:b9,righttrigger:b7,rightx:a2,righty:a3~,start:b14,x:b2,y:b3,platform:Mac OS X,
 03000000381000003014000000000000,SteelSeries Stratus Duo,a:b0,b:b1,back:b9,dpdown:b12,dpleft:b13,dpright:b14,dpup:b11,guide:b10,leftshoulder:b4,leftstick:b6,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b7,righttrigger:a5,rightx:a3,righty:a4,start:b8,x:b2,y:b3,platform:Mac OS X,
@@ -1249,12 +1253,16 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000c82d00001230000011010000,8BitDo Ultimate,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,paddle1:b2,paddle2:b5,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000c82d00001330000011010000,8BitDo Ultimate,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,misc1:b26,paddle1:b23,paddle2:b19,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 05000000c82d00001230000000010000,8BitDo Ultimate,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
+03000000c82d00001260000011010000,8BitDo Ultimate 2,a:b0,b:b1,x:b3,y:b4,back:b10,guide:b12,start:b11,leftstick:b13,rightstick:b14,leftshoulder:b6,rightshoulder:b7,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,paddle1:b17,paddle2:b16,paddle3:b2,paddle4:b5,rightx:a2,righty:a3,lefttrigger:a5,righttrigger:a4,platform:Linux,
+03000000c82d00000b31000014010000,8BitDo Ultimate 2,a:b0,b:b1,x:b2,y:b3,back:b6,guide:b8,start:b7,leftstick:b9,rightstick:b10,leftshoulder:b4,rightshoulder:b5,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,rightx:a3,righty:a4,lefttrigger:a2,righttrigger:a5,platform:Linux,
+05000000c82d00001260000001000000,8BitDo Ultimate 2,a:b0,b:b1,x:b3,y:b4,back:b10,guide:b12,start:b11,leftstick:b13,rightstick:b14,leftshoulder:b6,rightshoulder:b7,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,paddle1:b17,paddle2:b16,paddle3:b2,paddle4:b5,rightx:a2,righty:a3,lefttrigger:a5,righttrigger:a4,platform:Linux,
 03000000c82d00000a31000014010000,8BitDo Ultimate 2C,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 03000000c82d00001d30000011010000,8BitDo Ultimate 2C,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,paddle1:b5,paddle2:b2,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 05000000c82d00001b30000001000000,8BitDo Ultimate 2C,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,paddle1:b5,paddle2:b2,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000c82d00001530000011010000,8BitDo Ultimate C,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000c82d00001630000011010000,8BitDo Ultimate C,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000c82d00001730000011010000,8BitDo Ultimate C,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
+06000000c82d00000320000007010000,8BitDo Ultimate Wired Controller for Xbox,a:b0,b:b1,x:b2,y:b3,back:b6,guide:b8,start:b7,leftstick:b9,rightstick:b10,leftshoulder:b4,rightshoulder:b5,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,rightx:a3,righty:a4,lefttrigger:a2,righttrigger:a5,platform:Linux,
 03000000c82d00000121000011010000,8BitDo Xbox One SN30 Pro,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 05000000c82d00000121000000010000,8BitDo Xbox One SN30 Pro,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 05000000a00500003232000001000000,8BitDo Zero,a:b0,b:b1,back:b10,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,leftshoulder:b6,rightshoulder:b7,start:b11,x:b3,y:b4,platform:Linux,
@@ -1276,6 +1284,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000020500000913000010010000,Anbernic RG P01,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000373500000710000010010000,Anbernic RG P01,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 05000000373500004610000001000000,Anbernic RG P01,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,misc1:b15,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
+19000000010000000100000000010000,Anbernic RG28XX Controller,a:b4,b:b3,back:b9,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b11,leftshoulder:b7,misc1:b1,rightshoulder:b8,leftstick:b12,rightstick:b13,start:b10,x:b5,y:b6,platform:Linux,
 03000000190e00000110000010010000,Aquaplus Piece,a:b1,b:b0,back:b3,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,start:b2,platform:Linux,
 03000000790000003018000011010000,Arcade Fightstick F300,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,righttrigger:b7,start:b9,x:b0,y:b3,platform:Linux,
 03000000a30c00002700000011010000,Astro City Mini,a:b2,b:b1,back:b8,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,rightshoulder:b4,righttrigger:b5,start:b9,x:b3,y:b0,platform:Linux,
@@ -1348,6 +1357,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 05000000ac0500002d0200001b010000,GameSir G4s,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b33,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000ac0500007a05000011010000,GameSir G5,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:b8,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b16,righttrigger:b9,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000373500009710000001020000,GameSir Kaleid Flux,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,misc1:b15,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
+03000000373500005810000001020000,GameSir K1,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,misc1:b15,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000ac0500001a06000011010000,GameSir T3 2.02,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b15,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
 03000000373500000410000044010000,GameSir T4 Kaleid,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 03000000bc2000005656000011010000,GameSir T4w,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b0,y:b3,platform:Linux,
@@ -1446,6 +1456,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000242f00008a00000011010000,JYS Adapter,a:b1,b:b4,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:b8,rightshoulder:b7,rightstick:b14,righttrigger:b9,rightx:a2,righty:a3,start:b11,x:b0,y:b3,platform:Linux,
 03000000314100001935000000010000,LeadJoy Xeno Plus,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:+a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:+a5,rightx:a4,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 030000006f0e00000103000000020000,Logic3 Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
+030000006f0e000001f5000010010000,Logic3 Gamepad,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 030000006d040000d1ca000000000000,Logitech Chillstream,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b0,y:b3,platform:Linux,
 030000006d040000d1ca000011010000,Logitech Chillstream,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b0,y:b3,platform:Linux,
 030000006d04000016c2000010010000,Logitech Dual Action,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b0,y:b3,platform:Linux,
@@ -1536,10 +1547,10 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000853200000706000012010000,Nacon GC100,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 030000006b140000010c000010010000,Nacon GC400ES,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b2,y:b3,platform:Linux,
 05000000853200000503000000010000,Nacon MGX Pro,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
-0300000085320000170d000011010000,Nacon Revolution 5 Pro,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Linux,
-0300000085320000190d000011010000,Nacon Revolution 5 Pro,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Linux,
-0300000085320000180d000011010000,Nacon Revolution 5 Pro,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,x:b0,y:b3,platform:Linux,
 0300000085320000160d000011010000,Nacon Revolution 5 Pro,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Linux,
+0300000085320000170d000011010000,Nacon Revolution 5 Pro,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Linux,
+0300000085320000180d000011010000,Nacon Revolution 5 Pro,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,x:b0,y:b3,platform:Linux,
+0300000085320000190d000011010000,Nacon Revolution 5 Pro,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:a3,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:a4,rightx:a2,righty:a5,start:b9,touchpad:b13,x:b0,y:b3,platform:Linux,
 030000000d0f00000900000010010000,Natec Genesis P44,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b0,y:b3,platform:Linux,
 030000004f1f00000800000011010000,NeoGeo PS3,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,righttrigger:b7,start:b9,x:b0,y:b3,platform:Linux,
 0300000092120000474e000000010000,NeoGeo X Arcade Stick,a:b0,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,start:b9,x:b3,y:b2,platform:Linux,
@@ -1618,6 +1629,7 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000d62000000520000050010000,PowerA Xbox Series Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 03000000d62000000b20000001010000,PowerA Xbox Series Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 03000000d62000000f20000001010000,PowerA Xbox Series Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b7,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
+03000000d62000006520000002010000,PowerA Xbox Series Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 030000006d040000d2ca000011010000,Precision Controller,a:b1,b:b2,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b0,y:b3,platform:Linux,
 03000000250900000017000010010000,PS SS N64 Adapter,a:b1,b:b2,dpdown:b14,dpleft:b15,dpright:b13,dpup:b12,leftshoulder:b5,lefttrigger:b9,leftx:a0,lefty:a1,rightshoulder:b7,rightx:a2~,righty:a3,start:b8,platform:Linux,
 03000000ff1100004133000010010000,PS2 Controller,a:b2,b:b1,back:b8,leftshoulder:b6,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b7,righttrigger:b5,start:b9,x:b3,y:b0,platform:Linux,
@@ -1751,7 +1763,8 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 030000005e0400008e02000073050000,Speedlink Torid,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 030000005e0400008e02000020200000,SpeedLink Xeox Pro Analog,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 03000000de280000ff11000001000000,Steam Virtual Gamepad,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
-05000000381000001214000001000000,SteelSeries Free,a:b0,b:b1,x:b3,y:b4,back:b10,guide:b12,start:b11,leftstick:b13,rightstick:b14,leftshoulder:b6,rightshoulder:b7,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,rightx:a2,righty:a3,lefttrigger:a2,righttrigger:a3,platform:Linux,
+05000000381000001214000001000000,SteelSeries Free,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b6,leftstick:b13,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a3,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,
+05000000381000004114000000010000,SteelSeries Nimbus Cloud,a:b0,b:b1,x:b3,y:b4,back:b10,guide:b15,start:b11,leftstick:b13,rightstick:b14,leftshoulder:b6,rightshoulder:b7,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,paddle1:b5,paddle2:b2,leftx:a0,lefty:a1,rightx:a2,righty:a3,lefttrigger:a5,righttrigger:a4,platform:Linux,
 050000004e696d6275732b0000000000,SteelSeries Nimbus Plus,a:b0,b:b1,back:b10,guide:b11,leftshoulder:b4,leftstick:b8,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b9,righttrigger:b7,rightx:a2,righty:a3,start:b12,x:b2,y:b3,platform:Linux,
 03000000381000003014000075010000,SteelSeries Stratus Duo,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 03000000381000003114000075010000,SteelSeries Stratus Duo,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
@@ -1797,9 +1810,10 @@ xinput,XInput Controller,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,
 03000000e00d00000300000003000000,TRBot Virtual Joypad,a:b11,b:b12,back:b15,dpdown:b6,dpleft:b3,dpright:b4,dpup:b5,leftshoulder:b17,leftstick:b21,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b18,rightstick:b22,righttrigger:a2,rightx:a3,righty:a4,start:b16,x:b13,y:b14,platform:Linux,
 03000000f00600000300000003000000,TRBot Virtual Joypad,a:b11,b:b12,back:b15,dpdown:b6,dpleft:b3,dpright:b4,dpup:b5,leftshoulder:b17,leftstick:b21,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b18,rightstick:b22,righttrigger:a2,rightx:a3,righty:a4,start:b16,x:b13,y:b14,platform:Linux,
 030000005f140000c501000010010000,Trust Gamepad,a:b2,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b12,leftshoulder:b4,leftstick:b10,lefttrigger:b6,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b11,righttrigger:b7,rightx:a2,righty:a3,start:b9,x:b3,y:b0,platform:Linux,
+03000000f51000001370000000000000,Turtle Beach ReactR,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
+03000000f51000001370000000010000,Turtle Beach ReactR,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
 06000000f51000000870000003010000,Turtle Beach Recon,a:b0,b:b1,back:b6,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,guide:b8,leftshoulder:b4,leftstick:b9,lefttrigger:a2,leftx:a0,lefty:a1,rightshoulder:b5,rightstick:b10,righttrigger:a5,rightx:a3,righty:a4,start:b7,x:b2,y:b3,platform:Linux,
-03000000f51000001370000000010000,Turtle Beach ReactR,a:b0,b:b1,x:b2,y:b3,back:b6,guide:b8,start:b7,leftstick:b9,rightstick:b10,leftshoulder:b4,rightshoulder:b5,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,rightx:a3,righty:a4,lefttrigger:a2,righttrigger:a5,platform:Linux,
-03000000f51000001370000000000000,Turtle Beach ReactR,a:b0,b:b1,x:b2,y:b3,back:b6,guide:b8,start:b7,leftstick:b9,rightstick:b10,leftshoulder:b4,rightshoulder:b5,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,rightx:a3,righty:a4,lefttrigger:a2,righttrigger:a5,platform:Linux,
+05000000f51000008f01000001000000,Turtle Beach Rematch Nintendo Switch Controller,a:b0,b:b1,x:b2,y:b3,back:b8,guide:b12,start:b9,leftstick:b10,rightstick:b11,leftshoulder:b4,rightshoulder:b5,lefttrigger:b6,righttrigger:b7,dpup:h0.1,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,leftx:a0,lefty:a1,rightx:a2,righty:a3,platform:Linux,
 03000000100800000100000010010000,Twin PlayStation Adapter,a:b2,b:b1,back:b8,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftstick:b10,lefttrigger:b4,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b11,righttrigger:b5,rightx:a3,righty:a2,start:b9,x:b3,y:b0,platform:Linux,
 03000000c0160000e105000010010000,Ultimate Atari Fight Stick,a:b1,b:b2,back:b9,dpdown:+a1,dpleft:-a0,dpright:+a0,dpup:-a1,rightshoulder:b4,righttrigger:b5,start:b8,x:b0,y:b3,platform:Linux,
 03000000151900005678000010010000,Uniplay U6,a:b0,b:b1,back:b10,dpdown:h0.4,dpleft:h0.8,dpright:h0.2,dpup:h0.1,leftshoulder:b6,leftstick:b13,lefttrigger:a5,leftx:a0,lefty:a1,rightshoulder:b7,rightstick:b14,righttrigger:a4,rightx:a2,righty:a3,start:b11,x:b3,y:b4,platform:Linux,

bin/resources/icons/flags/az.svg

@@ -0,0 +1,8 @@
+<svg xmlns="http://www.w3.org/2000/svg" id="flag-icons-az" viewBox="0 0 640 480">
+  <path fill="#3f9c35" d="M.1 0h640v480H.1z"/>
+  <path fill="#ed2939" d="M.1 0h640v320H.1z"/>
+  <path fill="#00b9e4" d="M.1 0h640v160H.1z"/>
+  <circle cx="304" cy="240" r="72" fill="#fff"/>
+  <circle cx="320" cy="240" r="60" fill="#ed2939"/>
+  <path fill="#fff" d="m384 200 7.7 21.5 20.6-9.8-9.8 20.7L424 240l-21.5 7.7 9.8 20.6-20.6-9.8L384 280l-7.7-21.5-20.6 9.8 9.8-20.6L344 240l21.5-7.7-9.8-20.6 20.6 9.8z"/>
+</svg>

bin/resources/shaders/dx11/convert.fx

@@ -1,6 +1,8 @@
 // SPDX-FileCopyrightText: 2002-2026 PCSX2 Dev Team
 // SPDX-License-Identifier: GPL-3.0+
 
+#if defined(VERTEX_SHADER)
+
 struct VS_INPUT
 {
 	float4 p : POSITION;
@@ -15,6 +17,21 @@ struct VS_OUTPUT
 	float4 c : COLOR;
 };
 
+VS_OUTPUT vs_main(VS_INPUT input)
+{
+	VS_OUTPUT output;
+
+	output.p = input.p;
+	output.t = input.t;
+	output.c = input.c;
+
+	return output;
+}
+
+#endif // VERTEX_SHADER
+
+#if defined(PIXEL_SHADER)
+
 cbuffer cb0 : register(b0)
 {
 	float4 BGColor;
@@ -33,11 +50,22 @@ static const float3x3 rgb2yuv =
 Texture2D Texture;
 SamplerState TextureSampler;
 
+#if HAS_FLOAT32_INPUT
+
+float sample_c(float2 uv)
+{
+	return Texture.Sample(TextureSampler, uv).r;
+}
+
+#else
+
 float4 sample_c(float2 uv)
 {
 	return Texture.Sample(TextureSampler, uv);
 }
 
+#endif
+
 struct PS_INPUT
 {
 	float4 p : SV_Position;
@@ -45,36 +73,117 @@ struct PS_INPUT
 	float4 c : COLOR;
 };
 
+#if HAS_INTEGER_OUTPUT
+	#define OUTPUT_TYPE uint
+	#define OUTPUT_SV SV_Target
+#elif HAS_DEPTH_OUTPUT
+	#define OUTPUT_TYPE float
+	#define OUTPUT_SV SV_Depth
+#elif HAS_FLOAT32_OUTPUT
+	#define OUTPUT_TYPE float
+	#define OUTPUT_SV SV_Target
+#else
+	#define OUTPUT_TYPE float4
+	#define OUTPUT_SV SV_Target
+#endif
+
 struct PS_OUTPUT
 {
-	float4 c : SV_Target0;
+	OUTPUT_TYPE o : OUTPUT_SV;
 };
 
-VS_OUTPUT vs_main(VS_INPUT input)
+uint rgba8_to_uint(float4 c)
 {
-	VS_OUTPUT output;
-
-	output.p = input.p;
-	output.t = input.t;
-	output.c = input.c;
-
-	return output;
+	uint4 i = uint4(c * 255.5f) & 0xFFu;
+	return i.r | (i.g << 8) | (i.b << 16) | (i.a << 24);
 }
 
-PS_OUTPUT ps_copy(PS_INPUT input)
+uint rgb5a1_to_uint(float4 c)
 {
-	PS_OUTPUT output;
-	
-	output.c = sample_c(input.t);
-
-	return output;
+	uint4 i = uint4(c * 255.5f) & uint4(0xF8u, 0xF8u, 0xF8u, 0x80u);
+	return (i.r >> 3) | (i.g << 2) | (i.b << 7) | (i.a << 8);
 }
 
-float ps_depth_copy(PS_INPUT input) : SV_Depth
+uint depth_to_uint(float d)
 {
-	return sample_c(input.t).r;
+	return uint(d * exp2(32.0f));
 }
 
+float4 uint_to_rgba8(uint i)
+{
+	return float4((i & 0xFFu), ((i >> 8) & 0xFFu), ((i >> 16) & 0xFFu), ((i >> 24) & 0xFFu)) / 255.0f;
+}
+
+float4 uint_to_rgb5a1(uint i)
+{
+	return float4(uint4(i << 3, i >> 2, i >> 7, i >> 8) & uint4(0xF8u, 0xF8u, 0xF8u, 0x80u)) / 255.0f;
+}
+
+float uint_to_depth32(uint i)
+{
+	return float(i) * exp2(-32.0f);
+}
+
+float uint_to_depth24(uint i)
+{
+	return float(i & 0xFFFFFFu) * exp2(-32.0f);
+}
+
+float uint_to_depth16(uint i)
+{
+	return float(i & 0xFFFFu) * exp2(-32.0f);
+}
+
+float rgba8_to_depth32(float4 val)
+{
+	return uint_to_depth32(rgba8_to_uint(val));
+}
+
+float rgba8_to_depth24(float4 val)
+{
+	return uint_to_depth24(rgba8_to_uint(val));
+}
+
+float rgba8_to_depth16(float4 val)
+{
+	return uint_to_depth16(rgba8_to_uint(val));
+}
+
+float rgb5a1_to_depth16(float4 val)
+{
+	return uint_to_depth16(rgb5a1_to_uint(val));
+}
+
+float4 depth32_to_rgba8(float d)
+{
+	return uint_to_rgba8(depth_to_uint(d));
+}
+
+float4 depth16_to_rgb5a1(float d)
+{
+	return uint_to_rgb5a1(depth_to_uint(d));
+}
+
+float depth32_to_depth24(float d)
+{
+	return uint_to_depth24(depth_to_uint(d));
+}
+
+#if defined(__ps_copy__)
+OUTPUT_TYPE ps_copy(PS_INPUT input) : OUTPUT_SV
+{
+	return sample_c(input.t);
+}
+#endif
+
+#if defined(__ps_depth_copy__)
+OUTPUT_TYPE ps_depth_copy(PS_INPUT input) : OUTPUT_SV
+{
+	return sample_c(input.t);
+}
+#endif
+
+#if defined(__ps_downsample_copy__)
 PS_OUTPUT ps_downsample_copy(PS_INPUT input)
 {
 	int DownsampleFactor = DOFFSET;
@@ -85,179 +194,134 @@ PS_OUTPUT ps_downsample_copy(PS_INPUT input)
 	int2 coord = max(int2(input.p.xy) * DownsampleFactor, ClampMin);
 
 	PS_OUTPUT output;
-	output.c = (float4)0;
+	output.o = (float4)0;
 	for (int yoff = 0; yoff < DownsampleFactor; yoff++)
 	{
 		for (int xoff = 0; xoff < DownsampleFactor; xoff++)
-			output.c += Texture.Load(int3(coord + int2(xoff * step_multiplier, yoff * step_multiplier), 0));
+			output.o += Texture.Load(int3(coord + int2(xoff * step_multiplier, yoff * step_multiplier), 0));
 	}
-	output.c /= Weight;
+	output.o /= Weight;
 	return output;
 }
+#endif
 
+#if defined(__ps_filter_transparency__)
 PS_OUTPUT ps_filter_transparency(PS_INPUT input)
 {
 	PS_OUTPUT output;
 	float4 c = sample_c(input.t);
-	output.c = float4(c.rgb, 1.0);
+	output.o = float4(c.rgb, 1.0);
 	return output;
 }
+#endif
 
-// Need to be careful with precision here, it can break games like Spider-Man 3 and Dogs Life
-uint ps_convert_rgba8_16bits(PS_INPUT input) : SV_Target0
+#if defined(__ps_convert_rgb5a1_16bits__)
+OUTPUT_TYPE ps_convert_rgb5a1_16bits(PS_INPUT input) : OUTPUT_SV
 {
-	uint4 i = sample_c(input.t) * float4(255.5f, 255.5f, 255.5f, 255.5f);
-
-	return ((i.x & 0x00F8u) >> 3) | ((i.y & 0x00F8u) << 2) | ((i.z & 0x00f8u) << 7) | ((i.w & 0x80u) << 8);
+	// Need to be careful with precision here, it can break games like Spider-Man 3 and Dogs Life
+	return rgb5a1_to_uint(sample_c(input.t));
 }
+#endif
 
+#if defined(__ps_datm1__)
 void ps_datm1(PS_INPUT input)
 {
 	clip(sample_c(input.t).a - 127.5f / 255); // >= 0x80 pass
 }
+#endif
 
+#if defined(__ps_datm0__)
 void ps_datm0(PS_INPUT input)
 {
 	clip(127.5f / 255 - sample_c(input.t).a); // < 0x80 pass (== 0x80 should not pass)
 }
+#endif
 
+#if defined(__ps_datm1_rta_correction__)
 void ps_datm1_rta_correction(PS_INPUT input)
 {
 	clip(sample_c(input.t).a - 254.5f / 255); // >= 0x80 pass
 }
+#endif
 
+#if defined(__ps_datm0_rta_correction__)
 void ps_datm0_rta_correction(PS_INPUT input)
 {
 	clip(254.5f / 255 - sample_c(input.t).a); // < 0x80 pass (== 0x80 should not pass)
 }
+#endif
 
+#if defined(__ps_rta_correction__)
 PS_OUTPUT ps_rta_correction(PS_INPUT input)
 {
 	PS_OUTPUT output;
 	float4 value = sample_c(input.t);
-	output.c = float4(value.rgb, value.a / (128.25f / 255.0f));
+	output.o = float4(value.rgb, value.a / (128.25f / 255.0f));
 	return output;
 }
+#endif
 
+#if defined(__ps_rta_decorrection__)
 PS_OUTPUT ps_rta_decorrection(PS_INPUT input)
 {
 	PS_OUTPUT output;
 	float4 value = sample_c(input.t);
-	output.c = float4(value.rgb, value.a * (128.25f / 255.0f));
+	output.o = float4(value.rgb, value.a * (128.25f / 255.0f));
 	return output;
 }
+#endif
 
+#if defined(__ps_colclip_init__)
 PS_OUTPUT ps_colclip_init(PS_INPUT input)
 {
 	PS_OUTPUT output;
 	float4 value = sample_c(input.t);
-	output.c = float4(round(value.rgb * 255) / 65535, value.a);
+	output.o = float4(round(value.rgb * 255) / 65535, value.a);
 	return output;
 }
+#endif
 
+#if defined(__ps_colclip_resolve__)
 PS_OUTPUT ps_colclip_resolve(PS_INPUT input)
 {
 	PS_OUTPUT output;
 	float4 value = sample_c(input.t);
-	output.c = float4(float3(uint3(value.rgb * 65535.5) & 255) / 255, value.a);
+	output.o = float4(float3(uint3(value.rgb * 65535.5) & 255) / 255, value.a);
 	return output;
 }
+#endif
 
-uint ps_convert_float32_32bits(PS_INPUT input) : SV_Target0
+#if defined(__ps_convert_depth32_32bits__)
+OUTPUT_TYPE ps_convert_depth32_32bits(PS_INPUT input) : OUTPUT_SV
 {
-	// Convert a FLOAT32 depth texture into a 32 bits UINT texture
-	return uint(exp2(32.0f) * sample_c(input.t).r);
+	// Convert a depth texture into a 32 bits UINT texture
+	return depth_to_uint(sample_c(input.t));
 }
+#endif
 
-PS_OUTPUT ps_convert_float32_rgba8(PS_INPUT input)
+#if defined(__ps_convert_depth32_rgba8__)
+OUTPUT_TYPE ps_convert_depth32_rgba8(PS_INPUT input) : OUTPUT_SV
 {
-	PS_OUTPUT output;
-
-	// Convert a FLOAT32 depth texture into a RGBA color texture
-	uint d = uint(sample_c(input.t).r * exp2(32.0f));
-	output.c = float4(uint4((d & 0xFFu), ((d >> 8) & 0xFFu), ((d >> 16) & 0xFFu), (d >> 24))) / 255.0f;
-
-	return output;
+	// Convert a depth texture into a RGBA color texture
+	return depth32_to_rgba8(sample_c(input.t));
 }
+#endif
 
-PS_OUTPUT ps_convert_float16_rgb5a1(PS_INPUT input)
+#if defined(__ps_convert_depth16_rgb5a1__)
+OUTPUT_TYPE ps_convert_depth16_rgb5a1(PS_INPUT input) : OUTPUT_SV
 {
-	PS_OUTPUT output;
-
-	// Convert a FLOAT32 (only 16 lsb) depth into a RGB5A1 color texture
-	uint d = uint(sample_c(input.t).r * exp2(32.0f));
-	output.c = float4(uint4(d << 3, d >> 2, d >> 7, d >> 8) & uint4(0xf8, 0xf8, 0xf8, 0x80)) / 255.0f;
-	return output;
+	// Convert depth (only 16 lsb) into a RGB5A1 color texture
+	return depth16_to_rgb5a1(sample_c(input.t));
 }
+#endif
 
-float rgba8_to_depth32(float4 val)
-{
-	uint4 c = uint4(val * 255.5f);
-	return float(c.r | (c.g << 8) | (c.b << 16) | (c.a << 24)) * exp2(-32.0f);
-}
-
-float rgba8_to_depth24(float4 val)
-{
-	uint3 c = uint3(val.rgb * 255.5f);
-	return float(c.r | (c.g << 8) | (c.b << 16)) * exp2(-32.0f);
-}
-
-float rgba8_to_depth16(float4 val)
-{
-	uint2 c = uint2(val.rg * 255.5f);
-	return float(c.r | (c.g << 8)) * exp2(-32.0f);
-}
-
-float rgb5a1_to_depth16(float4 val)
-{
-	uint4 c = uint4(val * 255.5f);
-	return float(((c.r & 0xF8u) >> 3) | ((c.g & 0xF8u) << 2) | ((c.b & 0xF8u) << 7) | ((c.a & 0x80u) << 8)) * exp2(-32.0f);
-}
-
-float ps_convert_float32_depth_to_color(PS_INPUT input) : SV_Target0
-{
-	return sample_c(input.t).r;
-}
-
-float ps_convert_float32_color_to_depth(PS_INPUT input) : SV_Depth
-{
-	return sample_c(input.t).r;
-}
-
-float ps_convert_float32_float24(PS_INPUT input) : SV_Depth
+#if defined(__ps_convert_depth32_depth24__)
+OUTPUT_TYPE ps_convert_depth32_depth24(PS_INPUT input) : OUTPUT_SV
 {
 	// Truncates depth value to 24bits
-	uint d = uint(sample_c(input.t).r * exp2(32.0f)) & 0xFFFFFFu;
-	return float(d) * exp2(-32.0f);
-}
-
-float ps_convert_rgba8_float32(PS_INPUT input) : SV_Depth
-{
-	// Convert an RGBA texture into a float depth texture
-	return rgba8_to_depth32(sample_c(input.t));
-}
-
-float ps_convert_rgba8_float24(PS_INPUT input) : SV_Depth
-{
-	// Same as above but without the alpha channel (24 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
-	return rgba8_to_depth24(sample_c(input.t));
-}
-
-float ps_convert_rgba8_float16(PS_INPUT input) : SV_Depth
-{
-	// Same as above but without the A/B channels (16 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
-	return rgba8_to_depth16(sample_c(input.t));
-}
-
-float ps_convert_rgb5a1_float16(PS_INPUT input) : SV_Depth
-{
-	// Convert an RGB5A1 (saved as RGBA8) color to a 16 bit Z
-	return rgb5a1_to_depth16(sample_c(input.t));
+	return depth32_to_depth24(sample_c(input.t));
 }
+#endif
 
 #define SAMPLE_RGBA_DEPTH_BILN(CONVERT_FN) \
 	uint width, height; \
@@ -272,34 +336,57 @@ float ps_convert_rgb5a1_float16(PS_INPUT input) : SV_Depth
 	float depthBR = CONVERT_FN(Texture.Load(int3(coords.zw, 0))); \
 	return lerp(lerp(depthTL, depthTR, mix_vals.x), lerp(depthBL, depthBR, mix_vals.x), mix_vals.y);
 
-float ps_convert_rgba8_float32_biln(PS_INPUT input) : SV_Depth
+#if defined(__ps_convert_rgba8_depth32__)
+OUTPUT_TYPE ps_convert_rgba8_depth32(PS_INPUT input) : OUTPUT_SV
 {
 	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth32);
+#else
+	return rgba8_to_depth32(sample_c(input.t));
+#endif
 }
+#endif
 
-float ps_convert_rgba8_float24_biln(PS_INPUT input) : SV_Depth
+#if defined(__ps_convert_rgba8_depth24__)
+OUTPUT_TYPE ps_convert_rgba8_depth24(PS_INPUT input) : OUTPUT_SV
 {
 	// Same as above but without the alpha channel (24 bits Z)
-
 	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth24);
+#else
+	return rgba8_to_depth24(sample_c(input.t));
+#endif
 }
+#endif
 
-float ps_convert_rgba8_float16_biln(PS_INPUT input) : SV_Depth
+#if defined(__ps_convert_rgba8_depth16__)
+OUTPUT_TYPE ps_convert_rgba8_depth16(PS_INPUT input) : OUTPUT_SV
 {
 	// Same as above but without the A/B channels (16 bits Z)
-
 	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth16);
+#else
+	return rgba8_to_depth16(sample_c(input.t));
+#endif
 }
+#endif
 
-float ps_convert_rgb5a1_float16_biln(PS_INPUT input) : SV_Depth
+#if defined(__ps_convert_rgb5a1_depth16__)
+OUTPUT_TYPE ps_convert_rgb5a1_depth16(PS_INPUT input) : OUTPUT_SV
 {
 	// Convert an RGB5A1 (saved as RGBA8) color to a 16 bit Z
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgb5a1_to_depth16);
+#else
+	return rgb5a1_to_depth16(sample_c(input.t));
+#endif
 }
+#endif
 
+#if defined(__ps_convert_rgb5a1_8i__)
 PS_OUTPUT ps_convert_rgb5a1_8i(PS_INPUT input)
 {
 	PS_OUTPUT output;
@@ -406,7 +493,7 @@ PS_OUTPUT ps_convert_rgb5a1_8i(PS_INPUT input)
 		uint red = (denorm_c.r >> 3) & 0x1Fu;
 		uint green = (denorm_c.g >> 3) & 0x1Fu;
 		
-		output.c = (float4)(((float)(((green << 5) | red) & 0xFFu)) / 255.0f);
+		output.o = (float4)(((float)(((green << 5) | red) & 0xFFu)) / 255.0f);
 	}
 	else
 	{
@@ -414,11 +501,13 @@ PS_OUTPUT ps_convert_rgb5a1_8i(PS_INPUT input)
 		uint blue = (denorm_c.b >> 3) & 0x1Fu;
 		uint alpha = denorm_c.a & 0x80u;
 
-		output.c = (float4)(((float)((alpha | (blue << 2) | (green >> 3)) & 0xFFu)) / 255.0f);
+		output.o = (float4)(((float)((alpha | (blue << 2) | (green >> 3)) & 0xFFu)) / 255.0f);
 	}
 	return output;
 }
+#endif
 
+#if defined(__ps_convert_rgba_8i__)
 PS_OUTPUT ps_convert_rgba_8i(PS_INPUT input)
 {
 	PS_OUTPUT output;
@@ -462,10 +551,12 @@ PS_OUTPUT ps_convert_rgba_8i(PS_INPUT input)
 	float4 pixel = Texture.Load(int3(int2(coord), 0));
 	float2 sel0 = (pos.y & 2u) == 0u ? pixel.rb : pixel.ga;
 	float  sel1 = (pos.x & 8u) == 0u ? sel0.x : sel0.y;
-	output.c = (float4)(sel1); // Divide by something here?
+	output.o = (float4)(sel1); // Divide by something here?
 	return output;
 }
+#endif
 
+#if defined(__ps_convert_clut_4__)
 PS_OUTPUT ps_convert_clut_4(PS_INPUT input)
 {
 	// Borrowing the YUV constant buffer.
@@ -478,10 +569,12 @@ PS_OUTPUT ps_convert_clut_4(PS_INPUT input)
 
 	int2 final = int2(floor(float2(offset + pos) * scale));
 	PS_OUTPUT output;
-	output.c = Texture.Load(int3(final, 0), 0);
+	output.o = Texture.Load(int3(final, 0), 0);
 	return output;
 }
+#endif
 
+#if defined(__ps_convert_clut_8__)
 PS_OUTPUT ps_convert_clut_8(PS_INPUT input)
 {
 	float scale = BGColor.x;
@@ -497,10 +590,12 @@ PS_OUTPUT ps_convert_clut_8(PS_INPUT input)
 
 	int2 final = int2(floor(float2(offset + pos) * scale));
 	PS_OUTPUT output;
-	output.c = Texture.Load(int3(final, 0), 0);
+	output.o = Texture.Load(int3(final, 0), 0);
 	return output;
 }
+#endif
 
+#if defined(__ps_yuv__)
 PS_OUTPUT ps_yuv(PS_INPUT input)
 {
 	PS_OUTPUT output;
@@ -515,41 +610,43 @@ PS_OUTPUT ps_yuv(PS_INPUT input)
 	switch (EMODA)
 	{
 		case 0:
-			output.c.a = i.a;
+			output.o.a = i.a;
 			break;
 		case 1:
-			output.c.a = Y;
+			output.o.a = Y;
 			break;
 		case 2:
-			output.c.a = Y / 2.0f;
+			output.o.a = Y / 2.0f;
 			break;
 		case 3:
 		default:
-			output.c.a = 0.0f;
+			output.o.a = 0.0f;
 			break;
 	}
 
 	switch (EMODC)
 	{
 		case 0:
-			output.c.rgb = i.rgb;
+			output.o.rgb = i.rgb;
 			break;
 		case 1:
-			output.c.rgb = float3(Y, Y, Y);
+			output.o.rgb = float3(Y, Y, Y);
 			break;
 		case 2:
-			output.c.rgb = float3(Y, Cb, Cr);
+			output.o.rgb = float3(Y, Cb, Cr);
 			break;
 		case 3:
 		default:
-			output.c.rgb = float3(i.a, i.a, i.a);
+			output.o.rgb = float3(i.a, i.a, i.a);
 			break;
 	}
 
 	return output;
 }
+#endif
 
-float ps_stencil_image_init_0(PS_INPUT input) : SV_Target
+#if defined(__ps_primid_image_init_0__)
+float ps_primid_image_init_0(PS_INPUT input) : SV_Target
 {
 	float c;
 	if ((127.5f / 255.0f) < sample_c(input.t).a) // < 0x80 pass (== 0x80 should not pass)
@@ -558,8 +655,10 @@ float ps_stencil_image_init_0(PS_INPUT input) : SV_Target
 		c = float(0x7FFFFFFF);
 	return c;
 }
+#endif
 
-float ps_stencil_image_init_1(PS_INPUT input) : SV_Target
+#if defined(__ps_primid_image_init_1__)
+float ps_primid_image_init_1(PS_INPUT input) : SV_Target
 {
 	float c;
 	if (sample_c(input.t).a < (127.5f / 255.0f)) // >= 0x80 pass
@@ -568,9 +667,10 @@ float ps_stencil_image_init_1(PS_INPUT input) : SV_Target
 		c = float(0x7FFFFFFF);
 	return c;
 }
+#endif
 
-float ps_stencil_image_init_2(PS_INPUT input)
-	: SV_Target
+#if defined(__ps_primid_image_init_2__)
+float ps_primid_image_init_2(PS_INPUT input) : SV_Target
 {
 	float c;
 	if ((254.5f / 255.0f) < sample_c(input.t).a) // < 0x80 pass (== 0x80 should not pass)
@@ -579,9 +679,10 @@ float ps_stencil_image_init_2(PS_INPUT input)
 		c = float(0x7FFFFFFF);
 	return c;
 }
+#endif
 
-float ps_stencil_image_init_3(PS_INPUT input)
-	: SV_Target
+#if defined(__ps_primid_image_init_3__)
+float ps_primid_image_init_3(PS_INPUT input) : SV_Target
 {
 	float c;
 	if (sample_c(input.t).a < (254.5f / 255.0f)) // >= 0x80 pass
@@ -590,3 +691,6 @@ float ps_stencil_image_init_3(PS_INPUT input)
 		c = float(0x7FFFFFFF);
 	return c;
 }
+#endif
+
+#endif // PIXEL_SHADER

bin/resources/shaders/dx11/tfx.fx

@@ -50,6 +50,12 @@
 #define PS_AA1_TRIANGLE_SW_Z 3
 #endif
 
+#ifndef PS_ROV_DEPTH_NONE
+#define PS_ROV_DEPTH_NONE 0
+#define PS_ROV_DEPTH_READ_WRITE 1
+#define PS_ROV_DEPTH_READ_ONLY 2
+#endif
+
 #ifndef PS_FST
 #define PS_IIP 0
 #define PS_FST 0
@@ -111,6 +117,8 @@
 #define PS_TEX_IS_FB 0
 #define PS_AA1 0
 #define PS_ABE 0
+#define PS_ROV_COLOR 0
+#define PS_ROV_DEPTH 0
 #endif
 
 #ifndef VS_EXPAND_NONE
@@ -132,6 +140,12 @@
 #define SW_DEPTH (NEEDS_DEPTH_FOR_AFAIL || NEEDS_DEPTH_FOR_ZTST || NEEDS_DEPTH_FOR_AA1)
 #define ZWRITE (PS_ZFLOOR || PS_ZCLAMP || SW_DEPTH)
 
+#define PS_RETURN_COLOR_ROV (!PS_NO_COLOR && PS_ROV_COLOR)
+#define PS_RETURN_COLOR (!PS_NO_COLOR && !PS_ROV_COLOR)
+#define PS_RETURN_DEPTH_ROV (PS_ROV_DEPTH == PS_ROV_DEPTH_READ_WRITE)
+#define PS_RETURN_DEPTH (ZWRITE && !PS_ROV_DEPTH)
+#define PS_ROV_EARLYDEPTHSTENCIL (PS_ROV_COLOR && !PS_ROV_DEPTH && !ZWRITE)
+
 struct VS_INPUT
 {
 	float2 st : TEXCOORD0;
@@ -181,21 +195,26 @@ struct PS_INPUT
 struct PS_OUTPUT
 {
 #define NUM_RTS 0
-#if !PS_NO_COLOR
-#if PS_DATE == 1 || PS_DATE == 2
-	float c : SV_Target;
-#else
-	float4 c0 : SV_Target0;
-	#undef NUM_RTS
-	#define NUM_RTS 1
-#if !PS_NO_COLOR1
-	float4 c1 : SV_Target1;
+
+#if PS_RETURN_COLOR
+	#if PS_DATE == 1 || PS_DATE == 2
+		float c : SV_Target;
+	#else
+		
+		float4 c0 : SV_Target0;
+
+		#undef NUM_RTS
+		#define NUM_RTS 1
+		
+		#if !PS_NO_COLOR1
+			float4 c1 : SV_Target1;
+		#endif
+	#endif
 #endif
-#endif
-#endif
-#if ZWRITE
+
+#if PS_RETURN_DEPTH
 	// In DX12 we do depth feedback loops with a color copy.
-	#if SW_DEPTH && PS_NO_COLOR1 && DX12
+	#if SW_DEPTH && PS_NO_COLOR1 && PS_DEPTH_FEEDBACK_SUPPORT == 2
 		#if NUM_RTS > 0
 			float depth_color : SV_Target1;
 		#else
@@ -208,16 +227,31 @@ struct PS_OUTPUT
 		float depth : SV_Depth;
 	#endif
 #endif
+
 #undef NUM_RTS
 };
 
 Texture2D<float4> Texture : register(t0);
 Texture2D<float4> Palette : register(t1);
+#if !PS_ROV_COLOR
 Texture2D<float4> RtTexture : register(t2);
+#endif
 Texture2D<float> PrimMinTexture : register(t3);
+#if !PS_ROV_DEPTH
 Texture2D<float> DepthTexture : register(t4);
+#endif
 SamplerState TextureSampler : register(s0);
 
+#if PS_ROV_COLOR
+RasterizerOrderedTexture2D<unorm float4> RtTextureRov : register(u0);
+static float4 rov_rt_value;
+#endif
+
+#if PS_ROV_DEPTH
+RasterizerOrderedTexture2D<float> DepthTextureRov : register(u1);
+static float rov_depth_value;
+#endif
+
 #ifdef DX12
 cbuffer cb1 : register(b1)
 #else
@@ -242,8 +276,46 @@ cbuffer cb1
 	float4x4 DitherMatrix;
 	float ScaledScaleFactor;
 	float RcpScaleFactor;
+	float _pad0_cb1;
+	float _pad1_cb1;
+	float LineCovScale;
+	float _pad2_cb1;
+	float _pad3_cb1;
+	float _pad4_cb1;
 };
 
+float4 RtLoad(int2 xy)
+{
+#if PS_ROV_COLOR
+	return rov_rt_value;
+#else
+	return RtTexture.Load(int3(int2(xy), 0));
+#endif
+}
+
+float DepthLoad(int2 xy)
+{
+#if PS_ROV_DEPTH
+	return rov_depth_value;
+#else
+	return DepthTexture.Load(int3(int2(xy), 0));
+#endif
+}
+
+void RtWrite(int2 xy, float4 c)
+{
+#if PS_ROV_COLOR
+	RtTextureRov[xy] = c;
+#endif
+}
+
+void DepthWrite(int2 xy, float d)
+{
+#if PS_ROV_DEPTH
+	DepthTextureRov[xy] = d;
+#endif
+}
+
 #if (PS_AUTOMATIC_LOD != 1) && (PS_MANUAL_LOD == 1)
 float manual_lod(float uv_w)
 {
@@ -275,23 +347,28 @@ bool2 nan_or_inf(float2 xy)
 float4 sample_c_af(float2 uv, float uv_w)
 {
 	// HW sampler will reject bad UVs, match that here.
-	uv = any(nan_or_inf(uv)) ? float2(0, 0) : uv;
+	uv = any(nan_or_inf(uv)) ? float2(0.0f, 0.0f) : uv;
 
 	// Large floating point values risk NaN/Inf values.
 	// Above this value floats lose decimal precision, so seems a resonable limit for UVs.
 	uv = clamp(uv, -8388608.0f, 8388608.0f);
 
 	// Below taken from https://microsoft.github.io/DirectX-Specs/d3d/archive/D3D11_3_FunctionalSpec.htm#7.18.11%20LOD%20Calculations
+	// And https://registry.khronos.org/OpenGL/extensions/EXT/EXT_texture_filter_anisotropic.txt
 	// With guidance from https://pema.dev/2025/05/09/mipmaps-too-much-detail/ 
 	float2 sz;
 	Texture.GetDimensions(sz.x, sz.y);
 	float2 dX = ddx(uv) * sz;
 	float2 dY = ddy(uv) * sz;
 
+	float length_x = length(dX);
+	float length_y = length(dY);
+
 	// Calculate Ellipse Transform
-	bool d_zero = length(dX) == 0 || length(dY) == 0;
-	bool d_par = (dX.x * dY.y - dY.x * dX.y) == 0;
-	bool d_per = dot(dX, dY) == 0;
+	bool d_zero = length_x < 0.001f || length_y < 0.001f;
+	float f = (dX.x * dY.y - dX.y * dY.x);
+	bool d_par = f < 0.001f;
+	bool d_per = dot(dX, dY) < 0.001f;
 	bool d_inf_nan = any(nan_or_inf(dX) | nan_or_inf(dY));
 
 	if (!(d_zero || d_par || d_per || d_inf_nan))
@@ -299,42 +376,49 @@ float4 sample_c_af(float2 uv, float uv_w)
 		float A = dX.y * dX.y + dY.y * dY.y;
 		float B = -2 * (dX.x * dX.y + dY.x * dY.y);
 		float C = dX.x * dX.x + dY.x * dY.x;
-		float f = (dX.x * dY.y - dY.x * dX.y);
 		float F = f * f;
 
 		float p = A - C;
 		float q = A + C;
 		float t = sqrt(p * p + B * B);
 
+		float sqrt_num_plus  = sqrt(F * (t + p));
+		float sqrt_num_minus = sqrt(F * (t - p));
+
+		float inv_sqrt_denom_plus  = rsqrt(t * (q + t));
+		float inv_sqrt_denom_minus = rsqrt(t * (q - t));
+
+		float signB = sign(B);
+
 		float2 new_dX = float2(
-			sqrt(F * (t + p) / (t * (q + t))),
-			sqrt(F * (t - p) / (t * (q + t))) * sign(B)
+			sqrt_num_plus  * inv_sqrt_denom_plus,
+			sqrt_num_minus * inv_sqrt_denom_plus * signB
 		);
-		
+
 		float2 new_dY = float2(
-			sqrt(F * (t - p) / (t * (q - t))) * -sign(B),
-			sqrt(F * (t + p) / (t * (q - t)))
+			sqrt_num_minus * inv_sqrt_denom_minus * -signB,
+			sqrt_num_plus  * inv_sqrt_denom_minus
 		);
-		
+
 		d_inf_nan = any(nan_or_inf(new_dX) | nan_or_inf(new_dY));
 		if (!d_inf_nan)
 		{
 			dX = new_dX;
 			dY = new_dY;
+			length_x = length(dX);
+			length_y = length(dY);
 		}
 	}
 
 	// Compute AF values
-	float squared_length_x = dX.x * dX.x + dX.y * dX.y;
-	float squared_length_y = dY.x * dY.x + dY.y * dY.y;
-	float determinant = abs(dX.x * dY.y - dX.y * dY.x);
-	bool is_major_x = squared_length_x > squared_length_y;
-	float squared_length_major = is_major_x ? squared_length_x : squared_length_y;
-	float length_major = sqrt(squared_length_major);
+	bool is_major_x = length_x > length_y;
+	float length_major = is_major_x ? length_x : length_y;
+	float length_minor = is_major_x ? length_y : length_x;
 
 	float aniso_ratio;
 	float length_lod;
 	float2 aniso_line;
+
 	if (length_major <= 1.0f)
 	{
 		// A zero length_major would result in NaN Lod and break sampling.
@@ -342,57 +426,44 @@ float4 sample_c_af(float2 uv, float uv_w)
 		// Perform isotropic filtering instead.
 		aniso_ratio = 1.0f;
 		length_lod = length_major;
-		aniso_line = float2(0, 0);
+		aniso_line = float2(0.0f, 0.0f);
 	}
 	else
 	{
-		float norm_major = 1.0f / length_major;
-	
-		float2 aniso_line_dir = float2(
-			(is_major_x ? dX.x : dY.x) * norm_major,
-			(is_major_x ? dX.y : dY.y) * norm_major
-		);
-	
-		aniso_ratio = squared_length_major / determinant;
+		float2 aniso_line_dir = is_major_x ? dX : dY;
 
-		// Calculate the minor length of the ellipse for Lod, while also clamping the ratio of anisotropy.
-		if (aniso_ratio > PS_ANISOTROPIC_FILTERING)
-		{
-			// ratio is clamped - Lod is based on ratio (preserves area)
-			aniso_ratio = PS_ANISOTROPIC_FILTERING;
-			length_lod = length_major / PS_ANISOTROPIC_FILTERING;
-		}
-		else
-		{
-			// ratio not clamped - Lod is based on area
-			length_lod = determinant / length_major;
-		}
+		aniso_ratio = min(length_major / length_minor, PS_ANISOTROPIC_FILTERING);
+		length_lod = length_major / aniso_ratio;
 
 		// clamp to top Lod
 		if (length_lod < 1.0f)
 			aniso_ratio = max(1.0f, aniso_ratio * length_lod);
 
 		aniso_ratio = round(aniso_ratio);
-		aniso_line = aniso_line_dir * 0.5f * length_major * (1.0f / sz);
+
+		aniso_line = aniso_line_dir * 0.5f * (1.0f / sz);
 	}
-	
+
 #if PS_AUTOMATIC_LOD == 1
 	float lod = log2(length_lod);
 #elif PS_MANUAL_LOD == 1
 	float lod = manual_lod(uv_w);
 #else
-	float lod = 0; // No Lod
+	float lod = 0.0f; // No Lod
 #endif
-	
+
 	float4 colour;
 	if (aniso_ratio == 1.0f)
 		colour = Texture.SampleLevel(TextureSampler, uv, lod);
 	else
 	{
-		float4 num = float4(0, 0, 0, 0);
-		for (int i = 0; i < aniso_ratio; i++)
-		{		
-			float2 d = -aniso_line + (0.5f + i) * (2.0f * aniso_line) / aniso_ratio;	
+		float4 num = float4(0.0f, 0.0f, 0.0f, 0.0f);
+		float2 segment = (2.0f * aniso_line) / aniso_ratio;
+
+		int aniso_ratio_i = (int)aniso_ratio;
+		for (int i = 0; i < aniso_ratio_i; i++)
+		{
+			float2 d = -aniso_line + (0.5f + i) * segment;	
 			float2 uv_sample = uv + d;
 			float4 sample_colour = Texture.SampleLevel(TextureSampler, uv_sample, lod);
 			num += sample_colour;
@@ -407,7 +478,7 @@ float4 sample_c_af(float2 uv, float uv_w)
 float4 sample_c(float2 uv, float uv_w, int2 xy)
 {
 #if PS_TEX_IS_FB == 1
-	return RtTexture.Load(int3(int2(xy), 0));
+	return RtLoad(xy);
 #elif PS_REGION_RECT == 1
 	return Texture.Load(int3(int2(uv), 0));
 #else
@@ -602,7 +673,7 @@ float4x4 sample_4p(uint4 u)
 uint fetch_raw_depth(int2 xy)
 {
 #if PS_TEX_IS_FB == 1
-	float4 col = RtTexture.Load(int3(xy, 0));
+	float4 col = RtLoad(xy);
 #else
 	float4 col = Texture.Load(int3(xy, 0));
 #endif
@@ -612,7 +683,7 @@ uint fetch_raw_depth(int2 xy)
 float4 fetch_raw_color(int2 xy)
 {
 #if PS_TEX_IS_FB == 1
-	return RtTexture.Load(int3(xy, 0));
+	return RtLoad(xy);
 #else
 	return Texture.Load(int3(xy, 0));
 #endif
@@ -621,7 +692,7 @@ float4 fetch_raw_color(int2 xy)
 float4 fetch_c(int2 uv)
 {
 #if PS_TEX_IS_FB == 1
-	return RtTexture.Load(int3(uv, 0));
+	return RtLoad(uv);
 #else
 	return Texture.Load(int3(uv, 0));
 #endif
@@ -707,7 +778,7 @@ float4 sample_depth(float2 st, float2 pos)
 	}
 	else if (PS_DEPTH_FMT == 1)
 	{
-		// Based on ps_convert_float32_rgba8 of convert
+		// Based on ps_convert_depth32_rgba8 of convert
 
 		// Convert a FLOAT32 depth texture into a RGBA color texture
 		uint d = uint(fetch_c(uv).r * exp2(32.0f));
@@ -715,7 +786,7 @@ float4 sample_depth(float2 st, float2 pos)
 	}
 	else if (PS_DEPTH_FMT == 2)
 	{
-		// Based on ps_convert_float16_rgb5a1 of convert
+		// Based on ps_convert_depth16_rgb5a1 of convert
 
 		// Convert a FLOAT32 (only 16 lsb) depth into a RGB5A1 color texture
 		uint d = uint(fetch_c(uv).r * exp2(32.0f));
@@ -1027,7 +1098,7 @@ void ps_fbmask(inout float4 C, float2 pos_xy)
 	if (PS_FBMASK)
 	{
 		float multi = PS_COLCLIP_HW ? 65535.0f : 255.0f;
-		float4 RT = trunc(RtTexture.Load(int3(pos_xy, 0)) * multi + 0.1f);
+		float4 RT = trunc(RtLoad(int2(pos_xy)) * multi + 0.1f);
 		C = (float4)(((uint4)C & ~FbMask) | ((uint4)RT & FbMask));
 	}
 }
@@ -1103,7 +1174,7 @@ void ps_blend(inout float4 Color, inout float4 As_rgba, float2 pos_xy)
 			As_rgba.rgb = (float3)1.0f;
 		}
 
-		float4 RT = SW_BLEND_NEEDS_RT ? RtTexture.Load(int3(pos_xy, 0)) : (float4)0.0f;
+		float4 RT = SW_BLEND_NEEDS_RT ? RtLoad(int2(pos_xy)) : (float4)0.0f;
 
 		if (PS_SHUFFLE && SW_BLEND_NEEDS_RT)
 		{
@@ -1236,24 +1307,63 @@ void ps_blend(inout float4 Color, inout float4 As_rgba, float2 pos_xy)
 	}
 }
 
+#if PS_ROV_EARLYDEPTHSTENCIL
+[earlydepthstencil]
+#endif
+
+#if PS_ROV_COLOR || PS_ROV_DEPTH
+	#define DISCARD { rov_discard_color = true; rov_discard_depth = true; }
+	#define DISCARD_COLOR rov_discard_color = true
+	#define DISCARD_DEPTH rov_discard_depth = true
+#else
+	#define DISCARD discard
+	#define DISCARD_COLOR o_col0 = RtLoad(input.p.xy)
+	#define DISCARD_DEPTH input.p.z = DepthLoad(input.p.xy)
+#endif
+
+#if (PS_RETURN_COLOR || PS_RETURN_DEPTH)
 PS_OUTPUT ps_main(PS_INPUT input)
+#else
+void ps_main(PS_INPUT input)
+#endif
 {
 	// Must floor before depth testing.
 #if PS_ZFLOOR
 	input.p.z = floor(input.p.z * exp2(32.0f)) * exp2(-32.0f);
 #endif
 
-#if PS_ZTST == ZTST_GEQUAL
-	if (input.p.z < DepthTexture.Load(int3(input.p.xy, 0)).r)
-		discard;
-#elif PS_ZTST == ZTST_GREATER
-	if (input.p.z <= DepthTexture.Load(int3(input.p.xy, 0)).r)
-		discard;
+#if PS_ROV_COLOR
+	rov_rt_value = RtTextureRov[input.p.xy];
 #endif
+
+#if PS_ROV_DEPTH
+	rov_depth_value = DepthTextureRov[input.p.xy];
+#endif
+
+#if PS_ROV_COLOR || PS_ROV_DEPTH
+	bool rov_discard_color = false;
+	bool rov_discard_depth = false;
+#endif
+
+	// Use ROV discard macro for since we cannot do
+	// conditional discard based on value read from ROV.
+#if PS_ZTST == ZTST_GEQUAL
+	if (input.p.z < DepthLoad(input.p.xy))
+		DISCARD;
+#elif PS_ZTST == ZTST_GREATER
+	if (input.p.z <= DepthLoad(input.p.xy))
+		DISCARD;
+#endif
+
 	float4 C = ps_color(input);
 
 #if PS_AA1
-	float cov = clamp(1.0f - abs(input.inv_cov), 0.0f, 1.0f);
+	#if PS_AA1 == PS_AA1_LINE
+		// Blur only outer part of the line by scaling coverage.
+		float cov = clamp(LineCovScale * (1.0f - abs(input.inv_cov)), 0.0f, 1.0f);
+	#else
+		float cov = clamp(1.0f - abs(input.inv_cov), 0.0f, 1.0f);
+	#endif
 	#if PS_ABE
 		if (floor(C.a) == 128.0f) // The coverage is only used if the fragment alpha is 128.
 			C.a = 128.0f * cov;
@@ -1267,13 +1377,11 @@ PS_OUTPUT ps_main(PS_INPUT input)
 
 	bool atst_pass = atst(C);
 
-#if PS_AFAIL == AFAIL_KEEP
+#if PS_ATST != PS_ATST_NONE && PS_AFAIL == AFAIL_KEEP
 	if (!atst_pass)
 		discard;
 #endif
 
-	PS_OUTPUT output;
-
 	if (PS_SCANMSK & 2)
 	{
 		// fail depth test on prohibited lines
@@ -1284,7 +1392,7 @@ PS_OUTPUT ps_main(PS_INPUT input)
 	float4 alpha_blend = (float4)0.0f;
 	if (SW_AD_TO_HW)
 	{
-		float4 RT = PS_RTA_CORRECTION ? trunc(RtTexture.Load(int3(input.p.xy, 0)) * 128.0f + 0.1f) : trunc(RtTexture.Load(int3(input.p.xy, 0)) * 255.0f + 0.1f);
+		float4 RT = PS_RTA_CORRECTION ? trunc(RtLoad(input.p.xy) * 128.0f + 0.1f) : trunc(RtLoad(input.p.xy) * 255.0f + 0.1f);
 		alpha_blend = (float4)(RT.a / 128.0f);
 	}
 	else
@@ -1308,9 +1416,9 @@ PS_OUTPUT ps_main(PS_INPUT input)
 
 #if PS_WRITE_RG == 1
 	// Pseudo 16 bits access.
-	float rt_a = RtTexture.Load(int3(input.p.xy, 0)).g;
+	float rt_a = RtLoad(input.p.xy).g;
 #else
-	float rt_a = RtTexture.Load(int3(input.p.xy, 0)).a;
+	float rt_a = RtLoad(input.p.xy).a;
 #endif
 
 #if (PS_DATE & 3) == 1
@@ -1329,9 +1437,8 @@ PS_OUTPUT ps_main(PS_INPUT input)
 	#endif
 #endif
 
-	if (bad)
-		discard;
-
+if (bad)
+	discard;
 #endif
 
 #if PS_DATE == 3
@@ -1342,17 +1449,29 @@ PS_OUTPUT ps_main(PS_INPUT input)
 		discard;
 #endif
 
+	// Output values
+#if !PS_NO_COLOR
+	#if PS_DATE == 1 || PS_DATE == 2
+		float o_col0;
+	#else
+		float4 o_col0;
+		#if !PS_NO_COLOR1
+			float4 o_col1;
+		#endif
+	#endif
+#endif
+
 	// Get first primitive that will write a failling alpha value
 #if PS_DATE == 1
 	// DATM == 0
 	// Pixel with alpha equal to 1 will failed (128-255)
-	output.c = (C.a > 127.5f) ? float(input.primid) : float(0x7FFFFFFF);
+	o_col0 = (C.a > 127.5f) ? float(input.primid) : float(0x7FFFFFFF);
 
 #elif PS_DATE == 2
 
 	// DATM == 1
 	// Pixel with alpha equal to 0 will failed (0-127)
-	output.c = (C.a < 127.5f) ? float(input.primid) : float(0x7FFFFFFF);
+	o_col0 = (C.a < 127.5f) ? float(input.primid) : float(0x7FFFFFFF);
 
 #else
 	// Not primid DATE setup
@@ -1427,32 +1546,32 @@ PS_OUTPUT ps_main(PS_INPUT input)
 	alpha_blend.a = float(atst_pass);
 #endif
 
+	// Output color scaling
 #if !PS_NO_COLOR
-	output.c0.a = PS_RTA_CORRECTION ? C.a / 128.0f : C.a / 255.0f;
-	output.c0.rgb = PS_COLCLIP_HW ? float3(C.rgb / 65535.0f) : C.rgb / 255.0f;
+	o_col0.a = PS_RTA_CORRECTION ? C.a / 128.0f : C.a / 255.0f;
+	o_col0.rgb = PS_COLCLIP_HW ? float3(C.rgb / 65535.0f) : C.rgb / 255.0f;
 #if !PS_NO_COLOR1
-	output.c1 = alpha_blend;
+	o_col1 = alpha_blend;
 #endif
+#endif // !PS_NO_COLOR
 
 	// Alpha test with feedback
 #if PS_AFAIL == AFAIL_FB_ONLY
 	if (!atst_pass)
-		input.p.z = DepthTexture.Load(int3(input.p.xy, 0)).r;
+		DISCARD_DEPTH;
 #elif PS_AFAIL == AFAIL_ZB_ONLY
 	if (!atst_pass)
-		output.c0 = RtTexture.Load(int3(input.p.xy, 0));
+		DISCARD_COLOR;
 #elif PS_AFAIL == AFAIL_RGB_ONLY || PS_AFAIL == AFAIL_RGB_ONLY_SW_Z
 	if (!atst_pass)
 	{
-	output.c0.a = RtTexture.Load(int3(input.p.xy, 0)).a;
+		o_col0.a = RtLoad(input.p.xy).a; // discard alpha
 	#if PS_AFAIL == AFAIL_RGB_ONLY_SW_Z
-		input.p.z = DepthTexture.Load(int3(input.p.xy, 0)).r; 
+		DISCARD_DEPTH;
 	#endif
 	}
 #endif
 
-#endif // !PS_NO_COLOR
-
 #endif // PS_DATE != 1/2
 
 #if PS_ZCLAMP
@@ -1461,18 +1580,40 @@ PS_OUTPUT ps_main(PS_INPUT input)
 
 #if PS_AA1 == PS_AA1_TRIANGLE_SW_Z
 	if (!bool(input.interior))
-		input.p.z = DepthTexture.Load(int3(input.p.xy, 0)).r; // No depth update for triangle edges.
+		DISCARD_DEPTH; // No depth update for triangle edges.
 #endif
 
-#if ZWRITE
-#if SW_DEPTH && PS_NO_COLOR1 && DX12
-	// Output color clone for feedback as well as real depth.
-	output.depth_color = input.p.z;
+#if (PS_RETURN_COLOR || PS_RETURN_DEPTH)
+	PS_OUTPUT output;
 #endif
+
+	// Color write back
+#if PS_RETURN_COLOR
+	output.c0 = o_col0;
+	#if !PS_NO_COLOR1
+		output.c1 = o_col1;
+	#endif
+#elif PS_RETURN_COLOR_ROV
+	o_col0 = (FbMask == 0xFFu) ? RtLoad(input.p.xy) : o_col0; // channel masking
+	if (!rov_discard_color)
+		RtWrite(input.p.xy, o_col0);
+#endif
+
+	// Depth write back
+#if PS_RETURN_DEPTH
 	output.depth = input.p.z;
+	#if SW_DEPTH && PS_NO_COLOR1 && PS_DEPTH_FEEDBACK_SUPPORT == 2
+		// Output color clone for feedback.
+		output.depth_color = input.p.z;
+	#endif
+#elif PS_RETURN_DEPTH_ROV
+	if (!rov_discard_depth)
+		DepthWrite(input.p.xy, input.p.z);
 #endif
 
+#if (PS_RETURN_COLOR || PS_RETURN_DEPTH)
 	return output;
+#endif
 }
 
 #endif // PIXEL_SHADER
@@ -1495,7 +1636,7 @@ cbuffer cb0
 	float2 TextureOffset;
 	float2 PointSize;
 	uint MaxDepth;
-	uint _cb0_pad0;
+	float LineAA1Width;
 };
 
 #ifdef DX12
@@ -1738,11 +1879,9 @@ VS_OUTPUT vs_main_expand(uint vid : SV_VertexID)
 	// Use bottom minus top for delta regardless of which vertex we are expanding.
 	float2 line_delta = is_bottom ? (vtx.p.xy - other.p.xy) : (other.p.xy - vtx.p.xy);
 	float2 line_vector = normalize(line_delta / VertexScale);
-#if VS_EXPAND == VS_EXPAND_LINE
 	float2 line_expand = float2(line_vector.y, -line_vector.x);
-#elif VS_EXPAND == VS_EXPAND_LINE_AA1
-	// Expand in y direction for shallow lines and x direction for steep lines.
-	float2 line_expand = abs(line_vector.x) >= abs(line_vector.y) ? float2(0.0f, 2.0f) : float2(2.0f, 0.0f);
+#if VS_EXPAND == VS_EXPAND_LINE_AA1
+	line_expand *= 2.0f * LineAA1Width;
 #endif
 	float2 line_width = (line_expand * PointSize) / 2;
 	float2 offset = is_right ? line_width : -line_width;
@@ -1875,11 +2014,6 @@ VS_OUTPUT vs_main_expand(uint vid : SV_VertexID)
 		vtx.inv_cov = is_near_corner ? 0.0f : 1.0f; // Full coverage at near corner, otherwise none.
 	
 		vtx.interior = 0;
-
-		#if !VS_IIP
-			// Get the provoking vertex color (first vertex in DX)
-			vtx.c = i0 == 0 ? vtx.c : (i1 == 0 ? other.c : opposite.c);
-		#endif
 	}
 
 	return vtx;

bin/resources/shaders/opengl/convert.glsl

@@ -3,21 +3,12 @@
 
 //#version 420 // Keep it for editor detection
 
-
 #ifdef VERTEX_SHADER
 
 layout(location = 0) in vec2 POSITION;
 layout(location = 1) in vec2 TEXCOORD0;
 layout(location = 7) in vec4 COLOR;
 
-// FIXME set the interpolation (don't know what dx do)
-// flat means that there is no interpolation. The value given to the fragment shader is based on the provoking vertex conventions.
-//
-// noperspective means that there will be linear interpolation in window-space. This is usually not what you want, but it can have its uses.
-//
-// smooth, the default, means to do perspective-correct interpolation.
-//
-// The centroid qualifier only matters when multisampling. If this qualifier is not present, then the value is interpolated to the pixel's center, anywhere in the pixel, or to one of the pixel's samples. This sample may lie outside of the actual primitive being rendered, since a primitive can cover only part of a pixel's area. The centroid qualifier is used to prevent this; the interpolation point must fall within both the pixel's area and the primitive's area.
 out vec4 PSin_p;
 out vec2 PSin_t;
 out vec4 PSin_c;
@@ -40,44 +31,125 @@ in vec4 PSin_c;
 
 layout(binding = 0) uniform sampler2D TextureSampler;
 
-// Give a different name so I remember there is a special case!
-#if defined(ps_convert_rgba8_16bits) || defined(ps_convert_float32_32bits)
-layout(location = 0) out uint SV_Target1;
-#elif defined(ps_convert_float32_depth_to_color)
-layout(location = 0) out float SV_Target0;
-#elif !defined(ps_datm1) && \
-	!defined(ps_datm0) && \
-	!defined(ps_datm1_rta_correction) && \
-	!defined(ps_datm0_rta_correction) && \
-	!defined(ps_convert_rgba8_float32) && \
-	!defined(ps_convert_rgba8_float24) && \
-	!defined(ps_convert_rgba8_float16) && \
-	!defined(ps_convert_rgb5a1_float16) && \
-	!defined(ps_convert_rgba8_float32_biln) && \
-	!defined(ps_convert_rgba8_float24_biln) && \
-	!defined(ps_convert_rgba8_float16_biln) && \
-	!defined(ps_convert_rgb5a1_float16_biln) && \
-	!defined(ps_convert_float32_float24) && \
-	!defined(ps_depth_copy)
-layout(location = 0) out vec4 SV_Target0;
+#if HAS_INTEGER_OUTPUT
+	layout(location = 0) out uint o_col0;
+	#define OUTPUT o_col0
+#elif HAS_DEPTH_OUTPUT
+	out float gl_FragDepth;
+	#define OUTPUT gl_FragDepth
+#elif HAS_FLOAT32_OUTPUT
+	layout(location = 0) out float o_col0;
+	#define OUTPUT o_col0
+#elif HAS_STENCIL_OUTPUT
+#else
+	layout(location = 0) out vec4 o_col0;
+	#define OUTPUT o_col0
 #endif
 
+#if HAS_FLOAT32_INPUT
+
+float sample_c()
+{
+	return texture(TextureSampler, PSin_t).r;
+}
+
+#else
+
 vec4 sample_c()
 {
 	return texture(TextureSampler, PSin_t);
 }
 
+#endif
+
+uint rgba8_to_uint(vec4 c)
+{
+	uvec4 i = uvec4(c * 255.5f) & 0xFFu;
+	return i.r | (i.g << 8) | (i.b << 16) | (i.a << 24);
+}
+
+uint rgb5a1_to_uint(vec4 c)
+{
+	uvec4 i = uvec4(c * 255.5f) & uvec4(0xF8u, 0xF8u, 0xF8u, 0x80u);
+	return (i.r >> 3) | (i.g << 2) | (i.b << 7) | (i.a << 8);
+}
+
+uint depth_to_uint(float d)
+{
+	return uint(d * exp2(32.0f));
+}
+
+vec4 uint_to_rgba8(uint i)
+{
+	return vec4((i & 0xFFu), ((i >> 8) & 0xFFu), ((i >> 16) & 0xFFu), ((i >> 24) & 0xFFu)) / 255.0f;
+}
+
+vec4 uint_to_rgb5a1(uint i)
+{
+	return vec4(uvec4(i << 3, i >> 2, i >> 7, i >> 8) & uvec4(0xF8u, 0xF8u, 0xF8u, 0x80u)) / 255.0f;
+}
+
+float uint_to_depth32(uint i)
+{
+	return float(i) * exp2(-32.0f);
+}
+
+float uint_to_depth24(uint i)
+{
+	return float(i & 0xFFFFFFu) * exp2(-32.0f);
+}
+
+float uint_to_depth16(uint i)
+{
+	return float(i & 0xFFFFu) * exp2(-32.0f);
+}
+
+float rgba8_to_depth32(vec4 val)
+{
+	return uint_to_depth32(rgba8_to_uint(val));
+}
+
+float rgba8_to_depth24(vec4 val)
+{
+	return uint_to_depth24(rgba8_to_uint(val));
+}
+
+float rgba8_to_depth16(vec4 val)
+{
+	return uint_to_depth16(rgba8_to_uint(val));
+}
+
+float rgb5a1_to_depth16(vec4 val)
+{
+	return uint_to_depth16(rgb5a1_to_uint(val));
+}
+
+vec4 depth32_to_rgba8(float d)
+{
+	return uint_to_rgba8(depth_to_uint(d));
+}
+
+vec4 depth16_to_rgb5a1(float d)
+{
+	return uint_to_rgb5a1(depth_to_uint(d));
+}
+
+float depth32_to_depth24(float d)
+{
+	return uint_to_depth24(depth_to_uint(d));
+}
+
 #ifdef ps_copy
 void ps_copy()
 {
-	SV_Target0 = sample_c();
+	OUTPUT = sample_c();
 }
 #endif
 
 #ifdef ps_depth_copy
 void ps_depth_copy()
 {
-  gl_FragDepth = sample_c().r;
+	OUTPUT = sample_c();
 }
 #endif
 
@@ -96,126 +168,47 @@ void ps_downsample_copy()
 		for (int xoff = 0; xoff < DownsampleFactor; xoff++)
 			result += texelFetch(TextureSampler, coord + ivec2(xoff * StepMultiplier, yoff * StepMultiplier), 0);
 	}
-	SV_Target0 = result / Weight;
+	o_col0 = result / Weight;
 }
 #endif
 
-#ifdef ps_convert_rgba8_16bits
-// Need to be careful with precision here, it can break games like Spider-Man 3 and Dogs Life
-void ps_convert_rgba8_16bits()
+#ifdef ps_convert_rgb5a1_16bits
+void ps_convert_rgb5a1_16bits()
 {
-	highp uvec4 i = uvec4(sample_c() * vec4(255.5f, 255.5f, 255.5f, 255.5f));
-
-	SV_Target1 = ((i.x & 0x00F8u) >> 3) | ((i.y & 0x00F8u) << 2) | ((i.z & 0x00f8u) << 7) | ((i.w & 0x80u) << 8);
+	// Need to be careful with precision here, it can break games like Spider-Man 3 and Dogs Life
+	OUTPUT = rgb5a1_to_uint(sample_c());
 }
 #endif
 
-#ifdef ps_convert_float32_32bits
-void ps_convert_float32_32bits()
+#ifdef ps_convert_depth32_32bits
+void ps_convert_depth32_32bits()
 {
 	// Convert a GL_FLOAT32 depth texture into a 32 bits UINT texture
-	SV_Target1 = uint(exp2(32.0f) * sample_c().r);
+	OUTPUT = depth_to_uint(sample_c());
 }
 #endif
 
-#ifdef ps_convert_float32_rgba8
-void ps_convert_float32_rgba8()
+#ifdef ps_convert_depth32_rgba8
+void ps_convert_depth32_rgba8()
 {
 	// Convert a GL_FLOAT32 depth texture into a RGBA color texture
-	uint d = uint(sample_c().r * exp2(32.0f));
-	SV_Target0 = vec4(uvec4((d & 0xFFu), ((d >> 8) & 0xFFu), ((d >> 16) & 0xFFu), (d >> 24))) / vec4(255.0);
+	OUTPUT = depth32_to_rgba8(sample_c());
 }
 #endif
 
-#ifdef ps_convert_float16_rgb5a1
-void ps_convert_float16_rgb5a1()
+#ifdef ps_convert_depth16_rgb5a1
+void ps_convert_depth16_rgb5a1()
 {
 	// Convert a GL_FLOAT32 (only 16 lsb) depth into a RGB5A1 color texture
-	uint d = uint(sample_c().r * exp2(32.0f));
-	SV_Target0 = vec4(uvec4(d << 3, d >> 2, d >> 7, d >> 8) & uvec4(0xf8, 0xf8, 0xf8, 0x80)) / 255.0f;
+	OUTPUT = depth16_to_rgb5a1(sample_c());
 }
 #endif
 
-float rgba8_to_depth32(vec4 unorm)
-{
-	uvec4 c = uvec4(unorm * vec4(255.5f));
-	return float(c.r | (c.g << 8) | (c.b << 16) | (c.a << 24)) * exp2(-32.0f);
-}
-
-float rgba8_to_depth24(vec4 unorm)
-{
-	uvec3 c = uvec3(unorm.rgb * vec3(255.5f));
-	return float(c.r | (c.g << 8) | (c.b << 16)) * exp2(-32.0f);
-}
-
-float rgba8_to_depth16(vec4 unorm)
-{
-	uvec2 c = uvec2(unorm.rg * vec2(255.5f));
-	return float(c.r | (c.g << 8)) * exp2(-32.0f);
-}
-
-float rgb5a1_to_depth16(vec4 unorm)
-{
-	uvec4 c = uvec4(unorm * vec4(255.5f));
-	return float(((c.r & 0xF8u) >> 3) | ((c.g & 0xF8u) << 2) | ((c.b & 0xF8u) << 7) | ((c.a & 0x80u) << 8)) * exp2(-32.0f);
-}
-
-#ifdef ps_convert_float32_depth_to_color
-void ps_convert_float32_depth_to_color()
-{
-	SV_Target0 = sample_c().r;
-}
-#endif
-
-#ifdef ps_convert_float32_color_to_depth
-void ps_convert_float32_color_to_depth()
-{
-	gl_FragDepth = sample_c().r;
-}
-#endif
-
-#ifdef ps_convert_float32_float24
-void ps_convert_float32_float24()
+#ifdef ps_convert_depth32_depth24
+void ps_convert_depth32_depth24()
 {
 	// Truncates depth value to 24bits
-	uint d = uint(sample_c().r * exp2(32.0f)) & 0xFFFFFFu;
-	gl_FragDepth = float(d) * exp2(-32.0f);
-}
-#endif
-
-#ifdef ps_convert_rgba8_float32
-void ps_convert_rgba8_float32()
-{
-	// Convert an RGBA texture into a float depth texture
-	gl_FragDepth = rgba8_to_depth32(sample_c());
-}
-#endif
-
-#ifdef ps_convert_rgba8_float24
-void ps_convert_rgba8_float24()
-{
-	// Same as above but without the alpha channel (24 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
-	gl_FragDepth = rgba8_to_depth24(sample_c());
-}
-#endif
-
-#ifdef ps_convert_rgba8_float16
-void ps_convert_rgba8_float16()
-{
-	// Same as above but without the A/B channels (16 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
-	gl_FragDepth = rgba8_to_depth16(sample_c());
-}
-#endif
-
-#ifdef ps_convert_rgb5a1_float16
-void ps_convert_rgb5a1_float16()
-{
-	// Convert an RGB5A1 (saved as RGBA8) color to a 16 bit Z
-	gl_FragDepth = rgb5a1_to_depth16(sample_c());
+	OUTPUT = depth32_to_depth24(sample_c());
 }
 #endif
 
@@ -229,41 +222,55 @@ void ps_convert_rgb5a1_float16()
 	float depthTR = CONVERT_FN(texelFetch(TextureSampler, coords.zy, 0)); \
 	float depthBL = CONVERT_FN(texelFetch(TextureSampler, coords.xw, 0)); \
 	float depthBR = CONVERT_FN(texelFetch(TextureSampler, coords.zw, 0)); \
-	gl_FragDepth = mix(mix(depthTL, depthTR, mix_vals.x), mix(depthBL, depthBR, mix_vals.x), mix_vals.y);
+	OUTPUT = mix(mix(depthTL, depthTR, mix_vals.x), mix(depthBL, depthBR, mix_vals.x), mix_vals.y);
 
-#ifdef ps_convert_rgba8_float32_biln
-void ps_convert_rgba8_float32_biln()
+#ifdef ps_convert_rgba8_depth32
+void ps_convert_rgba8_depth32()
 {
 	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth32);
+#else
+	OUTPUT = rgba8_to_depth32(sample_c());
+#endif
 }
 #endif
 
-#ifdef ps_convert_rgba8_float24_biln
-void ps_convert_rgba8_float24_biln()
+#ifdef ps_convert_rgba8_depth24
+void ps_convert_rgba8_depth24()
 {
 	// Same as above but without the alpha channel (24 bits Z)
-
 	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth24);
+#else
+	OUTPUT = rgba8_to_depth24(sample_c());
+#endif
 }
 #endif
 
-#ifdef ps_convert_rgba8_float16_biln
-void ps_convert_rgba8_float16_biln()
+#ifdef ps_convert_rgba8_depth16
+void ps_convert_rgba8_depth16()
 {
 	// Same as above but without the A/B channels (16 bits Z)
-
 	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth16);
+#else
+	OUTPUT = rgba8_to_depth16(sample_c());
+#endif
 }
 #endif
 
-#ifdef ps_convert_rgb5a1_float16_biln
-void ps_convert_rgb5a1_float16_biln()
+#ifdef ps_convert_rgb5a1_depth16
+void ps_convert_rgb5a1_depth16()
 {
 	// Convert an RGB5A1 (saved as RGBA8) color to a 16 bit Z
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgb5a1_to_depth16);
+#else
+	OUTPUT = rgb5a1_to_depth16(sample_c());
+#endif
 }
 #endif
 
@@ -373,7 +380,7 @@ void ps_convert_rgb5a1_8i()
 		uint red = (denorm_c.r >> 3) & 0x1Fu;
 		uint green = (denorm_c.g >> 3) & 0x1Fu;
 
-		SV_Target0 = vec4(float(((green << 5) | red) & 0xFFu) / 255.0f);
+		o_col0 = vec4(float(((green << 5) | red) & 0xFFu) / 255.0f);
 	}
 	else
 	{
@@ -381,7 +388,7 @@ void ps_convert_rgb5a1_8i()
 		uint blue = (denorm_c.b >> 3) & 0x1Fu;
 		uint alpha = denorm_c.a & 0x80u;
 
-		SV_Target0 = vec4(float((alpha | (blue << 2) | (green >> 3)) & 0xFFu) / 255.0f);
+		o_col0 = vec4(float((alpha | (blue << 2) | (green >> 3)) & 0xFFu) / 255.0f);
 	}
 }
 #endif
@@ -430,7 +437,7 @@ void ps_convert_rgba_8i()
 	vec4 pixel = texelFetch(TextureSampler, ivec2(coord), 0);
 	vec2  sel0 = (pos.y & 2u) == 0u ? pixel.rb : pixel.ga;
 	float sel1 = (pos.x & 8u) == 0u ? sel0.x : sel0.y;
-	SV_Target0 = vec4(sel1);
+	o_col0 = vec4(sel1);
 }
 #endif
 
@@ -438,7 +445,7 @@ void ps_convert_rgba_8i()
 void ps_filter_transparency()
 {
 	vec4 c = sample_c();
-	SV_Target0 = vec4(c.rgb, 1.0);
+	o_col0 = vec4(c.rgb, 1.0);
 }
 #endif
 
@@ -486,7 +493,7 @@ void ps_datm0_rta_correction()
 void ps_rta_correction()
 {
 	vec4 value = sample_c();
-	SV_Target0 = vec4(value.rgb, value.a / (128.25f / 255.0f));
+	o_col0 = vec4(value.rgb, value.a / (128.25f / 255.0f));
 }
 #endif
 
@@ -494,7 +501,7 @@ void ps_rta_correction()
 void ps_rta_decorrection()
 {
 	vec4 value = sample_c();
-	SV_Target0 = vec4(value.rgb, value.a * (128.25f / 255.0f));
+	o_col0 = vec4(value.rgb, value.a * (128.25f / 255.0f));
 }
 #endif
 
@@ -502,7 +509,7 @@ void ps_rta_decorrection()
 void ps_colclip_init()
 {
 	vec4 value = sample_c();
-	SV_Target0 = vec4(round(value.rgb * 255.0f) / 65535.0f, value.a);
+	o_col0 = vec4(round(value.rgb * 255.0f) / 65535.0f, value.a);
 }
 #endif
 
@@ -510,7 +517,7 @@ void ps_colclip_init()
 void ps_colclip_resolve()
 {
 	vec4 value = sample_c();
-	SV_Target0 = vec4(vec3(uvec3(value.rgb * 65535.0f) & 255u) / 255.0f, value.a);
+	o_col0 = vec4(vec3(uvec3(value.rgb * 65535.0f) & 255u) / 255.0f, value.a);
 }
 #endif
 
@@ -525,7 +532,7 @@ void ps_convert_clut_4()
 	uvec2 pos = uvec2(index % 8u, index / 8u);
 
 	ivec2 final = ivec2(floor(vec2(offset.xy + pos) * vec2(scale)));
-	SV_Target0 = texelFetch(TextureSampler, final, 0);
+	o_col0 = texelFetch(TextureSampler, final, 0);
 }
 #endif
 
@@ -545,7 +552,7 @@ void ps_convert_clut_8()
 	pos.y = ((index / 32u) * 2u) + (subgroup % 2u);
 
 	ivec2 final = ivec2(floor(vec2(offset.xy + pos) * vec2(scale)));
-	SV_Target0 = texelFetch(TextureSampler, final, 0);
+	o_col0 = texelFetch(TextureSampler, final, 0);
 }
 #endif
 
@@ -600,31 +607,31 @@ void ps_yuv()
 			break;
 	}
 
-	SV_Target0 = o;
+	o_col0 = o;
 }
 #endif
 
-#if defined(ps_stencil_image_init_0) || defined(ps_stencil_image_init_1) || defined(ps_stencil_image_init_2) || defined(ps_stencil_image_init_3)
+#if defined(ps_primid_image_init_0) || defined(ps_primid_image_init_1) || defined(ps_primid_image_init_2) || defined(ps_primid_image_init_3)
 
 void main()
 {
-	SV_Target0 = vec4(0x7FFFFFFF);
+	o_col0 = vec4(0x7FFFFFFF);
 
-	#ifdef ps_stencil_image_init_0
+	#ifdef ps_primid_image_init_0
 		if((127.5f / 255.0f) < sample_c().a) // < 0x80 pass (== 0x80 should not pass)
-			SV_Target0 = vec4(-1);
+			o_col0 = vec4(-1);
 	#endif
-	#ifdef ps_stencil_image_init_1
+	#ifdef ps_primid_image_init_1
 		if(sample_c().a < (127.5f / 255.0f)) // >= 0x80 pass
-			SV_Target0 = vec4(-1);
+			o_col0 = vec4(-1);
 	#endif
-	#ifdef ps_stencil_image_init_2
+	#ifdef ps_primid_image_init_2
 		if((254.5f / 255.0f) < sample_c().a) // < 0x80 pass (== 0x80 should not pass)
-			SV_Target0 = vec4(-1);
+			o_col0 = vec4(-1);
 	#endif
-	#ifdef ps_stencil_image_init_3
+	#ifdef ps_primid_image_init_3
 		if(sample_c().a < (254.5f / 255.0f)) // >= 0x80 pass
-			SV_Target0 = vec4(-1);
+			o_col0 = vec4(-1);
 	#endif
 }
 #endif

bin/resources/shaders/opengl/tfx_fs.glsl

@@ -59,7 +59,7 @@
 #define NEEDS_DEPTH_FOR_ZTST (PS_ZTST == ZTST_GEQUAL || PS_ZTST == ZTST_GREATER)
 #define NEEDS_DEPTH_FOR_AA1 (PS_AA1 == PS_AA1_TRIANGLE_SW_Z)
 
-#define NEEDS_RT (NEEDS_RT_EARLY || NEEDS_RT_FOR_AFAIL || (!PS_PRIMID_INIT && (PS_FBMASK || SW_BLEND_NEEDS_RT || SW_AD_TO_HW)) || PS_COLOR_FEEDBACK)
+#define NEEDS_RT (NEEDS_RT_EARLY || NEEDS_RT_FOR_AFAIL || (!PS_PRIMID_INIT && (PS_FBMASK || SW_BLEND_NEEDS_RT || SW_AD_TO_HW)))
 #define NEEDS_TEX (PS_TFX != 4)
 #define SW_DEPTH (NEEDS_DEPTH_FOR_AFAIL || NEEDS_DEPTH_FOR_ZTST || NEEDS_DEPTH_FOR_AA1)
 #define ZWRITE (SW_DEPTH || PS_ZCLAMP || PS_ZFLOOR)
@@ -93,6 +93,13 @@ layout(std140, binding = 0) uniform cb21
 
 	float ScaledScaleFactor;
 	float RcpScaleFactor;
+	float _pad0_cb1;
+	float _pad1_cb1;
+
+	float LineCovScale;
+	float _pad2_cb1;
+	float _pad3_cb1;
+	float _pad4_cb1;
 };
 
 in SHADER
@@ -121,7 +128,7 @@ in SHADER
 	#if defined(GL_EXT_shader_framebuffer_fetch)
 		#undef TARGET_0_QUALIFIER
 		#define TARGET_0_QUALIFIER inout
-		#define LAST_FRAG_COLOR SV_Target0
+		#define LAST_FRAG_COLOR o_col0
 	#elif defined(GL_ARM_shader_framebuffer_fetch)
 		#define LAST_FRAG_COLOR gl_LastFragColorARM
 	#endif
@@ -129,20 +136,20 @@ in SHADER
 
 #if !PS_NO_COLOR && !PS_NO_COLOR1
 	// Same buffer but 2 colors for dual source blending
-	layout(location = 0, index = 0) TARGET_0_QUALIFIER vec4 SV_Target0;
-	layout(location = 0, index = 1) out vec4 SV_Target1;
+	layout(location = 0, index = 0) TARGET_0_QUALIFIER vec4 o_col0;
+	layout(location = 0, index = 1) out vec4 o_col1;
 #elif !PS_NO_COLOR
-	layout(location = 0) TARGET_0_QUALIFIER vec4 SV_Target0;
+	layout(location = 0) TARGET_0_QUALIFIER vec4 o_col0;
 #endif
 
 // Depth feedback mode 2 is for depth as color.
 // Use FB fetch for the feedback if it's available.
 #if SW_DEPTH && PS_NO_COLOR1 && (DEPTH_FEEDBACK_SUPPORT == 2)
-#if HAS_FRAMEBUFFER_FETCH
-	layout(location = 1) inout float SV_Target1;
-#else
-	layout(location = 1) out float SV_Target1;
-#endif
+	#if HAS_FRAMEBUFFER_FETCH
+		layout(location = 1) inout float o_col1;
+	#else
+		layout(location = 1) out float o_col1;
+	#endif
 #endif
 
 #if NEEDS_TEX
@@ -180,14 +187,14 @@ vec4 sample_from_rt()
 #endif
 }
 
-vec4 sample_from_depth()
+float sample_from_depth()
 {
 #if !SW_DEPTH
-	return vec4(0.0);
+	return 0.0f;
 #elif HAS_FRAMEBUFFER_FETCH && (DEPTH_FEEDBACK_SUPPORT == 2)
-	return SV_Target1;
+	return o_col1;
 #else
-	return texelFetch(DepthSampler, ivec2(gl_FragCoord.xy), 0);
+	return texelFetch(DepthSampler, ivec2(gl_FragCoord.xy), 0).r;
 #endif
 }
 
@@ -213,22 +220,27 @@ float manual_lod(float uv_w)
 vec4 sample_c_af(vec2 uv, float uv_w)
 {
 	// HW sampler will reject bad UVs, match that here.
-	uv = (any(isnan(uv)) || any(isinf(uv))) ? vec2(0, 0) : uv;
+	uv = (any(isnan(uv)) || any(isinf(uv))) ? vec2(0.0f, 0.0f) : uv;
 
 	// Large floating point values risk NaN/Inf values.
 	// Above this value floats lose decimal precision, so seems a resonable limit for UVs.
 	uv = clamp(uv, -8388608.0f, 8388608.0f);
 
 	// Below taken from https://microsoft.github.io/DirectX-Specs/d3d/archive/D3D11_3_FunctionalSpec.htm#7.18.11%20LOD%20Calculations
+	// And https://registry.khronos.org/OpenGL/extensions/EXT/EXT_texture_filter_anisotropic.txt
 	// With guidance from https://pema.dev/2025/05/09/mipmaps-too-much-detail/ 
 	vec2 sz = textureSize(TextureSampler, 0);
 	vec2 dX = dFdx(uv) * sz;
 	vec2 dY = dFdy(uv) * sz;
 
+	float length_x = length(dX);
+	float length_y = length(dY);
+
 	// Calculate Ellipse Transform
-	bool d_zero = length(dX) == 0 || length(dY) == 0;
-	bool d_par = (dX.x * dY.y - dY.x * dX.y) == 0;
-	bool d_per = dot(dX, dY) == 0;
+	bool d_zero = length_x < 0.001f || length_y < 0.001f;
+	float f = (dX.x * dY.y - dX.y * dY.x);
+	bool d_par = f < 0.001f;
+	bool d_per = dot(dX, dY) < 0.001f;
 	bool d_inf_nan = any(isinf(dX)) || any(isinf(dY)) || any(isnan(dX)) || any(isnan(dY));
 
 	if (!(d_zero || d_par || d_per || d_inf_nan))
@@ -236,38 +248,46 @@ vec4 sample_c_af(vec2 uv, float uv_w)
 		float A = dX.y * dX.y + dY.y * dY.y;
 		float B = -2 * (dX.x * dX.y + dY.x * dY.y);
 		float C = dX.x * dX.x + dY.x * dY.x;
-		float f = (dX.x * dY.y - dY.x * dX.y);
 		float F = f * f;
 
 		float p = A - C;
 		float q = A + C;
 		float t = sqrt(p * p + B * B);
 
+		float signB = sign(B);
+		float denom_plus  = t * (q + t);
+		float denom_minus = t * (q - t);
+
+		float sqrtA = sqrt(F * (t + p));
+		float sqrtB = sqrt(F * (t - p));
+
+		float inv_sqrt_denom_plus  = inversesqrt(denom_plus);
+		float inv_sqrt_denom_minus = inversesqrt(denom_minus);
+
 		vec2 new_dX = vec2(
-			sqrt(F * (t + p) / (t * (q + t))),
-			sqrt(F * (t - p) / (t * (q + t))) * sign(B)
+			sqrtA * inv_sqrt_denom_plus,
+			sqrtB * inv_sqrt_denom_plus * signB
 		);
-		
+
 		vec2 new_dY = vec2(
-			sqrt(F * (t - p) / (t * (q - t))) * -sign(B),
-			sqrt(F * (t + p) / (t * (q - t)))
+			sqrtB * inv_sqrt_denom_minus * -signB,
+			sqrtA * inv_sqrt_denom_minus
 		);
-		
+
 		d_inf_nan = any(isinf(new_dX)) || any(isinf(new_dY)) || any(isnan(new_dX)) || any(isnan(new_dY));
 		if (!d_inf_nan)
 		{
 			dX = new_dX;
 			dY = new_dY;
+			length_x = length(dX);
+			length_y = length(dY);
 		}
 	}
 
 	// Compute AF values
-	float squared_length_x = dX.x * dX.x + dX.y * dX.y;
-	float squared_length_y = dY.x * dY.x + dY.y * dY.y;
-	float determinant = abs(dX.x * dY.y - dX.y * dY.x);
-	bool is_major_x = squared_length_x > squared_length_y;
-	float squared_length_major = is_major_x ? squared_length_x : squared_length_y;
-	float length_major = sqrt(squared_length_major);
+	bool is_major_x = length_x > length_y;
+	float length_major = is_major_x ? length_x : length_y;
+	float length_minor = is_major_x ? length_y : length_x;
 
 	float aniso_ratio;
 	float length_lod;
@@ -279,57 +299,42 @@ vec4 sample_c_af(vec2 uv, float uv_w)
 		// Perform isotropic filtering instead.
 		aniso_ratio = 1.0f;
 		length_lod = length_major;
-		aniso_line = vec2(0, 0);
+		aniso_line = vec2(0.0f, 0.0f);
 	}
 	else
 	{
-		float norm_major = 1.0f / length_major;
-	
-		vec2 aniso_line_dir = vec2(
-			(is_major_x ? dX.x : dY.x) * norm_major,
-			(is_major_x ? dX.y : dY.y) * norm_major
-		);
-	
-		aniso_ratio = squared_length_major / determinant;
+		vec2 aniso_line_dir = is_major_x ? dX : dY;
 
-		// Calculate the minor length of the ellipse for Lod, while also clamping the ratio of anisotropy.
-		if (aniso_ratio > PS_ANISOTROPIC_FILTERING)
-		{
-			// ratio is clamped - Lod is based on ratio (preserves area)
-			aniso_ratio = PS_ANISOTROPIC_FILTERING;
-			length_lod = length_major / PS_ANISOTROPIC_FILTERING;
-		}
-		else
-		{
-			// ratio not clamped - Lod is based on area
-			length_lod = determinant / length_major;
-		}
+		aniso_ratio = min(length_major / length_minor, PS_ANISOTROPIC_FILTERING);
+		length_lod = length_major / aniso_ratio;
 
 		// clamp to top Lod
 		if (length_lod < 1.0f)
 			aniso_ratio = max(1.0f, aniso_ratio * length_lod);
 
 		aniso_ratio = round(aniso_ratio);
-		aniso_line = aniso_line_dir * 0.5f * length_major * (1.0f / sz);
+
+		aniso_line = aniso_line_dir * 0.5f * (1.0f / sz);
 	}
-	
+
 #if PS_AUTOMATIC_LOD == 1
 	float lod = log2(length_lod);
 #elif PS_MANUAL_LOD == 1
 	float lod = manual_lod(uv_w);
 #else
-	float lod = 0; // No Lod
+	float lod = 0.0f; // No Lod
 #endif
-	
+
 	vec4 colour;
 	if (aniso_ratio == 1.0f)
 		colour = textureLod(TextureSampler, uv, lod);
 	else
 	{
-		vec4 num = vec4(0, 0, 0, 0);
+		vec4 num = vec4(0.0f, 0.0f, 0.0f, 0.0f);
+		vec2 segment = (2.0f * aniso_line) / aniso_ratio;
 		for (int i = 0; i < aniso_ratio; i++)
-		{		
-			vec2 d = -aniso_line + (0.5f + i) * (2.0f * aniso_line) / aniso_ratio;	
+		{
+			vec2 d = -aniso_line + (0.5f + i) * segment;	
 			vec2 uv_sample = uv + d;
 			vec4 sample_colour = textureLod(TextureSampler, uv_sample, lod);
 			num += sample_colour;
@@ -621,13 +626,13 @@ vec4 sample_depth(vec2 st)
 
 
 #elif PS_DEPTH_FMT == 1
-	// Based on ps_convert_float32_rgba8 of convert
+	// Based on ps_convert_depth32_rgba8 of convert
 	// Convert a GL_FLOAT32 depth texture into a RGBA color texture
 	uint d = uint(fetch_c(uv).r * exp2(32.0f));
 	t = vec4(uvec4((d & 0xFFu), ((d >> 8) & 0xFFu), ((d >> 16) & 0xFFu), (d >> 24)));
 
 #elif PS_DEPTH_FMT == 2
-	// Based on ps_convert_float16_rgb5a1 of convert
+	// Based on ps_convert_depth16_rgb5a1 of convert
 	// Convert a GL_FLOAT32 (only 16 lsb) depth into a RGB5A1 color texture
 	uint d = uint(fetch_c(uv).r * exp2(32.0f));
 	t = vec4(uvec4((d & 0x1Fu), ((d >> 5) & 0x1Fu), ((d >> 10) & 0x1Fu), (d >> 15) & 0x01u)) * vec4(8.0f, 8.0f, 8.0f, 128.0f);
@@ -1193,10 +1198,10 @@ void ps_main()
 #endif
 
 #if PS_ZTST == ZTST_GEQUAL
-	if (input_z < sample_from_depth().r)
+	if (input_z < sample_from_depth())
 		discard;
 #elif PS_ZTST == ZTST_GREATER
-	if (input_z <= sample_from_depth().r)
+	if (input_z <= sample_from_depth())
 		discard;
 #endif
 
@@ -1250,7 +1255,12 @@ void ps_main()
 	vec4 C = ps_color();
 
 #if PS_AA1
-	float cov = clamp(1.0f - abs(PSin.inv_cov), 0.0f, 1.0f);
+	#if PS_AA1 == PS_AA1_LINE
+		// Blur only outer part of the line by scaling coverage.
+		float cov = clamp(LineCovScale * (1.0f - abs(PSin.inv_cov)), 0.0f, 1.0f);
+	#else
+		float cov = clamp(1.0f - abs(PSin.inv_cov), 0.0f, 1.0f);
+	#endif
 	#if PS_ABE
 		if (floor(C.a) == 128.0f) // The coverage is only used if the fragment alpha is 128.
 			C.a = 128.0f * cov;
@@ -1264,7 +1274,7 @@ void ps_main()
 
 	bool atst_pass = atst(C);
 
-#if PS_AFAIL == AFAIL_KEEP
+#if PS_ATST != PS_ATST_NONE && PS_AFAIL == AFAIL_KEEP
 	if (!atst_pass)
 		discard;
 #endif
@@ -1293,12 +1303,12 @@ void ps_main()
 #if PS_DATE == 1
 	// DATM == 0
 	// Pixel with alpha equal to 1 will failed (128-255)
-	SV_Target0 = (C.a > 127.5f) ? vec4(gl_PrimitiveID) : vec4(0x7FFFFFFF);
+	o_col0 = (C.a > 127.5f) ? vec4(gl_PrimitiveID) : vec4(0x7FFFFFFF);
 	return;
 #elif PS_DATE == 2
 	// DATM == 1
 	// Pixel with alpha equal to 0 will failed (0-127)
-	SV_Target0 = (C.a < 127.5f) ? vec4(gl_PrimitiveID) : vec4(0x7FFFFFFF);
+	o_col0 = (C.a < 127.5f) ? vec4(gl_PrimitiveID) : vec4(0x7FFFFFFF);
 	return;
 #endif
 
@@ -1372,7 +1382,7 @@ void ps_main()
 	// Alpha test with feedback
 	#if PS_AFAIL == AFAIL_FB_ONLY
 		if (!atst_pass)
-			input_z = sample_from_depth().r;
+			input_z = sample_from_depth();
 	#elif PS_AFAIL == AFAIL_ZB_ONLY
 		if (!atst_pass)
 			C = sample_from_rt();
@@ -1381,17 +1391,17 @@ void ps_main()
 		{
 			C.a = sample_from_rt().a;
 		#if PS_AFAIL == AFAIL_RGB_ONLY_SW_Z
-			input_z = sample_from_depth().r;
+			input_z = sample_from_depth();
 		#endif
 		}
 	#endif
 
-	// Warning: do not write SV_Target0 until the end since the value might be needed for
+	// Warning: do not write o_col0 until the end since the value might be needed for
 	// FB fetch in sample_from_rt().
-	SV_Target0 = C;
+	o_col0 = C;
 
 	#if !PS_NO_COLOR1
-		SV_Target1 = alpha_blend;
+		o_col1 = alpha_blend;
 	#endif
 #endif
 
@@ -1401,16 +1411,17 @@ void ps_main()
 
 #if PS_AA1 == PS_AA1_TRIANGLE_SW_Z
 	if (!bool(PSin.interior))
-		input_z = sample_from_depth().r; // No depth update for triangle edges.
+		input_z = sample_from_depth(); // No depth update for triangle edges.
 #endif
 
+// Writing back depth
 #if ZWRITE
 	#if SW_DEPTH && PS_NO_COLOR1 && (DEPTH_FEEDBACK_SUPPORT == 2)
 		// Depth as color write. For depth as color feedback we write to both
 		// color copy and real depth to avoid having to copy back to real depth.
-		// Warning: do not write SV_Target1 until the end since the value might
+		// Warning: do not write o_col1 until the end since the value might
 		// be needed for FB fetch in sample_from_depth().
-		SV_Target1 = input_z;
+		o_col1 = input_z;
 	#endif
 	// Standard depth write.
 	gl_FragDepth = input_z;

bin/resources/shaders/opengl/tfx_vgs.glsl

@@ -12,8 +12,9 @@ layout(std140, binding = 1) uniform cb20
 	vec2  TextureOffset;
 
 	vec2  PointSize;
+
 	uint  MaxDepth;
-	uint  pad_cb20;
+	float LineAA1Width;
 };
 
 #ifdef VERTEX_SHADER
@@ -83,7 +84,13 @@ void vs_main()
 	// example: 133.0625 (133 + 1/16) should start from line 134, ceil(133.0625 - 0.05) still above 133
 	gl_Position.xy = vec2(i_p) - vec2(0.05f, 0.05f);
 	gl_Position.xy = gl_Position.xy * VertexScale - VertexOffset;
-	gl_Position.z = float(z) * exp_min32;
+
+	#if HAS_CLIP_CONTROL
+		gl_Position.z = float(z) * exp_min32;
+	#else
+		gl_Position.z = (float(z) * exp_min32) * 2.0f - 1.0f;
+	#endif
+
 	gl_Position.w = 1.0f;
 
 	texture_coord();
@@ -160,7 +167,13 @@ ProcessedVertex load_vertex(uint index)
 	uint z = min(i_z, MaxDepth);
 	vtx.p.xy = vec2(i_p) - vec2(0.05f, 0.05f);
 	vtx.p.xy = vtx.p.xy * VertexScale - VertexOffset;
-	vtx.p.z = float(z) * exp_min32;
+
+	#if HAS_CLIP_CONTROL
+		vtx.p.z = float(z) * exp_min32;
+	#else
+		vtx.p.z = (float(z) * exp_min32) * 2.0f - 1.0f;
+	#endif
+
 	vtx.p.w = 1.0f;
 
 	vec2 uv = vec2(i_uv) - TextureOffset;
@@ -311,11 +324,9 @@ void main()
 	// Use bottom minus top for delta regardless of which vertex we are expanding.
 	vec2 line_delta = is_bottom ? (vtx.p.xy - other.p.xy) : (other.p.xy - vtx.p.xy);
 	vec2 line_vector = normalize(line_delta / VertexScale);
-#if VS_EXPAND == VS_EXPAND_LINE
 	vec2 line_expand = vec2(line_vector.y, -line_vector.x);
-#elif VS_EXPAND == VS_EXPAND_LINE_AA1
-	// Expand in y direction for shallow lines and x direction for steep lines.
-	vec2 line_expand = abs(line_vector.x) >= abs(line_vector.y) ? vec2(0.0f, 2.0f) : vec2(2.0f, 0.0f);
+#if VS_EXPAND == VS_EXPAND_LINE_AA1
+	line_expand *= 2.0f * LineAA1Width;
 #endif
 	vec2 line_width = (line_expand * PointSize) / 2;
 	vec2 offset = is_right ? line_width : -line_width;
@@ -444,11 +455,6 @@ void main()
 		VSout.inv_cov = is_near_corner ? 0.0f : 1.0f; // Full coverage at near corner, otherwise none.
 	
 		VSout.interior = 0;
-
-		#if !VS_IIP
-			// Get the provoking vertex color (last vertex in GL)
-			vtx.c = i0 == 2 ? vtx.c : (i1 == 2 ? other.c : opposite.c);
-		#endif
 	}
 
 #endif

bin/resources/shaders/vulkan/convert.glsl

@@ -20,45 +20,127 @@ void main()
 
 layout(location = 0) in vec2 v_tex;
 
-#if defined(ps_convert_rgba8_16bits) || defined(ps_convert_float32_32bits)
-layout(location = 0) out uint o_col0;
-#elif defined(ps_convert_float32_depth_to_color)
-layout(location = 0) out float o_col0;
-#elif !defined(ps_datm1) && \
-	!defined(ps_datm0) && \
-	!defined(ps_datm1_rta_correction) && \
-	!defined(ps_datm0_rta_correction) && \
-	!defined(ps_convert_rgba8_float32) && \
-	!defined(ps_convert_rgba8_float24) && \
-	!defined(ps_convert_rgba8_float16) && \
-	!defined(ps_convert_rgb5a1_float16) && \
-	!defined(ps_convert_rgba8_float32_biln) && \
-	!defined(ps_convert_rgba8_float24_biln) && \
-	!defined(ps_convert_rgba8_float16_biln) && \
-	!defined(ps_convert_rgb5a1_float16_biln) && \
-	!defined(ps_convert_float32_float24) && \
-	!defined(ps_depth_copy)
-layout(location = 0) out vec4 o_col0;
+#if HAS_INTEGER_OUTPUT
+	layout(location = 0) out uint o_col0;
+	#define OUTPUT o_col0
+#elif HAS_DEPTH_OUTPUT
+	out float gl_FragDepth;
+	#define OUTPUT gl_FragDepth
+#elif HAS_FLOAT32_OUTPUT
+	layout(location = 0) out float o_col0;
+	#define OUTPUT o_col0
+#elif HAS_STENCIL_OUTPUT
+#else
+	layout(location = 0) out vec4 o_col0;
+	#define OUTPUT o_col0
 #endif
 
 layout(set = 0, binding = 0) uniform sampler2D samp0;
 
+#if HAS_FLOAT32_INPUT
+
+float sample_c(vec2 uv)
+{
+	return texture(samp0, uv).r;
+}
+
+#else
+
 vec4 sample_c(vec2 uv)
 {
 	return texture(samp0, uv);
 }
 
+#endif
+
+uint rgba8_to_uint(vec4 c)
+{
+	uvec4 i = uvec4(c * 255.5f) & 0xFFu;
+	return i.r | (i.g << 8) | (i.b << 16) | (i.a << 24);
+}
+
+uint rgb5a1_to_uint(vec4 c)
+{
+	uvec4 i = uvec4(c * 255.5f) & uvec4(0xF8u, 0xF8u, 0xF8u, 0x80u);
+	return (i.r >> 3) | (i.g << 2) | (i.b << 7) | (i.a << 8);
+}
+
+uint depth_to_uint(float d)
+{
+	return uint(d * exp2(32.0f));
+}
+
+vec4 uint_to_rgba8(uint i)
+{
+	return vec4((i & 0xFFu), ((i >> 8) & 0xFFu), ((i >> 16) & 0xFFu), ((i >> 24) & 0xFFu)) / 255.0f;
+}
+
+vec4 uint_to_rgb5a1(uint i)
+{
+	return vec4(uvec4(i << 3, i >> 2, i >> 7, i >> 8) & uvec4(0xF8u, 0xF8u, 0xF8u, 0x80u)) / 255.0f;
+}
+
+float uint_to_depth32(uint i)
+{
+	return float(i) * exp2(-32.0f);
+}
+
+float uint_to_depth24(uint i)
+{
+	return float(i & 0xFFFFFFu) * exp2(-32.0f);
+}
+
+float uint_to_depth16(uint i)
+{
+	return float(i & 0xFFFFu) * exp2(-32.0f);
+}
+
+float rgba8_to_depth32(vec4 val)
+{
+	return uint_to_depth32(rgba8_to_uint(val));
+}
+
+float rgba8_to_depth24(vec4 val)
+{
+	return uint_to_depth24(rgba8_to_uint(val));
+}
+
+float rgba8_to_depth16(vec4 val)
+{
+	return uint_to_depth16(rgba8_to_uint(val));
+}
+
+float rgb5a1_to_depth16(vec4 val)
+{
+	return uint_to_depth16(rgb5a1_to_uint(val));
+}
+
+vec4 depth32_to_rgba8(float d)
+{
+	return uint_to_rgba8(depth_to_uint(d));
+}
+
+vec4 depth16_to_rgb5a1(float d)
+{
+	return uint_to_rgb5a1(depth_to_uint(d));
+}
+
+float depth32_to_depth24(float d)
+{
+	return uint_to_depth24(depth_to_uint(d));
+}
+
 #ifdef ps_copy
 void ps_copy()
 {
-	o_col0 = sample_c(v_tex);
+	OUTPUT = sample_c(v_tex);
 }
 #endif
 
 #ifdef ps_depth_copy
 void ps_depth_copy()
 {
-  gl_FragDepth = sample_c(v_tex).r;
+	OUTPUT = sample_c(v_tex);
 }
 #endif
 
@@ -83,7 +165,7 @@ void ps_downsample_copy()
 			result += texelFetch(samp0, coord + ivec2(xoff * step_multiplier, yoff * step_multiplier), 0);
 		}
 	}
-	o_col0 = result / Weight;
+	OUTPUT = result / Weight;
 }
 #endif
 
@@ -91,17 +173,15 @@ void ps_downsample_copy()
 void ps_filter_transparency()
 {
 	vec4 c = sample_c(v_tex);
-	o_col0 = vec4(c.rgb, 1.0);
+	OUTPUT = vec4(c.rgb, 1.0);
 }
 #endif
 
-#ifdef ps_convert_rgba8_16bits
-// Need to be careful with precision here, it can break games like Spider-Man 3 and Dogs Life
-void ps_convert_rgba8_16bits()
+#ifdef ps_convert_rgb5a1_16bits
+void ps_convert_rgb5a1_16bits()
 {
-	uvec4 i = uvec4(sample_c(v_tex) * vec4(255.5f, 255.5f, 255.5f, 255.5f));
-
-	o_col0 = ((i.x & 0x00F8u) >> 3) | ((i.y & 0x00F8u) << 2) | ((i.z & 0x00f8u) << 7) | ((i.w & 0x80u) << 8);
+	// Need to be careful with precision here, it can break games like Spider-Man 3 and Dogs Life
+	OUTPUT = rgb5a1_to_uint(sample_c(v_tex));
 }
 #endif
 
@@ -141,7 +221,7 @@ void ps_datm0_rta_correction()
 void ps_rta_correction()
 {
 	vec4 value = sample_c(v_tex);
-	o_col0 = vec4(value.rgb, value.a / (128.25f / 255.0f));
+	OUTPUT = vec4(value.rgb, value.a / (128.25f / 255.0f));
 }
 #endif
 
@@ -149,7 +229,7 @@ void ps_rta_correction()
 void ps_rta_decorrection()
 {
 	vec4 value = sample_c(v_tex);
-	o_col0 = vec4(value.rgb, value.a * (128.25f / 255.0f));
+	OUTPUT = vec4(value.rgb, value.a * (128.25f / 255.0f));
 }
 #endif
 
@@ -157,7 +237,7 @@ void ps_rta_decorrection()
 void ps_colclip_init()
 {
 	vec4 value = sample_c(v_tex);
-	o_col0 = vec4(roundEven(value.rgb * 255.0f) / 65535.0f, value.a);
+	OUTPUT = vec4(roundEven(value.rgb * 255.0f) / 65535.0f, value.a);
 }
 #endif
 
@@ -165,116 +245,39 @@ void ps_colclip_init()
 void ps_colclip_resolve()
 {
 	vec4 value = sample_c(v_tex);
-	o_col0 = vec4(vec3(uvec3(value.rgb * 65535.5f) & 255u) / 255.0f, value.a);
+	OUTPUT = vec4(vec3(uvec3(value.rgb * 65535.5f) & 255u) / 255.0f, value.a);
 }
 #endif
 
-#ifdef ps_convert_float32_depth_to_color
-void ps_convert_float32_depth_to_color()
-{
-	o_col0 = sample_c(v_tex).r;
-}
-#endif
-
-#ifdef ps_convert_float32_color_to_depth
-void ps_convert_float32_color_to_depth()
-{
-	gl_FragDepth = sample_c(v_tex).r;
-}
-#endif
-
-#ifdef ps_convert_float32_32bits
-void ps_convert_float32_32bits()
+#ifdef ps_convert_depth32_32bits
+void ps_convert_depth32_32bits()
 {
 	// Convert a vec32 depth texture into a 32 bits UINT texture
-	o_col0 = uint(exp2(32.0f) * sample_c(v_tex).r);
+	OUTPUT = depth_to_uint(sample_c(v_tex));
 }
 #endif
 
-#ifdef ps_convert_float32_rgba8
-void ps_convert_float32_rgba8()
+#ifdef ps_convert_depth32_rgba8
+void ps_convert_depth32_rgba8()
 {
 	// Convert a vec32 depth texture into a RGBA color texture
-	uint d = uint(sample_c(v_tex).r * exp2(32.0f));
-	o_col0 = vec4(uvec4((d & 0xFFu), ((d >> 8) & 0xFFu), ((d >> 16) & 0xFFu), (d >> 24))) / vec4(255.0);
+	OUTPUT = depth32_to_rgba8(sample_c(v_tex));
 }
 #endif
 
-#ifdef ps_convert_float16_rgb5a1
-void ps_convert_float16_rgb5a1()
+#ifdef ps_convert_depth16_rgb5a1
+void ps_convert_depth16_rgb5a1()
 {
 	// Convert a vec32 (only 16 lsb) depth into a RGB5A1 color texture
-	uint d = uint(sample_c(v_tex).r * exp2(32.0f));
-	o_col0 = vec4(uvec4(d << 3, d >> 2, d >> 7, d >> 8) & uvec4(0xf8, 0xf8, 0xf8, 0x80)) / 255.0f;
+	OUTPUT = depth16_to_rgb5a1(sample_c(v_tex));
 }
 #endif
 
-float rgba8_to_depth32(vec4 unorm)
-{
-	uvec4 c = uvec4(unorm * vec4(255.5f));
-	return float(c.r | (c.g << 8) | (c.b << 16) | (c.a << 24)) * exp2(-32.0f);
-}
-
-float rgba8_to_depth24(vec4 unorm)
-{
-	uvec3 c = uvec3(unorm.rgb * vec3(255.5f));
-	return float(c.r | (c.g << 8) | (c.b << 16)) * exp2(-32.0f);
-}
-
-float rgba8_to_depth16(vec4 unorm)
-{
-	uvec2 c = uvec2(unorm.rg * vec2(255.5f));
-	return float(c.r | (c.g << 8)) * exp2(-32.0f);
-}
-
-float rgb5a1_to_depth16(vec4 unorm)
-{
-	uvec4 c = uvec4(unorm * vec4(255.5f));
-	return float(((c.r & 0xF8u) >> 3) | ((c.g & 0xF8u) << 2) | ((c.b & 0xF8u) << 7) | ((c.a & 0x80u) << 8)) * exp2(-32.0f);
-}
-
-#ifdef ps_convert_float32_float24
-void ps_convert_float32_float24()
+#ifdef ps_convert_depth32_depth24
+void ps_convert_depth32_depth24()
 {
 	// Truncates depth value to 24bits
-	uint d = uint(sample_c(v_tex).r * exp2(32.0f)) & 0xFFFFFFu;
-	gl_FragDepth = float(d) * exp2(-32.0f);
-}
-#endif
-
-#ifdef ps_convert_rgba8_float32
-void ps_convert_rgba8_float32()
-{
-	// Convert an RGBA texture into a float depth texture
-	gl_FragDepth = rgba8_to_depth32(sample_c(v_tex));
-}
-#endif
-
-#ifdef ps_convert_rgba8_float24
-void ps_convert_rgba8_float24()
-{
-	// Same as above but without the alpha channel (24 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
-	gl_FragDepth = rgba8_to_depth24(sample_c(v_tex));
-}
-#endif
-
-#ifdef ps_convert_rgba8_float16
-void ps_convert_rgba8_float16()
-{
-	// Same as above but without the A/B channels (16 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
-	gl_FragDepth = rgba8_to_depth16(sample_c(v_tex));
-}
-#endif
-
-#ifdef ps_convert_rgb5a1_float16
-void ps_convert_rgb5a1_float16()
-{
-	// Convert an RGB5A1 (saved as RGBA8) color to a 16 bit Z
-	gl_FragDepth = rgb5a1_to_depth16(sample_c(v_tex));
+	OUTPUT = depth32_to_depth24(sample_c(v_tex));
 }
 #endif
 
@@ -288,41 +291,53 @@ void ps_convert_rgb5a1_float16()
 	float depthTR = CONVERT_FN(texelFetch(samp0, coords.zy, 0)); \
 	float depthBL = CONVERT_FN(texelFetch(samp0, coords.xw, 0)); \
 	float depthBR = CONVERT_FN(texelFetch(samp0, coords.zw, 0)); \
-	gl_FragDepth = mix(mix(depthTL, depthTR, mix_vals.x), mix(depthBL, depthBR, mix_vals.x), mix_vals.y);
+	OUTPUT = mix(mix(depthTL, depthTR, mix_vals.x), mix(depthBL, depthBR, mix_vals.x), mix_vals.y);
 
-#ifdef ps_convert_rgba8_float32_biln
-void ps_convert_rgba8_float32_biln()
+#ifdef ps_convert_rgba8_depth32
+void ps_convert_rgba8_depth32()
 {
 	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth32);
+#else
+	OUTPUT = rgba8_to_depth32(sample_c(v_tex));
+#endif
 }
 #endif
 
-#ifdef ps_convert_rgba8_float24_biln
-void ps_convert_rgba8_float24_biln()
+#ifdef ps_convert_rgba8_depth24
+void ps_convert_rgba8_depth24()
 {
 	// Same as above but without the alpha channel (24 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth24);
+#else
+	OUTPUT = rgba8_to_depth24(sample_c(v_tex));
+#endif
 }
 #endif
 
-#ifdef ps_convert_rgba8_float16_biln
-void ps_convert_rgba8_float16_biln()
+#ifdef ps_convert_rgba8_depth16
+void ps_convert_rgba8_depth16()
 {
 	// Same as above but without the A/B channels (16 bits Z)
-
-	// Convert an RGBA texture into a float depth texture
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgba8_to_depth16);
+#else
+	OUTPUT = rgba8_to_depth16(sample_c(v_tex));
+#endif
 }
 #endif
 
-#ifdef ps_convert_rgb5a1_float16_biln
-void ps_convert_rgb5a1_float16_biln()
+#ifdef ps_convert_rgb5a1_depth16
+void ps_convert_rgb5a1_depth16()
 {
 	// Convert an RGB5A1 (saved as RGBA8) color to a 16 bit Z
+#if HAS_BILN
 	SAMPLE_RGBA_DEPTH_BILN(rgb5a1_to_depth16);
+#else
+	OUTPUT = rgb5a1_to_depth16(sample_c(v_tex));
+#endif
 }
 #endif
 
@@ -609,25 +624,25 @@ void ps_yuv()
 }
 #endif
 
-#if defined(ps_stencil_image_init_0) || defined(ps_stencil_image_init_1) || defined(ps_stencil_image_init_2) || defined(ps_stencil_image_init_3)
+#if defined(ps_primid_image_init_0) || defined(ps_primid_image_init_1) || defined(ps_primid_image_init_2) || defined(ps_primid_image_init_3)
 
 void main()
 {
 	o_col0 = vec4(0x7FFFFFFF);
 
-	#ifdef ps_stencil_image_init_0
+	#ifdef ps_primid_image_init_0
 		if((127.5f / 255.0f) < sample_c(v_tex).a) // < 0x80 pass (== 0x80 should not pass)
 			o_col0 = vec4(-1);
 	#endif
-	#ifdef ps_stencil_image_init_1
+	#ifdef ps_primid_image_init_1
 		if(sample_c(v_tex).a < (127.5f / 255.0f)) // >= 0x80 pass
 			o_col0 = vec4(-1);
 	#endif
-	#ifdef ps_stencil_image_init_2
+	#ifdef ps_primid_image_init_2
 		if((254.5f / 255.0f) < sample_c(v_tex).a) // < 0x80 pass (== 0x80 should not pass)
 			o_col0 = vec4(-1);
 	#endif
-	#ifdef ps_stencil_image_init_3
+	#ifdef ps_primid_image_init_3
 		if(sample_c(v_tex).a < (254.5f / 255.0f)) // >= 0x80 pass
 			o_col0 = vec4(-1);
 	#endif

bin/resources/shaders/vulkan/tfx.glsl

@@ -24,7 +24,7 @@ layout(std140, set = 0, binding = 0) uniform cb0
 	vec2 TextureOffset;
 	vec2 PointSize;
 	uint MaxDepth;
-	uint pad_cb0;
+	float LineAA1Width;
 };
 
 layout(location = 0) out VSOutput
@@ -319,11 +319,9 @@ void main()
 	// Use bottom minus top for delta regardless of which vertex we are expanding.
 	vec2 line_delta = is_bottom ? (vtx.p.xy - other.p.xy) : (other.p.xy - vtx.p.xy);
 	vec2 line_vector = normalize(line_delta / VertexScale);
-#if VS_EXPAND == VS_EXPAND_LINE
 	vec2 line_expand = vec2(line_vector.y, -line_vector.x);
-#elif VS_EXPAND == VS_EXPAND_LINE_AA1
-	// Expand in y direction for shallow lines and x direction for steep lines.
-	vec2 line_expand = abs(line_vector.x) >= abs(line_vector.y) ? vec2(0.0f, 2.0f) : vec2(2.0f, 0.0f);
+#if VS_EXPAND == VS_EXPAND_LINE_AA1
+	line_expand *= 2.0f * LineAA1Width;
 #endif
 	vec2 line_width = (line_expand * PointSize) / 2;
 	vec2 offset = is_right ? line_width : -line_width;
@@ -452,11 +450,6 @@ void main()
 		vsOut.inv_cov = is_near_corner ? 0.0f : 1.0f; // Full coverage at near corner, otherwise none.
 	
 		vsOut.interior = 0;
-
-		#if !VS_IIP
-			// Get the provoking vertex color (last vertex in VK)
-			vtx.c = i0 == 2 ? vtx.c : (i1 == 2 ? other.c : opposite.c);
-		#endif
 	}
 
 #endif
@@ -522,6 +515,12 @@ void main()
 #define PS_AA1_TRIANGLE_SW_Z 3
 #endif
 
+#ifndef PS_ROV_DEPTH_NONE
+#define PS_ROV_DEPTH_NONE 0
+#define PS_ROV_DEPTH_READ_WRITE 1
+#define PS_ROV_DEPTH_READ_ONLY 2
+#endif
+
 #ifndef PS_FST
 #define PS_FST 0
 #define PS_WMS 0
@@ -574,6 +573,8 @@ void main()
 #define PS_NO_COLOR1 0
 #define PS_DATE 0
 #define PS_TEX_IS_FB 0
+#define PS_ROV_COLOR 0
+#define PS_ROV_DEPTH 0
 #endif
 
 #define SW_BLEND (PS_BLEND_A || PS_BLEND_B || PS_BLEND_D)
@@ -586,7 +587,13 @@ void main()
 
 #define PS_FEEDBACK_LOOP_IS_NEEDED_RT (PS_TEX_IS_FB == 1 || AFAIL_NEEDS_RT || PS_FBMASK || SW_BLEND_NEEDS_RT || SW_AD_TO_HW || (PS_DATE >= 5))
 #define PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH (AFAIL_NEEDS_DEPTH || ZTST_NEEDS_DEPTH || AA1_NEEDS_DEPTH)
-#define ZWRITE (PS_ZCLAMP || PS_ZFLOOR || PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH)
+#define ZWRITE (PS_ZCLAMP || PS_ZFLOOR || SW_DEPTH || PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH)
+
+#define PS_RETURN_COLOR_ROV (!PS_NO_COLOR && PS_ROV_COLOR)
+#define PS_RETURN_COLOR (!PS_NO_COLOR && !PS_ROV_COLOR)
+#define PS_RETURN_DEPTH_ROV (PS_ROV_DEPTH == PS_ROV_DEPTH_READ_WRITE)
+#define PS_RETURN_DEPTH (ZWRITE && !PS_ROV_DEPTH)
+#define PS_ROV_EARLYDEPTHSTENCIL (PS_ROV_COLOR && !PS_ROV_DEPTH && !ZWRITE)
 
 #define NEEDS_TEX (PS_TFX != 4)
 
@@ -610,6 +617,12 @@ layout(std140, set = 0, binding = 1) uniform cb1
 	mat4 DitherMatrix;
 	float ScaledScaleFactor;
 	float RcpScaleFactor;
+	float _pad0_cb1;
+	float _pad1_cb1;
+	float LineCovScale;
+	float _pad2_cb1;
+	float _pad3_cb1;
+	float _pad4_cb1;
 };
 
 layout(location = 0) in VSOutput
@@ -625,11 +638,27 @@ layout(location = 0) in VSOutput
 	flat uint interior; // 1 for triangle interior; 0 for edge;
 } vsIn;
 
-#if !PS_NO_COLOR && !PS_NO_COLOR1
-layout(location = 0, index = 0) out vec4 o_col0;
-layout(location = 0, index = 1) out vec4 o_col1;
-#elif !PS_NO_COLOR
-layout(location = 0) out vec4 o_col0;
+#if PS_RETURN_COLOR
+	#if !PS_NO_COLOR1
+		layout(location = 0, index = 0) out vec4 o_col0;
+		layout(location = 0, index = 1) out vec4 o_col1;
+	#elif !PS_NO_COLOR
+		layout(location = 0) out vec4 o_col0;
+	#endif
+#elif PS_RETURN_COLOR_ROV
+	vec4 o_col0;
+#endif
+
+#if PS_ROV_COLOR
+	layout(set = 1, binding = 5, rgba8) uniform restrict coherent image2D RtImageRov;
+	vec4 rov_rt_value;
+	vec4 sample_from_rt() { return rov_rt_value; }
+#endif
+
+#if PS_ROV_DEPTH
+	layout(set = 1, binding = 6, r32f) uniform restrict coherent image2D DepthImageRov;
+	float rov_depth_value;
+	float sample_from_depth() { return rov_depth_value; }
 #endif
 
 #if NEEDS_TEX
@@ -639,27 +668,27 @@ layout(set = 1, binding = 1) uniform texture2D Palette;
 
 #if PS_FEEDBACK_LOOP_IS_NEEDED_RT || PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH
 	#if defined(DISABLE_TEXTURE_BARRIER) || defined(HAS_FEEDBACK_LOOP_LAYOUT)
-		#if PS_FEEDBACK_LOOP_IS_NEEDED_RT
+		#if (PS_FEEDBACK_LOOP_IS_NEEDED_RT && !PS_ROV_COLOR)
 			layout(set = 1, binding = 2) uniform texture2D RtSampler;
 			vec4 sample_from_rt() { return texelFetch(RtSampler, ivec2(gl_FragCoord.xy), 0); }
 		#endif
-		#if PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH
+		#if (PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH && !PS_ROV_DEPTH)
 			layout(set = 1, binding = 4) uniform texture2D DepthSampler;
-			vec4 sample_from_depth() { return texelFetch(DepthSampler, ivec2(gl_FragCoord.xy), 0); }
+			float sample_from_depth() { return texelFetch(DepthSampler, ivec2(gl_FragCoord.xy), 0).r; }
 		#endif
 	#else
 		// Must consider each case separately since the input attachment indices must be consecutive.
-		#if PS_FEEDBACK_LOOP_IS_NEEDED_RT && PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH
+		#if (PS_FEEDBACK_LOOP_IS_NEEDED_RT && !PS_ROV_COLOR) && (PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH && !PS_ROV_DEPTH)
 			layout(input_attachment_index = 0, set = 1, binding = 2) uniform subpassInput RtSampler;
 			layout(input_attachment_index = 1, set = 1, binding = 4) uniform subpassInput DepthSampler;
 			vec4 sample_from_rt() { return subpassLoad(RtSampler); }
-			vec4 sample_from_depth() { return subpassLoad(DepthSampler); }
-		#elif PS_FEEDBACK_LOOP_IS_NEEDED_RT
+			float sample_from_depth() { return subpassLoad(DepthSampler).r; }
+		#elif (PS_FEEDBACK_LOOP_IS_NEEDED_RT && !PS_ROV_COLOR)
 			layout(input_attachment_index = 0, set = 1, binding = 2) uniform subpassInput RtSampler;
 			vec4 sample_from_rt() { return subpassLoad(RtSampler); }
-		#elif PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH
+		#elif (PS_FEEDBACK_LOOP_IS_NEEDED_DEPTH && !PS_ROV_DEPTH)
 			layout(input_attachment_index = 0, set = 1, binding = 4) uniform subpassInput DepthSampler;
-			vec4 sample_from_depth() { return subpassLoad(DepthSampler); }
+			float sample_from_depth() { return subpassLoad(DepthSampler).r; }
 		#endif
 	#endif
 #endif
@@ -694,22 +723,27 @@ float manual_lod(float uv_w)
 vec4 sample_c_af(vec2 uv, float uv_w)
 {
 	// HW sampler will reject bad UVs, match that here.
-	uv = (any(isnan(uv)) || any(isinf(uv))) ? vec2(0, 0) : uv;
+	uv = (any(isnan(uv)) || any(isinf(uv))) ? vec2(0.0f, 0.0f) : uv;
 
 	// Large floating point values risk NaN/Inf values.
 	// Above this value floats lose decimal precision, so seems a resonable limit for UVs.
 	uv = clamp(uv, -8388608.0f, 8388608.0f);
 
 	// Below taken from https://microsoft.github.io/DirectX-Specs/d3d/archive/D3D11_3_FunctionalSpec.htm#7.18.11%20LOD%20Calculations
+	// And https://registry.khronos.org/OpenGL/extensions/EXT/EXT_texture_filter_anisotropic.txt
 	// With guidance from https://pema.dev/2025/05/09/mipmaps-too-much-detail/ 
 	vec2 sz = textureSize(Texture, 0);
 	vec2 dX = dFdx(uv) * sz;
 	vec2 dY = dFdy(uv) * sz;
 
+	float length_x = length(dX);
+	float length_y = length(dY);
+
 	// Calculate Ellipse Transform
-	bool d_zero = length(dX) == 0 || length(dY) == 0;
-	bool d_par = (dX.x * dY.y - dY.x * dX.y) == 0;
-	bool d_per = dot(dX, dY) == 0;
+	bool d_zero = length_x < 0.001f || length_y < 0.001f;
+	float f = (dX.x * dY.y - dX.y * dY.x);
+	bool d_par = f < 0.001f;
+	bool d_per = dot(dX, dY) < 0.001f;
 	bool d_inf_nan = any(isinf(dX)) || any(isinf(dY)) || any(isnan(dX)) || any(isnan(dY));
 
 	if (!(d_zero || d_par || d_per || d_inf_nan))
@@ -724,31 +758,40 @@ vec4 sample_c_af(vec2 uv, float uv_w)
 		float q = A + C;
 		float t = sqrt(p * p + B * B);
 
+		float signB = sign(B);
+		float denom_plus  = t * (q + t);
+		float denom_minus = t * (q - t);
+
+		float sqrtA = sqrt(F * (t + p));
+		float sqrtB = sqrt(F * (t - p));
+
+		float inv_sqrt_denom_plus  = inversesqrt(denom_plus);
+		float inv_sqrt_denom_minus = inversesqrt(denom_minus);
+
 		vec2 new_dX = vec2(
-			sqrt(F * (t + p) / (t * (q + t))),
-			sqrt(F * (t - p) / (t * (q + t))) * sign(B)
+			sqrtA * inv_sqrt_denom_plus,
+			sqrtB * inv_sqrt_denom_plus * signB
 		);
-		
+
 		vec2 new_dY = vec2(
-			sqrt(F * (t - p) / (t * (q - t))) * -sign(B),
-			sqrt(F * (t + p) / (t * (q - t)))
+			sqrtB * inv_sqrt_denom_minus * -signB,
+			sqrtA * inv_sqrt_denom_minus
 		);
-		
+
 		d_inf_nan = any(isinf(new_dX)) || any(isinf(new_dY)) || any(isnan(new_dX)) || any(isnan(new_dY));
 		if (!d_inf_nan)
 		{
 			dX = new_dX;
 			dY = new_dY;
+			length_x = length(dX);
+			length_y = length(dY);
 		}
 	}
 
 	// Compute AF values
-	float squared_length_x = dX.x * dX.x + dX.y * dX.y;
-	float squared_length_y = dY.x * dY.x + dY.y * dY.y;
-	float determinant = abs(dX.x * dY.y - dX.y * dY.x);
-	bool is_major_x = squared_length_x > squared_length_y;
-	float squared_length_major = is_major_x ? squared_length_x : squared_length_y;
-	float length_major = sqrt(squared_length_major);
+	bool is_major_x = length_x > length_y;
+	float length_major = is_major_x ? length_x : length_y;
+	float length_minor = is_major_x ? length_y : length_x;
 
 	float aniso_ratio;
 	float length_lod;
@@ -760,57 +803,42 @@ vec4 sample_c_af(vec2 uv, float uv_w)
 		// Perform isotropic filtering instead.
 		aniso_ratio = 1.0f;
 		length_lod = length_major;
-		aniso_line = vec2(0, 0);
+		aniso_line = vec2(0.0f, 0.0f);
 	}
 	else
 	{
-		float norm_major = 1.0f / length_major;
-	
-		vec2 aniso_line_dir = vec2(
-			(is_major_x ? dX.x : dY.x) * norm_major,
-			(is_major_x ? dX.y : dY.y) * norm_major
-		);
-	
-		aniso_ratio = squared_length_major / determinant;
+		vec2 aniso_line_dir = is_major_x ? dX : dY;
 
-		// Calculate the minor length of the ellipse for Lod, while also clamping the ratio of anisotropy.
-		if (aniso_ratio > PS_ANISOTROPIC_FILTERING)
-		{
-			// ratio is clamped - Lod is based on ratio (preserves area)
-			aniso_ratio = PS_ANISOTROPIC_FILTERING;
-			length_lod = length_major / PS_ANISOTROPIC_FILTERING;
-		}
-		else
-		{
-			// ratio not clamped - Lod is based on area
-			length_lod = determinant / length_major;
-		}
+		aniso_ratio = min(length_major / length_minor, PS_ANISOTROPIC_FILTERING);
+		length_lod = length_major / aniso_ratio;
 
 		// clamp to top Lod
 		if (length_lod < 1.0f)
 			aniso_ratio = max(1.0f, aniso_ratio * length_lod);
 
 		aniso_ratio = round(aniso_ratio);
-		aniso_line = aniso_line_dir * 0.5f * length_major * (1.0f / sz);
+
+		aniso_line = aniso_line_dir * 0.5f * (1.0f / sz);
 	}
-	
+
 #if PS_AUTOMATIC_LOD == 1
 	float lod = log2(length_lod);
 #elif PS_MANUAL_LOD == 1
 	float lod = manual_lod(uv_w);
 #else
-	float lod = 0; // No Lod
+	float lod = 0.0f; // No Lod
 #endif
-	
+
 	vec4 colour;
 	if (aniso_ratio == 1.0f)
 		colour = textureLod(Texture, uv, lod);
 	else
 	{
-		vec4 num = vec4(0, 0, 0, 0);
+		vec4 num = vec4(0.0f, 0.0f, 0.0f, 0.0f);
+		vec2 segment = (2.0f * aniso_line) / aniso_ratio;
 		for (int i = 0; i < aniso_ratio; i++)
-		{		
-			vec2 d = -aniso_line + (0.5f + i) * (2.0f * aniso_line) / aniso_ratio;	
+		{
+			vec2 d = -aniso_line + (0.5f + i) * segment;	
 			vec2 uv_sample = uv + d;
 			vec4 sample_colour = textureLod(Texture, uv_sample, lod);
 			num += sample_colour;
@@ -1114,7 +1142,7 @@ vec4 sample_depth(vec2 st, ivec2 pos)
 	}
 	#elif (PS_DEPTH_FMT == 1)
 	{
-		// Based on ps_convert_float32_rgba8 of convert
+		// Based on ps_convert_depth32_rgba8 of convert
 
 		// Convert a vec32 depth texture into a RGBA color texture
 		uint d = uint(fetch_c(uv).r * exp2(32.0f));
@@ -1122,7 +1150,7 @@ vec4 sample_depth(vec2 st, ivec2 pos)
 	}
 	#elif (PS_DEPTH_FMT == 2)
 	{
-		// Based on ps_convert_float16_rgb5a1 of convert
+		// Based on ps_convert_depth16_rgb5a1 of convert
 
 		// Convert a vec32 (only 16 lsb) depth into a RGB5A1 color texture
 		uint d = uint(fetch_c(uv).r * exp2(32.0f));
@@ -1604,7 +1632,7 @@ void ps_blend(inout vec4 Color, inout vec4 As_rgba)
 		#elif PS_BLEND_MIX == 1
 			Color.rgb = ((A - B) * C_clamped + D) - (124.0f/256.0f);
 		#else
-				Color.rgb = trunc((A - B) * C + D);
+			Color.rgb = trunc((A - B) * C + D);
 		#endif
 
 		#if PS_BLEND_HW == 1
@@ -1678,6 +1706,24 @@ void ps_blend(inout vec4 Color, inout vec4 As_rgba)
 	#endif
 }
 
+#if PS_ROV_COLOR || PS_ROV_DEPTH
+layout(pixel_interlock_ordered) in;
+#endif
+
+#if PS_ROV_EARLYDEPTHSTENCIL
+layout(early_fragment_tests) in;
+#endif
+
+#if PS_ROV_COLOR || PS_ROV_DEPTH
+	#define DISCARD { rov_discard_color = true; rov_discard_depth = true; }
+	#define DISCARD_COLOR rov_discard_color = true
+	#define DISCARD_DEPTH rov_discard_depth = true
+#else
+	#define DISCARD discard
+	#define DISCARD_COLOR o_col0 = sample_from_rt()
+	#define DISCARD_DEPTH input_z = sample_from_depth()
+#endif
+
 void main()
 {
 	float input_z = gl_FragCoord.z;
@@ -1687,18 +1733,35 @@ void main()
 	input_z = floor(input_z * exp2(32.0f)) * exp2(-32.0f);
 #endif
 
+#if PS_ROV_COLOR || PS_ROV_DEPTH
+	beginInvocationInterlockARB();
+#endif
+
+#if PS_ROV_COLOR
+	rov_rt_value = imageLoad(RtImageRov, ivec2(gl_FragCoord.xy));
+#endif
+
+#if PS_ROV_DEPTH
+	rov_depth_value = imageLoad(DepthImageRov, ivec2(gl_FragCoord.xy)).r;
+#endif
+
+#if PS_ROV_COLOR || PS_ROV_DEPTH
+	bool rov_discard_color = gl_HelperInvocation;
+	bool rov_discard_depth = gl_HelperInvocation;
+#endif
+
 #if PS_ZTST == ZTST_GEQUAL
-	if (input_z < sample_from_depth().r)
-		discard;
+	if (input_z < sample_from_depth())
+		DISCARD;
 #elif PS_ZTST == ZTST_GREATER
-	if (input_z <= sample_from_depth().r)
-		discard;
+	if (input_z <= sample_from_depth())
+		DISCARD;
 #endif
 
 #if PS_SCANMSK & 2
 	// fail depth test on prohibited lines
 	if ((int(gl_FragCoord.y) & 1) == (PS_SCANMSK & 1))
-		discard;
+		DISCARD;
 #endif
 #if PS_DATE >= 5
 
@@ -1726,7 +1789,7 @@ void main()
 #endif
 
 	if (bad) {
-		discard;
+		DISCARD;
 	}
 
 #endif		// PS_DATE >= 5
@@ -1737,14 +1800,19 @@ void main()
 	// the bad alpha value so we must keep it.
 
 	if (gl_PrimitiveID > stencil_ceil) {
-		discard;
+		DISCARD;
 	}
 #endif
 
 	vec4 C = ps_color();
 
 #if PS_AA1
-	float cov = clamp(1.0f - abs(vsIn.inv_cov), 0.0f, 1.0f);
+	#if PS_AA1 == PS_AA1_LINE
+		// Blur only outer part of the line by scaling coverage.
+		float cov = clamp(LineCovScale * (1.0f - abs(vsIn.inv_cov)), 0.0f, 1.0f);
+	#else
+		float cov = clamp(1.0f - abs(vsIn.inv_cov), 0.0f, 1.0f);
+	#endif
 	#if PS_ABE
 		if (floor(C.a) == 128.0f) // The coverage is only used if the fragment alpha is 128.
 			C.a = 128.0f * cov;
@@ -1758,9 +1826,10 @@ void main()
 
 	bool atst_pass = atst(C);
 
-#if PS_AFAIL == AFAIL_KEEP
-	if (!atst_pass)
-		discard;
+#if PS_ATST != PS_ATST_NONE && PS_AFAIL == AFAIL_KEEP
+	if (!atst_pass) {
+		DISCARD;
+	}
 #endif
 
 #if SW_AD_TO_HW
@@ -1852,6 +1921,7 @@ void main()
 		alpha_blend.a = float(atst_pass);
 	#endif
 
+	// Output color scaling
 	#if !PS_NO_COLOR
 		#if PS_RTA_CORRECTION
 			o_col0.a = C.a / 128.0f;
@@ -1870,16 +1940,16 @@ void main()
 		// Alpha test with feedback
 		#if PS_AFAIL == AFAIL_FB_ONLY
 			if (!atst_pass)
-				input_z = sample_from_depth().r;
+				DISCARD_DEPTH;
 		#elif PS_AFAIL == AFAIL_ZB_ONLY
 			if (!atst_pass)
-				o_col0 = sample_from_rt();
+				DISCARD_COLOR;
 		#elif (PS_AFAIL == AFAIL_RGB_ONLY || PS_AFAIL == AFAIL_RGB_ONLY_SW_Z)
 			if (!atst_pass)
 			{
-				o_col0.a = sample_from_rt().a;
+				o_col0.a = sample_from_rt().a; // discard alpha
 			#if PS_AFAIL == AFAIL_RGB_ONLY_SW_Z
-				input_z = sample_from_depth().r;
+				DISCARD_DEPTH;
 			#endif
 			}
 		#endif
@@ -1891,11 +1961,27 @@ void main()
 	
 	#if PS_AA1 == PS_AA1_TRIANGLE_SW_Z
 		if (!bool(vsIn.interior))
-			input_z = sample_from_depth().r; // No depth update for triangle edges.
+			DISCARD_DEPTH; // No depth update for triangle edges.
 	#endif
 	
-	#if ZWRITE
+	// Writing back color (result already written to o_col0 for non-ROV)
+	#if PS_RETURN_COLOR_ROV
+		o_col0 = mix(o_col0, sample_from_rt(), equal(FbMask, uvec4(0xFFu))); // channel masking
+
+		if (!rov_discard_color)
+			imageStore(RtImageRov, ivec2(gl_FragCoord.xy), o_col0);
+	#endif
+	
+	// Writing back depth
+	#if PS_RETURN_DEPTH
 		gl_FragDepth = input_z;
+	#elif PS_RETURN_DEPTH_ROV
+		if (!rov_discard_depth)
+			imageStore(DepthImageRov, ivec2(gl_FragCoord.xy), vec4(input_z, 0, 0, 1.0f));
+	#endif
+
+	#if PS_ROV_COLOR || PS_ROV_DEPTH
+		endInvocationInterlockARB();
 	#endif
 #endif // PS_DATE
 }

cmake/BuildParameters.cmake

@@ -300,8 +300,8 @@ else()
 	if(CMAKE_INTERPROCEDURAL_OPTIMIZATION)
 		message(WARNING
 			"The CMAKE_INTERPROCEDURAL_OPTIMIZATION option is enabled but the "
-			"CMAKE_POSITION_INDEPENDENT_CODE option is disabled. This has been "
-			"found to result in broken builds on certain platforms.")
+			"POSITION_INDEPENDENT_CODE option is disabled. This has been found "
+			"to result in broken builds on certain platforms.")
 	endif()
 
 	set(CMAKE_POSITION_INDEPENDENT_CODE OFF)

cmake/SearchForStuff.cmake

@@ -22,6 +22,9 @@ find_package(Freetype 2.10 REQUIRED) # 2.10 is the first with COLRv0 support, wh
 find_package(plutovg 1.1.0 REQUIRED)
 find_package(plutosvg 0.0.7 REQUIRED)
 find_package(ryml REQUIRED)
+if (WIN32)
+	find_package(DirectX-Headers 1.618.1 REQUIRED)
+endif()
 
 if(USE_VULKAN)
 	find_package(Shaderc REQUIRED)

common/Console.cpp

@@ -98,14 +98,25 @@ __ri void Log::WriteToConsole(LOGLEVEL level, ConsoleColors color, std::string_v
 
 	static constexpr size_t BUFFER_SIZE = 512;
 
+#ifdef _WIN32
+	constexpr bool supports_color_codes = true;
+#else
+	const int fd = (level <= LOGLEVEL_WARNING) ? STDERR_FILENO : STDOUT_FILENO;
+	const bool supports_color_codes = static_cast<bool>(isatty(fd));
+#endif
+
 	SmallStackString<BUFFER_SIZE> buffer;
 	buffer.reserve(static_cast<u32>(32 + message.length()));
-	buffer.append(s_ansi_color_codes[color]);
+	if (supports_color_codes)
+		buffer.append(s_ansi_color_codes[color]);
 
 	if (s_log_timestamps)
 		buffer.append_format(TIMESTAMP_FORMAT_STRING, Log::GetCurrentMessageTime());
 
 	buffer.append(message);
+
+	if (supports_color_codes)
+		buffer.append(s_ansi_color_codes[Color_Default]);
 	buffer.append('\n');
 
 #ifdef _WIN32
@@ -134,7 +145,6 @@ cleanup:
 	if (wmessage_buf != wbuf)
 		std::free(wmessage_buf);
 #else
-	const int fd = (level <= LOGLEVEL_WARNING) ? STDERR_FILENO : STDOUT_FILENO;
 	write(fd, buffer.data(), buffer.length());
 #endif
 }
@@ -157,7 +167,7 @@ void Log::SetConsoleOutputLevel(LOGLEVEL level)
 	if (was_enabled == now_enabled)
 		return;
 
-		// Worst that happens here is we write to a bad handle..
+	// Worst that happens here is we write to a bad handle..
 
 #if defined(_WIN32)
 	static constexpr auto enable_virtual_terminal_processing = [](HANDLE hConsole) {

common/MemoryInterface.cpp

@@ -34,6 +34,34 @@ Value MemoryInterface::Read(u32 address, bool* valid)
 		return Value(0);
 }
 
+std::optional<std::string> MemoryInterface::ReadString(u32 address, u32 max_size, ReadStringFlags flags)
+{
+	std::string string;
+
+	for (u32 i = 0; i < max_size; i++)
+	{
+		bool valid;
+		char c = Read8(address + i, &valid);
+		if (!valid)
+			return std::nullopt;
+
+		if (c == '\0')
+			return string;
+		else if (!(flags & ALLOW_NON_PRINTABLE_CHARACTERS) && (c < ' ' || c > '~'))
+			return std::nullopt;
+
+		string += c;
+	}
+
+	if (flags & ALLOW_LONG_STRINGS)
+	{
+		string += '~';
+		return string;
+	}
+
+	return std::nullopt;
+}
+
 template <MemoryAccessType Value>
 bool MemoryInterface::Write(u32 address, Value value)
 {
@@ -62,6 +90,17 @@ bool MemoryInterface::Write(u32 address, Value value)
 		return false;
 }
 
+bool MemoryInterface::WriteString(u32 address, std::string_view string)
+{
+	if (!WriteBytes(address, string.data(), static_cast<u32>(string.size())))
+		return false;
+
+	if (!Write8(address + static_cast<u32>(string.size()), '\0'))
+		return false;
+
+	return true;
+}
+
 bool MemoryInterface::IdempotentWrite8(u32 address, u8 value)
 {
 	bool valid;

common/MemoryInterface.h

@@ -5,6 +5,9 @@
 
 #include "Pcsx2Types.h"
 
+#include <optional>
+#include <string>
+#include <string_view>
 #include <type_traits>
 
 template <typename Value>
@@ -31,16 +34,27 @@ public:
 	template <MemoryAccessType Value>
 	Value Read(u32 address, bool* valid = nullptr);
 
+	enum ReadStringFlags
+	{
+		NO_FLAGS = 0,
+		ALLOW_LONG_STRINGS = 1 << 0,
+		ALLOW_NON_PRINTABLE_CHARACTERS = 1 << 1
+	};
+
+	std::optional<std::string> ReadString(u32 address, u32 max_size = 256, ReadStringFlags flags = NO_FLAGS);
+
 	virtual bool Write8(u32 address, u8 value) = 0;
 	virtual bool Write16(u32 address, u16 value) = 0;
 	virtual bool Write32(u32 address, u32 value) = 0;
 	virtual bool Write64(u32 address, u64 value) = 0;
 	virtual bool Write128(u32 address, u128 value) = 0;
-	virtual bool WriteBytes(u32 address, void* src, u32 size) = 0;
+	virtual bool WriteBytes(u32 address, const void* src, u32 size) = 0;
 
 	template <MemoryAccessType Value>
 	bool Write(u32 address, Value value);
 
+	bool WriteString(u32 address, std::string_view string);
+
 	bool IdempotentWrite8(u32 address, u8 value);
 	bool IdempotentWrite16(u32 address, u16 value);
 	bool IdempotentWrite32(u32 address, u32 value);
@@ -51,5 +65,5 @@ public:
 	template <MemoryAccessType Value>
 	bool IdempotentWrite(u32 address, Value value);
 
-	virtual bool CompareBytes(u32 address, void* src, u32 size) = 0;
+	virtual bool CompareBytes(u32 address, const void* src, u32 size) = 0;
 };

common/YAML.cpp

@@ -17,7 +17,7 @@ struct RapidYAMLContext
 	Error* error = nullptr;
 };
 
-std::optional<ryml::Tree> ParseYAMLFromString(ryml::csubstr yaml, ryml::csubstr file_name, Error* error)
+std::optional<ryml::Tree> ParseYAMLFromString(ryml::csubstr yaml, ryml::csubstr file_name, Error* error, bool resolve_anchors)
 {
 	RapidYAMLContext context;
 	context.error = error;
@@ -28,40 +28,93 @@ std::optional<ryml::Tree> ParseYAMLFromString(ryml::csubstr yaml, ryml::csubstr
 	callbacks.set_user_data(static_cast<void*>(&context));
 
 	callbacks.set_error_basic([](ryml::csubstr msg, const ryml::ErrorDataBasic& errdata, void* user_data) {
-		std::string description;
-		auto callback = [&description](ryml::csubstr string) {
-			description.append(string.str, string.len);
-		};
-		ryml::err_basic_format(std::move(callback), msg, errdata);
+		RapidYAMLContext* context = static_cast<RapidYAMLContext*>(user_data);
+		// This scope needs to stay, so all objects destruct before std::longjump
+		{
+			std::string description;
+			auto callback = [&description](ryml::csubstr string) {
+				description.append(string.str, string.len);
+			};
+			ryml::err_basic_format(std::move(callback), msg, errdata);
 
-		// We might have already returned, so don't try to recover.
-		pxFailRel(description.c_str());
-		std::abort();
+			if (context != nullptr)
+			{
+				Error::SetString(context->error, std::move(description));
+			}
+			else
+			{
+				pxFailRel(description.c_str());
+			}
+		}
+
+		if (context != nullptr)
+		{
+			std::longjmp(context->env, 1);
+		}
+		else
+		{
+			std::terminate();
+		}
 	});
 
 	callbacks.set_error_parse([](ryml::csubstr msg, const ryml::ErrorDataParse& errdata, void* user_data) {
 		RapidYAMLContext* context = static_cast<RapidYAMLContext*>(user_data);
+		// This scope needs to stay, so all objects destruct before std::longjump
+		{
+			std::string description;
+			auto callback = [&description](ryml::csubstr string) {
+				description.append(string.str, string.len);
+			};
+			ryml::err_parse_format(std::move(callback), msg, errdata);
 
-		std::string description;
-		auto callback = [&description](ryml::csubstr string) {
-			description.append(string.str, string.len);
-		};
-		ryml::err_parse_format(std::move(callback), msg, errdata);
+			if (context != nullptr)
+			{
+				Error::SetString(context->error, std::move(description));
+			}
+			else
+			{
+				pxFailRel(description.c_str());
+			}
+		}
 
-		Error::SetString(context->error, std::move(description));
-		std::longjmp(context->env, 1);
+		if (context != nullptr)
+		{
+			std::longjmp(context->env, 2);
+		}
+		else
+		{
+			std::terminate();
+		}
 	});
 
 	callbacks.set_error_visit([](ryml::csubstr msg, const ryml::ErrorDataVisit& errdata, void* user_data) {
-		std::string description;
-		auto callback = [&description](ryml::csubstr string) {
-			description.append(string.str, string.len);
-		};
-		ryml::err_visit_format(std::move(callback), msg, errdata);
+		RapidYAMLContext* context = static_cast<RapidYAMLContext*>(user_data);
+		// This scope needs to stay, so all objects destruct before std::longjump
+		{
+			std::string description;
+			auto callback = [&description](ryml::csubstr string) {
+				description.append(string.str, string.len);
+			};
+			ryml::err_visit_format(std::move(callback), msg, errdata);
 
-		// We've probably already returned, so don't try to recover.
-		pxFailRel(description.c_str());
-		std::abort();
+			if (context != nullptr)
+			{
+				Error::SetString(context->error, std::move(description));
+			}
+			else
+			{
+				pxFailRel(description.c_str());
+			}
+		}
+
+		if (context != nullptr)
+		{
+			std::longjmp(context->env, 3);
+		}
+		else
+		{
+			std::terminate();
+		}
 	});
 #else
 	callbacks.m_user_data = static_cast<void*>(&context);
@@ -76,15 +129,24 @@ std::optional<ryml::Tree> ParseYAMLFromString(ryml::csubstr yaml, ryml::csubstr
 	ryml::EventHandlerTree event_handler(callbacks);
 	ryml::Parser parser(&event_handler);
 
-	ryml::Tree tree;
+	ryml::Tree tree(callbacks);
 
 	// The only options RapidYAML provides for recovering from errors are
 	// throwing an exception or using setjmp/longjmp. Since we have exceptions
 	// disabled we have to use the latter option.
-	if (setjmp(context.env))
+	if (setjmp(context.env) != 0)
 		return std::nullopt;
 
 	ryml::parse_in_arena(&parser, file_name, yaml, &tree);
+	if (resolve_anchors)
+	{
+		tree.resolve();
+	}
+
+	// Callbacks passed to ryml::Tree are used for value parsing errors later,
+	// so we need to clear the context before it goes out of scope.
+	callbacks.set_user_data(nullptr);
+	tree.callbacks(callbacks);
 
 	return tree;
 }

common/YAML.h

@@ -14,4 +14,6 @@
 /// parsing errors (as is recommended by the documentation for cases where
 /// exceptions are disabled). The file_name parameter is only used for error
 /// messages, which are returned via the error parameter.
-std::optional<ryml::Tree> ParseYAMLFromString(ryml::csubstr yaml, ryml::csubstr file_name, Error* error);
+/// If resolve_anchors is set to true, YAML anchors and aliases will be resolved.
+/// It's a potentially slow operation, so it's opt-in.
+std::optional<ryml::Tree> ParseYAMLFromString(ryml::csubstr yaml, ryml::csubstr file_name, Error* error, bool resolve_anchors = false);

common/common.vcxproj

@@ -163,6 +163,7 @@
     <ClInclude Include="WindowInfo.h" />
     <ClInclude Include="YAML.h" />
     <ClInclude Include="Threading.h" />
+    <ClInclude Include="EnumOps.h" />
     <ClInclude Include="emitter\implement\avx.h" />
     <ClInclude Include="emitter\implement\bmi.h" />
     <ClInclude Include="emitter\instructions.h" />

common/emitter/internal.h

@@ -30,8 +30,6 @@ namespace x86Emitter
 	extern void EmitRex(SIMDInstructionInfo info, const xRegisterBase& reg1, const xRegisterBase& reg2);
 	extern void EmitRex(SIMDInstructionInfo info, const xRegisterBase& reg1, const xIndirectVoid& sib);
 
-	extern void _xMovRtoR(const xRegisterInt& to, const xRegisterInt& from);
-
 	template <typename T>
 	inline void xWrite(T val)
 	{

common/emitter/movs.cpp

@@ -22,7 +22,7 @@
 namespace x86Emitter
 {
 
-	void _xMovRtoR(const xRegisterInt& to, const xRegisterInt& from)
+	void xImpl_Mov::operator()(const xRegisterInt& to, const xRegisterInt& from) const
 	{
 		pxAssert(to.GetOperandSize() == from.GetOperandSize());
 
@@ -32,12 +32,6 @@ namespace x86Emitter
 		xOpWrite(from.GetPrefix16(), from.Is8BitOp() ? 0x88 : 0x89, from, to);
 	}
 
-	void xImpl_Mov::operator()(const xRegisterInt& to, const xRegisterInt& from) const
-	{
-		// FIXME WTF?
-		_xMovRtoR(to, from);
-	}
-
 	void xImpl_Mov::operator()(const xIndirectVoid& dest, const xRegisterInt& from) const
 	{
 		// mov eax has a special from when writing directly to a DISP32 address

common/emitter/x86emitter.cpp

@@ -1045,7 +1045,7 @@ const xRegister32
 			}
 			else if (displacement_size == 0)
 			{
-				_xMovRtoR(to, src.Index.MatchSizeTo(to));
+				xMOV(to, src.Index.MatchSizeTo(to));
 				return;
 			}
 			else if (!preserve_flags)
@@ -1053,7 +1053,7 @@ const xRegister32
 				// encode as MOV and ADD combo.  Make sure to use the immediate on the
 				// ADD since it can encode as an 8-bit sign-extended value.
 
-				_xMovRtoR(to, src.Index.MatchSizeTo(to));
+				xMOV(to, src.Index.MatchSizeTo(to));
 				xADD(to, src.Displacement);
 				return;
 			}
@@ -1071,7 +1071,7 @@ const xRegister32
 					// (this does not apply to older model P4s with the broken barrel shifter,
 					//  but we currently aren't optimizing for that target anyway).
 
-					_xMovRtoR(to, src.Index);
+					xMOV(to, src.Index);
 					xSHL(to, src.Scale);
 					return;
 				}
@@ -1085,14 +1085,14 @@ const xRegister32
 						if (src.Index == rsp)
 						{
 							// ESP is not encodable as an index (ix86 ignores it), thus:
-							_xMovRtoR(to, src.Base.MatchSizeTo(to)); // will do the trick!
+							xMOV(to, src.Base.MatchSizeTo(to)); // will do the trick!
 							if (src.Displacement)
 								xADD(to, src.Displacement);
 							return;
 						}
 						else if (src.Displacement == 0)
 						{
-							_xMovRtoR(to, src.Base.MatchSizeTo(to));
+							xMOV(to, src.Base.MatchSizeTo(to));
 							xADD(to, src.Index.MatchSizeTo(to));
 							return;
 						}
@@ -1102,7 +1102,7 @@ const xRegister32
 						// special case handling of ESP as Index, which is replaceable with
 						// a single MOV even when preserve_flags is set! :D
 
-						_xMovRtoR(to, src.Base.MatchSizeTo(to));
+						xMOV(to, src.Base.MatchSizeTo(to));
 						return;
 					}
 				}

common/vsprops/DepsDir.props

@@ -0,0 +1,10 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <PropertyGroup>
+    <DepsRootDir Condition="'$(Platform)'=='x64'">$(SolutionDir)deps\</DepsRootDir>
+    <DepsRootDir Condition="'$(Platform)'=='ARM64'">$(SolutionDir)deps-arm64\</DepsRootDir>
+    <DepsBinDir>$(DepsRootDir)bin\</DepsBinDir>
+    <DepsLibDir>$(DepsRootDir)lib\</DepsLibDir>
+    <DepsIncludeDir>$(DepsRootDir)include\</DepsIncludeDir>
+  </PropertyGroup>
+</Project>

common/vsprops/common.props

@@ -1,5 +1,8 @@
 <?xml version="1.0" encoding="utf-8"?>
 <Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ImportGroup Label="PropertySheets">
+    <Import Project="DepsDir.props" />
+  </ImportGroup>
   <PropertyGroup Label="UserMacros">
     <SSEString />
     <SSEString Condition="$(Configuration.Contains(AVX2))">-avx2</SSEString>
@@ -15,11 +18,6 @@
   <PropertyGroup>
     <OutDir>$(SolutionDir)build\lib-$(PlatformName)-$(Configuration)\</OutDir>
     <IntDir>$(SolutionDir)build\obj-$(ProjectName)-$(PlatformName)-$(Configuration)\</IntDir>
-    <DepsRootDir Condition="'$(Platform)'=='x64'">$(SolutionDir)deps\</DepsRootDir>
-    <DepsRootDir Condition="'$(Platform)'=='ARM64'">$(SolutionDir)deps-arm64\</DepsRootDir>
-    <DepsBinDir>$(DepsRootDir)bin\</DepsBinDir>
-    <DepsLibDir>$(DepsRootDir)lib\</DepsLibDir>
-    <DepsIncludeDir>$(DepsRootDir)include\</DepsIncludeDir>
   </PropertyGroup>
   <PropertyGroup>
     <LinkIncremental>false</LinkIncremental>

pcsx2-gsrunner/Main.cpp

@@ -91,6 +91,9 @@ static double s_last_barriers = 0;
 static double s_last_copies = 0;
 static double s_last_uploads = 0;
 static double s_last_readbacks = 0;
+static double s_last_depth_copies_rov = 0;
+static double s_last_draws_rov = 0;
+static double s_last_barriers_rov = 0;
 static u64 s_total_internal_draws = 0;
 static u64 s_total_draws = 0;
 static u64 s_total_render_passes = 0;
@@ -98,6 +101,9 @@ static u64 s_total_barriers = 0;
 static u64 s_total_copies = 0;
 static u64 s_total_uploads = 0;
 static u64 s_total_readbacks = 0;
+static u64 s_total_copies_rov = 0;
+static u64 s_total_draws_rov = 0;
+static u64 s_total_barriers_rov = 0;
 static u32 s_total_frames = 0;
 static u32 s_total_drawn_frames = 0;
 
@@ -216,6 +222,11 @@ bool Host::CopyTextToClipboard(const std::string_view text)
 	return false;
 }
 
+std::string Host::GetTextFromClipboard()
+{
+	return std::string();
+}
+
 void Host::BeginTextInput()
 {
 	// noop
@@ -287,6 +298,9 @@ void Host::BeginPresentFrame()
 		update_stat(GSPerfMon::TextureCopies, s_total_copies, s_last_copies);
 		update_stat(GSPerfMon::TextureUploads, s_total_uploads, s_last_uploads);
 		update_stat(GSPerfMon::Readbacks, s_total_readbacks, s_last_readbacks);
+		update_stat(GSPerfMon::TextureCopiesROV, s_total_copies_rov, s_last_depth_copies_rov);
+		update_stat(GSPerfMon::DrawCallsROV, s_total_draws_rov, s_last_draws_rov);
+		update_stat(GSPerfMon::BarriersROV, s_total_barriers_rov, s_last_barriers_rov);
 
 		const bool idle_frame = s_total_frames && (last_draws == s_total_internal_draws && last_uploads == s_total_uploads);
 
@@ -419,16 +433,6 @@ void Host::OnAchievementsRefreshed()
 	// noop
 }
 
-void Host::OnCoverDownloaderOpenRequested()
-{
-	// noop
-}
-
-void Host::OnCreateMemoryCardOpenRequested()
-{
-	// noop
-}
-
 bool Host::InBatchMode()
 {
 	return false;
@@ -913,6 +917,9 @@ void GSRunner::DumpStats()
 	Console.WriteLn(fmt::format("@HWSTAT@ Copies: {} (avg {})", s_total_copies, static_cast<u64>(std::ceil(s_total_copies / static_cast<double>(s_total_drawn_frames)))));
 	Console.WriteLn(fmt::format("@HWSTAT@ Uploads: {} (avg {})", s_total_uploads, static_cast<u64>(std::ceil(s_total_uploads / static_cast<double>(s_total_drawn_frames)))));
 	Console.WriteLn(fmt::format("@HWSTAT@ Readbacks: {} (avg {})", s_total_readbacks, static_cast<u64>(std::ceil(s_total_readbacks / static_cast<double>(s_total_drawn_frames)))));
+	Console.WriteLn(fmt::format("@HWSTAT@ Copies (ROV): {} (avg {})", s_total_copies_rov, static_cast<u64>(std::ceil(s_total_copies_rov / static_cast<double>(s_total_drawn_frames)))));
+	Console.WriteLn(fmt::format("@HWSTAT@ Draws Calls (ROV): {} (avg {})", s_total_draws_rov, static_cast<u64>(std::ceil(s_total_draws_rov / static_cast<double>(s_total_drawn_frames)))));
+	Console.WriteLn(fmt::format("@HWSTAT@ Barriers (ROV): {} (avg {})", s_total_barriers_rov, static_cast<u64>(std::ceil(s_total_barriers_rov / static_cast<double>(s_total_drawn_frames)))));
 	if (s_perf_enable)
 	{
 		Console.WriteLn(fmt::format("@HWSTAT@ Minimum Frame Time: {:.3f} ms ({:.3f} FPS)", PerformanceMetrics::GetMinimumFrameTime(), 1000.0f / PerformanceMetrics::GetMinimumFrameTime()));

pcsx2-qt/Debugger/Memory/MemoryView.h

@@ -50,15 +50,15 @@ private:
 	QWidget* parent;
 	MemoryViewType displayType = MemoryViewType::BYTE;
 	bool littleEndian = true;
-	u32 rowCount;
-	u32 rowVisible;
-	s32 rowHeight;
+	u32 rowCount = 0;
+	u32 rowVisible = 0;
+	s32 rowHeight = 0;
 
 	// Stuff used for selection handling
 	// This gets set every paint and depends on the window size / current display mode (1byte,2byte,etc)
-	s32 valuexAxis; // Where the hexadecimal view begins
-	s32 textXAxis; // Where the text view begins
-	s32 row1YAxis; // Where the first row starts
+	s32 valuexAxis = 0; // Where the hexadecimal view begins
+	s32 textXAxis = 0; // Where the text view begins
+	s32 row1YAxis = 0; // Where the first row starts
 	s32 segmentXAxis[16]; // Where the segments begin
 	bool selectedText = false; // Whether the user has clicked on text or hex
 
@@ -89,9 +89,9 @@ public:
 	{
 	}
 
-	u32 startAddress;
-	u32 selectedAddress;
-	s32 selectedIndex;
+	u32 startAddress = 0;
+	u32 selectedAddress = 0;
+	s32 selectedIndex = 0;
 
 	void UpdateStartAddress(u32 start);
 	void UpdateSelectedAddress(u32 selected, bool page = false);

pcsx2-qt/Debugger/SymbolTree/SymbolTreeDelegates.cpp

@@ -7,6 +7,8 @@
 #include "Debugger/SymbolTree/SymbolTreeModel.h"
 #include "Debugger/SymbolTree/TypeString.h"
 
+#include "common/StringUtil.h"
+
 #include <QtWidgets/QCheckBox>
 #include <QtWidgets/QComboBox>
 #include <QtWidgets/QDoubleSpinBox>
@@ -44,12 +46,28 @@ QWidget* SymbolTreeValueDelegate::createEditor(QWidget* parent, const QStyleOpti
 		const ccc::ast::Node& physical_type = *logical_type->physical_type(database).first;
 		QVariant value = node->readValueAsVariant(physical_type, m_cpu, database);
 
-		const ccc::ast::Node& type = *logical_type->physical_type(database).first;
-		switch (type.descriptor)
+		switch (physical_type.descriptor)
 		{
+			case ccc::ast::ARRAY:
+			{
+				const ccc::ast::Array& array = physical_type.as<ccc::ast::Array>();
+				const ccc::ast::Node& element_type = *array.element_type->physical_type(database).first;;
+
+				if (array.element_count > 0 &&
+					element_type.name == "char" &&
+					node->location.type == SymbolTreeLocation::MEMORY &&
+					!value.isNull())
+				{
+					QLineEdit* editor = new QLineEdit(parent);
+					editor->setText(value.toString());
+					result = editor;
+				}
+
+				break;
+			}
 			case ccc::ast::BUILTIN:
 			{
-				const ccc::ast::BuiltIn& builtin = type.as<ccc::ast::BuiltIn>();
+				const ccc::ast::BuiltIn& builtin = physical_type.as<ccc::ast::BuiltIn>();
 
 				switch (builtin.bclass)
 				{
@@ -111,7 +129,7 @@ QWidget* SymbolTreeValueDelegate::createEditor(QWidget* parent, const QStyleOpti
 			}
 			case ccc::ast::ENUM:
 			{
-				const ccc::ast::Enum& enumeration = type.as<ccc::ast::Enum>();
+				const ccc::ast::Enum& enumeration = physical_type.as<ccc::ast::Enum>();
 
 				QComboBox* combo_box = new QComboBox(parent);
 				bool named = false;
@@ -186,12 +204,30 @@ void SymbolTreeValueDelegate::setModelData(QWidget* editor, QAbstractItemModel*
 		if (!logical_type)
 			return;
 
-		const ccc::ast::Node& type = *logical_type->physical_type(database).first;
-		switch (type.descriptor)
+		const ccc::ast::Node& physical_type = *logical_type->physical_type(database).first;
+
+		switch (physical_type.descriptor)
 		{
+			case ccc::ast::ARRAY:
+			{
+				const ccc::ast::Array& array = physical_type.as<ccc::ast::Array>();
+				const ccc::ast::Node& element_type = *array.element_type->physical_type(database).first;
+
+				QLineEdit* line_edit = qobject_cast<QLineEdit*>(editor);
+
+				if (array.element_count > 0 &&
+					element_type.name == "char" &&
+					node->location.type == SymbolTreeLocation::MEMORY &&
+					line_edit)
+				{
+					value = line_edit->text();
+				}
+
+				break;
+			}
 			case ccc::ast::BUILTIN:
 			{
-				const ccc::ast::BuiltIn& builtin = type.as<ccc::ast::BuiltIn>();
+				const ccc::ast::BuiltIn& builtin = physical_type.as<ccc::ast::BuiltIn>();
 
 				switch (builtin.bclass)
 				{

pcsx2-qt/Debugger/SymbolTree/SymbolTreeNode.cpp

@@ -67,7 +67,7 @@ bool SymbolTreeNode::writeToVM(
 	if (logical_type)
 	{
 		const ccc::ast::Node& physical_type = *logical_type->physical_type(database).first;
-		writeValueFromVariant(m_value, physical_type, cpu);
+		writeValueFromVariant(m_value, physical_type, cpu, database);
 	}
 
 	data_changed |= updateDisplayString(cpu, database, display_options);
@@ -76,10 +76,29 @@ bool SymbolTreeNode::writeToVM(
 	return data_changed;
 }
 
-QVariant SymbolTreeNode::readValueAsVariant(const ccc::ast::Node& physical_type, DebugInterface& cpu, const ccc::SymbolDatabase& database) const
+QVariant SymbolTreeNode::readValueAsVariant(
+	const ccc::ast::Node& physical_type,
+	DebugInterface& cpu,
+	const ccc::SymbolDatabase& database) const
 {
 	switch (physical_type.descriptor)
 	{
+		case ccc::ast::ARRAY:
+		{
+			const ccc::ast::Array& array = physical_type.as<ccc::ast::Array>();
+			const ccc::ast::Node& element_type = *array.element_type->physical_type(database).first;
+
+			if (array.element_count > 0 &&
+				element_type.name == "char" &&
+				location.type == SymbolTreeLocation::MEMORY)
+			{
+				std::optional<std::string> string = cpu.ReadString(location.address, 256);
+				if (string.has_value())
+					return QString::fromStdString(*string);
+			}
+
+			break;
+		}
 		case ccc::ast::BUILTIN:
 		{
 			const ccc::ast::BuiltIn& builtIn = physical_type.as<ccc::ast::BuiltIn>();
@@ -134,10 +153,36 @@ QVariant SymbolTreeNode::readValueAsVariant(const ccc::ast::Node& physical_type,
 	return QVariant();
 }
 
-bool SymbolTreeNode::writeValueFromVariant(QVariant value, const ccc::ast::Node& physical_type, DebugInterface& cpu) const
+bool SymbolTreeNode::writeValueFromVariant(
+	QVariant value,
+	const ccc::ast::Node& physical_type,
+	DebugInterface& cpu,
+	const ccc::SymbolDatabase& database) const
 {
+	if (value.isNull())
+		return false;
+
 	switch (physical_type.descriptor)
 	{
+		case ccc::ast::ARRAY:
+		{
+			const ccc::ast::Array& array = physical_type.as<ccc::ast::Array>();
+			const ccc::ast::Node& element_type = *array.element_type->physical_type(database).first;
+
+			if (array.element_count > 0 &&
+				element_type.name == "char" &&
+				location.type == SymbolTreeLocation::MEMORY)
+			{
+				QByteArray byte_array = value.toString().toLatin1();
+				std::string_view view(
+					byte_array.data(),
+					std::min(static_cast<s32>(byte_array.size()), array.element_count - 1));
+				cpu.WriteString(location.address, view);
+				break;
+			}
+
+			return false;
+		}
 		case ccc::ast::BUILTIN:
 		{
 			const ccc::ast::BuiltIn& built_in = physical_type.as<ccc::ast::BuiltIn>();
@@ -191,6 +236,7 @@ bool SymbolTreeNode::writeValueFromVariant(QVariant value, const ccc::ast::Node&
 					return false;
 				}
 			}
+
 			break;
 		}
 		case ccc::ast::ENUM:
@@ -266,11 +312,14 @@ QString SymbolTreeNode::generateDisplayString(
 			const ccc::ast::Array& array = physical_type.as<ccc::ast::Array>();
 			const ccc::ast::Node& element_type = *array.element_type->physical_type(database).first;
 
-			if (element_type.name == "char" && location.type == SymbolTreeLocation::MEMORY)
+			if (array.element_count > 0 &&
+				element_type.name == "char" &&
+				location.type == SymbolTreeLocation::MEMORY)
 			{
-				char* string = cpu.stringFromPointer(location.address);
-				if (string)
-					return QString("\"%1\"").arg(string);
+				std::optional<std::string> string = cpu.ReadString(
+					location.address, 256, MemoryInterface::ALLOW_LONG_STRINGS);
+				if (string.has_value())
+					return QString("\"%1\"").arg(*string);
 			}
 
 			QString result;
@@ -412,9 +461,10 @@ QString SymbolTreeNode::generateDisplayString(
 
 			if (pointer_or_reference.is_pointer && value_type.name == "char")
 			{
-				const char* string = cpu.stringFromPointer(address);
-				if (string)
-					result += QString(" \"%1\"").arg(string);
+				std::optional<std::string> string = cpu.ReadString(
+					address, 256, MemoryInterface::ALLOW_LONG_STRINGS);
+				if (string.has_value())
+					result += QString(" \"%1\"").arg(*string);
 			}
 			else if (depth == 0)
 			{

pcsx2-qt/Debugger/SymbolTree/SymbolTreeNode.h

@@ -62,11 +62,20 @@ public:
 		const ccc::SymbolDatabase& database,
 		const SymbolTreeDisplayOptions& display_options);
 
-	QVariant readValueAsVariant(const ccc::ast::Node& physical_type, DebugInterface& cpu, const ccc::SymbolDatabase& database) const;
-	bool writeValueFromVariant(QVariant value, const ccc::ast::Node& physical_type, DebugInterface& cpu) const;
+	QVariant readValueAsVariant(
+		const ccc::ast::Node& physical_type,
+		DebugInterface& cpu,
+		const ccc::SymbolDatabase& database) const;
+
+	bool writeValueFromVariant(
+		QVariant value,
+		const ccc::ast::Node& physical_type,
+		DebugInterface& cpu,
+		const ccc::SymbolDatabase& database) const;
 
 	bool updateDisplayString(
 		DebugInterface& cpu, const ccc::SymbolDatabase& database, const SymbolTreeDisplayOptions& display);
+
 	QString generateDisplayString(
 		const ccc::ast::Node& physical_type,
 		DebugInterface& cpu,

pcsx2-qt/GameList/GameListModel.cpp

@@ -73,9 +73,10 @@ const char* GameListModel::getColumnName(const Column col)
 	return s_column_names[static_cast<int>(col)];
 }
 
-GameListModel::GameListModel(const float cover_scale, const bool show_cover_titles, const qreal dpr, QObject* parent /* = nullptr */)
+GameListModel::GameListModel(const float cover_scale, const bool show_cover_titles, bool show_full_cover_titles, const qreal dpr, QObject* parent /* = nullptr */)
 	: QAbstractTableModel(parent)
 	, m_show_titles_for_covers(show_cover_titles)
+	, m_show_full_titles_for_covers(show_full_cover_titles)
 	, m_dpr{dpr}
 {
 	loadSettings();
@@ -315,9 +316,33 @@ QVariant GameListModel::data(const QModelIndex& index, const int role) const
 			switch (index.column())
 			{
 				case Column_Cover:
-					return QSize(static_cast<int>(static_cast<float>(SIZE_HINT_WIDTH) * m_cover_scale),
-						static_cast<int>(static_cast<float>(m_show_titles_for_covers ? SIZE_HINT_HEIGHT_TITLES : SIZE_HINT_HEIGHT) * m_cover_scale));
+				{
+					const int cover_width = static_cast<int>(static_cast<float>(SIZE_HINT_WIDTH) * m_cover_scale);
+					const int cover_height = static_cast<int>(static_cast<float>(SIZE_HINT_HEIGHT) * m_cover_scale);
 
+					if (m_show_full_titles_for_covers && m_show_titles_for_covers)
+					{
+						// Find height needed for the longest title at the current font/width
+						const int text_width = static_cast<int>(static_cast<float>(COVER_ART_WIDTH) * m_cover_scale);
+						QFont font;
+						font.setPointSizeF(20.0f * m_cover_scale); 
+						const QFontMetrics fm(font);
+						int max_text_height = fm.height();
+						const auto lock = GameList::GetLock();
+						const u32 count = GameList::GetEntryCount();
+						for (u32 i = 0; i < count; i++)
+						{
+							const GameList::Entry* entry = GameList::GetEntryByIndex(i);
+							if (!entry) continue;
+							const QString title = QString::fromStdString(entry->GetTitle(m_prefer_english_titles));
+							const QRect bound = fm.boundingRect(QRect(0, 0, text_width, 0),
+								Qt::TextWordWrap | Qt::AlignHCenter, title);
+							max_text_height = std::max(max_text_height, bound.height());
+						}
+						return QSize(cover_width, cover_height + max_text_height);
+					}
+					return QSize(cover_width, static_cast<int>(static_cast<float>(m_show_titles_for_covers ? SIZE_HINT_HEIGHT_TITLES : SIZE_HINT_HEIGHT) * m_cover_scale));
+				}
 				default:
 					return QVariant();
 			}
@@ -354,7 +379,7 @@ bool GameListModel::titlesLessThan(const int left_row, const int right_row) cons
 	const GameList::Entry* left = GameList::GetEntryByIndex(left_row);
 	const GameList::Entry* right = GameList::GetEntryByIndex(right_row);
 	return QtHost::LocaleSensitiveCompare(QString::fromStdString(left->GetTitleSort(m_prefer_english_titles)),
-			   QString::fromStdString(right->GetTitleSort(m_prefer_english_titles))) < 0;
+				QString::fromStdString(right->GetTitleSort(m_prefer_english_titles))) < 0;
 }
 
 bool GameListModel::lessThan(const QModelIndex& left_index, const QModelIndex& right_index, const int column) const

pcsx2-qt/GameList/GameListModel.h

@@ -43,7 +43,7 @@ public:
 	static QIcon getIconForType(GameList::EntryType type);
 	static QIcon getIconForRegion(GameList::Region region);
 
-	GameListModel(float cover_scale, bool show_cover_titles, qreal dpr, QObject* parent = nullptr);
+	GameListModel(float cover_scale, bool show_cover_titles, bool show_full_cover_titles, qreal dpr, QObject* parent = nullptr);
 	~GameListModel();
 
 	int rowCount(const QModelIndex& parent = QModelIndex()) const override;
@@ -63,6 +63,9 @@ public:
 	bool getShowCoverTitles() const { return m_show_titles_for_covers; }
 	void setShowCoverTitles(bool enabled) { m_show_titles_for_covers = enabled; }
 
+	bool getShowFullCoverTitles() const { return m_show_full_titles_for_covers; }
+	void setShowFullCoverTitles(bool enabled) { m_show_full_titles_for_covers = enabled; }
+
 	float getCoverScale() const { return m_cover_scale; }
 	void setCoverScale(float scale);
 	int getCoverArtWidth() const;
@@ -89,6 +92,7 @@ private:
 	float m_cover_scale = 0.0f;
 	std::atomic<u32> m_cover_scale_counter{0};
 	bool m_show_titles_for_covers = false;
+	bool m_show_full_titles_for_covers = false;
 	bool m_prefer_english_titles = false;
 
 	std::array<QString, Column_Count> m_column_display_names;

pcsx2-qt/GameList/GameListWidget.cpp

@@ -22,7 +22,10 @@
 #include <QtCore/QSortFilterProxyModel>
 #include <QtCore/QDir>
 #include <QtCore/QString>
+#include <QtGui/QColor>
+#include <QtGui/QImage>
 #include <QtGui/QPainter>
+#include <QtGui/QPalette>
 #include <QtGui/QPixmap>
 #include <QtGui/QPixmapCache>
 #include <QtGui/QWheelEvent>
@@ -191,6 +194,29 @@ namespace
 
 				painter->drawPixmap(rect.topLeft() + icon_top_left, highlighted_icon);
 			}
+			// Recolor the icon based on the custom background color
+			else if (index.column() == GameListModel::Column_Type)
+			{
+				// Fetch pixmap from cache or construct a new one.
+				const QColor color = option.palette.color(QPalette::Text);
+				const QString key = QString::fromStdString(fmt::format("type-{:016X}-{:08X}", icon.cacheKey(), color.rgba()));
+
+				QPixmap tinted_icon;
+				if (!QPixmapCache::find(key, &tinted_icon))
+				{
+					QImage img = icon.toImage().convertToFormat(QImage::Format_ARGB32_Premultiplied);
+
+					QPainter tinted_painter(&img);
+					tinted_painter.setCompositionMode(QPainter::CompositionMode_SourceAtop);
+					tinted_painter.fillRect(0, 0, img.width(), img.height(), color);
+					tinted_painter.end();
+
+					tinted_icon = QPixmap(QPixmap::fromImage(img));
+					QPixmapCache::insert(key, tinted_icon);
+				}
+
+				painter->drawPixmap(rect.topLeft() + icon_top_left, tinted_icon);
+			}
 			else
 			{
 				painter->drawPixmap(rect.topLeft() + icon_top_left, icon);
@@ -213,7 +239,8 @@ void GameListWidget::initialize()
 {
 	const float cover_scale = Host::GetBaseFloatSettingValue("UI", "GameListCoverArtScale", 0.45f);
 	const bool show_cover_titles = Host::GetBaseBoolSettingValue("UI", "GameListShowCoverTitles", true);
-	m_model = new GameListModel(cover_scale, show_cover_titles, devicePixelRatioF(), this);
+	const bool show_full_cover_titles = Host::GetBaseBoolSettingValue("UI", "GameListShowFullCoverTitles", true);
+	m_model = new GameListModel(cover_scale, show_cover_titles, show_full_cover_titles, devicePixelRatioF(), this);
 	m_model->updateCacheSize(width(), height());
 
 	m_sort_model = new GameListSortModel(m_model);
@@ -242,6 +269,7 @@ void GameListWidget::initialize()
 	connect(m_ui.viewGameGrid, &QPushButton::clicked, this, &GameListWidget::showGameGrid);
 	connect(m_ui.gridScale, &QSlider::valueChanged, this, &GameListWidget::gridIntScale);
 	connect(m_ui.viewGridTitles, &QPushButton::toggled, this, &GameListWidget::setShowCoverTitles);
+	connect(m_ui.viewFullGridTitles, &QPushButton::toggled, this, &GameListWidget::setShowFullCoverTitles);
 	connect(m_ui.filterType, &QComboBox::currentIndexChanged, this, [this](int index) {
 		m_sort_model->setFilterType((index == 0) ? GameList::EntryType::Count : static_cast<GameList::EntryType>(index - 1));
 	});
@@ -322,6 +350,7 @@ void GameListWidget::initialize()
 	m_list_view->setFrameStyle(QFrame::NoFrame);
 	m_list_view->setVerticalScrollMode(QAbstractItemView::ScrollMode::ScrollPerPixel);
 	m_list_view->verticalScrollBar()->setSingleStep(15);
+	m_list_view->setWordWrap(show_full_cover_titles);
 	onCoverScaleChanged();
 
 	connect(m_list_view->selectionModel(), &QItemSelectionModel::currentChanged, this,
@@ -381,8 +410,14 @@ void GameListWidget::setCustomBackground()
 			delete m_background_movie;
 			m_background_movie = nullptr;
 		}
+		// Cache all frames for small images so loops don't keep re-decoding
+		else if (const s64 file_size = FileSystem::GetPathFileSize(path.c_str()); file_size > 0 && file_size < 64 * 1024 * 1024)
+			m_background_movie->setCacheMode(QMovie::CacheAll);
 	}
 
+	// Invalidate frame cache so the next animated frame triggers full reprocessing
+	m_background_last_size = QSize();
+
 	// If there is no valid background then reset fallback to default UI state
 	if (!m_background_movie)
 	{
@@ -391,6 +426,11 @@ void GameListWidget::setCustomBackground()
 		m_table_view->viewport()->setAutoFillBackground(true);
 		m_list_view->viewport()->setAutoFillBackground(true);
 
+		m_ui.stack->setPalette(QPalette());
+		m_background_text_color = QColor();
+		m_empty_widget->setPalette(QPalette());
+		m_empty_widget->setAutoFillBackground(false);
+
 		m_ui.stack->update();
 		m_table_view->setAlternatingRowColors(true);
 		return;
@@ -415,45 +455,79 @@ void GameListWidget::setCustomBackground()
 	m_background_opacity = Host::GetBaseFloatSettingValue("UI", "GameListBackgroundOpacity", 100.0f);
 
 	// Selected Custom background is valid, connect the signals and start animation in gamelist
-	connect(m_background_movie, &QMovie::frameChanged, this, &GameListWidget::processBackgroundFrames, Qt::UniqueConnection);
+	connect(m_background_movie, &QMovie::frameChanged, this, &GameListWidget::processBackgroundFrames);
 	m_ui.stack->setAutoFillBackground(false);
 
 	m_table_view->viewport()->setAutoFillBackground(false);
 	m_list_view->viewport()->setAutoFillBackground(false);
-	updateCustomBackgroundState(true);
+	m_background_movie->start();
+	updateCustomBackgroundState();
 	m_table_view->setAlternatingRowColors(false);
 	processBackgroundFrames();
 }
 
-void GameListWidget::updateCustomBackgroundState(const bool force_start)
+void GameListWidget::updateCustomBackgroundState()
 {
 	if (m_background_movie && m_background_movie->isValid())
-	{
-		if ((isVisible() && (isActiveWindow() || force_start)) && qGuiApp->applicationState() == Qt::ApplicationActive)
-			m_background_movie->setPaused(false);
-		else
-			m_background_movie->setPaused(true);
-	}
+		m_background_movie->setPaused(!(isVisible() && qGuiApp->applicationState() == Qt::ApplicationActive));
 }
 
 void GameListWidget::processBackgroundFrames()
 {
-	if (m_background_movie && m_background_movie->isValid() && isVisible())
-	{
-		const int widget_width = m_ui.stack->width();
-		const int widget_height = m_ui.stack->height();
+	if (!m_background_movie || !m_background_movie->isValid() || !isVisible())
+		return;
 
-		if (widget_width <= 0 || widget_height <= 0)
-			return;
+	const QSize widget_size(m_ui.stack->width(), m_ui.stack->height());
+	if (widget_size.isEmpty())
+		return;
 
-		QPixmap pm = m_background_movie->currentPixmap();
-		const qreal dpr = devicePixelRatioF();
+	const int frame_number = m_background_movie->currentFrameNumber();
+	const qreal dpr = devicePixelRatioF();
 
-		QtUtils::resizeAndScalePixmap(&pm, widget_width, widget_height, dpr, m_background_scaling, m_background_opacity);
+	if (frame_number == m_background_last_frame && widget_size == m_background_last_size && qFuzzyCompare(dpr, m_background_last_dpr))
+		return;
 
-		m_background_pixmap = std::move(pm);
-		m_ui.stack->update();
-	}
+	QPixmap pm = m_background_movie->currentPixmap();
+	updateBackgroundTextColor(pm);
+	QtUtils::resizeAndScalePixmap(&pm, widget_size.width(), widget_size.height(), dpr, m_background_scaling, m_background_opacity);
+
+	m_background_pixmap = std::move(pm);
+	m_background_last_frame = frame_number;
+	m_background_last_size = widget_size;
+	m_background_last_dpr = dpr;
+	m_ui.stack->update();
+}
+
+void GameListWidget::updateBackgroundTextColor(const QPixmap& frame)
+{
+	if (frame.isNull())
+		return;
+
+	const QImage sampled = frame.scaled(32, 32, Qt::IgnoreAspectRatio, Qt::FastTransformation).toImage();
+	const QColor average = sampled.scaled(1, 1, Qt::IgnoreAspectRatio, Qt::SmoothTransformation).pixelColor(0, 0);
+	const QColor base = qApp->palette().color(QPalette::Base);
+	const qreal coverage = average.alphaF() * std::clamp(m_background_opacity / 100.0f, 0.0f, 1.0f);
+	const qreal brightness = qGray(average.rgb()) * coverage + qGray(base.rgb()) * (1.0 - coverage);
+	const QColor text_color = (brightness > 127.5) ? Qt::black : Qt::white;
+
+	if (m_background_text_color == text_color)
+		return;
+	m_background_text_color = text_color;
+
+	QColor highlight_color = qApp->palette().color(QPalette::Highlight);
+	highlight_color.setAlpha(128);
+	const QColor empty_backdrop_color = (text_color == Qt::black) ? QColor(255, 255, 255, 128) : QColor(0, 0, 0, 128);
+
+	QPalette palette;
+	palette.setColor(QPalette::Text, text_color);
+	palette.setColor(QPalette::WindowText, text_color);
+	palette.setColor(QPalette::Highlight, highlight_color);
+	m_ui.stack->setPalette(palette);
+
+	QPalette empty_palette;
+	empty_palette.setColor(QPalette::Window, empty_backdrop_color);
+	m_empty_widget->setPalette(empty_palette);
+	m_empty_widget->setAutoFillBackground(true);
 }
 
 bool GameListWidget::isShowingGameList() const
@@ -471,6 +545,11 @@ bool GameListWidget::getShowGridCoverTitles() const
 	return m_model->getShowCoverTitles();
 }
 
+bool GameListWidget::getShowGridFullCoverTitles() const
+{
+	return m_model->getShowFullCoverTitles();
+}
+
 void GameListWidget::refresh(bool invalidate_cache, bool popup_on_error)
 {
 	cancelRefresh();
@@ -499,6 +578,8 @@ void GameListWidget::cancelRefresh()
 void GameListWidget::reloadThemeSpecificImages()
 {
 	m_model->reloadThemeSpecificImages();
+	m_background_last_size = QSize();
+	processBackgroundFrames();
 }
 
 void GameListWidget::onRefreshProgress(const QString& status, int current, int total)
@@ -654,6 +735,9 @@ void GameListWidget::refreshGridCovers()
 
 void GameListWidget::showGameList()
 {
+	Host::SetBaseBoolSettingValue("UI", "GameListGridView", false);
+	Host::CommitBaseSettingChanges();
+
 	if (m_ui.stack->currentIndex() == 0 || m_model->rowCount() == 0)
 	{
 		// We can click the toolbar multiple times, so keep it correct.
@@ -661,8 +745,6 @@ void GameListWidget::showGameList()
 		return;
 	}
 
-	Host::SetBaseBoolSettingValue("UI", "GameListGridView", false);
-	Host::CommitBaseSettingChanges();
 	m_ui.stack->setCurrentIndex(0);
 	setFocusProxy(m_ui.stack->currentWidget());
 	resizeTableViewColumnsToFit();
@@ -672,6 +754,9 @@ void GameListWidget::showGameList()
 
 void GameListWidget::showGameGrid()
 {
+	Host::SetBaseBoolSettingValue("UI", "GameListGridView", true);
+	Host::CommitBaseSettingChanges();
+
 	if (m_ui.stack->currentIndex() == 1 || m_model->rowCount() == 0)
 	{
 		// We can click the toolbar multiple times, so keep it correct.
@@ -679,8 +764,6 @@ void GameListWidget::showGameGrid()
 		return;
 	}
 
-	Host::SetBaseBoolSettingValue("UI", "GameListGridView", true);
-	Host::CommitBaseSettingChanges();
 	m_ui.stack->setCurrentIndex(1);
 	setFocusProxy(m_ui.stack->currentWidget());
 	updateToolbar();
@@ -701,9 +784,24 @@ void GameListWidget::setShowCoverTitles(bool enabled)
 	emit layoutChange();
 }
 
+void GameListWidget::setShowFullCoverTitles(bool enabled)
+{
+	if (m_model->getShowFullCoverTitles() == enabled)
+		return;
+
+	Host::SetBaseBoolSettingValue("UI", "GameListShowFullCoverTitles", enabled);
+	Host::CommitBaseSettingChanges();
+	m_model->setShowFullCoverTitles(enabled);
+	m_list_view->setWordWrap(enabled); 
+	if (isShowingGameGrid())
+		m_model->refresh();
+	updateToolbar();
+	emit layoutChange();
+}
+
 void GameListWidget::updateToolbar()
 {
-	const bool grid_view = isShowingGameGrid();
+	const bool grid_view = Host::GetBaseBoolSettingValue("UI", "GameListGridView", false);
 	{
 		QSignalBlocker sb(m_ui.viewGameGrid);
 		m_ui.viewGameGrid->setChecked(grid_view);
@@ -716,12 +814,17 @@ void GameListWidget::updateToolbar()
 		QSignalBlocker sb(m_ui.viewGridTitles);
 		m_ui.viewGridTitles->setChecked(m_model->getShowCoverTitles());
 	}
+	{
+		QSignalBlocker sb(m_ui.viewFullGridTitles);
+		m_ui.viewFullGridTitles->setChecked(m_model->getShowFullCoverTitles());
+	}
 	{
 		QSignalBlocker sb(m_ui.gridScale);
 		m_ui.gridScale->setValue(static_cast<int>(m_model->getCoverScale() * 100.0f));
 	}
 
 	m_ui.viewGridTitles->setEnabled(grid_view);
+	m_ui.viewFullGridTitles->setEnabled(grid_view && m_model->getShowCoverTitles());
 	m_ui.gridScale->setEnabled(grid_view);
 }
 
@@ -729,6 +832,7 @@ void GameListWidget::showEvent(QShowEvent* event)
 {
 	QWidget::showEvent(event);
 	updateCustomBackgroundState();
+	processBackgroundFrames();
 }
 
 void GameListWidget::hideEvent(QHideEvent* event)
@@ -765,11 +869,8 @@ bool GameListWidget::eventFilter(QObject* watched, QEvent* event)
 		if (!m_background_pixmap.isNull())
 		{
 			QPainter painter(m_ui.stack);
-			const auto* paint_event = static_cast<QPaintEvent*>(event);
-			painter.save();
-			painter.setClipRect(paint_event->rect());
-			painter.drawTiledPixmap(m_ui.stack->rect(), m_background_pixmap);
-			painter.restore();
+			painter.setClipRect(static_cast<QPaintEvent*>(event)->rect());
+			painter.drawPixmap(0, 0, m_background_pixmap);
 			return true;
 		}
 	}

pcsx2-qt/GameList/GameListWidget.h

@@ -9,6 +9,7 @@
 
 #include "pcsx2/GameList.h"
 
+#include <QtGui/QColor>
 #include <QtGui/QMovie>
 #include <QtGui/QPixmap>
 #include <QtWidgets/QListView>
@@ -53,12 +54,14 @@ public:
 	void cancelRefresh();
 	void reloadThemeSpecificImages();
 	void setCustomBackground();
-	void updateCustomBackgroundState(const bool force_start = false);
+	void updateCustomBackgroundState();
 	void processBackgroundFrames();
+	void updateBackgroundTextColor(const QPixmap& frame);
 
 	bool isShowingGameList() const;
 	bool isShowingGameGrid() const;
 	bool getShowGridCoverTitles() const;
+	bool getShowGridFullCoverTitles() const;
 
 	std::optional<GameList::Entry> getSelectedEntry() const;
 
@@ -93,6 +96,7 @@ public Q_SLOTS:
 	void showGameList();
 	void showGameGrid();
 	void setShowCoverTitles(bool enabled);
+	void setShowFullCoverTitles(bool enabled);
 	void gridZoomIn();
 	void gridZoomOut();
 	void gridIntScale(int int_scale);
@@ -128,6 +132,10 @@ private:
 
 	QMovie* m_background_movie = nullptr;
 	QPixmap m_background_pixmap;
+	QColor m_background_text_color;
 	QtUtils::ScalingMode m_background_scaling = QtUtils::ScalingMode::Fit;
 	float m_background_opacity = 100.0f;
+	int m_background_last_frame = -1;
+	QSize m_background_last_size;
+	qreal m_background_last_dpr = 0.0;
 };

pcsx2-qt/GameList/GameListWidget.ui

@@ -120,6 +120,31 @@
          </property>
         </widget>
        </item>
+       <item>
+        <widget class="QToolButton" name="viewFullGridTitles">
+         <property name="minimumSize">
+          <size>
+           <width>32</width>
+           <height>0</height>
+          </size>
+         </property>
+         <property name="toolTip">
+          <string>Show Full Titles</string>
+         </property>
+         <property name="text">
+          <string>Show Full Titles</string>
+         </property>
+         <property name="icon">
+          <iconset theme="price-tag-2-line"/>
+         </property>
+         <property name="checkable">
+          <bool>true</bool>
+         </property>
+         <property name="autoRaise">
+          <bool>true</bool>
+         </property>
+        </widget>
+       </item>
        <item>
         <widget class="QSlider" name="gridScale">
          <property name="minimumSize">
@@ -218,6 +243,7 @@
   <tabstop>viewGameList</tabstop>
   <tabstop>viewGameGrid</tabstop>
   <tabstop>viewGridTitles</tabstop>
+  <tabstop>viewFullGridTitles</tabstop>
   <tabstop>gridScale</tabstop>
   <tabstop>filterType</tabstop>
   <tabstop>filterRegion</tabstop>