Update spirv-tools to v2024.2 (dd4b663)

2024-06-27 10:23:14 -07:00
873 changed files with 53491 additions and 44882 deletions
@@ -1,17 +0,0 @@
-name: 'Android Continuous'
-inputs:
-  build-abi:
-    description: 'The target platform ABI'
-    required: true
-    default: 'armeabi-v7a'
-runs:
-  using: "composite"
-  steps:
-    - uses: actions/setup-java@v3
-      with:
-        distribution: 'temurin'
-        java-version: '17'
-    - name: Run build script
-      run: |
-        cd build/android && printf "y" | ./build.sh continuous ${{ inputs.build-abi }}
-      shell: bash
@@ -8,35 +8,32 @@ on:
      - rc/**

 jobs:
-  build-android-armv7:
-    name: build-android-armv7
+  build-android:
+    name: build-android
    runs-on: macos-14

    steps:
      - uses: actions/checkout@v4.1.6
-      - name: Run Android Continuous
-        uses: ./.github/actions/android-continuous
+      - uses: actions/setup-java@v3
        with:
-          build-abi: armeabi-v7a
-
-  build-android-armv8a:
-    name: build-android-armv8a
-    runs-on: macos-14
-
-    steps:
-      - uses: actions/checkout@v4.1.6
-      - name: Run Android Continuous
-        uses: ./.github/actions/android-continuous
+          distribution: 'temurin'
+          java-version: '17'
+      - name: Run build script
+        run: |
+          cd build/android && printf "y" | ./build.sh continuous
+      - uses: actions/upload-artifact@v1.0.0
        with:
-          build-abi: arm64-v8a
-
-  build-android-x86_64:
-    name: build-android-x86_64
-    runs-on: macos-14
-
-    steps:
-      - uses: actions/checkout@v4.1.6
-      - name: Run Android Continuous
-        uses: ./.github/actions/android-continuous
+          name: filament-android
+          path: out/filament-android-release.aar
+      - uses: actions/upload-artifact@v1.0.0
        with:
-          build-abi: x86_64
+          name: filamat-android-full
+          path: out/filamat-android-release.aar
+      - uses: actions/upload-artifact@v1.0.0
+        with:
+          name: gltfio-android-release
+          path: out/gltfio-android-release.aar
+      - uses: actions/upload-artifact@v1.0.0
+        with:
+          name: filament-utils-android-release
+          path: out/filament-utils-android-release.aar
@@ -17,7 +17,7 @@ jobs:
      - name: Run build script
        run: |
          cd build/ios && printf "y" | ./build.sh continuous
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@v1.0.0
        with:
          name: filament-ios
          path: out/filament-release-ios.tgz
@@ -17,7 +17,7 @@ jobs:
      - name: Run build script
        run: |
          cd build/linux && printf "y" | ./build.sh continuous
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@v1.0.0
        with:
          name: filament-linux
          path: out/filament-release-linux.tgz
@@ -17,7 +17,7 @@ jobs:
      - name: Run build script
        run: |
          cd build/mac && printf "y" | ./build.sh continuous
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@v1.0.0
        with:
          name: filament-mac
          path: out/filament-release-darwin.tgz
@@ -15,7 +15,7 @@ jobs:

    strategy:
      matrix:
-        os: [macos-14-xlarge, ubuntu-22.04-16core]
+        os: [macos-14, ubuntu-22.04]

    steps:
      - uses: actions/checkout@v4.1.6
@@ -29,7 +29,7 @@ jobs:

  build-windows:
    name: build-windows
-    runs-on: windows-2019-32core
+    runs-on: windows-2019

    steps:
      - uses: actions/checkout@v4.1.6
@@ -40,7 +40,7 @@ jobs:

  build-android:
    name: build-android
-    runs-on: ubuntu-22.04-16core
+    runs-on: macos-14

    steps:
      - uses: actions/checkout@v4.1.6
@@ -49,14 +49,12 @@ jobs:
          distribution: 'temurin'
          java-version: '17'
      - name: Run build script
-        # Only build 1 64 bit target during presubmit to cut down build times during presubmit
-        # Continuous builds will build everything
        run: |
-          cd build/android && printf "y" | ./build.sh presubmit arm64-v8a
+          cd build/android && printf "y" | ./build.sh presubmit

  build-ios:
    name: build-iOS
-    runs-on: macos-14-xlarge
+    runs-on: macos-14

    steps:
      - uses: actions/checkout@v4.1.6
@@ -69,7 +67,7 @@ jobs:

  build-web:
    name: build-web
-    runs-on: ubuntu-22.04-16core
+    runs-on: macos-14

    steps:
      - uses: actions/checkout@v4.1.6
@@ -120,7 +120,7 @@ jobs:
        env:
          TAG: ${{ steps.git_ref.outputs.tag }}
        run: |
-          cd build/android && printf "y" | ./build.sh release armeabi-v7a,arm64-v8a,x86,x86_64
+          cd build/android && printf "y" | ./build.sh release
          cd ../..
          mv out/filament-android-release.aar out/filament-${TAG}-android.aar
          mv out/filamat-android-release.aar out/filamat-${TAG}-android.aar
@@ -17,7 +17,7 @@ jobs:
      - name: Run build script
        run: |
          cd build/web && printf "y" | ./build.sh continuous
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@v1.0.0
        with:
          name: filament-web
          path: out/filament-release-web.tgz
@@ -10,7 +10,7 @@ on:
 jobs:
  build-windows:
    name: build-windows
-    runs-on: windows-2019-32core
+    runs-on: windows-2019

    steps:
      - uses: actions/checkout@v4.1.6
@@ -18,7 +18,7 @@ jobs:
        run: |
          build\windows\build-github.bat continuous
        shell: cmd
-      - uses: actions/upload-artifact@v4
+      - uses: actions/upload-artifact@v1.0.0
        with:
          name: filament-windows
          path: out/filament-windows.tgz
@@ -31,7 +31,7 @@ repositories {
 }

 dependencies {
-    implementation 'com.google.android.filament:filament-android:1.54.5'
+    implementation 'com.google.android.filament:filament-android:1.53.1'
 }
 ```

@@ -51,9 +51,19 @@ Here are all the libraries available in the group `com.google.android.filament`:
 iOS projects can use CocoaPods to install the latest release:

 ```shell
-pod 'Filament', '~> 1.54.5'
+pod 'Filament', '~> 1.53.1'
 ```

+### Snapshots
+
+If you prefer to live on the edge, you can download a continuous build by following the following
+steps:
+
+1. Find the [commit](https://github.com/google/filament/commits/main) you're interested in.
+2. Click the green check mark under the commit message.
+3. Click on the _Details_ link for the platform you're interested in.
+4. On the top left click _Summary_, then in the _Artifacts_ section choose the desired artifact.
+
 ## Documentation

 - [Filament](https://google.github.io/filament/Filament.html), an in-depth explanation of
@@ -7,40 +7,6 @@ A new header is inserted each time a *tag* is created.
 Instead, if you are authoring a PR for the main branch, add your release note to
 [NEW_RELEASE_NOTES.md](./NEW_RELEASE_NOTES.md).

-## v1.54.5
-
-
-## v1.54.4
-
- Add support for multi-layered render target with array textures.
-
-## v1.54.3
-
-
-## v1.54.2
-
- Add a `name` API to Filament objects for debugging handle use-after-free assertions
-
-## v1.54.1
-
-
-## v1.54.0
-
- materials: add a new `stereoscopicType` material parameter. [⚠️ **New Material Version**]
- Fix a crash when compiling shaders on IMG devices
-
-## v1.53.5
-
- engine: Fix bug causing certain sampler parameters to not be applied correctly in GLES 2.0 and on
-  certain GLES 3.0 drivers.
-
-## v1.53.4
-
-
-## v1.53.3
-
- Add drag and drop support for IBL files for desktop gltf_viewer.
-
 ## v1.53.2


@@ -253,10 +253,16 @@ Java_com_google_android_filament_RenderableManager_nBuilderMorphingStandard(JNIE
 }

 extern "C" JNIEXPORT void JNICALL
-Java_com_google_android_filament_RenderableManager_nBuilderSetMorphTargetBufferOffsetAt(JNIEnv*, jclass,
-        jlong nativeBuilder, int level, int primitiveIndex, int offset) {
+Java_com_google_android_filament_RenderableManager_nBuilderSetMorphTargetBufferAt(JNIEnv*, jclass,
+        jlong nativeBuilder, int level, int primitiveIndex, jlong nativeMorphTargetBuffer,
+        int offset, int count) {
    RenderableManager::Builder *builder = (RenderableManager::Builder *) nativeBuilder;
-    builder->morphing(level, primitiveIndex, offset);
+    if (nativeMorphTargetBuffer) {
+        MorphTargetBuffer *morphTargetBuffer = (MorphTargetBuffer *) nativeMorphTargetBuffer;
+        builder->morphing(level, primitiveIndex, morphTargetBuffer, offset, count);
+    } else {
+        builder->morphing(level, primitiveIndex, offset, count);
+    }
 }

 extern "C" JNIEXPORT void JNICALL
@@ -328,12 +334,18 @@ Java_com_google_android_filament_RenderableManager_nSetMorphWeights(JNIEnv* env,
 }

 extern "C" JNIEXPORT void JNICALL
-Java_com_google_android_filament_RenderableManager_nSetMorphTargetBufferOffsetAt(JNIEnv*,
+Java_com_google_android_filament_RenderableManager_nSetMorphTargetBufferAt(JNIEnv*,
        jclass, jlong nativeRenderableManager, jint i, int level, jint primitiveIndex,
-        jlong, jint offset) {
+        jlong nativeMorphTargetBuffer, jint offset, jint count) {
    RenderableManager *rm = (RenderableManager *) nativeRenderableManager;
-    rm->setMorphTargetBufferOffsetAt((RenderableManager::Instance) i, (uint8_t) level,
-            (size_t) primitiveIndex, (size_t) offset);
+    if (nativeMorphTargetBuffer) {
+        MorphTargetBuffer *morphTargetBuffer = (MorphTargetBuffer *) nativeMorphTargetBuffer;
+        rm->setMorphTargetBufferAt((RenderableManager::Instance) i, (uint8_t) level,
+                (size_t) primitiveIndex, morphTargetBuffer, (size_t) offset, (size_t) count);
+    } else {
+        rm->setMorphTargetBufferAt((RenderableManager::Instance) i, (uint8_t) level,
+                (size_t) primitiveIndex, (size_t) offset, (size_t) count);
+    }
 }

 extern "C" JNIEXPORT jint JNICALL
@@ -28,14 +28,6 @@
 using namespace filament;
 using namespace backend;

-
-extern "C" JNIEXPORT void JNICALL
-Java_com_google_android_filament_Renderer_nSkipFrame(JNIEnv *, jclass, jlong nativeRenderer,
-        jlong vsyncSteadyClockTimeNano) {
-    Renderer *renderer = (Renderer *) nativeRenderer;
-    renderer->skipFrame(uint64_t(vsyncSteadyClockTimeNano));
-}
-
 extern "C" JNIEXPORT jboolean JNICALL
 Java_com_google_android_filament_Renderer_nBeginFrame(JNIEnv *, jclass, jlong nativeRenderer,
        jlong nativeSwapChain, jlong frameTimeNanos) {
@@ -531,12 +531,3 @@ Java_com_google_android_filament_View_nGetFogEntity(JNIEnv *env, jclass clazz,
    View *view = (View *) nativeView;
    return (jint)view->getFogEntity().getId();
 }
-
-extern "C"
-JNIEXPORT void JNICALL
-Java_com_google_android_filament_View_nClearFrameHistory(JNIEnv *env, jclass clazz,
-        jlong nativeView, jlong nativeEngine) {
-    View *view = (View *) nativeView;
-    Engine *engine = (Engine *) nativeEngine;
-    view->clearFrameHistory(*engine);
-}
@@ -425,12 +425,9 @@ public class Engine {
        public long resourceAllocatorCacheSizeMB = 64;

        /*
-         * This value determines how many frames texture entries are kept for in the cache. This
-         * is a soft limit, meaning some texture older than this are allowed to stay in the cache.
-         * Typically only one texture is evicted per frame.
-         * The default is 1.
+         * This value determines for how many frames are texture entries kept in the cache.
         */
-        public long resourceAllocatorCacheMaxAge = 1;
+        public long resourceAllocatorCacheMaxAge = 2;

        /*
         * Disable backend handles use-after-free checks.
@@ -74,7 +74,7 @@ public class MorphTargetBuffer {
         *
         * @exception IllegalStateException if the MorphTargetBuffer could not be created
         *
-         * @see #setMorphTargetBufferOffsetAt
+         * @see #setMorphTargetBufferAt
         */
        @NonNull
        public MorphTargetBuffer build(@NonNull Engine engine) {
@@ -569,13 +569,40 @@ public class RenderableManager {
         *
         * @param level the level of detail (lod), only 0 can be specified
         * @param primitiveIndex zero-based index of the primitive, must be less than the count passed to Builder constructor
+         * @param morphTargetBuffer specifies the morph target buffer
         * @param offset specifies where in the morph target buffer to start reading (expressed as a number of vertices)
+         * @param count number of vertices in the morph target buffer to read, must equal the geometry's count (for triangles, this should be a multiple of 3)
         */
        @NonNull
        public Builder morphing(@IntRange(from = 0) int level,
                                @IntRange(from = 0) int primitiveIndex,
-                                @IntRange(from = 0) int offset) {
-            nBuilderSetMorphTargetBufferOffsetAt(mNativeBuilder, level, primitiveIndex, offset);
+                                @IntRange(from = 0) int offset,
+                                @IntRange(from = 0) int count) {
+            nBuilderSetMorphTargetBufferAt(mNativeBuilder, level, primitiveIndex, 0, offset, count);
+            return this;
+        }
+
+        /** @deprecated */
+        @Deprecated
+        @NonNull
+        public Builder morphing(@IntRange(from = 0) int level,
+                                @IntRange(from = 0) int primitiveIndex,
+                                @NonNull MorphTargetBuffer morphTargetBuffer,
+                                @IntRange(from = 0) int offset,
+                                @IntRange(from = 0) int count) {
+            nBuilderSetMorphTargetBufferAt(mNativeBuilder, level, primitiveIndex,
+                    morphTargetBuffer.getNativeObject(), offset, count);
+            return this;
+        }
+
+        /** @deprecated */
+        @Deprecated
+        @NonNull
+        public Builder morphing(@IntRange(from = 0) int level,
+                                @IntRange(from = 0) int primitiveIndex,
+                                @NonNull MorphTargetBuffer morphTargetBuffer) {
+            nBuilderSetMorphTargetBufferAt(mNativeBuilder, level, primitiveIndex,
+                    morphTargetBuffer.getNativeObject(), 0, morphTargetBuffer.getVertexCount());
            return this;
        }

@@ -678,11 +705,34 @@ public class RenderableManager {
     *
     * @see Builder#morphing
     */
-    public void setMorphTargetBufferOffsetAt(@EntityInstance int i,
+    public void setMorphTargetBufferAt(@EntityInstance int i,
                                       @IntRange(from = 0) int level,
                                       @IntRange(from = 0) int primitiveIndex,
-                                       @IntRange(from = 0) int offset) {
-        nSetMorphTargetBufferOffsetAt(mNativeObject, i, level, primitiveIndex, 0, offset);
+                                       @IntRange(from = 0) int offset,
+                                       @IntRange(from = 0) int count) {
+        nSetMorphTargetBufferAt(mNativeObject, i, level, primitiveIndex, 0, offset, count);
+    }
+
+    /** @deprecated */
+    @Deprecated
+    public void setMorphTargetBufferAt(@EntityInstance int i,
+                                       @IntRange(from = 0) int level,
+                                       @IntRange(from = 0) int primitiveIndex,
+                                       @NonNull MorphTargetBuffer morphTargetBuffer,
+                                       @IntRange(from = 0) int offset,
+                                       @IntRange(from = 0) int count) {
+        nSetMorphTargetBufferAt(mNativeObject, i, level, primitiveIndex,
+                morphTargetBuffer.getNativeObject(), offset, count);
+    }
+
+    /** @deprecated */
+    @Deprecated
+    public void setMorphTargetBufferAt(@EntityInstance int i,
+                                       @IntRange(from = 0) int level,
+                                       @IntRange(from = 0) int primitiveIndex,
+                                       @NonNull MorphTargetBuffer morphTargetBuffer) {
+        nSetMorphTargetBufferAt(mNativeObject, i, level, primitiveIndex,
+                morphTargetBuffer.getNativeObject(), 0, morphTargetBuffer.getVertexCount());
    }

    /**
@@ -983,7 +1033,7 @@ public class RenderableManager {
    private static native void nBuilderSkinningBuffer(long nativeBuilder, long nativeSkinningBuffer, int boneCount, int offset);
    private static native void nBuilderMorphing(long nativeBuilder, int targetCount);
    private static native void nBuilderMorphingStandard(long nativeBuilder, long nativeMorphTargetBuffer);
-    private static native void nBuilderSetMorphTargetBufferOffsetAt(long nativeBuilder, int level, int primitiveIndex, int offset);
+    private static native void nBuilderSetMorphTargetBufferAt(long nativeBuilder, int level, int primitiveIndex, long nativeMorphTargetBuffer, int offset, int count);
    private static native void nBuilderEnableSkinningBuffers(long nativeBuilder, boolean enabled);
    private static native void nBuilderFog(long nativeBuilder, boolean enabled);
    private static native void nBuilderLightChannel(long nativeRenderableManager, int channel, boolean enable);
@@ -993,7 +1043,7 @@ public class RenderableManager {
    private static native int nSetBonesAsMatrices(long nativeObject, int i, Buffer matrices, int remaining, int boneCount, int offset);
    private static native int nSetBonesAsQuaternions(long nativeObject, int i, Buffer quaternions, int remaining, int boneCount, int offset);
    private static native void nSetMorphWeights(long nativeObject, int instance, float[] weights, int offset);
-    private static native void nSetMorphTargetBufferOffsetAt(long nativeObject, int i, int level, int primitiveIndex, long nativeMorphTargetBuffer, int offset);
+    private static native void nSetMorphTargetBufferAt(long nativeObject, int i, int level, int primitiveIndex, long nativeMorphTargetBuffer, int offset, int count);
    private static native int nGetMorphTargetCount(long nativeObject, int i);
    private static native void nSetAxisAlignedBoundingBox(long nativeRenderableManager, int i, float cx, float cy, float cz, float ex, float ey, float ez);
    private static native void nSetLayerMask(long nativeRenderableManager, int i, int select, int value);
@@ -297,20 +297,6 @@ public class Renderer {
        nSetVsyncTime(getNativeObject(), steadyClockTimeNano);
    }

-    /**
-     * Call skipFrame when momentarily skipping frames, for instance if the content of the
-     * scene doesn't change.
-     *
-     * @param vsyncSteadyClockTimeNano The time in nanoseconds when the frame started being rendered,
-     *                       in the {@link System#nanoTime()} timebase. Divide this value by 1000000 to
-     *                       convert it to the {@link android.os.SystemClock#uptimeMillis()}
-     *                       time base. This typically comes from
-     *                       {@link android.view.Choreographer.FrameCallback}.
-     */
-    public void skipFrame(long vsyncSteadyClockTimeNano) {
-        nSkipFrame(getNativeObject(), vsyncSteadyClockTimeNano);
-    }
-
    /**
     * Sets up a frame for this <code>Renderer</code>.
     * <p><code>beginFrame</code> manages frame pacing, and returns whether or not a frame should be
@@ -730,7 +716,6 @@ public class Renderer {

    private static native void nSetPresentationTime(long nativeObject, long monotonicClockNanos);
    private static native void nSetVsyncTime(long nativeObject, long steadyClockTimeNano);
-    private static native void nSkipFrame(long nativeObject, long vsyncSteadyClockTimeNano);
    private static native boolean nBeginFrame(long nativeRenderer, long nativeSwapChain, long frameTimeNanos);
    private static native void nEndFrame(long nativeRenderer);
    private static native void nRender(long nativeRenderer, long nativeView);
@@ -1233,18 +1233,6 @@ public class View {
        return nGetFogEntity(getNativeObject());
    }

-    /**
-     * When certain temporal features are used (e.g.: TAA or Screen-space reflections), the view
-     * keeps a history of previous frame renders associated with the Renderer the view was last
-     * used with. When switching Renderer, it may be necessary to clear that history by calling
-     * this method. Similarly, if the whole content of the screen change, like when a cut-scene
-     * starts, clearing the history might be needed to avoid artifacts due to the previous frame
-     * being very different.
-     */
-    public void clearFrameHistory(Engine engine) {
-        nClearFrameHistory(getNativeObject(), engine.getNativeObject());
-    }
-
    public long getNativeObject() {
        if (mNativeObject == 0) {
            throw new IllegalStateException("Calling method on destroyed View");
@@ -1306,7 +1294,7 @@ public class View {
    private static native void nSetMaterialGlobal(long nativeView, int index, float x, float y, float z, float w);
    private static native void nGetMaterialGlobal(long nativeView, int index, float[] out);
    private static native int nGetFogEntity(long nativeView);
-    private static native void nClearFrameHistory(long nativeView, long nativeEngine);
+

    /**
     * List of available ambient occlusion techniques.
@@ -125,11 +125,6 @@ extern "C" JNIEXPORT void Java_com_google_android_filament_utils_Manipulator_nBu
    builder->groundPlane(a, b, c, d);
 }

-extern "C" JNIEXPORT void Java_com_google_android_filament_utils_Manipulator_nBuilderPanning(JNIEnv*, jclass, jlong nativeBuilder, jboolean enabled) {
-    Builder* builder = (Builder*) nativeBuilder;
-    builder->panning(enabled);
-}
-
 extern "C" JNIEXPORT long Java_com_google_android_filament_utils_Manipulator_nBuilderBuild(JNIEnv*, jclass, jlong nativeBuilder, jint mode) {
    Builder* builder = (Builder*) nativeBuilder;
    return (jlong) builder->build((Mode) mode);
@@ -274,17 +274,6 @@ public class Manipulator {
            return this;
        }

-        /**
-         * Sets whether panning is enabled in the manipulator.
-         *
-         * @return this <code>Builder</code> object for chaining calls
-         */
-        @NonNull
-        public Builder panning(Boolean enabled) {
-            nBuilderPanning(mNativeBuilder, enabled);
-            return this;
-        }
-
        /**
         * Creates and returns the <code>Manipulator</code> object.
         *
@@ -494,7 +483,6 @@ public class Manipulator {
    private static native void nBuilderFlightPanSpeed(long nativeBuilder, float x, float y);
    private static native void nBuilderFlightMoveDamping(long nativeBuilder, float damping);
    private static native void nBuilderGroundPlane(long nativeBuilder, float a, float b, float c, float d);
-    private static native void nBuilderPanning(long nativeBuilder, Boolean enabled);
    private static native long nBuilderBuild(long nativeBuilder, int mode);

    private static native void nDestroyManipulator(long nativeManip);
@@ -1,5 +1,5 @@
 GROUP=com.google.android.filament
-VERSION_NAME=1.54.5
+VERSION_NAME=1.53.1

 POM_DESCRIPTION=Real-time physically based rendering engine for Android.

@@ -60,7 +60,13 @@ if [[ ! -d "${ANDROID_HOME}/ndk/$FILAMENT_NDK_VERSION" ]]; then
    yes | ${ANDROID_HOME}/cmdline-tools/latest/bin/sdkmanager --licenses
    ${ANDROID_HOME}/cmdline-tools/latest/bin/sdkmanager "ndk;$FILAMENT_NDK_VERSION"
 fi
+
+# Only build 1 64 bit target during presubmit to cut down build times during presubmit
+# Continuous builds will build everything
 ANDROID_ABIS=
+if [[ "$TARGET" == "presubmit" ]]; then
+  ANDROID_ABIS="-q arm64-v8a"
+fi

 # Build the Android sample-gltf-viewer APK during release.
 BUILD_SAMPLES=
@@ -68,19 +74,5 @@ if [[ "$TARGET" == "release" ]]; then
    BUILD_SAMPLES="-k sample-gltf-viewer"
 fi

-function build_android() {
-    local ABI=$1
-
-    # Do the following in two steps so that we do not run out of space
-    if [[ -n "${BUILD_DEBUG}" ]]; then
-        FILAMENT_NDK_VERSION=${FILAMENT_NDK_VERSION} ./build.sh -p android -q ${ABI} -c ${BUILD_SAMPLES} ${GENERATE_ARCHIVES} ${BUILD_DEBUG}
-        rm -rf out/cmake-android-debug-*
-    fi
-    if [[ -n "${BUILD_RELEASE}" ]]; then
-        FILAMENT_NDK_VERSION=${FILAMENT_NDK_VERSION} ./build.sh -p android -q ${ABI} -c ${BUILD_SAMPLES} ${GENERATE_ARCHIVES} ${BUILD_RELEASE}
-        rm -rf out/cmake-android-release-*
-    fi
-}
-
 pushd `dirname $0`/../.. > /dev/null
-build_android $2
+FILAMENT_NDK_VERSION=${FILAMENT_NDK_VERSION} ./build.sh -p android $ANDROID_ABIS -c $BUILD_SAMPLES $GENERATE_ARCHIVES $BUILD_DEBUG $BUILD_RELEASE
@@ -1,6 +1,7 @@
 #!/bin/bash
 if [ `uname` == "Linux" ];then
-    source `dirname $0`/../linux/ci-common.sh
+    curl -OL https://github.com/ninja-build/ninja/releases/download/v1.10.2/ninja-linux.zip
+    unzip -q ninja-linux.zip
 elif [ `uname` == "Darwin" ];then
    curl -OL https://github.com/ninja-build/ninja/releases/download/v1.10.2/ninja-mac.zip
    unzip -q ninja-mac.zip
@@ -12,6 +13,9 @@ fi
 chmod +x ninja
 export PATH="$PWD:$PATH"

+# FIXME: kokoro machines have node and npm but currently they are symlinked to non-existent files
+# npm install -g typescript
+
 # Install emscripten.
 curl -L https://github.com/emscripten-core/emsdk/archive/refs/tags/3.1.15.zip > emsdk.zip
 unzip emsdk.zip ; mv emsdk-* emsdk ; cd emsdk
@@ -1308,12 +1308,7 @@ Description
      declare a variable called `eyeDirection` you can access it in the fragment shader using
      `variable_eyeDirection`. In the vertex shader, the interpolant name is simply a member of
      the `MaterialVertexInputs` structure (`material.eyeDirection` in your example). Each
-      interpolant is of type `float4` (`vec4`) in the shaders. By default the precision of the
-      interpolant is `highp` in *both* the vertex and fragment shaders.
-      An alternate syntax can be used to specify both the name and precision of the interpolant.
-      In this case the specified precision is used as-is in both fragment and vertex stages, in
-      particular if `default` is specified the default precision is used is the fragment shader
-      (`mediump`) and in the vertex shader (`highp`).
+      interpolant is of type `float4` (`vec4`) in the shaders.

 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ JSON
 material {
@@ -1325,11 +1320,7 @@ material {
        }
    ],
    variables : [
-         eyeDirection,
-         {
-            name : eyeColor,
-            precision : medium
-         }
+         eyeDirection
    ],
    vertexDomain : device,
    depthWrite : false,
@@ -61,7 +61,6 @@ set(SRCS
        src/Engine.cpp
        src/Exposure.cpp
        src/Fence.cpp
-        src/FilamentBuilder.cpp
        src/FrameInfo.cpp
        src/FrameSkipper.cpp
        src/Froxelizer.cpp
@@ -1058,7 +1058,7 @@ struct RasterState {
            bool inverseFrontFaces                      : 1;        // 31

            //! padding, must be 0
-            bool depthClamp                             : 1;        // 32
+            uint8_t padding                             : 1;        // 32
        };
        uint32_t u = 0;
    };
@@ -1244,7 +1244,7 @@ enum class Workaround : uint16_t {
    ADRENO_UNIFORM_ARRAY_CRASH,
    // Workaround a Metal pipeline compilation error with the message:
    // "Could not statically determine the target of a texture". See light_indirect.fs
-    METAL_STATIC_TEXTURE_TARGET_ERROR,
+    A8X_STATIC_TEXTURE_TARGET_ERROR,
    // Adreno drivers sometimes aren't able to blit into a layer of a texture array.
    DISABLE_BLIT_INTO_TEXTURE_ARRAY,
    // Multiple workarounds needed for PowerVR GPUs
@@ -162,10 +162,6 @@ DECL_DRIVER_API_0(finish)
 // reset state tracking, if the driver does any state tracking (e.g. GL)
 DECL_DRIVER_API_0(resetState)

-DECL_DRIVER_API_N(setDebugTag,
-        backend::HandleBase::HandleId, handleId,
-        utils::CString, tag)
-
 /*
 * Creating driver objects
 * -----------------------
@@ -309,7 +305,6 @@ DECL_DRIVER_API_SYNCHRONOUS_0(bool, isParallelShaderCompileSupported)
 DECL_DRIVER_API_SYNCHRONOUS_0(bool, isDepthStencilResolveSupported)
 DECL_DRIVER_API_SYNCHRONOUS_N(bool, isDepthStencilBlitSupported, backend::TextureFormat, format)
 DECL_DRIVER_API_SYNCHRONOUS_0(bool, isProtectedTexturesSupported)
-DECL_DRIVER_API_SYNCHRONOUS_0(bool, isDepthClampSupported)
 DECL_DRIVER_API_SYNCHRONOUS_0(uint8_t, getMaxDrawBuffers)
 DECL_DRIVER_API_SYNCHRONOUS_0(size_t, getMaxUniformBufferSize)
 DECL_DRIVER_API_SYNCHRONOUS_0(math::float2, getClipSpaceParams)
@@ -444,10 +439,12 @@ DECL_DRIVER_API_N(setPushConstant,
        backend::PushConstantVariant, value)

 DECL_DRIVER_API_N(insertEventMarker,
-        const char*, string)
+        const char*, string,
+        uint32_t, len = 0)

 DECL_DRIVER_API_N(pushGroupMarker,
-        const char*, string)
+        const char*, string,
+        uint32_t, len = 0)

 DECL_DRIVER_API_0(popGroupMarker)

@@ -20,12 +20,11 @@
 #include <backend/Handle.h>

 #include <utils/Allocator.h>
-#include <utils/CString.h>
 #include <utils/Log.h>
-#include <utils/Panic.h>
 #include <utils/compiler.h>
 #include <utils/debug.h>
 #include <utils/ostream.h>
+#include <utils/Panic.h>

 #include <tsl/robin_map.h>

@@ -40,7 +39,7 @@

 #define HandleAllocatorGL  HandleAllocator<32,  64, 136>    // ~4520 / pool / MiB
 #define HandleAllocatorVK  HandleAllocator<64, 160, 312>    // ~1820 / pool / MiB
-#define HandleAllocatorMTL HandleAllocator<32,  64, 552>    // ~1660 / pool / MiB
+#define HandleAllocatorMTL HandleAllocator<32,  48, 552>    // ~1660 / pool / MiB

 namespace filament::backend {

@@ -174,10 +173,8 @@ public:
                uint8_t const age = (tag & HANDLE_AGE_MASK) >> HANDLE_AGE_SHIFT;
                auto const pNode = static_cast<typename Allocator::Node*>(p);
                uint8_t const expectedAge = pNode[-1].age;
-                // getHandleTag() is only called if the check fails.
-                FILAMENT_CHECK_POSTCONDITION(expectedAge == age)
-                        << "use-after-free of Handle with id=" << handle.getId()
-                        << ", tag=" << getHandleTag(handle.getId()).c_str_safe();
+                FILAMENT_CHECK_POSTCONDITION(expectedAge == age) <<
+                        "use-after-free of Handle with id=" << handle.getId();
            }
        }

@@ -204,29 +201,6 @@ public:
        return handle_cast<Dp>(const_cast<Handle<B>&>(handle));
    }

-    void associateTagToHandle(HandleBase::HandleId id, utils::CString&& tag) noexcept {
-        // TODO: for now, only pool handles check for use-after-free, so we only keep tags for
-        // those
-        if (isPoolHandle(id)) {
-            // Truncate the age to get the debug tag
-            uint32_t const key = id & ~(HANDLE_DEBUG_TAG_MASK ^ HANDLE_AGE_MASK);
-            // This line is the costly part. In the future, we could potentially use a custom
-            // allocator.
-            mDebugTags[key] = std::move(tag);
-        }
-    }
-
-    utils::CString getHandleTag(HandleBase::HandleId id) const noexcept {
-        if (!isPoolHandle(id)) {
-            return "(no tag)";
-        }
-        uint32_t const key = id & ~(HANDLE_DEBUG_TAG_MASK ^ HANDLE_AGE_MASK);
-        if (auto pos = mDebugTags.find(key); pos != mDebugTags.end()) {
-            return pos->second;
-        }
-        return "(no tag)";
-    }
-
 private:

    template<typename D>
@@ -344,24 +318,12 @@ private:
        }
    }

-    // number if bits allotted to the handle's age (currently 4 max)
-    static constexpr uint32_t HANDLE_AGE_BIT_COUNT = 4;
-    // number if bits allotted to the handle's debug tag (HANDLE_AGE_BIT_COUNT max)
-    static constexpr uint32_t HANDLE_DEBUG_TAG_BIT_COUNT = 2;
-    // bit shift for both the age and debug tag
-    static constexpr uint32_t HANDLE_AGE_SHIFT = 27;
-    // mask for the heap (vs pool) flag
-    static constexpr uint32_t HANDLE_HEAP_FLAG = 0x80000000u;
-    // mask for the age
-    static constexpr uint32_t HANDLE_AGE_MASK =
-            ((1 << HANDLE_AGE_BIT_COUNT) - 1) << HANDLE_AGE_SHIFT;
-    // mask for the debug tag
-    static constexpr uint32_t HANDLE_DEBUG_TAG_MASK =
-            ((1 << HANDLE_DEBUG_TAG_BIT_COUNT) - 1) << HANDLE_AGE_SHIFT;
-    // mask for the index
-    static constexpr uint32_t HANDLE_INDEX_MASK = 0x07FFFFFFu;
-
-    static_assert(HANDLE_DEBUG_TAG_BIT_COUNT <= HANDLE_AGE_BIT_COUNT);
+    // we handle a 4 bits age per address
+    static constexpr uint32_t HANDLE_HEAP_FLAG      = 0x80000000u;      // pool vs heap handle
+    static constexpr uint32_t HANDLE_AGE_MASK       = 0x78000000u;      // handle's age
+    static constexpr uint32_t HANDLE_INDEX_MASK     = 0x07FFFFFFu;      // handle index
+    static constexpr uint32_t HANDLE_TAG_MASK       = HANDLE_AGE_MASK;
+    static constexpr uint32_t HANDLE_AGE_SHIFT      = 27;

    static bool isPoolHandle(HandleBase::HandleId id) noexcept {
        return (id & HANDLE_HEAP_FLAG) == 0u;
@@ -376,7 +338,7 @@ private:
        // a non-pool handle.
        if (UTILS_LIKELY(isPoolHandle(id))) {
            char* const base = (char*)mHandleArena.getArea().begin();
-            uint32_t const tag = id & HANDLE_AGE_MASK;
+            uint32_t const tag = id & HANDLE_TAG_MASK;
            size_t const offset = (id & HANDLE_INDEX_MASK) * Allocator::getAlignment();
            return { static_cast<void*>(base + offset), tag };
        }
@@ -391,7 +353,7 @@ private:
        size_t const offset = (char*)p - base;
        assert_invariant((offset % Allocator::getAlignment()) == 0);
        auto id = HandleBase::HandleId(offset / Allocator::getAlignment());
-        id |= tag & HANDLE_AGE_MASK;
+        id |= tag & HANDLE_TAG_MASK;
        assert_invariant((id & HANDLE_HEAP_FLAG) == 0);
        return id;
    }
@@ -401,7 +363,6 @@ private:
    // Below is only used when running out of space in the HandleArena
    mutable utils::Mutex mLock;
    tsl::robin_map<HandleBase::HandleId, void*> mOverflowMap;
-    tsl::robin_map<HandleBase::HandleId, utils::CString> mDebugTags;
    HandleBase::HandleId mId = 0;
    bool mUseAfterFreeCheckDisabled = false;
 };
@@ -24,7 +24,7 @@
 #include <utils/debug.h>
 #include <utils/ostream.h>

-#if !defined(WIN32) && !defined(__EMSCRIPTEN__)
+#if !defined(WIN32) && !defined(__EMSCRIPTEN__) && !defined(IOS)
 #    include <sys/mman.h>
 #    include <unistd.h>
 #    define HAS_MMAP 1
@@ -32,11 +32,10 @@
 #    define HAS_MMAP 0
 #endif

-#include <stddef.h>
 #include <stdint.h>
-#include <stdio.h>
+#include <stddef.h>
 #include <stdlib.h>
-#include <string.h>
+#include <stdio.h>

 using namespace utils;

@@ -82,9 +81,6 @@ void* CircularBuffer::alloc(size_t size) noexcept {
            // map the circular buffer once...
            vaddr = mmap(reserve_vaddr, size, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
            if (vaddr != MAP_FAILED) {
-                // populate the address space with pages (because this is a circular buffer,
-                // all the pages will be allocated eventually, might as well do it now)
-                memset(vaddr, 0, size);
                // and map the circular buffer again, behind the previous copy...
                vaddr_shadow = mmap((char*)vaddr + size, size,
                        PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
@@ -105,7 +101,7 @@ void* CircularBuffer::alloc(size_t size) noexcept {
    if (UTILS_UNLIKELY(mAshmemFd < 0)) {
        // ashmem failed
        if (vaddr_guard != MAP_FAILED) {
-            munmap(vaddr_guard, BLOCK_SIZE);
+            munmap(vaddr_guard, size);
        }

        if (vaddr_shadow != MAP_FAILED) {
@@ -123,11 +119,12 @@ void* CircularBuffer::alloc(size_t size) noexcept {
        data = mmap(nullptr, size * 2 + BLOCK_SIZE,
                PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);

-        FILAMENT_CHECK_POSTCONDITION(data != MAP_FAILED) <<
+        FILAMENT_CHECK_POSTCONDITION(data) <<
                "couldn't allocate " << (size * 2 / 1024) <<
                " KiB of virtual address space for the command buffer";

-        slog.w << "Using 'soft' CircularBuffer (" << (size * 2 / 1024) << " KiB)" << io::endl;
+        slog.d << "WARNING: Using soft CircularBuffer (" << (size * 2 / 1024) << " KiB)"
+               << io::endl;

        // guard page at the end
        void* guard = (void*)(uintptr_t(data) + size * 2);
@@ -101,8 +101,9 @@ void CommandBufferQueue::flush() noexcept {
    size_t const used = std::distance(
            static_cast<char const*>(begin), static_cast<char const*>(end));

-
    std::unique_lock<utils::Mutex> lock(mLock);
+    mCommandBuffersToExecute.push_back({ begin, end });
+    mCondition.notify_one();

    // circular buffer is too small, we corrupted the stream
    FILAMENT_CHECK_POSTCONDITION(used <= mFreeSpace) <<
@@ -111,13 +112,11 @@ void CommandBufferQueue::flush() noexcept {
            "Space used at this time: " << used <<
            " bytes, overflow: " << used - mFreeSpace << " bytes";

-    mFreeSpace -= used;
-    mCommandBuffersToExecute.push_back({ begin, end });
-    mCondition.notify_one();
-
    // wait until there is enough space in the buffer
+    mFreeSpace -= used;
    if (UTILS_UNLIKELY(mFreeSpace < requiredSize)) {

+
 #ifndef NDEBUG
        size_t const totalUsed = circularBuffer.size() - mFreeSpace;
        slog.d << "CommandStream used too much space (will block): "
@@ -154,10 +153,8 @@ std::vector<CommandBufferQueue::Range> CommandBufferQueue::waitForCommands() con
 }

 void CommandBufferQueue::releaseBuffer(CommandBufferQueue::Range const& buffer) {
-    size_t const used = std::distance(
-            static_cast<char const*>(buffer.begin), static_cast<char const*>(buffer.end));
    std::lock_guard<utils::Mutex> const lock(mLock);
-    mFreeSpace += used;
+    mFreeSpace += uintptr_t(buffer.end) - uintptr_t(buffer.begin);
    mCondition.notify_one();
 }

@@ -20,16 +20,11 @@
 #include <utils/CallStack.h>
 #endif

-#include <utils/compiler.h>
 #include <utils/Log.h>
-#include <utils/ostream.h>
 #include <utils/Profiler.h>
 #include <utils/Systrace.h>

-#include <cstddef>
 #include <functional>
-#include <string>
-#include <utility>

 #ifdef __ANDROID__
 #include <sys/system_properties.h>
@@ -79,8 +74,8 @@ CommandStream::CommandStream(Driver& driver, CircularBuffer& buffer) noexcept
 }

 void CommandStream::execute(void* buffer) {
-    // NOTE: we can't use SYSTRACE_CALL() or similar here because, execute() below, also
-    // uses systrace BEGIN/END and the END is not guaranteed to be happening in this scope.
+    SYSTRACE_CALL();
+    SYSTRACE_CONTEXT();

    Profiler profiler;

@@ -105,7 +100,6 @@ void CommandStream::execute(void* buffer) {
            // we want to remove all this when tracing is completely disabled
            profiler.stop();
            UTILS_UNUSED Profiler::Counters const counters = profiler.readCounters();
-            SYSTRACE_CONTEXT();
            SYSTRACE_VALUE32("GLThread (I)", counters.getInstructions());
            SYSTRACE_VALUE32("GLThread (C)", counters.getCpuCycles());
            SYSTRACE_VALUE32("GLThread (CPI x10)", counters.getCPI() * 10);
@@ -80,9 +80,6 @@ HandleAllocator<P0, P1, P2>::HandleAllocator(const char* name, size_t size,
        bool disableUseAfterFreeCheck) noexcept
    : mHandleArena(name, size, disableUseAfterFreeCheck),
      mUseAfterFreeCheckDisabled(disableUseAfterFreeCheck) {
-    // Reserve initial space for debug tags. This prevents excessive calls to malloc when the first
-    // few tags are set.
-    mDebugTags.reserve(512);
 }

 template <size_t P0, size_t P1, size_t P2>
@@ -160,8 +160,6 @@ public:
         size_t size, bool forceGpuBuffer = false);
    ~MetalBuffer();

-    [[nodiscard]] bool wasAllocationSuccessful() const noexcept { return mBuffer || mCpuBuffer; }
-
    MetalBuffer(const MetalBuffer& rhs) = delete;
    MetalBuffer& operator=(const MetalBuffer& rhs) = delete;

@@ -53,8 +53,8 @@ MetalBuffer::MetalBuffer(MetalContext& context, BufferObjectBinding bindingType,
        mBuffer = { [context.device newBufferWithLength:size options:MTLResourceStorageModePrivate],
            TrackedMetalBuffer::Type::GENERIC };
    }
-    // mBuffer might fail to be allocated. Clients can check for this by calling
-    // wasAllocationSuccessful().
+    FILAMENT_CHECK_POSTCONDITION(mBuffer)
+            << "Could not allocate Metal buffer of size " << size << ".";
 }

 MetalBuffer::~MetalBuffer() {
@@ -145,7 +145,6 @@ std::pair<id<MTLBuffer>, size_t> MetalBumpAllocator::allocateStagingArea(size_t
    // We're finished with the current allocation.
    mCurrentUploadBuffer = { [mDevice newBufferWithLength:mCapacity options:MTLStorageModeShared],
        TrackedMetalBuffer::Type::STAGING };
-    mHead = size;

    return { mCurrentUploadBuffer.get(), 0 };
 }
@@ -93,7 +93,7 @@ struct MetalContext {
    } highestSupportedGpuFamily;

    struct {
-        bool staticTextureTargetError;
+        bool a8xStaticTextureTargetError;
    } bugs;

    // sampleCountLookup[requestedSamples] gives a <= sample count supported by the device.
@@ -112,7 +112,6 @@ struct MetalContext {
    std::array<BufferState, MAX_SSBO_COUNT> ssboState;
    CullModeStateTracker cullModeState;
    WindingStateTracker windingState;
-    DepthClampStateTracker depthClampState;
    Handle<HwRenderPrimitive> currentRenderPrimitive;

    // State caches.
@@ -57,6 +57,7 @@ class MetalDriver final : public DriverBase {

 public:
    static Driver* create(MetalPlatform* platform, const Platform::DriverConfig& driverConfig);
+    void runAtNextTick(const std::function<void()>& fn) noexcept;

 private:

@@ -72,12 +73,10 @@ private:

    /*
     * Tasks run regularly on the driver thread.
-     * Not thread-safe; tasks are run from the driver thead and must be enqueued from the driver
-     * thread.
     */
-    void runAtNextTick(const std::function<void()>& fn) noexcept;
    void executeTickOps() noexcept;
    std::vector<std::function<void()>> mTickOps;
+    std::mutex mTickOpsLock;

    /*
     * Driver interface
@@ -53,14 +53,14 @@ Driver* MetalDriverFactory::create(MetalPlatform* const platform, const Platform
    //    MetalRenderPrimitive         :  24       many
    //    MetalVertexBuffer            :  32       moderate
    // -- less than or equal 32 bytes
+    //    MetalIndexBuffer             :  40       moderate
    //    MetalFence                   :  48       few
-    //    MetalIndexBuffer             :  56       moderate
-    //    MetalBufferObject            :  64       many
-    // -- less than or equal 64 bytes
+    //    MetalBufferObject            :  48       many
+    // -- less than or equal 48 bytes
    //    MetalSamplerGroup            : 112       few
    //    MetalProgram                 : 152       moderate
    //    MetalTexture                 : 152       moderate
-    //    MetalSwapChain               : 208       few
+    //    MetalSwapChain               : 184       few
    //    MetalRenderTarget            : 272       few
    //    MetalVertexBufferInfo        : 552       moderate
    // -- less than or equal to 552 bytes
@@ -162,10 +162,8 @@ MetalDriver::MetalDriver(MetalPlatform* platform, const Platform::DriverConfig&
        sc[s] = [mContext->device supportsTextureSampleCount:s] ? s : sc[s - 1];
    }

-    mContext->bugs.staticTextureTargetError =
-            [mContext->device.name containsString:@"Apple A8X GPU"] ||
-            [mContext->device.name containsString:@"Apple A8 GPU"] ||
-            [mContext->device.name containsString:@"Apple A7 GPU"];
+    mContext->bugs.a8xStaticTextureTargetError =
+            [mContext->device.name containsString:@"Apple A8X GPU"];

    mContext->commandQueue = mPlatform.createCommandQueue(mContext->device);
    mContext->pipelineStateCache.setDevice(mContext->device);
@@ -320,40 +318,17 @@ void MetalDriver::createVertexBufferR(Handle<HwVertexBuffer> vbh,
        uint32_t vertexCount, Handle<HwVertexBufferInfo> vbih) {
    MetalVertexBufferInfo const* const vbi = handle_cast<const MetalVertexBufferInfo>(vbih);
    construct_handle<MetalVertexBuffer>(vbh, *mContext, vertexCount, vbi->bufferCount, vbih);
-    // No actual GPU memory is allocated here, so no need to check for allocation success.
 }

 void MetalDriver::createIndexBufferR(Handle<HwIndexBuffer> ibh, ElementType elementType,
        uint32_t indexCount, BufferUsage usage) {
-    auto elementSize = (uint8_t)getElementTypeSize(elementType);
-    auto* indexBuffer =
-            construct_handle<MetalIndexBuffer>(ibh, *mContext, usage, elementSize, indexCount);
-    auto& buffer = indexBuffer->buffer;
-    // If the allocation was not successful, postpone the error message until the next tick, to give
-    // Filament a chance to call setDebugTag on the handle; this way we get a nicer error message.
-    if (UTILS_UNLIKELY(!buffer.wasAllocationSuccessful())) {
-        const size_t byteCount = buffer.getSize();
-        runAtNextTick([byteCount, this, ibh]() {
-            FILAMENT_CHECK_POSTCONDITION(false)
-                    << "Could not allocate Metal index buffer of size " << byteCount
-                    << ", tag=" << mHandleAllocator.getHandleTag(ibh.getId()).c_str_safe();
-        });
-    }
+    auto elementSize = (uint8_t) getElementTypeSize(elementType);
+    construct_handle<MetalIndexBuffer>(ibh, *mContext, usage, elementSize, indexCount);
 }

 void MetalDriver::createBufferObjectR(Handle<HwBufferObject> boh, uint32_t byteCount,
        BufferObjectBinding bindingType, BufferUsage usage) {
-    auto* bufferObject =
-            construct_handle<MetalBufferObject>(boh, *mContext, bindingType, usage, byteCount);
-    // If the allocation was not successful, postpone the error message until the next tick, to give
-    // Filament a chance to call setDebugTag on the handle; this way we get a nicer error message.
-    if (UTILS_UNLIKELY(!bufferObject->getBuffer()->wasAllocationSuccessful())) {
-        runAtNextTick([byteCount, this, boh]() {
-            FILAMENT_CHECK_POSTCONDITION(false)
-                    << "Could not allocate Metal buffer of size " << byteCount
-                    << ", tag=" << mHandleAllocator.getHandleTag(boh.getId()).c_str_safe();
-        });
-    }
+    construct_handle<MetalBufferObject>(boh, *mContext, bindingType, usage, byteCount);
 }

 void MetalDriver::createTextureR(Handle<HwTexture> th, SamplerType target, uint8_t levels,
@@ -857,10 +832,6 @@ bool MetalDriver::isProtectedTexturesSupported() {
    return false;
 }

-bool MetalDriver::isDepthClampSupported() {
-    return true;
-}
-
 bool MetalDriver::isWorkaroundNeeded(Workaround workaround) {
    switch (workaround) {
        case Workaround::SPLIT_EASU:
@@ -869,8 +840,8 @@ bool MetalDriver::isWorkaroundNeeded(Workaround workaround) {
            return true;
        case Workaround::ADRENO_UNIFORM_ARRAY_CRASH:
            return false;
-        case Workaround::METAL_STATIC_TEXTURE_TARGET_ERROR:
-            return mContext->bugs.staticTextureTargetError;
+        case Workaround::A8X_STATIC_TEXTURE_TARGET_ERROR:
+            return mContext->bugs.a8xStaticTextureTargetError;
        case Workaround::DISABLE_BLIT_INTO_TEXTURE_ARRAY:
            return false;
        default:
@@ -1302,11 +1273,11 @@ void MetalDriver::setPushConstant(backend::ShaderStage stage, uint8_t index,
    pushConstants.setPushConstant(value, index);
 }

-void MetalDriver::insertEventMarker(const char* string) {
+void MetalDriver::insertEventMarker(const char* string, uint32_t len) {

 }

-void MetalDriver::pushGroupMarker(const char* string) {
+void MetalDriver::pushGroupMarker(const char* string, uint32_t len) {
    mContext->groupMarkers.push(string);
 }

@@ -1778,13 +1749,6 @@ void MetalDriver::bindPipeline(PipelineState const& ps) {
        [mContext->currentRenderPassEncoder setFrontFacingWinding:winding];
    }

-    // depth clip mode
-    MTLDepthClipMode depthClipMode = rs.depthClamp ? MTLDepthClipModeClamp : MTLDepthClipModeClip;
-    mContext->depthClampState.updateState(depthClipMode);
-    if (mContext->depthClampState.stateChanged()) {
-        [mContext->currentRenderPassEncoder setDepthClipMode:depthClipMode];
-    }
-
    // Set the depth-stencil state, if a state change is needed.
    DepthStencilState depthState;
    if (depthAttachment) {
@@ -2095,17 +2059,16 @@ void MetalDriver::enumerateBoundBuffers(BufferObjectBinding bindingType,
 void MetalDriver::resetState(int) {
 }

-void MetalDriver::setDebugTag(HandleBase::HandleId handleId, utils::CString tag) {
-    mHandleAllocator.associateTagToHandle(handleId, std::move(tag));
-}
-
 void MetalDriver::runAtNextTick(const std::function<void()>& fn) noexcept {
+    std::lock_guard<std::mutex> const lock(mTickOpsLock);
    mTickOps.push_back(fn);
 }

 void MetalDriver::executeTickOps() noexcept {
    std::vector<std::function<void()>> ops;
+    mTickOpsLock.lock();
    std::swap(ops, mTickOps);
+    mTickOpsLock.unlock();
    for (const auto& f : ops) {
        f();
    }
@@ -257,6 +257,10 @@ void MetalSwapChain::present() {
    }
 }

+#ifndef FILAMENT_RELEASE_PRESENT_DRAWABLE_MAIN_THREAD
+#define FILAMENT_RELEASE_PRESENT_DRAWABLE_MAIN_THREAD 1
+#endif
+
 class PresentDrawableData {
 public:
    PresentDrawableData() = delete;
@@ -275,10 +279,14 @@ public:
           [that->mDrawable present];
        }

+#if FILAMENT_RELEASE_PRESENT_DRAWABLE_MAIN_THREAD == 1
        // mDrawable is acquired on the driver thread. Typically, we would release this object on
        // the same thread, but after receiving consistent crash reports from within
        // [CAMetalDrawable dealloc], we suspect this object requires releasing on the main thread.
        dispatch_async(dispatch_get_main_queue(), ^{ cleanupAndDestroy(that); });
+#else
+        that->mDriver->runAtNextTick([that]() { cleanupAndDestroy(that); });
+#endif
    }

 private:
@@ -548,6 +556,8 @@ MetalTexture::MetalTexture(MetalContext& context, SamplerType target, uint8_t le
            descriptor.usage = getMetalTextureUsage(usage);
            descriptor.storageMode = MTLStorageModePrivate;
            texture = [context.device newTextureWithDescriptor:descriptor];
+            FILAMENT_CHECK_POSTCONDITION(texture != nil)
+                    << "Could not create Metal texture. Out of memory?";
            break;
        case SamplerType::SAMPLER_CUBEMAP:
        case SamplerType::SAMPLER_CUBEMAP_ARRAY:
@@ -562,6 +572,8 @@ MetalTexture::MetalTexture(MetalContext& context, SamplerType target, uint8_t le
            descriptor.usage = getMetalTextureUsage(usage);
            descriptor.storageMode = MTLStorageModePrivate;
            texture = [context.device newTextureWithDescriptor:descriptor];
+            FILAMENT_CHECK_POSTCONDITION(texture != nil)
+                    << "Could not create Metal texture. Out of memory?";
            break;
        case SamplerType::SAMPLER_3D:
            descriptor = [MTLTextureDescriptor new];
@@ -574,6 +586,8 @@ MetalTexture::MetalTexture(MetalContext& context, SamplerType target, uint8_t le
            descriptor.usage = getMetalTextureUsage(usage);
            descriptor.storageMode = MTLStorageModePrivate;
            texture = [context.device newTextureWithDescriptor:descriptor];
+            FILAMENT_CHECK_POSTCONDITION(texture != nil)
+                    << "Could not create Metal texture. Out of memory?";
            break;
        case SamplerType::SAMPLER_EXTERNAL:
            // If we're using external textures (CVPixelBufferRefs), we don't need to make any
@@ -582,12 +596,6 @@ MetalTexture::MetalTexture(MetalContext& context, SamplerType target, uint8_t le
            break;
    }

-    FILAMENT_CHECK_POSTCONDITION(target == SamplerType::SAMPLER_EXTERNAL || texture != nil)
-            << "Could not create Metal texture (SamplerType = " << int(target)
-            << ", levels = " << int(levels) << ", MTLPixelFormat = " << int(devicePixelFormat)
-            << ", width = " << width << ", height = " << height << ", depth = " << depth
-            << "). Out of memory?";
-
    // If swizzling is set, set up a swizzled texture view that we'll use when sampling this texture.
    const bool isDefaultSwizzle =
            r == TextureSwizzle::CHANNEL_0 &&
@@ -382,7 +382,6 @@ using SamplerStateCache = StateCache<SamplerState, id<MTLSamplerState>, SamplerS

 using CullModeStateTracker = StateTracker<MTLCullMode>;
 using WindingStateTracker = StateTracker<MTLWinding>;
-using DepthClampStateTracker = StateTracker<MTLDepthClipMode>;

 // Argument encoder

@@ -202,10 +202,6 @@ bool NoopDriver::isProtectedTexturesSupported() {
    return true;
 }

-bool NoopDriver::isDepthClampSupported() {
-    return false;
-}
-
 bool NoopDriver::isWorkaroundNeeded(Workaround) {
    return false;
 }
@@ -320,10 +316,10 @@ void NoopDriver::setPushConstant(backend::ShaderStage stage, uint8_t index,
        backend::PushConstantVariant value) {
 }

-void NoopDriver::insertEventMarker(char const* string) {
+void NoopDriver::insertEventMarker(char const* string, uint32_t len) {
 }

-void NoopDriver::pushGroupMarker(char const* string) {
+void NoopDriver::pushGroupMarker(char const* string,  uint32_t len) {
 }

 void NoopDriver::popGroupMarker(int) {
@@ -392,7 +388,4 @@ void NoopDriver::endTimerQuery(Handle<HwTimerQuery> tqh) {
 void NoopDriver::resetState(int) {
 }

-void NoopDriver::setDebugTag(HandleBase::HandleId handleId, utils::CString tag) {
-}
-
 } // namespace filament
@@ -676,7 +676,6 @@ void OpenGLContext::initExtensionsGLES(Extensions* ext, GLint major, GLint minor
 #ifndef __EMSCRIPTEN__
    ext->EXT_debug_marker = exts.has("GL_EXT_debug_marker"sv);
 #endif
-    ext->EXT_depth_clamp = exts.has("GL_EXT_depth_clamp"sv);
    ext->EXT_discard_framebuffer = exts.has("GL_EXT_discard_framebuffer"sv);
 #ifndef __EMSCRIPTEN__
    ext->EXT_disjoint_timer_query = exts.has("GL_EXT_disjoint_timer_query"sv);
@@ -747,7 +746,6 @@ void OpenGLContext::initExtensionsGL(Extensions* ext, GLint major, GLint minor)
    ext->EXT_color_buffer_half_float = true;  // Assumes core profile.
    ext->EXT_clip_cull_distance = true;
    ext->EXT_debug_marker = exts.has("GL_EXT_debug_marker"sv);
-    ext->EXT_depth_clamp = true;
    ext->EXT_discard_framebuffer = false;
    ext->EXT_disjoint_timer_query = true;
    ext->EXT_multisampled_render_to_texture = false;
@@ -220,9 +220,8 @@ public:
        bool EXT_color_buffer_float;
        bool EXT_color_buffer_half_float;
        bool EXT_debug_marker;
-        bool EXT_depth_clamp;
-        bool EXT_discard_framebuffer;
        bool EXT_disjoint_timer_query;
+        bool EXT_discard_framebuffer;
        bool EXT_multisampled_render_to_texture2;
        bool EXT_multisampled_render_to_texture;
        bool EXT_protected_textures;
@@ -240,10 +239,10 @@ public:
        bool KHR_parallel_shader_compile;
        bool KHR_texture_compression_astc_hdr;
        bool KHR_texture_compression_astc_ldr;
-        bool OES_EGL_image_external_essl3;
-        bool OES_depth24;
        bool OES_depth_texture;
+        bool OES_depth24;
        bool OES_packed_depth_stencil;
+        bool OES_EGL_image_external_essl3;
        bool OES_rgb8_rgba8;
        bool OES_standard_derivatives;
        bool OES_texture_npot;
@@ -628,7 +627,6 @@ constexpr size_t OpenGLContext::getIndexForCap(GLenum cap) noexcept { //NOLINT
 #ifdef BACKEND_OPENGL_VERSION_GL
        case GL_PROGRAM_POINT_SIZE:             index = 10; break;
 #endif
-        case GL_DEPTH_CLAMP:                    index = 11; break;
        default: break;
    }
    assert_invariant(index < state.enables.caps.size());
@@ -83,20 +83,12 @@
 #define HAS_MAPBUFFERS 1
 #endif

-#define DEBUG_GROUP_MARKER_NONE       0x00    // no debug marker
-#define DEBUG_GROUP_MARKER_OPENGL     0x01    // markers in the gl command queue (req. driver support)
-#define DEBUG_GROUP_MARKER_BACKEND    0x02    // markers on the backend side (systrace)
-#define DEBUG_GROUP_MARKER_ALL        0x03    // all markers
+#define DEBUG_MARKER_NONE       0x00    // no debug marker
+#define DEBUG_MARKER_OPENGL     0x01    // markers in the gl command queue (req. driver support)
+#define DEBUG_MARKER_BACKEND    0x02    // markers on the backend side (systrace)
+#define DEBUG_MARKER_ALL        0x03    // all markers

-#define DEBUG_MARKER_NONE             0x00    // no debug marker
-#define DEBUG_MARKER_OPENGL           0x01    // markers in the gl command queue (req. driver support)
-#define DEBUG_MARKER_BACKEND          0x02    // markers on the backend side (systrace)
-#define DEBUG_MARKER_ALL              0x03    // all markers
-
-// set to the desired debug marker level (for user markers [default: All])
-#define DEBUG_GROUP_MARKER_LEVEL      DEBUG_GROUP_MARKER_ALL
-
-// set to the desired debug level (for internal debugging [Default: None])
+// set to the desired debug marker level
 #define DEBUG_MARKER_LEVEL      DEBUG_MARKER_NONE

 #if DEBUG_MARKER_LEVEL > DEBUG_MARKER_NONE
@@ -204,37 +196,11 @@ Driver* OpenGLDriver::create(OpenGLPlatform* const platform,

 OpenGLDriver::DebugMarker::DebugMarker(OpenGLDriver& driver, const char* string) noexcept
        : driver(driver) {
-#ifndef __EMSCRIPTEN__
-#ifdef GL_EXT_debug_marker
-#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_OPENGL
-    if (UTILS_LIKELY(driver.getContext().ext.EXT_debug_marker)) {
-        glPushGroupMarkerEXT(GLsizei(strlen(string)), string);
-    }
-#endif
-#endif
-
-#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_BACKEND
-    SYSTRACE_CONTEXT();
-    SYSTRACE_NAME_BEGIN(string);
-#endif
-#endif
+    driver.pushGroupMarker(string, strlen(string));
 }

 OpenGLDriver::DebugMarker::~DebugMarker() noexcept {
-#ifndef __EMSCRIPTEN__
-#ifdef GL_EXT_debug_marker
-#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_OPENGL
-    if (UTILS_LIKELY(driver.getContext().ext.EXT_debug_marker)) {
-        glPopGroupMarkerEXT();
-    }
-#endif
-#endif
-
-#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_BACKEND
-    SYSTRACE_CONTEXT();
-    SYSTRACE_NAME_END();
-#endif
-#endif
+    driver.popGroupMarker();
 }

 // ------------------------------------------------------------------------------------------------
@@ -451,14 +417,6 @@ void OpenGLDriver::setRasterState(RasterState rs) noexcept {
    } else {
        gl.disable(GL_SAMPLE_ALPHA_TO_COVERAGE);
    }
-
-    if (gl.ext.EXT_depth_clamp) {
-        if (rs.depthClamp) {
-            gl.enable(GL_DEPTH_CLAMP);
-        } else {
-            gl.disable(GL_DEPTH_CLAMP);
-        }
-    }
 }

 void OpenGLDriver::setStencilState(StencilState ss) noexcept {
@@ -1148,7 +1106,6 @@ void OpenGLDriver::framebufferTexture(TargetBufferInfo const& binfo,
    if (any(t->usage & TextureUsage::SAMPLEABLE)) {
        switch (t->target) {
            case SamplerType::SAMPLER_2D:
-            case SamplerType::SAMPLER_3D:
            case SamplerType::SAMPLER_2D_ARRAY:
            case SamplerType::SAMPLER_CUBEMAP_ARRAY:
                // this could be GL_TEXTURE_2D_MULTISAMPLE or GL_TEXTURE_2D_ARRAY
@@ -1159,9 +1116,7 @@ void OpenGLDriver::framebufferTexture(TargetBufferInfo const& binfo,
                target = getCubemapTarget(binfo.layer);
                // note: cubemaps can't be multi-sampled
                break;
-            case SamplerType::SAMPLER_EXTERNAL:
-                // This is an error. We have asserted in debug build.
-                target = t->gl.target;
+            default:
                break;
        }
    }
@@ -1179,7 +1134,7 @@ void OpenGLDriver::framebufferTexture(TargetBufferInfo const& binfo,

    if (rt->gl.samples <= 1 ||
        (rt->gl.samples > 1 && t->samples > 1 && gl.features.multisample_texture)) {
-        // On GL3.2 / GLES3.1 and above multisample is handled when creating the texture.
+        // on GL3.2 / GLES3.1 and above multisample is handled when creating the texture.
        // If multisampled textures are not supported and we end-up here, things should
        // still work, albeit without MSAA.
        gl.bindFramebuffer(GL_FRAMEBUFFER, rt->gl.fbo);
@@ -1203,7 +1158,6 @@ void OpenGLDriver::framebufferTexture(TargetBufferInfo const& binfo,
                            GL_RENDERBUFFER, t->gl.id);
                }
                break;
-            case GL_TEXTURE_3D:
            case GL_TEXTURE_2D_ARRAY:
            case GL_TEXTURE_CUBE_MAP_ARRAY:
 #ifndef FILAMENT_SILENCE_NOT_SUPPORTED_BY_ES2
@@ -1317,7 +1271,6 @@ void OpenGLDriver::framebufferTexture(TargetBufferInfo const& binfo,
                            GL_RENDERBUFFER, t->gl.id);
                }
                break;
-            case GL_TEXTURE_3D:
            case GL_TEXTURE_2D_ARRAY:
            case GL_TEXTURE_CUBE_MAP_ARRAY:
 #ifndef FILAMENT_SILENCE_NOT_SUPPORTED_BY_ES2
@@ -2132,10 +2085,6 @@ bool OpenGLDriver::isProtectedTexturesSupported() {
    return getContext().ext.EXT_protected_textures;
 }

-bool OpenGLDriver::isDepthClampSupported() {
-    return getContext().ext.EXT_depth_clamp;
-}
-
 bool OpenGLDriver::isWorkaroundNeeded(Workaround workaround) {
    switch (workaround) {
        case Workaround::SPLIT_EASU:
@@ -2227,10 +2176,6 @@ void OpenGLDriver::makeCurrent(Handle<HwSwapChain> schDraw, Handle<HwSwapChain>
 // Updating driver objects
 // ------------------------------------------------------------------------------------------------

-void OpenGLDriver::setDebugTag(HandleBase::HandleId handleId, utils::CString tag) {
-    mHandleAllocator.associateTagToHandle(handleId, std::move(tag));
-}
-
 void OpenGLDriver::setVertexBufferObject(Handle<HwVertexBuffer> vbh,
        uint32_t index, Handle<HwBufferObject> boh) {
   DEBUG_MARKER()
@@ -2384,7 +2329,9 @@ void OpenGLDriver::updateSamplerGroup(Handle<HwSamplerGroup> sbh,
    GLSamplerGroup* const sb = handle_cast<GLSamplerGroup *>(sbh);
    assert_invariant(sb->textureUnitEntries.size() == data.size / sizeof(SamplerDescriptor));

+#ifndef FILAMENT_SILENCE_NOT_SUPPORTED_BY_ES2
    bool const es2 = context.isES2();
+#endif

    auto const* const pSamplers = (SamplerDescriptor const*)data.buffer;
    for (size_t i = 0, c = sb->textureUnitEntries.size(); i < c; i++) {
@@ -2394,15 +2341,6 @@ void OpenGLDriver::updateSamplerGroup(Handle<HwSamplerGroup> sbh,
            GLTexture const* const t = handle_cast<const GLTexture*>(th);
            assert_invariant(t);

-            if (UTILS_UNLIKELY(es2)
-#if defined(GL_EXT_texture_filter_anisotropic)
-                    || UTILS_UNLIKELY(anisotropyWorkaround)
- #endif
-                    ) {
-                // We must set texture parameters on the texture itself.
-                bindTexture(OpenGLContext::DUMMY_TEXTURE_BINDING, t);
-            }
-
            SamplerParams params = pSamplers[i].s;
            if (UTILS_UNLIKELY(t->target == SamplerType::SAMPLER_EXTERNAL)) {
                // From OES_EGL_image_external spec:
@@ -3199,28 +3137,28 @@ void OpenGLDriver::bindSamplers(uint32_t index, Handle<HwSamplerGroup> sbh) {
    CHECK_GL_ERROR(utils::slog.e)
 }

-void OpenGLDriver::insertEventMarker(char const* string) {
+void OpenGLDriver::insertEventMarker(char const* string, uint32_t len) {
 #ifndef __EMSCRIPTEN__
 #ifdef GL_EXT_debug_marker
    auto& gl = mContext;
    if (gl.ext.EXT_debug_marker) {
-        glInsertEventMarkerEXT(GLsizei(strlen(string)), string);
+        glInsertEventMarkerEXT(GLsizei(len ? len : strlen(string)), string);
    }
 #endif
 #endif
 }

-void OpenGLDriver::pushGroupMarker(char const* string) {
+void OpenGLDriver::pushGroupMarker(char const* string, uint32_t len) {
 #ifndef __EMSCRIPTEN__
 #ifdef GL_EXT_debug_marker
-#if DEBUG_GROUP_MARKER_LEVEL & DEBUG_GROUP_MARKER_OPENGL
+#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_OPENGL
    if (UTILS_LIKELY(mContext.ext.EXT_debug_marker)) {
-        glPushGroupMarkerEXT(GLsizei(strlen(string)), string);
+        glPushGroupMarkerEXT(GLsizei(len ? len : strlen(string)), string);
    }
 #endif
 #endif

-#if DEBUG_GROUP_MARKER_LEVEL & DEBUG_GROUP_MARKER_BACKEND
+#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_BACKEND
    SYSTRACE_CONTEXT();
    SYSTRACE_NAME_BEGIN(string);
 #endif
@@ -3230,14 +3168,14 @@ void OpenGLDriver::pushGroupMarker(char const* string) {
 void OpenGLDriver::popGroupMarker(int) {
 #ifndef __EMSCRIPTEN__
 #ifdef GL_EXT_debug_marker
-#if DEBUG_GROUP_MARKER_LEVEL & DEBUG_GROUP_MARKER_OPENGL
+#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_OPENGL
    if (UTILS_LIKELY(mContext.ext.EXT_debug_marker)) {
        glPopGroupMarkerEXT();
    }
 #endif
 #endif

-#if DEBUG_GROUP_MARKER_LEVEL & DEBUG_GROUP_MARKER_BACKEND
+#if DEBUG_MARKER_LEVEL & DEBUG_MARKER_BACKEND
    SYSTRACE_CONTEXT();
    SYSTRACE_NAME_END();
 #endif
@@ -3943,7 +3881,6 @@ void OpenGLDriver::bindRenderPrimitive(Handle<HwRenderPrimitive> rph) {
 }

 void OpenGLDriver::draw2(uint32_t indexOffset, uint32_t indexCount, uint32_t instanceCount) {
-    DEBUG_MARKER()
    GLRenderPrimitive const* const rp = mBoundRenderPrimitive;
    if (UTILS_UNLIKELY(!rp || !mValidProgram)) {
        return;
@@ -3965,7 +3902,6 @@ void OpenGLDriver::draw2(uint32_t indexOffset, uint32_t indexCount, uint32_t ins
 }

 void OpenGLDriver::draw2GLES2(uint32_t indexOffset, uint32_t indexCount, uint32_t instanceCount) {
-    DEBUG_MARKER()
    GLRenderPrimitive const* const rp = mBoundRenderPrimitive;
    if (UTILS_UNLIKELY(!rp || !mValidProgram)) {
        return;
@@ -3986,7 +3922,6 @@ void OpenGLDriver::draw2GLES2(uint32_t indexOffset, uint32_t indexCount, uint32_
 }

 void OpenGLDriver::scissor(Viewport scissor) {
-    DEBUG_MARKER()
    setScissor(scissor);
 }

@@ -4006,7 +3941,6 @@ void OpenGLDriver::draw(PipelineState state, Handle<HwRenderPrimitive> rph,
 }

 void OpenGLDriver::dispatchCompute(Handle<HwProgram> program, math::uint3 workGroupCount) {
-    DEBUG_MARKER()
    getShaderCompilerService().tick();

    OpenGLProgram* const p = handle_cast<OpenGLProgram*>(program);
@@ -81,7 +81,7 @@ public:
            const Platform::DriverConfig& driverConfig) noexcept;

    class DebugMarker {
-        UTILS_UNUSED OpenGLDriver& driver;
+        OpenGLDriver& driver;
    public:
        DebugMarker(OpenGLDriver& driver, const char* string) noexcept;
        ~DebugMarker() noexcept;
@@ -143,25 +143,21 @@ void TimerQueryNativeFactory::endTimeElapsedQuery(OpenGLDriver& driver, GLTimerQ

    driver.runEveryNowAndThen([&context = mContext, weak]() -> bool {
        auto state = weak.lock();
-        if (!state) {
-            // The timer query state has been destroyed on the way, very likely due to the IBL
-            // prefilter context destruction. We still return true to get this element removed from
-            // the query list.
-            return true;
+        if (state) {
+            GLuint available = 0;
+            context.procs.getQueryObjectuiv(state->gl.query, GL_QUERY_RESULT_AVAILABLE, &available);
+            CHECK_GL_ERROR(utils::slog.e)
+            if (!available) {
+                // we need to try this one again later
+                return false;
+            }
+            GLuint64 elapsedTime = 0;
+            // we won't end-up here if we're on ES and don't have GL_EXT_disjoint_timer_query
+            context.procs.getQueryObjectui64v(state->gl.query, GL_QUERY_RESULT, &elapsedTime);
+            state->elapsed.store((int64_t)elapsedTime, std::memory_order_relaxed);
+        } else {
+            state->elapsed.store(int64_t(TimerQueryResult::ERROR), std::memory_order_relaxed);
        }
-
-        GLuint available = 0;
-        context.procs.getQueryObjectuiv(state->gl.query, GL_QUERY_RESULT_AVAILABLE, &available);
-        CHECK_GL_ERROR(utils::slog.e)
-        if (!available) {
-            // we need to try this one again later
-            return false;
-        }
-        GLuint64 elapsedTime = 0;
-        // we won't end-up here if we're on ES and don't have GL_EXT_disjoint_timer_query
-        context.procs.getQueryObjectui64v(state->gl.query, GL_QUERY_RESULT, &elapsedTime);
-        state->elapsed.store((int64_t)elapsedTime, std::memory_order_relaxed);
-
        return true;
    });
 }
@@ -26,28 +26,16 @@

 #include <utils/compiler.h>
 #include <utils/CString.h>
-#include <utils/debug.h>
-#include <utils/FixedCapacityVector.h>
 #include <utils/JobSystem.h>
 #include <utils/Log.h>
-#include <utils/ostream.h>
-#include <utils/Panic.h>
 #include <utils/Systrace.h>

-#include <array>
 #include <cctype>
 #include <chrono>
-#include <mutex>
-#include <memory>
 #include <string>
 #include <string_view>
-#include <thread>
-#include <utility>
 #include <variant>

-#include <stddef.h>
-#include <stdint.h>
-
 namespace filament::backend {

 using namespace utils;
@@ -488,12 +476,6 @@ GLuint ShaderCompilerService::initialize(program_token_t& token) noexcept {
    // check status of program linking and shader compilation, logs error and free all resources
    // in case of error.
    bool const success = checkProgramStatus(token);
-
-    // Unless we have matdbg, we panic if a program is invalid. Otherwise, we'd get a UB.
-    // The compilation error has been logged to log.e by this point.
-    FILAMENT_CHECK_POSTCONDITION(FILAMENT_ENABLE_MATDBG || success)
-            << "OpenGL program " << token->name.c_str_safe() << " failed to link or compile";
-
    if (UTILS_LIKELY(success)) {
        program = token->gl.program;
        // no need to keep the shaders around
@@ -597,31 +579,24 @@ void ShaderCompilerService::compileShaders(OpenGLContext& context,
                version = "#version 310 es\n";
            }

-            std::array<std::string_view, 5> sources = {
-                version,
-                prolog,
-                specializationConstantString,
-                packingFunctions,
-                { body.data(), body.size() - 1 }  // null-terminated
+            const std::array<const char*, 5> sources = {
+                    version.data(),
+                    prolog.data(),
+                    specializationConstantString.c_str(),
+                    packingFunctions.data(),
+                    body.data()
            };

-            // Some of the sources may be zero-length. Remove them as to avoid passing lengths of
-            // zero to glShaderSource(). glShaderSource should work with lengths of zero, but some
-            // drivers instead interpret zero as a sentinel for a null-terminated string.
-            auto partitionPoint = std::stable_partition(
-                    sources.begin(), sources.end(), [](std::string_view s) { return !s.empty(); });
-            size_t count = std::distance(sources.begin(), partitionPoint);
-
-            std::array<const char*, 5> shaderStrings;
-            std::array<GLint, 5> lengths;
-            for (size_t i = 0; i < count; i++) {
-                shaderStrings[i] = sources[i].data();
-                lengths[i] = sources[i].size();
-            }
+            const std::array<GLint, 5> lengths = {
+                    (GLint)version.length(),
+                    (GLint)prolog.length(),
+                    (GLint)specializationConstantString.length(),
+                    (GLint)packingFunctions.length(),
+                    (GLint)body.length() - 1 // null terminated
+            };

            GLuint const shaderId = glCreateShader(glShaderType);
-            glShaderSource(shaderId, count, shaderStrings.data(), lengths.data());
-
+            glShaderSource(shaderId, sources.size(), sources.data(), lengths.data());
            glCompileShader(shaderId);

 #ifndef NDEBUG
@@ -810,24 +785,25 @@ mediump vec4 unpackSnorm4x8(highp uint v) {
 // - extensions
 // - everything else
 std::array<std::string_view, 3> ShaderCompilerService::splitShaderSource(std::string_view source) noexcept {
-    auto version_start = source.find("#version");
-    assert_invariant(version_start != std::string_view::npos);
+    auto start = source.find("#version");
+    assert_invariant(start != std::string_view::npos);

-    auto version_eol = source.find('\n', version_start) + 1;
+    auto version_eol = source.find('\n', start) + 1;
    assert_invariant(version_eol != std::string_view::npos);

-    auto prolog_start = version_eol;
-    auto prolog_eol = source.rfind("\n#extension"); // last #extension line
-    if (prolog_eol == std::string_view::npos) {
-        prolog_eol = prolog_start;
+    auto pos = source.rfind("\n#extension");
+    if (pos == std::string_view::npos) {
+        pos = version_eol;
    } else {
-        prolog_eol = source.find('\n', prolog_eol + 1) + 1;
+        ++pos;
    }
-    auto body_start = prolog_eol;

-    std::string_view const version = source.substr(version_start, version_eol - version_start);
-    std::string_view const prolog = source.substr(prolog_start, prolog_eol - prolog_start);
-    std::string_view const body = source.substr(body_start, source.length() - body_start);
+    auto eol = source.find('\n', pos) + 1;
+    assert_invariant(eol != std::string_view::npos);
+
+    std::string_view const version = source.substr(start, version_eol - start);
+    std::string_view const prolog = source.substr(version_eol, eol - version_eol);
+    std::string_view const body = source.substr(eol, source.length() - eol);
    return { version, prolog, body };
 }

@@ -201,12 +201,6 @@ using namespace glext;
 #   define GL_CLIP_DISTANCE1                        0x3001
 #endif

-#if defined(GL_EXT_depth_clamp)
-#   define GL_DEPTH_CLAMP                           GL_DEPTH_CLAMP_EXT
-#else
-#   define GL_DEPTH_CLAMP                           0x864F
-#endif
-
 #if defined(GL_KHR_debug)
 #   define GL_DEBUG_OUTPUT                          GL_DEBUG_OUTPUT_KHR
 #   define GL_DEBUG_OUTPUT_SYNCHRONOUS              GL_DEBUG_OUTPUT_SYNCHRONOUS_KHR
@@ -28,7 +28,6 @@ VulkanBuffer::VulkanBuffer(VmaAllocator allocator, VulkanStagePool& stagePool,
    : mAllocator(allocator),
      mStagePool(stagePool),
      mUsage(usage),
-	  mUpdatedOffset(0),
      mUpdatedBytes(0) {
    // for now make sure that only 1 bit is set in usage
    // (because loadFromCpu() assumes that somewhat)
@@ -81,7 +80,6 @@ void VulkanBuffer::loadFromCpu(VkCommandBuffer cmdbuf, const void* cpuData, uint
            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
            .buffer = mGpuBuffer,
-			.offset = mUpdatedOffset,
            .size = mUpdatedBytes,
        };
        vkCmdPipelineBarrier(cmdbuf, srcStage, VK_PIPELINE_STAGE_TRANSFER_BIT, 0, 0, nullptr, 1,
@@ -95,7 +93,6 @@ void VulkanBuffer::loadFromCpu(VkCommandBuffer cmdbuf, const void* cpuData, uint
    };
    vkCmdCopyBuffer(cmdbuf, stage->buffer, mGpuBuffer, 1, &region);

-	mUpdatedOffset = byteOffset;
    mUpdatedBytes = numBytes;

    // Firstly, ensure that the copy finishes before the next draw call.
@@ -42,7 +42,6 @@ private:
    VmaAllocation mGpuMemory = VK_NULL_HANDLE;
    VkBuffer mGpuBuffer = VK_NULL_HANDLE;
    VkBufferUsageFlags mUsage = {};
-	uint32_t mUpdatedOffset = 0;
    uint32_t mUpdatedBytes = 0;
 };

@@ -59,11 +59,7 @@ VkExtent2D VulkanAttachment::getExtent2D() const {

 VkImageView VulkanAttachment::getImageView() {
    assert_invariant(texture);
-    VkImageSubresourceRange range = getSubresourceRange();
-    if (range.layerCount > 1) {
-        return texture->getMultiviewAttachmentView(range);
-    }
-    return texture->getAttachmentView(range);
+    return texture->getAttachmentView(getSubresourceRange());
 }

 bool VulkanAttachment::isDepth() const {
@@ -77,7 +73,7 @@ VkImageSubresourceRange VulkanAttachment::getSubresourceRange() const {
            .baseMipLevel = uint32_t(level),
            .levelCount = 1,
            .baseArrayLayer = uint32_t(layer),
-            .layerCount = layerCount,
+            .layerCount = 1,
    };
 }

@@ -43,8 +43,6 @@ struct VulkanCommandBuffer;
 struct VulkanAttachment {
    VulkanTexture* texture = nullptr;
    uint8_t level = 0;
-    uint8_t baseViewIndex = 0;
-    uint8_t layerCount = 1;
    uint16_t layer = 0;

    bool isDepth() const;
@@ -125,10 +123,6 @@ public:
        return mPhysicalDeviceFeatures.imageCubeArray == VK_TRUE;
    }

-    inline bool isDepthClampSupported() const noexcept {
-        return mPhysicalDeviceFeatures.depthClamp == VK_TRUE;
-    }
-
    inline bool isDebugMarkersSupported() const noexcept {
        return mDebugMarkersSupported;
    }
@@ -595,8 +595,6 @@ void VulkanDriver::createRenderTargetR(Handle<HwRenderTarget> rth,
            colorTargets[i] = {
                .texture = mResourceAllocator.handle_cast<VulkanTexture*>(color[i].handle),
                .level = color[i].level,
-                .baseViewIndex = color[i].baseViewIndex,
-                .layerCount = layerCount,
                .layer = color[i].layer,
            };
            UTILS_UNUSED_IN_RELEASE VkExtent2D extent = colorTargets[i].getExtent2D();
@@ -611,8 +609,6 @@ void VulkanDriver::createRenderTargetR(Handle<HwRenderTarget> rth,
        depthStencil[0] = {
            .texture = mResourceAllocator.handle_cast<VulkanTexture*>(depth.handle),
            .level = depth.level,
-            .baseViewIndex = depth.baseViewIndex,
-            .layerCount = layerCount,
            .layer = depth.layer,
        };
        UTILS_UNUSED_IN_RELEASE VkExtent2D extent = depthStencil[0].getExtent2D();
@@ -625,8 +621,6 @@ void VulkanDriver::createRenderTargetR(Handle<HwRenderTarget> rth,
        depthStencil[1] = {
            .texture = mResourceAllocator.handle_cast<VulkanTexture*>(stencil.handle),
            .level = stencil.level,
-            .baseViewIndex = stencil.baseViewIndex,
-            .layerCount = layerCount,
            .layer = stencil.layer,
        };
        UTILS_UNUSED_IN_RELEASE VkExtent2D extent = depthStencil[1].getExtent2D();
@@ -643,7 +637,7 @@ void VulkanDriver::createRenderTargetR(Handle<HwRenderTarget> rth,

    auto renderTarget = mResourceAllocator.construct<VulkanRenderTarget>(rth, mPlatform->getDevice(),
            mPlatform->getPhysicalDevice(), mContext, mAllocator, &mCommands, width, height,
-            samples, colorTargets, depthStencil, mStagePool, layerCount);
+            samples, colorTargets, depthStencil, mStagePool);
    mResourceManager.acquire(renderTarget);
 }

@@ -944,10 +938,6 @@ bool VulkanDriver::isProtectedTexturesSupported() {
    return false;
 }

-bool VulkanDriver::isDepthClampSupported() {
-    return mContext.isDepthClampSupported();
-}
-
 bool VulkanDriver::isWorkaroundNeeded(Workaround workaround) {
    switch (workaround) {
        case Workaround::SPLIT_EASU: {
@@ -996,15 +986,15 @@ FeatureLevel VulkanDriver::getFeatureLevel() {

    // If the max sampler counts do not meet FL2 standards, then this is an FL1 device.
    const auto& fl2 = FEATURE_LEVEL_CAPS[+FeatureLevel::FEATURE_LEVEL_2];
-    if (limits.maxPerStageDescriptorSamplers < fl2.MAX_VERTEX_SAMPLER_COUNT  ||
-        limits.maxPerStageDescriptorSamplers < fl2.MAX_FRAGMENT_SAMPLER_COUNT) {
+    if (fl2.MAX_VERTEX_SAMPLER_COUNT < limits.maxPerStageDescriptorSamplers ||
+        fl2.MAX_FRAGMENT_SAMPLER_COUNT < limits.maxPerStageDescriptorSamplers) {
        return FeatureLevel::FEATURE_LEVEL_1;
    }

    // If the max sampler counts do not meet FL3 standards, then this is an FL2 device.
    const auto& fl3 = FEATURE_LEVEL_CAPS[+FeatureLevel::FEATURE_LEVEL_3];
-    if (limits.maxPerStageDescriptorSamplers < fl3.MAX_VERTEX_SAMPLER_COUNT||
-        limits.maxPerStageDescriptorSamplers < fl3.MAX_FRAGMENT_SAMPLER_COUNT) {
+    if (fl3.MAX_VERTEX_SAMPLER_COUNT < limits.maxPerStageDescriptorSamplers ||
+        fl3.MAX_FRAGMENT_SAMPLER_COUNT < limits.maxPerStageDescriptorSamplers) {
        return FeatureLevel::FEATURE_LEVEL_2;
    }

@@ -1277,8 +1267,6 @@ void VulkanDriver::beginRenderPass(Handle<HwRenderTarget> rth, const RenderPassP
        }
    }

-    uint8_t const renderTargetLayerCount = rt->getLayerCount();
-
    // Create the VkRenderPass or fetch it from cache.
    VulkanFboCache::RenderPassKey rpkey = {
        .initialColorLayoutMask = 0,
@@ -1289,12 +1277,10 @@ void VulkanDriver::beginRenderPass(Handle<HwRenderTarget> rth, const RenderPassP
        .discardEnd = discardEndVal,
        .samples = rt->getSamples(),
        .subpassMask = uint8_t(params.subpassMask),
-        .viewCount = renderTargetLayerCount,
    };
    for (int i = 0; i < MRT::MAX_SUPPORTED_RENDER_TARGET_COUNT; i++) {
        const VulkanAttachment& info = rt->getColor(i);
        if (info.texture) {
-            assert_invariant(info.layerCount == renderTargetLayerCount);
            rpkey.initialColorLayoutMask |= 1 << i;
            rpkey.colorFormat[i] = info.getFormat();
            if (rpkey.samples > 1 && info.texture->samples == 1) {
@@ -1606,13 +1592,13 @@ void VulkanDriver::setPushConstant(backend::ShaderStage stage, uint8_t index,
            value);
 }

-void VulkanDriver::insertEventMarker(char const* string) {
+void VulkanDriver::insertEventMarker(char const* string, uint32_t len) {
 #if FVK_ENABLED(FVK_DEBUG_GROUP_MARKERS)
-    mCommands.insertEventMarker(string, strlen(string));
+    mCommands.insertEventMarker(string, len);
 #endif
 }

-void VulkanDriver::pushGroupMarker(char const* string) {
+void VulkanDriver::pushGroupMarker(char const* string, uint32_t) {
    // Turns out all the markers are 0-terminated, so we can just pass it without len.
 #if FVK_ENABLED(FVK_DEBUG_GROUP_MARKERS)
    mCommands.pushGroupMarker(string);
@@ -1820,7 +1806,6 @@ void VulkanDriver::bindPipeline(PipelineState const& pipelineState) {
        .dstAlphaBlendFactor = getBlendFactor(rasterState.blendFunctionDstAlpha),
        .colorWriteMask = (VkColorComponentFlags) (rasterState.colorWrite ? 0xf : 0x0),
        .rasterizationSamples = rt->getSamples(),
-        .depthClamp = rasterState.depthClamp,
        .colorTargetCount = rt->getColorTargetCount(mCurrentRenderPass),
        .colorBlendOp = rasterState.blendEquationRGB,
        .alphaBlendOp =  rasterState.blendEquationAlpha,
@@ -2037,10 +2022,6 @@ void VulkanDriver::debugCommandBegin(CommandStream* cmds, bool synchronous, cons
 void VulkanDriver::resetState(int) {
 }

-void VulkanDriver::setDebugTag(HandleBase::HandleId handleId, utils::CString tag) {
-    mResourceAllocator.associateHandle(handleId, std::move(tag));
-}
-
 // explicit instantiation of the Dispatcher
 template class ConcreteDispatcher<VulkanDriver>;

@@ -43,7 +43,6 @@ bool VulkanFboCache::RenderPassEq::operator()(const RenderPassKey& k1,
    if (k1.samples != k2.samples) return false;
    if (k1.needsResolveMask != k2.needsResolveMask) return false;
    if (k1.subpassMask != k2.subpassMask) return false;
-    if (k1.viewCount != k2.viewCount) return false;
    return true;
 }

@@ -187,25 +186,6 @@ VkRenderPass VulkanFboCache::getRenderPass(RenderPassKey config) noexcept {
        .pDependencies = dependencies
    };

-    VkRenderPassMultiviewCreateInfo multiviewCreateInfo = {};
-    uint32_t const subpassViewMask = (1 << config.viewCount) - 1;
-    // Prepare a view mask array for the maximum number of subpasses. All subpasses have all views
-    // activated.
-    uint32_t const viewMasks[2] = {subpassViewMask, subpassViewMask};
-    if (config.viewCount > 1) {
-        // Fill the multiview create info.
-        multiviewCreateInfo.sType = VK_STRUCTURE_TYPE_RENDER_PASS_MULTIVIEW_CREATE_INFO;
-        multiviewCreateInfo.pNext = nullptr;
-        multiviewCreateInfo.subpassCount = hasSubpasses ? 2u : 1u;
-        multiviewCreateInfo.pViewMasks = viewMasks;
-        multiviewCreateInfo.dependencyCount = 0;
-        multiviewCreateInfo.pViewOffsets = nullptr;
-        multiviewCreateInfo.correlationMaskCount = 1;
-        multiviewCreateInfo.pCorrelationMasks = &subpassViewMask;
-
-        renderPassInfo.pNext = &multiviewCreateInfo;
-    }
-
    int attachmentIndex = 0;

    // Populate the Color Attachments.
@@ -64,7 +64,7 @@ public:
        uint8_t samples; // 1 byte
        uint8_t needsResolveMask; // 1 byte
        uint8_t subpassMask; // 1 byte
-        uint8_t viewCount; // 1 byte
+        uint8_t padding2; // 1 byte
    };
    struct RenderPassVal {
        VkRenderPass handle;
@@ -16,6 +16,7 @@

 #include "VulkanHandles.h"

+#include "VulkanDriver.h"
 #include "VulkanConstants.h"
 #include "VulkanDriver.h"
 #include "VulkanMemory.h"
@@ -322,12 +323,11 @@ VulkanRenderTarget::VulkanRenderTarget(VkDevice device, VkPhysicalDevice physica
        VulkanContext const& context, VmaAllocator allocator, VulkanCommands* commands,
        uint32_t width, uint32_t height, uint8_t samples,
        VulkanAttachment color[MRT::MAX_SUPPORTED_RENDER_TARGET_COUNT],
-        VulkanAttachment depthStencil[2], VulkanStagePool& stagePool, uint8_t layerCount)
+        VulkanAttachment depthStencil[2], VulkanStagePool& stagePool)
    : HwRenderTarget(width, height),
      VulkanResource(VulkanResourceType::RENDER_TARGET),
      mOffscreen(true),
-      mSamples(samples),
-      mLayerCount(layerCount) {
+      mSamples(samples) {
    for (int index = 0; index < MRT::MAX_SUPPORTED_RENDER_TARGET_COUNT; index++) {
        mColor[index] = color[index];
    }
@@ -304,7 +304,7 @@ struct VulkanRenderTarget : private HwRenderTarget, VulkanResource {
            VulkanContext const& context, VmaAllocator allocator,
            VulkanCommands* commands, uint32_t width, uint32_t height,
            uint8_t samples, VulkanAttachment color[MRT::MAX_SUPPORTED_RENDER_TARGET_COUNT],
-            VulkanAttachment depthStencil[2], VulkanStagePool& stagePool, uint8_t layerCount);
+            VulkanAttachment depthStencil[2], VulkanStagePool& stagePool);

    // Creates a special "default" render target (i.e. associated with the swap chain)
    explicit VulkanRenderTarget();
@@ -319,7 +319,6 @@ struct VulkanRenderTarget : private HwRenderTarget, VulkanResource {
    VulkanAttachment& getMsaaDepth();
    uint8_t getColorTargetCount(const VulkanRenderPass& pass) const;
    uint8_t getSamples() const { return mSamples; }
-    uint8_t getLayerCount() const { return mLayerCount; }
    bool hasDepth() const { return mDepth.texture; }
    bool isSwapChain() const { return !mOffscreen; }
    void bindToSwapChain(VulkanSwapChain& surf);
@@ -331,7 +330,6 @@ private:
    VulkanAttachment mMsaaDepthAttachment = {};
    const bool mOffscreen : 1;
    uint8_t mSamples : 7;
-    uint8_t mLayerCount = 1;
 };

 struct VulkanBufferObject;
@@ -184,7 +184,6 @@ VulkanPipelineCache::PipelineCacheEntry* VulkanPipelineCache::createPipeline() n
    vkRaster.polygonMode = VK_POLYGON_MODE_FILL;
    vkRaster.cullMode = raster.cullMode;
    vkRaster.frontFace = raster.frontFace;
-    vkRaster.depthClampEnable = raster.depthClamp;
    vkRaster.depthBiasEnable = raster.depthBiasEnable;
    vkRaster.depthBiasConstantFactor = raster.depthBiasConstantFactor;
    vkRaster.depthBiasClamp = 0.0f;
@@ -90,8 +90,7 @@ public:
        VkBlendFactor         srcAlphaBlendFactor : 5;
        VkBlendFactor         dstAlphaBlendFactor : 5;
        VkColorComponentFlags colorWriteMask : 4;
-        uint8_t               rasterizationSamples : 4;// offset = 4 bytes
-        uint8_t               depthClamp : 4;
+        uint8_t               rasterizationSamples;    // offset = 4 bytes
        uint8_t               colorTargetCount;        // offset = 5 bytes
        BlendEquation         colorBlendOp : 4;        // offset = 6 bytes
        BlendEquation         alphaBlendOp : 4;
@@ -105,10 +105,6 @@ public:
        mHandleAllocatorImpl.deallocate(handle, obj);
    }

-    inline void associateHandle(HandleBase::HandleId id, utils::CString&& tag) noexcept {
-        mHandleAllocatorImpl.associateTagToHandle(id, std::move(tag));
-    }
-
 private:
    AllocatorImpl mHandleAllocatorImpl;

@@ -97,11 +97,13 @@ VulkanTexture::VulkanTexture(VkDevice device, VkPhysicalDevice physicalDevice,
        imageInfo.extent.depth = 1;
    }

-    if (any(usage & TextureUsage::BLIT_SRC)) {
-        imageInfo.usage |= VK_IMAGE_USAGE_TRANSFER_SRC_BIT;
-    }
-    if (any(usage & TextureUsage::BLIT_DST)) {
-        imageInfo.usage |= VK_IMAGE_USAGE_TRANSFER_DST_BIT;
+    // Filament expects blit() to work with any texture, so we almost always set these usage flags.
+    // TODO: investigate performance implications of setting these flags.
+    constexpr VkImageUsageFlags blittable = VK_IMAGE_USAGE_TRANSFER_DST_BIT |
+            VK_IMAGE_USAGE_TRANSFER_SRC_BIT;
+
+    if (any(usage & (TextureUsage::BLIT_DST | TextureUsage::BLIT_SRC))) {
+        imageInfo.usage |= blittable;
    }

    if (any(usage & TextureUsage::SAMPLEABLE)) {
@@ -119,7 +121,7 @@ VulkanTexture::VulkanTexture(VkDevice device, VkPhysicalDevice physicalDevice,
        imageInfo.usage |= VK_IMAGE_USAGE_SAMPLED_BIT;
    }
    if (any(usage & TextureUsage::COLOR_ATTACHMENT)) {
-        imageInfo.usage |= VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT;
+        imageInfo.usage |= VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT | blittable;
        if (any(usage & TextureUsage::SUBPASS_INPUT)) {
            imageInfo.usage |= VK_IMAGE_USAGE_INPUT_ATTACHMENT_BIT;
        }
@@ -128,9 +130,10 @@ VulkanTexture::VulkanTexture(VkDevice device, VkPhysicalDevice physicalDevice,
        imageInfo.usage |= VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT;
    }
    if (any(usage & TextureUsage::UPLOADABLE)) {
-        imageInfo.usage |= VK_IMAGE_USAGE_TRANSFER_DST_BIT;
+        imageInfo.usage |= blittable;
    }
    if (any(usage & TextureUsage::DEPTH_ATTACHMENT)) {
+        imageInfo.usage |= blittable;
        imageInfo.usage |= VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT;

        // Depth resolves uses a custom shader and therefore needs to be sampleable.
@@ -384,15 +387,12 @@ void VulkanTexture::setPrimaryRange(uint32_t minMiplevel, uint32_t maxMiplevel)
 }

 VkImageView VulkanTexture::getAttachmentView(VkImageSubresourceRange range) {
+    // Attachments should only have one mipmap level and one layer.
    range.levelCount = 1;
    range.layerCount = 1;
    return getImageView(range, VK_IMAGE_VIEW_TYPE_2D, {});
 }

-VkImageView VulkanTexture::getMultiviewAttachmentView(VkImageSubresourceRange range) {
-    return getImageView(range, VK_IMAGE_VIEW_TYPE_2D_ARRAY, {});
-}
-
 VkImageView VulkanTexture::getViewForType(VkImageSubresourceRange const& range, VkImageViewType type) {
    return getImageView(range, type, mSwizzle);
 }
@@ -73,10 +73,6 @@ struct VulkanTexture : public HwTexture, VulkanResource {
    // and the identity swizzle.
    VkImageView getAttachmentView(VkImageSubresourceRange range);

-    // Gets or creates a cached VkImageView for a single subresource that can be used as a render
-    // target attachment when rendering with multiview.
-    VkImageView getMultiviewAttachmentView(VkImageSubresourceRange range);
-
    // This is a workaround for the first few frames where we're waiting for the texture to actually
    // be uploaded.  In that case, we bind the sampler to an empty texture, but the corresponding
    // imageView needs to be of the right type. Hence, we provide an option to indicate the
@@ -172,7 +172,7 @@ public:
        UTILS_UNUSED VkResult result = vkAllocateDescriptorSets(mDevice, &allocInfo, &vkSet);
        FILAMENT_CHECK_POSTCONDITION(result == VK_SUCCESS)
                << "Failed to allocate descriptor set code=" << result << " size=" << mSize
-                << " capacity=" << mCapacity << " count=" << mUnusedCount;
+                << " capacity=" << mCapacity << " count=" << "%s";
        mSize++;
        return createSet(layout->bitmask, vkSet);
    }
@@ -32,12 +32,12 @@ VkPipelineLayout VulkanPipelineLayoutCache::getLayout(
    }

    // build the push constant layout key
-    uint32_t const pushConstantRangeCount = program->getPushConstantRangeCount();
-    auto const& pushConstantRanges = program->getPushConstantRanges();
+    uint32_t pushConstantRangeCount = program->getPushConstantRangeCount();
+    auto const& pushCostantRanges = program->getPushConstantRanges();    
    if (pushConstantRangeCount > 0) {
        assert_invariant(pushConstantRangeCount <= Program::SHADER_TYPE_COUNT);
        for (uint8_t i = 0; i < pushConstantRangeCount; ++i) {
-            auto const& range = pushConstantRanges[i];
+            auto const& range = pushCostantRanges[i];
            auto& pushConstant = key.pushConstant[i];
            if (range.stageFlags &  VK_SHADER_STAGE_VERTEX_BIT) {
                pushConstant.stage = static_cast<uint8_t>(ShaderStage::VERTEX);
@@ -52,8 +52,9 @@ VkPipelineLayout VulkanPipelineLayoutCache::getLayout(
        }
    }

-    if (auto iter = mPipelineLayouts.find(key); iter != mPipelineLayouts.end()) {
-        PipelineLayoutCacheEntry& entry = iter->second;
+    auto iter = mPipelineLayouts.find(key);
+    if (iter != mPipelineLayouts.end()) {
+        PipelineLayoutCacheEntry& entry = mPipelineLayouts[key];
        entry.lastUsed = mTimestamp++;
        return entry.handle;
    }
@@ -64,7 +65,7 @@ VkPipelineLayout VulkanPipelineLayoutCache::getLayout(
        .setLayoutCount = (uint32_t) descSetLayoutCount,
        .pSetLayouts = key.descSetLayouts.data(),
        .pushConstantRangeCount = pushConstantRangeCount,
-        .pPushConstantRanges = pushConstantRanges,
+        .pPushConstantRanges = pushCostantRanges,
    };

    VkPipelineLayout layout;
@@ -22,7 +22,7 @@

 #include <utils/Hash.h>

-#include <unordered_map>
+#include <tsl/robin_map.h>

 namespace filament::backend {

@@ -36,8 +36,8 @@ public:
    void terminate() noexcept;

    struct PushConstantKey {
-        uint8_t stage = 0;// We have one set of push constant per shader stage (fragment, vertex, etc).
-        uint8_t size = 0;
+        uint8_t stage;// We have one set of push constant per shader stage (fragment, vertex, etc).
+        uint8_t size;
        // Note that there is also an offset parameter for push constants, but
        // we always assume our update range will have the offset 0.
    };
@@ -73,7 +73,7 @@ private:
        }
    };

-    using PipelineLayoutMap = std::unordered_map<PipelineLayoutKey, PipelineLayoutCacheEntry,
+    using PipelineLayoutMap = tsl::robin_map<PipelineLayoutKey, PipelineLayoutCacheEntry,
            PipelineLayoutKeyHashFn, PipelineLayoutKeyEqual>;

    VkDevice mDevice;
@@ -82,6 +82,6 @@ private:
    PipelineLayoutMap mPipelineLayouts;
 };

-} // filament::backend
+}

 #endif // TNT_FILAMENT_BACKEND_VULKANPIPELINECACHE_H
@@ -325,7 +325,6 @@ VkDevice createLogicalDevice(VkPhysicalDevice physicalDevice,
    // We could simply enable all supported features, but since that may have performance
    // consequences let's just enable the features we need.
    VkPhysicalDeviceFeatures enabledFeatures{
-            .depthClamp = features.depthClamp,
            .samplerAnisotropy = features.samplerAnisotropy,
            .textureCompressionETC2 = features.textureCompressionETC2,
            .textureCompressionBC = features.textureCompressionBC,
@@ -64,15 +64,14 @@ First determine the Filament material and variant that causes the problem.
 What we want is the "raw" GLSL version of the shader, before any optimizations / cross-compilation
 happens.

-We can use the `--save-raw-variants` debug flag in matc to export each GLSL
-shader to a file. For example:
+The easiest way I've found to do this is by adding a `std::cout` statement inside `MaterialBuilder`,
+right after `sg.createFragmentProgram` or `sg.createVertexProgram` is called. For example,

 ```
-matc --save-raw-variants --optimize-size --variant-filter fog,ssr,vsm,stereo \
-        -a all -p all -o mymaterial.filamat mymaterial.mat
+if (v.variant.key == 8u) { std::cout << shader << std::endl; }
 ```

-Files will be named like `mymaterial_0x05.frag` or `mymaterial_0x05.vert`.
+Then, I invoke `matc` and redirect stdout to a .frag or .vert file.

 Note that gltfio material "templates" first go through a build step. After building gltfio, the
 gltfio Filament materials are output to:
@@ -54,7 +54,7 @@ public:
    using BufferDescriptor = backend::BufferDescriptor;
    using BindingType = backend::BufferObjectBinding;

-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;
    public:
        Builder() noexcept;
@@ -78,21 +78,6 @@ public:
         */
        Builder& bindingType(BindingType bindingType) noexcept;

-        /**
-         * Associate an optional name with this BufferObject for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this BufferObject
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-        // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the BufferObject and returns a pointer to it. After creation, the buffer
         * object is uninitialized. Use BufferObject::setBuffer() to initialize it.
@@ -350,12 +350,9 @@ public:
        uint32_t resourceAllocatorCacheSizeMB = 64;

        /*
-         * This value determines how many frames texture entries are kept for in the cache. This
-         * is a soft limit, meaning some texture older than this are allowed to stay in the cache.
-         * Typically only one texture is evicted per frame.
-         * The default is 1.
+         * This value determines for how many frames are texture entries kept in the cache.
         */
-        uint32_t resourceAllocatorCacheMaxAge = 1;
+        uint32_t resourceAllocatorCacheMaxAge = 2;

        /*
         * Disable backend handles use-after-free checks.
@@ -896,29 +893,6 @@ public:
    /** Tells whether an InstanceBuffer object is valid */
    bool isValid(const InstanceBuffer* UTILS_NULLABLE p) const;

-    /**
-     * Retrieve the count of each resource tracked by Engine.
-     * This is intended for debugging.
-     * @{
-     */
-    size_t getBufferObjectCount() const noexcept;
-    size_t getViewCount() const noexcept;
-    size_t getSceneCount() const noexcept;
-    size_t getSwapChainCount() const noexcept;
-    size_t getStreamCount() const noexcept;
-    size_t getIndexBufferCount() const noexcept;
-    size_t getSkinningBufferCount() const noexcept;
-    size_t getMorphTargetBufferCount() const noexcept;
-    size_t getInstanceBufferCount() const noexcept;
-    size_t getVertexBufferCount() const noexcept;
-    size_t getIndirectLightCount() const noexcept;
-    size_t getMaterialCount() const noexcept;
-    size_t getTextureCount() const noexcept;
-    size_t getSkyboxeCount() const noexcept;
-    size_t getColorGradingCount() const noexcept;
-    size_t getRenderTargetCount() const noexcept;
-    /**  @} */
-
    /**
     * Kicks the hardware thread (e.g. the OpenGL, Vulkan or Metal thread) and blocks until
     * all commands to this point are executed. Note that does guarantee that the
@@ -19,7 +19,6 @@

 #include <utils/compiler.h>
 #include <utils/PrivateImplementation.h>
-#include <utils/CString.h>

 #include <stddef.h>

@@ -55,23 +54,6 @@ public:
 template<typename T>
 using BuilderBase = utils::PrivateImplementation<T>;

-// This needs to be public because it is used in the following template.
-UTILS_PUBLIC void builderMakeName(utils::CString& outName, const char* name, size_t len) noexcept;
-
-template <typename Builder>
-class UTILS_PUBLIC BuilderNameMixin {
-public:
-    Builder& name(const char* name, size_t len) noexcept {
-        builderMakeName(mName, name, len);
-        return static_cast<Builder&>(*this);
-    }
-
-    utils::CString const& getName() const noexcept { return mName; }
-
-private:
-    utils::CString mName;
-};
-
 } // namespace filament

 #endif // TNT_FILAMENT_FILAMENTAPI_H
@@ -59,7 +59,7 @@ public:
        UINT = uint8_t(backend::ElementType::UINT),      //!< 32-bit indices
    };

-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;
    public:
        Builder() noexcept;
@@ -83,21 +83,6 @@ public:
         */
        Builder& bufferType(IndexType indexType) noexcept;

-        /**
-         * Associate an optional name with this IndexBuffer for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this IndexBuffer
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-        // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the IndexBuffer object and returns a pointer to it. After creation, the index
         * buffer is uninitialized. Use IndexBuffer::setBuffer() to initialize the IndexBuffer.
@@ -38,7 +38,7 @@ class UTILS_PUBLIC InstanceBuffer : public FilamentAPI {
    struct BuilderDetails;

 public:
-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;

    public:
@@ -70,21 +70,6 @@ public:
         */
        Builder& localTransforms(math::mat4f const* UTILS_NULLABLE localTransforms) noexcept;

-        /**
-         * Associate an optional name with this InstanceBuffer for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this InstanceBuffer
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-        // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the InstanceBuffer object and returns a pointer to it.
         */
@@ -39,7 +39,7 @@ class UTILS_PUBLIC MorphTargetBuffer : public FilamentAPI {
    struct BuilderDetails;

 public:
-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;
    public:
        Builder() noexcept;
@@ -63,21 +63,6 @@ public:
         */
        Builder& count(size_t count) noexcept;

-        /**
-         * Associate an optional name with this MorphTargetBuffer for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this MorphTargetBuffer
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-        // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the MorphTargetBuffer object and returns a pointer to it.
         *
@@ -489,20 +489,43 @@ public:
         * For standard morphing, A MorphTargetBuffer must be provided.
         * Standard morphing supports up to \c CONFIG_MAX_MORPH_TARGET_COUNT morph targets.
         *
+         * For legacy morphing, the attached VertexBuffer must provide data in the
+         * appropriate VertexAttribute slots (\c MORPH_POSITION_0 etc). Legacy morphing only
+         * supports up to 4 morph targets and will be deprecated in the future. Legacy morphing must
+         * be enabled on the material definition: either via the legacyMorphing material attribute
+         * or by calling filamat::MaterialBuilder::useLegacyMorphing().
+         *
         * See also RenderableManager::setMorphWeights(), which can be called on a per-frame basis
         * to advance the animation.
         */
        Builder& morphing(MorphTargetBuffer* UTILS_NONNULL morphTargetBuffer) noexcept;

+        /**
+         * @deprecated Use morphing(uint8_t level, size_t primitiveIndex, size_t offset, size_t count) instead
+         */
+        Builder& morphing(uint8_t level, size_t primitiveIndex,
+                MorphTargetBuffer* UTILS_NONNULL morphTargetBuffer,
+                size_t offset, size_t count) noexcept;
+
+        /**
+         * @deprecated Use morphing(uint8_t level, size_t primitiveIndex, size_t offset, size_t count) instead
+         */
+        inline Builder& morphing(uint8_t level, size_t primitiveIndex,
+                MorphTargetBuffer* UTILS_NONNULL morphTargetBuffer) noexcept {
+            return morphing(level, primitiveIndex, morphTargetBuffer, 0,
+                    morphTargetBuffer->getVertexCount());
+        }
+
        /**
         * Specifies the the range of the MorphTargetBuffer to use with this primitive.
         *
         * @param level the level of detail (lod), only 0 can be specified
         * @param primitiveIndex zero-based index of the primitive, must be less than the count passed to Builder constructor
         * @param offset specifies where in the morph target buffer to start reading (expressed as a number of vertices)
+         * @param count number of vertices in the morph target buffer to read, must equal the geometry's count (for triangles, this should be a multiple of 3)
         */
-        RenderableManager::Builder& morphing(uint8_t level,
-                size_t primitiveIndex, size_t offset) noexcept;
+        Builder& morphing(uint8_t level, size_t primitiveIndex,
+                size_t offset, size_t count) noexcept;


        /**
@@ -763,13 +786,25 @@ public:
    /**
     * Associates a MorphTargetBuffer to the given primitive.
     */
-    void setMorphTargetBufferOffsetAt(Instance instance, uint8_t level, size_t primitiveIndex,
-            size_t offset);
+    void setMorphTargetBufferAt(Instance instance, uint8_t level, size_t primitiveIndex,
+            size_t offset, size_t count);
+
+    /** @deprecated */
+    void setMorphTargetBufferAt(Instance instance, uint8_t level, size_t primitiveIndex,
+            MorphTargetBuffer* UTILS_NONNULL morphTargetBuffer, size_t offset, size_t count);
+
+    /** @deprecated */
+    inline void setMorphTargetBufferAt(Instance instance, uint8_t level, size_t primitiveIndex,
+            MorphTargetBuffer* UTILS_NONNULL morphTargetBuffer) {
+        setMorphTargetBufferAt(instance, level, primitiveIndex, morphTargetBuffer, 0,
+                morphTargetBuffer->getVertexCount());
+    }

    /**
-     * Get a MorphTargetBuffer to the given renderable or null if it doesn't exist.
+     * Get a MorphTargetBuffer to the given primitive or null if it doesn't exist.
     */
-    MorphTargetBuffer* UTILS_NULLABLE getMorphTargetBuffer(Instance instance) const noexcept;
+    MorphTargetBuffer* UTILS_NULLABLE getMorphTargetBufferAt(Instance instance,
+            uint8_t level, size_t primitiveIndex) const noexcept;

    /**
     * Gets the number of morphing in the given entity.
@@ -22,11 +22,9 @@
 #include <filament/FilamentAPI.h>

 #include <utils/compiler.h>
-#include <utils/FixedCapacityVector.h>

 #include <math/vec4.h>

-#include <stddef.h>
 #include <stdint.h>

 namespace filament {
@@ -83,37 +81,6 @@ public:
        UTILS_DEPRECATED uint64_t vsyncOffsetNanos = 0;
    };

-    /**
-     * Timing information about a frame
-     * @see getFrameInfoHistory()
-     */
-    struct FrameInfo {
-        using time_point_ns = int64_t;
-        using duration_ns = int64_t;
-        uint32_t frameId;                   //!< monotonically increasing frame identifier
-        duration_ns frameTime;              //!< frame duration on the GPU in nanosecond [ns]
-        duration_ns denoisedFrameTime;      //!< denoised frame duration on the GPU in [ns]
-        time_point_ns beginFrame;           //!< Renderer::beginFrame() time since epoch [ns]
-        time_point_ns endFrame;             //!< Renderer::endFrame() time since epoch [ns]
-        time_point_ns backendBeginFrame;    //!< Backend thread time of frame start since epoch [ns]
-        time_point_ns backendEndFrame;      //!< Backend thread time of frame end since epoch [ns]
-    };
-
-    /**
-     * Retrieve an historic of frame timing information. The maximum frame history size is
-     * given by getMaxFrameHistorySize().
-     * @param historySize requested history size. The returned vector could be smaller.
-     * @return A vector of FrameInfo.
-     */
-    utils::FixedCapacityVector<Renderer::FrameInfo> getFrameInfoHistory(
-            size_t historySize = 1) const noexcept;
-
-    /**
-     * @return the maximum supported frame history size.
-     * @see getFrameInfoHistory()
-     */
-    size_t getMaxFrameHistorySize() const noexcept;
-
    /**
     * Use FrameRateOptions to set the desired frame rate and control how quickly the system
     * reacts to GPU load changes.
@@ -271,14 +238,6 @@ public:
     */
    void setVsyncTime(uint64_t steadyClockTimeNano) noexcept;

-    /**
-     * Call skipFrame when momentarily skipping frames, for instance if the content of the
-     * scene doesn't change.
-     *
-     * @param vsyncSteadyClockTimeNano
-     */
-    void skipFrame(uint64_t vsyncSteadyClockTimeNano = 0u);
-
    /**
     * Set-up a frame for this Renderer.
     *
@@ -39,7 +39,7 @@ class UTILS_PUBLIC SkinningBuffer : public FilamentAPI {
    struct BuilderDetails;

 public:
-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;
    public:
        Builder() noexcept;
@@ -69,21 +69,6 @@ public:
         */
        Builder& initialize(bool initialize = true) noexcept;

-        /**
-         * Associate an optional name with this SkinningBuffer for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this SkinningBuffer
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-        // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the SkinningBuffer object and returns a pointer to it.
         *
@@ -94,7 +94,7 @@ public:
     *
     * To create a NATIVE stream, call the <pre>stream</pre> method on the builder.
     */
-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;
    public:
        Builder() noexcept;
@@ -136,21 +136,6 @@ public:
         */
        Builder& height(uint32_t height) noexcept;

-        /**
-         * Associate an optional name with this Stream for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this Stream
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-        // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the Stream object and returns a pointer to it.
         *
@@ -112,7 +112,7 @@ public:


    //! Use Builder to construct a Texture object instance
-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;
    public:
        Builder() noexcept;
@@ -202,21 +202,6 @@ public:
         */
        Builder& swizzle(Swizzle r, Swizzle g, Swizzle b, Swizzle a) noexcept;

-        /**
-         * Associate an optional name with this Texture for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this Texture
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-         // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the Texture object and returns a pointer to it.
         *
@@ -61,7 +61,7 @@ public:
    using AttributeType = backend::ElementType;
    using BufferDescriptor = backend::BufferDescriptor;

-    class Builder : public BuilderBase<BuilderDetails>, public BuilderNameMixin<Builder> {
+    class Builder : public BuilderBase<BuilderDetails> {
        friend struct BuilderDetails;
    public:
        Builder() noexcept;
@@ -158,21 +158,6 @@ public:
         */
        Builder& advancedSkinning(bool enabled) noexcept;

-        /**
-         * Associate an optional name with this VertexBuffer for debugging purposes.
-         *
-         * name will show in error messages and should be kept as short as possible. The name is
-         * truncated to a maximum of 128 characters.
-         *
-         * The name string is copied during this method so clients may free its memory after
-         * the function returns.
-         *
-         * @param name A string to identify this VertexBuffer
-         * @param len Length of name, should be less than or equal to 128
-         * @return This Builder, for chaining calls.
-         */
-        // Builder& name(const char* UTILS_NONNULL name, size_t len) noexcept; // inherited
-
        /**
         * Creates the VertexBuffer object and returns a pointer to it.
         *
@@ -24,7 +24,6 @@

 #include <utils/compiler.h>
 #include <utils/Entity.h>
-#include <utils/FixedCapacityVector.h>

 #include <math/mathfwd.h>

@@ -41,7 +40,6 @@ class CallbackHandler;

 class Camera;
 class ColorGrading;
-class Engine;
 class MaterialInstance;
 class RenderTarget;
 class Scene;
@@ -570,13 +568,6 @@ public:
     */
    void setShadowType(ShadowType shadow) noexcept;

-    /**
-     * Returns the shadow mapping technique used by this View.
-     *
-     * @return value set by setShadowType().
-     */
-    ShadowType getShadowType() const noexcept;
-
    /**
     * Sets VSM shadowing options that apply across the entire View.
     *
@@ -735,7 +726,7 @@ public:
    void setDebugCamera(Camera* UTILS_NULLABLE camera) noexcept;

    //! debugging: returns a Camera from the point of view of *the* dominant directional light used for shadowing.
-    utils::FixedCapacityVector<Camera const*> getDirectionalShadowCameras() const noexcept;
+    Camera const* UTILS_NULLABLE getDirectionalShadowCamera() const noexcept;


    /** Result of a picking query */
@@ -887,17 +878,6 @@ public:
     */
    utils::Entity getFogEntity() const noexcept;

-
-    /**
-     * When certain temporal features are used (e.g.: TAA or Screen-space reflections), the view
-     * keeps a history of previous frame renders associated with the Renderer the view was last
-     * used with. When switching Renderer, it may be necessary to clear that history by calling
-     * this method. Similarly, if the whole content of the screen change, like when a cut-scene
-     * starts, clearing the history might be needed to avoid artifacts due to the previous frame
-     * being very different.
-     */
-    void clearFrameHistory(Engine& engine) noexcept;
-
    /**
     * List of available ambient occlusion techniques
     * @deprecated use AmbientOcclusionOptions::enabled instead
@@ -16,8 +16,6 @@

 #include "details/Engine.h"

-#include "ResourceAllocator.h"
-
 #include "details/BufferObject.h"
 #include "details/Camera.h"
 #include "details/Fence.h"
@@ -268,71 +266,6 @@ bool Engine::isValid(const InstanceBuffer* p) const {
    return downcast(this)->isValid(downcast(p));
 }

-size_t Engine::getBufferObjectCount() const noexcept {
-    return downcast(this)->getBufferObjectCount();
-}
-
-size_t Engine::getViewCount() const noexcept {
-    return downcast(this)->getViewCount();
-}
-
-size_t Engine::getSceneCount() const noexcept {
-    return downcast(this)->getSceneCount();
-}
-
-size_t Engine::getSwapChainCount() const noexcept {
-    return downcast(this)->getSwapChainCount();
-}
-
-size_t Engine::getStreamCount() const noexcept {
-    return downcast(this)->getStreamCount();
-}
-
-size_t Engine::getIndexBufferCount() const noexcept {
-    return downcast(this)->getIndexBufferCount();
-}
-
-size_t Engine::getSkinningBufferCount() const noexcept {
-    return downcast(this)->getSkinningBufferCount();
-}
-
-size_t Engine::getMorphTargetBufferCount() const noexcept {
-    return downcast(this)->getMorphTargetBufferCount();
-}
-
-size_t Engine::getInstanceBufferCount() const noexcept {
-    return downcast(this)->getInstanceBufferCount();
-}
-
-size_t Engine::getVertexBufferCount() const noexcept {
-    return downcast(this)->getVertexBufferCount();
-}
-
-size_t Engine::getIndirectLightCount() const noexcept {
-    return downcast(this)->getIndirectLightCount();
-}
-
-size_t Engine::getMaterialCount() const noexcept {
-    return downcast(this)->getMaterialCount();
-}
-
-size_t Engine::getTextureCount() const noexcept {
-    return downcast(this)->getTextureCount();
-}
-
-size_t Engine::getSkyboxeCount() const noexcept {
-    return downcast(this)->getSkyboxeCount();
-}
-
-size_t Engine::getColorGradingCount() const noexcept {
-    return downcast(this)->getColorGradingCount();
-}
-
-size_t Engine::getRenderTargetCount() const noexcept {
-    return downcast(this)->getRenderTargetCount();
-}
-
-
 void Engine::flushAndWait() {
    downcast(this)->flushAndWait();
 }
@@ -1,31 +0,0 @@
-/*
- * Copyright (C) 2024 The Android Open Source Project
- *
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at
- *
- *      http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-#include <filament/FilamentAPI.h>
-
-#include <algorithm>
-
-namespace filament {
-
-void builderMakeName(utils::CString& outName, const char* name, size_t len) noexcept {
-    if (!name) {
-        return;
-    }
-    size_t const length = std::min(len, size_t { 128u });
-    outName = utils::CString(name, length);
-}
-
-} // namespace filament
@@ -21,9 +21,6 @@
 #include <fg/FrameGraphTexture.h>

 #include <math/mat4.h>
-#include <math/vec2.h>
-
-#include <stdint.h>

 namespace filament {

@@ -16,34 +16,30 @@

 #include "FrameInfo.h"

-#include <filament/Renderer.h>
-
-#include <backend/DriverEnums.h>
-
-#include <utils/compiler.h>
-#include <utils/debug.h>
-#include <utils/FixedCapacityVector.h>
 #include <utils/Log.h>
-#include <utils/ostream.h>
 #include <utils/Systrace.h>

-#include <algorithm>
-#include <atomic>
-#include <chrono>
-#include <memory>
-#include <ratio>
+#include <math/scalar.h>

-#include <stdint.h>
-#include <stddef.h>
+#include <cmath>

 namespace filament {

 using namespace utils;
 using namespace backend;

+// this is to avoid a call to memmove
+template<class InputIterator, class OutputIterator>
+static inline
+void move_backward(InputIterator first, InputIterator last, OutputIterator result) {
+    while (first != last) {
+        *--result = *--last;
+    }
+}
+
 FrameInfoManager::FrameInfoManager(DriverApi& driver) noexcept {
    for (auto& query : mQueries) {
-        query.handle = driver.createTimerQuery();
+        query = driver.createTimerQuery();
    }
 }

@@ -51,155 +47,65 @@ FrameInfoManager::~FrameInfoManager() noexcept = default;

 void FrameInfoManager::terminate(DriverApi& driver) noexcept {
    for (auto& query : mQueries) {
-        driver.destroyTimerQuery(query.handle);
+        driver.destroyTimerQuery(query);
    }
 }

-void FrameInfoManager::beginFrame(DriverApi& driver, Config const& config, uint32_t frameId) noexcept {
-    auto& history = mFrameTimeHistory;
-    // don't exceed the capacity, drop the oldest entry
-    if (UTILS_LIKELY(history.size() == history.capacity())) {
-        history.pop_back();
-    }
-
-    // create a new entry
-    auto& front = history.emplace_front(frameId);
-
-    // store the current time
-    front.beginFrame = std::chrono::steady_clock::now();
-
-    // references are not invalidated by CircularQueue<>, so we can associate a reference to
-    // the slot we created to the timer query used to find the frame time.
-    mQueries[mIndex].pInfo = std::addressof(front);
-    // issue the timer query
-    driver.beginTimerQuery(mQueries[mIndex].handle);
-    // issue the custom backend command to get the backend time
-    driver.queueCommand([&front](){
-        front.backendBeginFrame = std::chrono::steady_clock::now();
-    });
-
-    // now is a good time to check the oldest active query
-    while (mLast != mIndex) {
-        uint64_t elapsed = 0;
-        TimerQueryResult const result = driver.getTimerQueryValue(mQueries[mLast].handle, &elapsed);
-        switch (result) {
-            case TimerQueryResult::NOT_READY:
-                // nothing to do
-                break;
-            case TimerQueryResult::ERROR:
-                mLast = (mLast + 1) % POOL_COUNT;
-                break;
-            case TimerQueryResult::AVAILABLE: {
-                SYSTRACE_CONTEXT();
-                SYSTRACE_VALUE32("FrameInfo::elapsed", uint32_t(elapsed));
-                // conversion to our duration happens here
-                pFront = mQueries[mLast].pInfo;
-                pFront->frameTime = std::chrono::duration<uint64_t, std::nano>(elapsed);
-                mLast = (mLast + 1) % POOL_COUNT;
-                denoiseFrameTime(history, config);
-                break;
-            }
-        }
-        if (result != TimerQueryResult::AVAILABLE) {
+void FrameInfoManager::beginFrame(DriverApi& driver,Config const& config, uint32_t) noexcept {
+    driver.beginTimerQuery(mQueries[mIndex]);
+    uint64_t elapsed = 0;
+    TimerQueryResult const result = driver.getTimerQueryValue(mQueries[mLast], &elapsed);
+    switch (result) {
+        case TimerQueryResult::NOT_READY:
+            // nothing to do
+            break;
+        case TimerQueryResult::ERROR:
+            mLast = (mLast + 1) % POOL_COUNT;
+            break;
+        case TimerQueryResult::AVAILABLE:
+            mLast = (mLast + 1) % POOL_COUNT;
+            // conversion to our duration happens here
+            mFrameTime = std::chrono::duration<uint64_t, std::nano>(elapsed);
            break;
-        }
-        // read the pending timer queries until we find one that's not ready
-    }
-
-    // keep this just for debugging
-    if constexpr (false) {
-        using namespace utils;
-        auto h = getFrameInfoHistory(1);
-        if (!h.empty()) {
-            slog.d << frameId << ": "
-                   << h[0].frameId << " (" << frameId - h[0].frameId << ")"
-                   << ", Dm=" << h[0].endFrame - h[0].beginFrame
-                   << ", L =" << h[0].backendBeginFrame - h[0].beginFrame
-                   << ", Db=" << h[0].backendEndFrame - h[0].backendBeginFrame
-                   << ", T =" << h[0].frameTime
-                   << io::endl;
-        }
    }
+    update(config, mFrameTime);
 }

 void FrameInfoManager::endFrame(DriverApi& driver) noexcept {
-    auto& front = mFrameTimeHistory.front();
-    // close the timer query
-    driver.endTimerQuery(mQueries[mIndex].handle);
-    // queue custom backend command to query the current time
-    driver.queueCommand([&front](){
-        // backend frame end-time
-        front.backendEndFrame = std::chrono::steady_clock::now();
-        // signal that the data is available
-        front.ready.store(true, std::memory_order_release);
-    });
-    // and finally acquire the time on the main thread
-    front.endFrame = std::chrono::steady_clock::now();
+    driver.endTimerQuery(mQueries[mIndex]);
    mIndex = (mIndex + 1) % POOL_COUNT;
 }

-void FrameInfoManager::denoiseFrameTime(FrameHistoryQueue& history, Config const& config) noexcept {
-    assert_invariant(!history.empty());
+void FrameInfoManager::update(Config const& config,
+        FrameInfoManager::duration lastFrameTime) noexcept {
+    // keep an history of frame times
+    auto& history = mFrameTimeHistory;

-    // find the first slot that has a valid frame duration
-    size_t first = history.size();
-    for (size_t i = 0, c = history.size(); i < c; ++i) {
-        if (history[i].frameTime != duration(0)) {
-            first = i;
-            break;
-        }
+    // this is like doing { pop_back(); push_front(); }
+    filament::move_backward(history.begin(), history.end() - 1, history.end());
+    history[0].frameTime = lastFrameTime;
+
+    mFrameTimeHistorySize = std::min(++mFrameTimeHistorySize, uint32_t(MAX_FRAMETIME_HISTORY));
+    if (UTILS_UNLIKELY(mFrameTimeHistorySize < 3)) {
+        // not enough history to do anything useful
+        history[0].valid = false;
+        return;
    }
-    assert_invariant(first != history.size());

-    // we need at least 3 valid frame time to calculate the median
-    if (history.size() >= first + 3) {
-        // apply a median filter to get a good representation of the frame time of the last
-        // N frames.
-        std::array<duration, MAX_FRAMETIME_HISTORY> median; // NOLINT -- it's initialized below
-        size_t const size = std::min({
-            history.size() - first,
-            median.size(),
-            size_t(config.historySize) });
+    // apply a median filter to get a good representation of the frame time of the last
+    // N frames.
+    std::array<duration, MAX_FRAMETIME_HISTORY> median; // NOLINT -- it's initialized below
+    size_t const size = std::min(mFrameTimeHistorySize,
+            std::min(config.historySize, (uint32_t)median.size()));
+    for (size_t i = 0; i < size; ++i) {
+        median[i] = history[i].frameTime;
+    }
+    std::sort(median.begin(), median.begin() + size);
+    duration const denoisedFrameTime = median[size / 2];

-        for (size_t i = 0; i < size; ++i) {
-            median[i] = history[first + i].frameTime;
-        }
-        std::sort(median.begin(), median.begin() + size);
-        duration const denoisedFrameTime = median[size / 2];
-
-        history[first].denoisedFrameTime = denoisedFrameTime;
-        history[first].valid = true;
-     }
+    history[0].denoisedFrameTime = denoisedFrameTime;
+    history[0].valid = true;
 }

-utils::FixedCapacityVector<Renderer::FrameInfo> FrameInfoManager::getFrameInfoHistory(
-        size_t historySize) const noexcept {
-    auto result = utils::FixedCapacityVector<Renderer::FrameInfo>::with_capacity(MAX_FRAMETIME_HISTORY);
-    auto const& history = mFrameTimeHistory;
-    size_t i = 0;
-    size_t const c = history.size();
-    for (; i < c; ++i) {
-        auto const& entry = history[i];
-        if (entry.ready.load(std::memory_order_acquire) && entry.valid) {
-            // once we found an entry ready,
-            // we know by construction that all following ones are too
-            break;
-        }
-    }
-    for (; i < c && historySize; ++i, --historySize) {
-        auto const& entry = history[i];
-        using namespace std::chrono;
-        result.push_back({
-                entry.frameId,
-                duration_cast<nanoseconds>(entry.frameTime).count(),
-                duration_cast<nanoseconds>(entry.denoisedFrameTime).count(),
-                duration_cast<nanoseconds>(entry.beginFrame.time_since_epoch()).count(),
-                duration_cast<nanoseconds>(entry.endFrame.time_since_epoch()).count(),
-                duration_cast<nanoseconds>(entry.backendBeginFrame.time_since_epoch()).count(),
-                duration_cast<nanoseconds>(entry.backendEndFrame.time_since_epoch()).count()
-        });
-    }
-    return result;
-}

 } // namespace filament
@@ -17,136 +17,30 @@
 #ifndef TNT_FILAMENT_FRAMEINFO_H
 #define TNT_FILAMENT_FRAMEINFO_H

-#include <filament/Renderer.h>
-
-#include <backend/Handle.h>
-
+#include "backend/Handle.h"
 #include <private/backend/DriverApi.h>

-#include <utils/compiler.h>
-#include <utils/debug.h>
-#include <utils/FixedCapacityVector.h>
-
 #include <array>
-#include <atomic>
 #include <chrono>
-#include <ratio>
-#include <type_traits>

 #include <stdint.h>
-#include <stddef.h>

 namespace filament {
 class FEngine;

-namespace details {
 struct FrameInfo {
    using duration = std::chrono::duration<float, std::milli>;
    duration frameTime{};            // frame period
    duration denoisedFrameTime{};    // frame period (median filter)
-    bool valid = false;              // true if the data of the structure is valid
-};
-} // namespace details
-
-struct FrameInfoImpl : public details::FrameInfo {
-    using clock = std::chrono::steady_clock;
-    using time_point = clock::time_point;
-    uint32_t const frameId;
-    time_point beginFrame;           // main thread beginFrame time
-    time_point endFrame;             // main thread endFrame time
-    time_point backendBeginFrame;    // backend thread beginFrame time (makeCurrent time)
-    time_point backendEndFrame;      // backend thread endFrame time (present time)
-    std::atomic_bool ready{};        // true once backend thread has populated its data
-    explicit FrameInfoImpl(uint32_t frameId) noexcept
-        : frameId(frameId) {
-    }
-};
-
-template<typename T, size_t CAPACITY>
-class CircularQueue {
-public:
-    using value_type = T;
-    using reference = value_type&;
-    using const_reference = value_type const&;
-
-    size_t capacity() const {
-        return CAPACITY;
-    }
-
-    size_t size() const {
-        return mSize;
-    }
-
-    bool empty() const noexcept {
-        return !size();
-    }
-
-    void pop_back() noexcept {
-        assert_invariant(!empty());
-        --mSize;
-        std::destroy_at(&mStorage[(mFront - mSize) % CAPACITY]);
-    }
-
-    void push_front(T const& v) noexcept {
-        assert_invariant(size() < CAPACITY);
-        mFront = advance(mFront);
-        new(&mStorage[mFront]) T(v);
-        ++mSize;
-    }
-
-    void push_front(T&& v) noexcept {
-        assert_invariant(size() < CAPACITY);
-        mFront = advance(mFront);
-        new(&mStorage[mFront]) T(std::move(v));
-        ++mSize;
-    }
-
-    template<typename ...Args>
-    T& emplace_front(Args&&... args) noexcept {
-        assert_invariant(size() < CAPACITY);
-        mFront = advance(mFront);
-        new(&mStorage[mFront]) T(std::forward<Args>(args)...);
-        ++mSize;
-        return front();
-    }
-
-    T& operator[](size_t pos) noexcept {
-        assert_invariant(pos < size());
-        size_t const index = (mFront + CAPACITY - pos) % CAPACITY;
-        return *std::launder(reinterpret_cast<T*>(&mStorage[index]));
-    }
-
-    T const& operator[](size_t pos) const noexcept {
-        return const_cast<CircularQueue&>(*this)[pos];
-    }
-
-    T const& front() const noexcept {
-        assert_invariant(!empty());
-        return operator[](0);
-    }
-
-    T& front() noexcept {
-        assert_invariant(!empty());
-        return operator[](0);
-    }
-
-private:
-    using Storage = std::aligned_storage_t<sizeof(T), alignof(T)>;
-    Storage mStorage[CAPACITY];
-    uint32_t mFront = 0;    // always index 0
-    uint32_t mSize = 0;
-    [[nodiscard]] inline uint32_t advance(uint32_t v) noexcept {
-        return (v + 1) % CAPACITY;
-    }
+    bool valid = false;
 };

 class FrameInfoManager {
    static constexpr size_t POOL_COUNT = 4;
-    static constexpr size_t MAX_FRAMETIME_HISTORY = 16u;
+    static constexpr size_t MAX_FRAMETIME_HISTORY = 31u;

 public:
-    using duration = FrameInfoImpl::duration;
-    using clock = FrameInfoImpl::clock;
+    using duration = FrameInfo::duration;

    struct Config {
        uint32_t historySize;
@@ -163,25 +57,23 @@ public:
    // call this immediately before "swap buffers"
    void endFrame(backend::DriverApi& driver) noexcept;

-    details::FrameInfo getLastFrameInfo() const noexcept {
-        // if pFront is not set yet, return FrameInfo(). But the `valid` field will be false in this case.
-        return pFront ? *pFront : details::FrameInfo{};
+    FrameInfo const& getLastFrameInfo() const noexcept {
+        return mFrameTimeHistory[0];
    }

-    utils::FixedCapacityVector<Renderer::FrameInfo> getFrameInfoHistory(size_t historySize) const noexcept;
+    duration getLastFrameTime() const noexcept {
+        return getLastFrameInfo().frameTime;
+    }

 private:
-    using FrameHistoryQueue = CircularQueue<FrameInfoImpl, MAX_FRAMETIME_HISTORY>;
-    static void denoiseFrameTime(FrameHistoryQueue& history, Config const& config) noexcept;
-    struct Query {
-        backend::Handle<backend::HwTimerQuery> handle{};
-        FrameInfoImpl* pInfo = nullptr;
-    };
-    std::array<Query, POOL_COUNT> mQueries;
-    uint32_t mIndex = 0;                // index of current query
-    uint32_t mLast = 0;                 // index of oldest query still active
-    FrameInfoImpl* pFront = nullptr;    // the most recent slot with a valid frame time
-    FrameHistoryQueue mFrameTimeHistory;
+    void update(Config const& config, duration lastFrameTime) noexcept;
+    backend::Handle<backend::HwTimerQuery> mQueries[POOL_COUNT];
+    duration mFrameTime{};
+    uint32_t mIndex = 0;
+    uint32_t mLast = 0;
+
+    std::array<FrameInfo, MAX_FRAMETIME_HISTORY> mFrameTimeHistory;
+    uint32_t mFrameTimeHistorySize = 0;
 };


@@ -16,22 +16,17 @@

 #include "FrameSkipper.h"

-#include <backend/DriverEnums.h>
-
-#include <utils/compiler.h>
+#include <utils/Log.h>
 #include <utils/debug.h>

-#include <algorithm>
-
-#include <stddef.h>
-
 namespace filament {

 using namespace utils;
 using namespace backend;

 FrameSkipper::FrameSkipper(size_t latency) noexcept
-        : mLast(std::clamp(latency, size_t(1), MAX_FRAME_LATENCY) - 1) {
+        : mLast(latency - 1) {
+    assert_invariant(latency <= MAX_FRAME_LATENCY);
 }

 FrameSkipper::~FrameSkipper() noexcept = default;
@@ -46,32 +41,31 @@ void FrameSkipper::terminate(DriverApi& driver) noexcept {

 bool FrameSkipper::beginFrame(DriverApi& driver) noexcept {
    auto& fences = mDelayedFences;
-    if (fences.front()) {
-        // Do we have a latency old fence?
-        auto status = driver.getFenceStatus(fences.front());
-        if (UTILS_UNLIKELY(status == FenceStatus::TIMEOUT_EXPIRED)) {
-            // The fence hasn't signaled yet, skip this frame
+    auto fence = fences.front();
+    if (fence) {
+        auto status = driver.getFenceStatus(fence);
+        if (status == FenceStatus::TIMEOUT_EXPIRED) {
+            // Sync not ready, skip frame
            return false;
        }
        assert_invariant(status == FenceStatus::CONDITION_SATISFIED);
+        driver.destroyFence(fence);
    }
+    // shift all fences down by 1
+    std::move(fences.begin() + 1, fences.end(), fences.begin());
+    fences.back() = {};
    return true;
 }

 void FrameSkipper::endFrame(DriverApi& driver) noexcept {
-    auto& fences = mDelayedFences;
-    size_t const last = mLast;
-
-    // pop the oldest fence and advance the other ones
-    if (fences.front()) {
-        driver.destroyFence(fences.front());
+    // If the user produced a new frame despite the fact that the previous one wasn't finished
+    // (i.e. FrameSkipper::beginFrame() returned false), we need to make sure to replace
+    // a fence that might be here already)
+    auto& fence = mDelayedFences[mLast];
+    if (fence) {
+        driver.destroyFence(fence);
    }
-    std::move(fences.begin() + 1, fences.end(), fences.begin());
-
-    // add a new fence to the end
-    assert_invariant(!fences[last]);
-
-    fences[last] = driver.createFence();
+    fence = driver.createFence();
 }

 } // namespace filament
@@ -22,9 +22,6 @@

 #include <array>

-#include <stddef.h>
-#include <stdint.h>
-
 namespace filament {

 /*
@@ -32,18 +29,7 @@ namespace filament {
 * outrun the GPU.
 */
 class FrameSkipper {
-    /*
-     * The maximum frame latency acceptable on ANDROID is 2 because higher latencies will be
-     * throttled anyway in BufferQueueProducer::dequeueBuffer(), because ANDROID is generally
-     * triple-buffered no more; that case is actually pretty bad because the GL thread can block
-     * anywhere (usually inside the first draw command that touches the swapchain).
-     *
-     * A frame latency of 1 has the benefit of reducing render latency,
-     * but the drawback of preventing CPU / GPU overlap.
-     *
-     * Generally a frame latency of 2 is the best compromise.
-     */
-    static constexpr size_t MAX_FRAME_LATENCY = 2;
+    static constexpr size_t MAX_FRAME_LATENCY = 3;
 public:
    /*
     * The latency parameter defines how many unfinished frames we want to accept before we start
@@ -74,7 +60,7 @@ public:
 private:
    using Container = std::array<backend::Handle<backend::HwFence>, MAX_FRAME_LATENCY>;
    mutable Container mDelayedFences{};
-    uint8_t const mLast;
+    size_t mLast;
 };

 } // namespace filament
@@ -26,7 +26,6 @@
 #include <filament/Viewport.h>

 #include <utils/BinaryTreeArray.h>
-#include <utils/Log.h>
 #include <utils/Systrace.h>
 #include <utils/debug.h>

@@ -354,12 +354,6 @@ bool MaterialParser::getSpecularAntiAliasingThreshold(float* value) const noexce
    return mImpl.getFromSimpleChunk(ChunkType::MaterialSpecularAntiAliasingThreshold, value);
 }

-bool MaterialParser::getStereoscopicType(backend::StereoscopicType* value) const noexcept {
-    static_assert(sizeof(StereoscopicType) == sizeof(uint8_t),
-            "StereoscopicType expected size is wrong");
-    return mImpl.getFromSimpleChunk(ChunkType::MaterialStereoscopicType, reinterpret_cast<uint8_t*>(value));
-}
-
 bool MaterialParser::getRequiredAttributes(AttributeBitset* value) const noexcept {
    uint32_t rawAttributes = 0;
    if (!mImpl.getFromSimpleChunk(ChunkType::MaterialRequiredAttributes, &rawAttributes)) {
@@ -122,7 +122,6 @@ public:
    bool hasSpecularAntiAliasing(bool* value) const noexcept;
    bool getSpecularAntiAliasingVariance(float* value) const noexcept;
    bool getSpecularAntiAliasingThreshold(float* value) const noexcept;
-    bool getStereoscopicType(backend::StereoscopicType*) const noexcept;

    bool getShader(filaflat::ShaderContent& shader, backend::ShaderModel shaderModel,
            Variant variant, backend::ShaderStage stage) noexcept;
@@ -459,12 +459,10 @@ void PerViewUniforms::bind(backend::DriverApi& driver) noexcept {

 void PerViewUniforms::unbindSamplers() noexcept {
    auto& samplerGroup = mSamplers;
-    samplerGroup.clearSampler(PerViewSib::SHADOW_MAP);
-    samplerGroup.clearSampler(PerViewSib::IBL_SPECULAR);
    samplerGroup.clearSampler(PerViewSib::SSAO);
    samplerGroup.clearSampler(PerViewSib::SSR);
    samplerGroup.clearSampler(PerViewSib::STRUCTURE);
-    samplerGroup.clearSampler(PerViewSib::FOG);
+    samplerGroup.clearSampler(PerViewSib::SHADOW_MAP);
 }

 } // namespace filament
@@ -494,6 +494,7 @@ PostProcessManager::StructurePassOutput PostProcessManager::structure(FrameGraph

    struct StructureMipmapData {
        FrameGraphId<FrameGraphTexture> depth;
+        uint32_t rt[8];
    };

    fg.addPass<StructureMipmapData>("StructureMipmap",
@@ -504,7 +505,7 @@ PostProcessManager::StructurePassOutput PostProcessManager::structure(FrameGraph
                            .level = uint8_t(i)
                    });
                    out = builder.write(out, FrameGraphTexture::Usage::DEPTH_ATTACHMENT);
-                    builder.declareRenderPass("Structure mip target", {
+                    data.rt[i - 1] = builder.declareRenderPass("Structure mip target", {
                            .attachments = { .depth = out }
                    });
                }
@@ -1964,8 +1965,8 @@ PostProcessManager::BloomPassOutput PostProcessManager::bloom(FrameGraph& fg,
        // - visible bloom size changes with dynamic resolution in non-homogenous mode
        // This allows us to use the 9 sample downsampling filter (instead of 13)
        // for at least 4 levels.
-        uint32_t width  = std::max(16u, uint32_t(std::floor(bloomWidth)));
-        uint32_t height = std::max(16u, uint32_t(std::floor(bloomHeight)));
+        uint32_t width  = std::max(1u, uint32_t(std::floor(bloomWidth)));
+        uint32_t height = std::max(1u, uint32_t(std::floor(bloomHeight)));
        width  &= ~((1 << 4) - 1);  // at least 4 levels
        height &= ~((1 << 4) - 1);
        bloomWidth  = float(width);
@@ -1977,8 +1978,6 @@ PostProcessManager::BloomPassOutput PostProcessManager::bloom(FrameGraph& fg,
    // we don't need to do the fireflies reduction if we have TAA (it already does it)
    bool fireflies = threshold && !taaOptions.enabled;

-    assert_invariant(bloomWidth && bloomHeight);
-
    while (2 * bloomWidth < float(desc.width) || 2 * bloomHeight < float(desc.height)) {
        if (inoutBloomOptions.quality == QualityLevel::LOW ||
            inoutBloomOptions.quality == QualityLevel::MEDIUM) {
@@ -2945,10 +2944,6 @@ FrameGraphId<FrameGraphTexture> PostProcessManager::upscale(FrameGraph& fg, bool
                                    1.0f / outputDesc.height});
                    }

-                    if (blitterNames[index] == "blitLow") {
-                        mi->setParameter("levelOfDetail", 0.0f);
-                    }
-
                    mi->setParameter("viewport", float4{
                            (float)vp.left   / inputDesc.width,
                            (float)vp.bottom / inputDesc.height,
@@ -3007,8 +3002,9 @@ FrameGraphId<FrameGraphTexture> PostProcessManager::blit(FrameGraph& fg, bool tr
        SamplerMagFilter filterMag,
        SamplerMinFilter filterMin) noexcept {

-    uint32_t const layer = fg.getSubResourceDescriptor(input).layer;
-    float const levelOfDetail = fg.getSubResourceDescriptor(input).level;
+    // TODO: add support for sub-resources
+    assert_invariant(fg.getSubResourceDescriptor(input).layer == 0);
+    assert_invariant(fg.getSubResourceDescriptor(input).level == 0);

    struct QuadBlitData {
        FrameGraphId<FrameGraphTexture> input;
@@ -3034,8 +3030,7 @@ FrameGraphId<FrameGraphTexture> PostProcessManager::blit(FrameGraph& fg, bool tr
                // --------------------------------------------------------------------------------
                // set uniforms

-                PostProcessMaterial const& material =
-                        getPostProcessMaterial(layer ? "blitArray" : "blitLow");
+                PostProcessMaterial const& material = getPostProcessMaterial("blitLow");
                auto* mi = material.getMaterialInstance(mEngine);
                mi->setParameter("color", color, {
                        .filterMag = filterMag,
@@ -3047,10 +3042,6 @@ FrameGraphId<FrameGraphTexture> PostProcessManager::blit(FrameGraph& fg, bool tr
                        float(vp.width)  / inputDesc.width,
                        float(vp.height) / inputDesc.height
                });
-                mi->setParameter("levelOfDetail", levelOfDetail);
-                if (layer) {
-                    mi->setParameter("layerIndex", layer);
-                }
                mi->commit(driver);
                mi->use(driver);

@@ -3193,7 +3184,6 @@ FrameGraphId<FrameGraphTexture> PostProcessManager::resolve(FrameGraph& fg,
                assert_invariant(dst);
                assert_invariant(srcDesc.format == dstDesc.format);
                assert_invariant(srcDesc.width == dstDesc.width && srcDesc.height == dstDesc.height);
-                assert_invariant(srcDesc.samples > 1 && dstDesc.samples <= 1);
                driver.resolve(
                        dst, dstSubDesc.level, dstSubDesc.layer,
                        src, srcSubDesc.level, srcSubDesc.layer);
@@ -38,7 +38,6 @@
 #include <backend/PipelineState.h>

 #include "private/backend/CircularBuffer.h"
-#include "private/backend/CommandStream.h"

 #include <utils/compiler.h>
 #include <utils/debug.h>
@@ -118,11 +117,10 @@ RenderPass::RenderPass(FEngine& engine, RenderPassBuilder const& builder) noexce
    uint32_t const customCommandCount =
            builder.mCustomCommands.has_value() ? builder.mCustomCommands->size() : 0;

-    Command* const commandBegin = builder.mArena.alloc<Command>(commandCount + customCommandCount);
-    Command* commandEnd = commandBegin + (commandCount + customCommandCount);
-    assert_invariant(commandBegin);
+    Command* const curr = builder.mArena.alloc<Command>(commandCount + customCommandCount);
+    assert_invariant(curr);

-    if (UTILS_UNLIKELY(builder.mArena.getAllocator().isHeapAllocation(commandBegin))) {
+    if (UTILS_UNLIKELY(builder.mArena.getAllocator().isHeapAllocation(curr))) {
        static bool sLogOnce = true;
        if (UTILS_UNLIKELY(sLogOnce)) {
            sLogOnce = false;
@@ -131,7 +129,10 @@ RenderPass::RenderPass(FEngine& engine, RenderPassBuilder const& builder) noexce
        }
    }

-    appendCommands(engine, { commandBegin, commandCount },
+    mCommandBegin = curr;
+    mCommandEnd = curr + commandCount + customCommandCount;
+
+    appendCommands(engine, { curr, commandCount },
            builder.mUboHandle,
            builder.mVisibleRenderables,
            builder.mCommandTypeFlags,
@@ -142,36 +143,33 @@ RenderPass::RenderPass(FEngine& engine, RenderPassBuilder const& builder) noexce
            builder.mCameraForwardVector);

    if (builder.mCustomCommands.has_value()) {
-        Command* p = commandBegin + commandCount;
+        Command* p = curr + commandCount;
        for (auto [channel, passId, command, order, fn]: builder.mCustomCommands.value()) {
            appendCustomCommand(p++, channel, passId, command, order, fn);
        }
    }

    // sort commands once we're done adding commands
-    commandEnd = resize(builder.mArena,
-            RenderPass::sortCommands(commandBegin, commandEnd));
+    sortCommands(builder.mArena);

    if (engine.isAutomaticInstancingEnabled()) {
-        int32_t stereoscopicEyeCount = 1;
+        uint32_t stereoscopicEyeCount = 1;
        if (builder.mFlags & IS_INSTANCED_STEREOSCOPIC) {
            stereoscopicEyeCount *= engine.getConfig().stereoscopicEyeCount;
        }
-        commandEnd = resize(builder.mArena,
-                instanceify(engine, commandBegin, commandEnd, stereoscopicEyeCount));
+        instanceify(engine, builder.mArena, stereoscopicEyeCount);
    }
-
-    // these are `const` from this point on...
-    mCommandBegin = commandBegin;
-    mCommandEnd = commandEnd;
 }

 // this destructor is actually heavy because it inlines ~vector<>
-RenderPass::~RenderPass() noexcept = default;
+RenderPass::~RenderPass() noexcept {
+}

-RenderPass::Command* RenderPass::resize(Arena& arena, Command* const last) noexcept {
-    arena.rewind(last);
-    return last;
+void RenderPass::resize(Arena& arena, size_t count) noexcept {
+    if (mCommandBegin) {
+        mCommandEnd = mCommandBegin + count;
+        arena.rewind(mCommandEnd);
+    }
 }

 void RenderPass::appendCommands(FEngine& engine,
@@ -223,7 +221,7 @@ void RenderPass::appendCommands(FEngine& engine,
        work(vr.first, vr.size());
    } else {
        auto* jobCommandsParallel = jobs::parallel_for(js, nullptr, vr.first, (uint32_t)vr.size(),
-                std::cref(work), jobs::CountSplitter<JOBS_PARALLEL_FOR_COMMANDS_COUNT>());
+                std::cref(work), jobs::CountSplitter<JOBS_PARALLEL_FOR_COMMANDS_COUNT, 5>());
        js.runAndWait(jobCommandsParallel);
    }

@@ -262,19 +260,18 @@ void RenderPass::appendCustomCommand(Command* commands,
    commands->key = cmd;
 }

-RenderPass::Command* RenderPass::sortCommands(
-        Command* const begin, Command* const end) noexcept {
-    SYSTRACE_NAME("sort commands");
+void RenderPass::sortCommands(Arena& arena) noexcept {
+    SYSTRACE_NAME("sort and trim commands");

-    std::sort(begin, end);
+    std::sort(mCommandBegin, mCommandEnd);

    // find the last command
-    Command* const last = std::partition_point(begin, end,
+    Command const* const last = std::partition_point(mCommandBegin, mCommandEnd,
            [](Command const& c) {
                return c.key != uint64_t(Pass::SENTINEL);
            });

-    return last;
+    resize(arena, uint32_t(last - mCommandBegin));
 }

 void RenderPass::execute(RenderPass const& pass,
@@ -287,9 +284,7 @@ void RenderPass::execute(RenderPass const& pass,
    driver.endRenderPass();
 }

-RenderPass::Command* RenderPass::instanceify(FEngine& engine,
-        Command* curr, Command* const last,
-        int32_t eyeCount) const noexcept {
+void RenderPass::instanceify(FEngine& engine, Arena& arena, int32_t eyeCount) noexcept {
    SYSTRACE_NAME("instanceify");

    // instanceify works by scanning the **sorted** command stream, looking for repeat draw
@@ -302,12 +297,14 @@ RenderPass::Command* RenderPass::instanceify(FEngine& engine,

    UTILS_UNUSED uint32_t drawCallsSavedCount = 0;

+    Command* curr = mCommandBegin;
+    Command* const last = mCommandEnd;
+
    Command* firstSentinel = nullptr;
    PerRenderableData const* uboData = nullptr;
    PerRenderableData* stagingBuffer = nullptr;
    uint32_t stagingBufferSize = 0;
    uint32_t instancedPrimitiveOffset = 0;
-    size_t const count = last - curr;

    // TODO: for the case of instancing we could actually use 128 instead of 64 instances
    constexpr size_t maxInstanceCount = CONFIG_MAX_INSTANCES;
@@ -346,6 +343,7 @@ RenderPass::Command* RenderPass::instanceify(FEngine& engine,
            if (UTILS_UNLIKELY(!stagingBuffer)) {

                // create a temporary UBO for instancing
+                size_t const count = mCommandEnd - mCommandBegin;
                mInstancedUboHandle = BufferObjectSharedHandle{
                        engine.getDriverApi().createBufferObject(
                                count * sizeof(PerRenderableData) + sizeof(PerRenderableUib),
@@ -387,7 +385,7 @@ RenderPass::Command* RenderPass::instanceify(FEngine& engine,

    if (UTILS_UNLIKELY(firstSentinel)) {
        //slog.d << "auto-instancing, saving " << drawCallsSavedCount << " draw calls, out of "
-        //       << count << io::endl;
+        //       << mCommandEnd - mCommandBegin << io::endl;

        // we have instanced primitives
        DriverApi& driver = engine.getDriverApi();
@@ -403,15 +401,14 @@ RenderPass::Command* RenderPass::instanceify(FEngine& engine,
        stagingBuffer = nullptr;

        // remove all the canceled commands
-        auto lastCommand = std::remove_if(firstSentinel, last, [](auto const& command) {
+        auto lastCommand = std::remove_if(firstSentinel, mCommandEnd, [](auto const& command) {
            return command.key == uint64_t(Pass::SENTINEL);
        });

-        return lastCommand;
+        resize(arena, uint32_t(lastCommand - mCommandBegin));
    }

    assert_invariant(stagingBuffer == nullptr);
-    return last;
 }


@@ -419,8 +416,7 @@ RenderPass::Command* RenderPass::instanceify(FEngine& engine,
 UTILS_ALWAYS_INLINE // This function exists only to make the code more readable. we want it inlined.
 inline              // and we don't need it in the compilation unit
 void RenderPass::setupColorCommand(Command& cmdDraw, Variant variant,
-        FMaterialInstance const* const UTILS_RESTRICT mi,
-        bool inverseFrontFaces, bool hasDepthClamp) noexcept {
+        FMaterialInstance const* const UTILS_RESTRICT mi, bool inverseFrontFaces) noexcept {

    FMaterial const * const UTILS_RESTRICT ma = mi->getMaterial();
    variant = Variant::filterVariant(variant, ma->isVariantLit());
@@ -461,7 +457,6 @@ void RenderPass::setupColorCommand(Command& cmdDraw, Variant variant,
    cmdDraw.info.rasterState.colorWrite = mi->isColorWriteEnabled();
    cmdDraw.info.rasterState.depthWrite = mi->isDepthWriteEnabled();
    cmdDraw.info.rasterState.depthFunc = mi->getDepthFunc();
-    cmdDraw.info.rasterState.depthClamp = hasDepthClamp;
    cmdDraw.info.materialVariant = variant;
    // we keep "RasterState::colorWrite" to the value set by material (could be disabled)
 }
@@ -560,9 +555,6 @@ RenderPass::Command* RenderPass::generateCommandsImpl(RenderPass::CommandTypeFla
    bool const hasInstancedStereo =
            renderFlags & IS_INSTANCED_STEREOSCOPIC;

-    bool const hasDepthClamp =
-            renderFlags & HAS_DEPTH_CLAMP;
-
    float const cameraPositionDotCameraForward = dot(cameraPosition, cameraForward);

    auto const* const UTILS_RESTRICT soaWorldAABBCenter = soa.data<FScene::WORLD_AABB_CENTER>();
@@ -582,7 +574,6 @@ RenderPass::Command* RenderPass::generateCommandsImpl(RenderPass::CommandTypeFla
        cmd.info.rasterState.depthWrite = true;
        cmd.info.rasterState.depthFunc = RasterState::DepthFunc::GE;
        cmd.info.rasterState.alphaToCoverage = false;
-        cmd.info.rasterState.depthClamp = hasDepthClamp;
    }

    for (uint32_t i = range.first; i < range.last; ++i) {
@@ -616,7 +607,7 @@ RenderPass::Command* RenderPass::generateCommandsImpl(RenderPass::CommandTypeFla
        //   Here, objects close to the camera (but behind) will be drawn first.
        // An alternative that keeps the mathematical ordering is given here:
        //   distanceBits ^= ((int32_t(distanceBits) >> 31) | 0x80000000u);
-        float const distance = -(dot(soaWorldAABBCenter[i], cameraForward) - cameraPositionDotCameraForward);
+        float const distance = -dot(soaWorldAABBCenter[i], cameraForward) - cameraPositionDotCameraForward;
        uint32_t const distanceBits = reinterpret_cast<uint32_t const&>(distance);

        // calculate the per-primitive face winding order inversion
@@ -680,6 +671,7 @@ RenderPass::Command* RenderPass::generateCommandsImpl(RenderPass::CommandTypeFla
         */
        for (size_t pi = 0, c = primitives.size(); pi < c; ++pi) {
            auto const& primitive = primitives[pi];
+            auto const& morphTargets = morphing.targets[pi];
            FMaterialInstance const* const mi = primitive.getMaterialInstance();
            FMaterial const* const ma = mi->getMaterial();

@@ -692,13 +684,11 @@ RenderPass::Command* RenderPass::generateCommandsImpl(RenderPass::CommandTypeFla
            cmd.info.indexOffset = primitive.getIndexOffset();
            cmd.info.indexCount = primitive.getIndexCount();
            cmd.info.type = primitive.getPrimitiveType();
-            cmd.info.morphTargetBuffer = morphing.morphTargetBuffer ?
-                    morphing.morphTargetBuffer->getHwHandle() : SamplerGroupHandle{};
-            cmd.info.morphingOffset = primitive.getMorphingBufferOffset();
+            cmd.info.morphTargetBuffer = morphTargets.buffer->getHwHandle();
+            cmd.info.morphingOffset = morphTargets.offset;

            if constexpr (isColorPass) {
-                RenderPass::setupColorCommand(cmd, renderableVariant, mi,
-                        inverseFrontFaces, hasDepthClamp);
+                RenderPass::setupColorCommand(cmd, renderableVariant, mi, inverseFrontFaces);
                const bool blendPass = Pass(cmd.key & PASS_MASK) == Pass::BLENDED;
                if (blendPass) {
                    // TODO: at least for transparent objects, AABB should be per primitive
@@ -895,31 +885,14 @@ void RenderPass::Executor::execute(FEngine& engine,

        // Maximum space occupied in the CircularBuffer by a single `Command`. This must be
        // reevaluated when the inner loop below adds DriverApi commands or when we change the
-        // CommandStream protocol. Currently, the maximum is 320 bytes.
-        // The batch size is calculated by adding the size of all commands that can possibly be
-        // emitted per draw call:
-        constexpr size_t const maxCommandSizeInBytes =
-                sizeof(CustomCommand) +
-                sizeof(COMMAND_TYPE(scissor)) +
-                sizeof(COMMAND_TYPE(bindUniformBuffer)) +
-                sizeof(COMMAND_TYPE(bindSamplers)) +
-                sizeof(COMMAND_TYPE(bindBufferRange)) +
-                sizeof(COMMAND_TYPE(bindBufferRange)) +
-                sizeof(COMMAND_TYPE(bindSamplers)) +
-                sizeof(COMMAND_TYPE(bindSamplers)) +
-                sizeof(COMMAND_TYPE(bindUniformBuffer)) +
-                sizeof(COMMAND_TYPE(bindSamplers)) +
-                sizeof(COMMAND_TYPE(bindSamplers)) +
-                sizeof(COMMAND_TYPE(bindPipeline)) +
-                sizeof(COMMAND_TYPE(setPushConstant)) +
-                sizeof(COMMAND_TYPE(bindRenderPrimitive)) +
-                sizeof(COMMAND_TYPE(draw2));
-
+        // CommandStream protocol. Currently, the maximum is 240 bytes, and we use 256 to be on
+        // the safer side.
+        size_t const maxCommandSizeInBytes = 256;

        // Number of Commands that can be issued and guaranteed to fit in the current
        // CircularBuffer allocation. In practice, we'll have tons of headroom especially if
        // skinning and morphing aren't used. With a 2 MiB buffer (the default) a batch is
-        // 6553 commands (i.e. draw calls).
+        // 8192 commands (i.e. draw calls).
        size_t const batchCommandCount = capacity / maxCommandSizeInBytes;
        while(first != last) {
            Command const* const batchLast = std::min(first + batchCommandCount, last);
@@ -284,7 +284,6 @@ public:
    static constexpr RenderFlags HAS_SHADOWING             = 0x01;
    static constexpr RenderFlags HAS_INVERSE_FRONT_FACES   = 0x02;
    static constexpr RenderFlags IS_INSTANCED_STEREOSCOPIC = 0x04;
-    static constexpr RenderFlags HAS_DEPTH_CLAMP           = 0x08;

    // Arena used for commands
    using Arena = utils::Arena<
@@ -409,19 +408,17 @@ private:
            uint8_t channel, Pass pass, CustomCommand custom, uint32_t order,
            Executor::CustomCommandFn command);

-    static Command* resize(Arena& arena, Command* const last) noexcept;
+    void resize(Arena& arena, size_t count) noexcept;

    // sorts commands then trims sentinels
-    static Command* sortCommands(
-            Command* begin, Command* end) noexcept;
+    void sortCommands(Arena& arena) noexcept;

    // instanceify commands then trims sentinels
-    RenderPass::Command* instanceify(FEngine& engine,
-            Command* begin, Command* end,
-            int32_t eyeCount) const noexcept;
+    void instanceify(FEngine& engine, Arena& arena, int32_t eyeCount) noexcept;

    // We choose the command count per job to minimize JobSystem overhead.
-    static constexpr size_t JOBS_PARALLEL_FOR_COMMANDS_COUNT = 128;
+    // On a Pixel 4, 2048 commands is about half a millisecond of processing.
+    static constexpr size_t JOBS_PARALLEL_FOR_COMMANDS_COUNT = 2048;
    static constexpr size_t JOBS_PARALLEL_FOR_COMMANDS_SIZE  =
            sizeof(Command) * JOBS_PARALLEL_FOR_COMMANDS_COUNT;

@@ -445,17 +442,17 @@ private:
            uint8_t instancedStereoEyeCount) noexcept;

    static void setupColorCommand(Command& cmdDraw, Variant variant,
-            FMaterialInstance const* mi, bool inverseFrontFaces, bool hasDepthClamp) noexcept;
+            FMaterialInstance const* mi, bool inverseFrontFaces) noexcept;

    static void updateSummedPrimitiveCounts(
            FScene::RenderableSoa& renderableData, utils::Range<uint32_t> vr) noexcept;

    FScene::RenderableSoa const& mRenderableSoa;
    backend::Viewport const mScissorViewport;
-    Command const* /* const */ mCommandBegin = nullptr;   // Pointer to the first command
-    Command const* /* const */ mCommandEnd = nullptr;     // Pointer to one past the last command
+    Command* mCommandBegin = nullptr;   // Pointer to the first command
+    Command* mCommandEnd = nullptr;     // Pointer to one past the last command
    // a UBO for instanced primitives
-    mutable BufferObjectSharedHandle mInstancedUboHandle;
+    BufferObjectSharedHandle mInstancedUboHandle;
    // a vector for our custom commands
    using CustomCommandVector = std::vector<Executor::CustomCommandFn,
            utils::STLAllocator<Executor::CustomCommandFn, LinearAllocatorArena>>;
@@ -529,9 +526,7 @@ public:
    // like above but allows to set specific flags
    RenderPassBuilder& renderFlags(
            RenderPass::RenderFlags mask, RenderPass::RenderFlags value) noexcept {
-        value &= mask;
-        mFlags &= ~mask;
-        mFlags |= value;
+        mFlags = (mFlags & mask) | (value & mask);
        return *this;
    }

@@ -56,7 +56,6 @@ public:
    backend::VertexBufferInfoHandle getVertexBufferInfoHandle() const { return mVertexBufferInfoHandle; }
    uint32_t getIndexOffset() const noexcept { return mIndexOffset; }
    uint32_t getIndexCount() const noexcept { return mIndexCount; }
-    uint32_t getMorphingBufferOffset() const noexcept { return mMorphingBufferOffset; }

    backend::PrimitiveType getPrimitiveType() const noexcept { return mPrimitiveType; }
    AttributeBitset getEnabledAttributes() const noexcept { return mEnabledAttributes; }
@@ -73,10 +72,6 @@ public:
        mGlobalBlendOrderEnabled = enabled;
    }

-    void setMorphingBufferOffset(uint32_t offset) noexcept {
-        mMorphingBufferOffset = offset;
-    }
-
 private:
    // These first fields are dereferences from PrimitiveInfo, keep them together
    struct {
@@ -85,7 +80,6 @@ private:
        backend::Handle<backend::HwVertexBufferInfo> mVertexBufferInfoHandle = {};
        uint32_t mIndexOffset = 0;
        uint32_t mIndexCount = 0;
-        uint32_t mMorphingBufferOffset = 0;
    };

    AttributeBitset mEnabledAttributes = {};
@@ -158,14 +158,20 @@ void RenderableManager::setMorphWeights(Instance instance, float const* weights,
    downcast(this)->setMorphWeights(instance, weights, count, offset);
 }

-void RenderableManager::setMorphTargetBufferOffsetAt(Instance instance, uint8_t level,
-        size_t primitiveIndex,
-        size_t offset) {
-    downcast(this)->setMorphTargetBufferOffsetAt(instance, level, primitiveIndex, offset);
+void RenderableManager::setMorphTargetBufferAt(Instance instance, uint8_t level, size_t primitiveIndex,
+        MorphTargetBuffer* morphTargetBuffer, size_t offset, size_t count) {
+    downcast(this)->setMorphTargetBufferAt(instance, level, primitiveIndex,
+            downcast(morphTargetBuffer), offset, count);
 }

-MorphTargetBuffer* RenderableManager::getMorphTargetBuffer(Instance instance) const noexcept {
-    return downcast(this)->getMorphTargetBuffer(instance);
+void RenderableManager::setMorphTargetBufferAt(
+        Instance instance, uint8_t level, size_t primitiveIndex, size_t offset, size_t count) {
+    downcast(this)->setMorphTargetBufferAt(instance, level, primitiveIndex, offset, count);
+}
+
+MorphTargetBuffer* RenderableManager::getMorphTargetBufferAt(Instance instance, uint8_t level,
+        size_t primitiveIndex) const noexcept {
+    return downcast(this)->getMorphTargetBufferAt(instance, level, primitiveIndex);
 }

 size_t RenderableManager::getMorphTargetCount(Instance instance) const noexcept {
@@ -14,21 +14,11 @@
 * limitations under the License.
 */

-#include <filament/Renderer.h>
-
-#include "ResourceAllocator.h"
+#include "details/Renderer.h"

 #include "details/Engine.h"
-#include "details/Renderer.h"
 #include "details/View.h"

-#include <utils/FixedCapacityVector.h>
-
-#include <utility>
-
-#include <stddef.h>
-#include <stdint.h>
-
 namespace filament {

 using namespace math;
@@ -46,10 +36,6 @@ void Renderer::setPresentationTime(int64_t monotonic_clock_ns) {
    downcast(this)->setPresentationTime(monotonic_clock_ns);
 }

-void Renderer::skipFrame(uint64_t vsyncSteadyClockTimeNano) {
-    downcast(this)->skipFrame(vsyncSteadyClockTimeNano);
-}
-
 bool Renderer::beginFrame(SwapChain* swapChain, uint64_t vsyncSteadyClockTimeNano) {
    return downcast(this)->beginFrame(downcast(swapChain), vsyncSteadyClockTimeNano);
 }
@@ -107,12 +93,4 @@ void Renderer::setVsyncTime(uint64_t steadyClockTimeNano) noexcept {
    downcast(this)->setVsyncTime(steadyClockTimeNano);
 }

-utils::FixedCapacityVector<Renderer::FrameInfo> Renderer::getFrameInfoHistory(size_t historySize) const noexcept {
-    return downcast(this)->getFrameInfoHistory(historySize);
-}
-
-size_t Renderer::getMaxFrameHistorySize() const noexcept {
-    return downcast(this)->getMaxFrameHistorySize();
-}
-
 } // namespace filament
--- a/Show More
+++ b/Show More