fix typo in max_triangles_per_leaf option.

Ultimately the root problem here is that when we do the binning stage we use a
different method to test the first stage and the second. In the first stage we
use bin() to get an integer bin associated with a point in space. In the second
we get the floating point division point on the axis we have chosen to bin on
and do FP comparisons instead. Under most circumstances this is fine.

Occasionally these binning differences will result in unprofitable splits
happening but still building a usable tree.

In the test case I've given, the unprofitable splits we erroneously choose
fill up the split limit and we don't end up splitting the triangles the
previous stage had intended to choose. As a result, we can't split those
triangles. Then when we do the real bifurcation, everything is on the same
side. This resulted in an assertion failure in production (1).

(1) We never noticed that the RelWithDebInfo build in MSVC still has assertions
enabled. I can provide another change disabling these since I think it's
unintentional.

.github/workflows/continuous.yml

@@ -3,665 +3,531 @@
 
 name: continuous
 
+permissions: read-all
+
 on: [push, workflow_dispatch]
 
 jobs:
-
-  ####################################################################
-  # Setup dpcpp and gfx if not already on NAS
-  ####################################################################
-  setup_dpcpp:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-    with:
-      image: embree/ubuntu:22.04
-      runs-on: '[ "Linux", "docker", "build" ]'
-      cmd: |
-        DPCPP_VERSION=$(cat .github/workflows/dpcpp-sycl-nightly.env | grep DPCPP_VERSION | sed 's/.*=//')
-        echo "Setup DPCPP ${DPCPP_VERSION}"
-        ./scripts/download_dpcpp.sh ${DPCPP_VERSION} ${NAS_LINUX}/dpcpp-compiler-linux LINUX ${PUBLIC_GITHUB_TOKEN}
-        ./scripts/download_dpcpp.sh ${DPCPP_VERSION} ${NAS_LINUX}/dpcpp-compiler-win WIN ${PUBLIC_GITHUB_TOKEN}
-
-  setup_gfx:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-    with:
-      image: embree/ubuntu:22.04
-      runs-on: '[ "Linux", "docker", "build" ]'
-      cmd: |
-        GFX_DRIVER_VERSION=$(cat .github/workflows/gfx-ubuntu22-embargo.env | grep GFX_DRIVER_VERSION | sed 's/.*=//')
-        echo "Setup embargo GFX driver ${GFX_DRIVER_VERSION} for Linux"
-        ./scripts/download_gfx.sh ${GFX_DRIVER_VERSION} ${NAS_LINUX}/gfx-driver-linux/ ${ARTIFACTORY_TOKEN}
-
-        GFX_DRIVER_VERSION=$(cat .github/workflows/gfx-ubuntu22-internal.env | grep GFX_DRIVER_VERSION | sed 's/.*=//')
-        echo "Setup internal GFX driver ${GFX_DRIVER_VERSION} for Linux"
-        ./scripts/download_gfx.sh ${GFX_DRIVER_VERSION} ${NAS_LINUX}/gfx-driver-linux/ ${ARTIFACTORY_TOKEN}
-
-        GFX_DRIVER_VERSION=$(cat .github/workflows/gfx-ubuntu22-public.env | grep GFX_DRIVER_VERSION | sed 's/.*=//')
-        echo "Setup public GFX driver ${GFX_DRIVER_VERSION} for Linux"
-        ./scripts/download_gfx.sh ${GFX_DRIVER_VERSION} ${NAS_LINUX}/gfx-driver-linux/ ${ARTIFACTORY_TOKEN}
-
-        GFX_DRIVER_VERSION=$(cat .github/workflows/gfx-ubuntu22-release.env | grep GFX_DRIVER_VERSION | sed 's/.*=//')
-        echo "Setup release GFX driver ${GFX_DRIVER_VERSION} for Linux"
-        ./scripts/download_gfx.sh ${GFX_DRIVER_VERSION} ${NAS_LINUX}/gfx-driver-linux/ ${ARTIFACTORY_TOKEN}
-
-        GFX_DRIVER_VERSION=$(cat .github/workflows/gfx-windows-internal.env | grep GFX_DRIVER_VERSION | sed 's/.*=//')
-        echo "Setup internal GFX driver ${GFX_DRIVER_VERSION} for Windows"
-        ./scripts/download_gfx.sh ${GFX_DRIVER_VERSION} ${NAS_LINUX}/gfx-driver-win/ ${ARTIFACTORY_TOKEN}
-
-        GFX_DRIVER_VERSION=$(cat .github/workflows/gfx-windows-public.env | grep GFX_DRIVER_VERSION | sed 's/.*=//')
-        echo "Setup public GFX driver ${GFX_DRIVER_VERSION} for Windows"
-        ./scripts/download_gfx.sh ${GFX_DRIVER_VERSION} ${NAS_LINUX}/gfx-driver-win/ ${ARTIFACTORY_TOKEN}
-
-        GFX_DRIVER_VERSION=$(cat .github/workflows/gfx-windows-release.env | grep GFX_DRIVER_VERSION | sed 's/.*=//')
-        echo "Setup release GFX driver ${GFX_DRIVER_VERSION} for Windows"
-        ./scripts/download_gfx.sh ${GFX_DRIVER_VERSION} ${NAS_LINUX}/gfx-driver-win/ ${ARTIFACTORY_TOKEN}
-
   ################################################################################
   ################################################################################
   # GPU tests
   ################################################################################
   ###############################################################################
-
-  ########################################
-  # ze_raytracing
-  ########################################
-
-  linux-ze_raytracing:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
-    with:
-      image: embree/ubuntu:22.04
-      runs-on: '[ "Linux", "docker", "build" ]'
-      project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
-      artifact-out: linux-ze_raytracing
-      artifact-path: ./build
-      cmd: |
-        source $DPCPP_ROOT/startup.sh
-        mkdir build
-        cd build
-        cmake -G Ninja -D CMAKE_CXX_COMPILER=clang++ -D CMAKE_C_COMPILER=clang ../kernels/rthwif
-        cmake --build . --target package
-
-  linux-ze_raytracing-test-DG2:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: ["linux-ze_raytracing"]
-    with:
-      image: embree/ubuntu:22.04
-      options: --device=/dev/dri:/dev/dri
-      runs-on: '[ "Linux", "docker", "dg2" ]'
-      project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
-      artifact-in: linux-ze_raytracing
-      cmd: |
-        cd build
-        ctest
-         
-  windows-ze_raytracing:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
-    with:
-      runs-on: '[ "Windows", "NAS", "dg2" ]'
-      project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-windows-internal.env
-      artifact-path: ./build
-      artifact-out: windows-ze_raytracing
-      shell: cmd
-      cmd: |
-        call %DPCPP_ROOT%\\startup.bat
-        mkdir build
-        cd build
-        cmake -G Ninja -D CMAKE_BUILD_TYPE=Release -D CMAKE_CXX_COMPILER=clang++ -D CMAKE_C_COMPILER=clang ../kernels/rthwif
-        cmake --build . --target package
-        ctest
-
+  
   ########################################
   # Linux DG2
   ########################################
-  
-  linux-DG2-JIT-INTERNAL:
+  linux-DG2-JIT-INTERNAL-build:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
       image: embree/ubuntu:22.04
       runs-on: '[ "Linux", "docker", "build" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
-      artifact-out: linux-DG2-JIT-INTERNAL
-      artifact-path: ./build .ctest_conf
+      dpcpp-version: intel-llvm/sycl-rel_5_2_0-rk
+      artifact-out: linux-DG2-JIT-INTERNAL-build
+      artifact-path: ./build/*.tar.gz
       cmd: |
-        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.18.1 isa:SSE2 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON sycl:none tasking:TBB2021.7.0 intensity:4
-        scripts/test.py build --debug
-        scripts/test.py build
+        module load cmake/3.25.3
+        cmake --preset linux-DG2-JIT-INTERNAL -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target build
 
   linux-DG2-JIT-INTERNAL-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: ["linux-DG2-JIT-INTERNAL"]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["linux-DG2-JIT-INTERNAL-build"]
     with:
       image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "dg2" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
-      artifact-in: linux-DG2-JIT-INTERNAL
+      env-from-files: ./.github/workflows/gfx-ubuntu22-internal.env
+      dpcpp-version: intel-llvm/sycl-rel_5_2_0-rk
+      artifact-in: linux-DG2-JIT-INTERNAL-build
       cmd: |
-        scripts/test.py test --debug
-        scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset linux-DG2-JIT-INTERNAL -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-  linux-DG2-JIT-PUBLIC:
+  linux-DG2-JIT-PUBLIC-build:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
       image: embree/ubuntu:22.04
       runs-on: '[ "Linux", "docker", "build" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-public.env
-      artifact-out: linux-DG2-JIT-PUBLIC
-      artifact-path: ./build .ctest_conf
+      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env
+      artifact-out: linux-DG2-JIT-PUBLIC-build
+      artifact-path: ./build/*.tar.gz
       cmd: |
-        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.18.1 isa:SSE2 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON sycl:none tasking:TBB2021.7.0 intensity:4
-        scripts/test.py build --debug
-        scripts/test.py build
+        module load cmake/3.25.3
+        cmake --preset linux-DG2-JIT-PUBLIC -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target build
 
   linux-DG2-JIT-PUBLIC-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: ["linux-DG2-JIT-PUBLIC"]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["linux-DG2-JIT-PUBLIC-build"]
     with:
       image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "dg2" ]'
       project: embree
       env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-public.env
-      artifact-in: linux-DG2-JIT-PUBLIC
+      artifact-in: linux-DG2-JIT-PUBLIC-build
       cmd: |
-        scripts/test.py test --debug
-        scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset linux-DG2-JIT-PUBLIC -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-  linux-DG2-JIT-EMBARGO:
+  linux-DG2-JIT-PUBLIC-ICX-build:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
       image: embree/ubuntu:22.04
       runs-on: '[ "Linux", "docker", "build" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-embargo.env
-      artifact-out: linux-DG2-JIT-EMBARGO
-      artifact-path: ./build .ctest_conf
+      env-from-files: ./.github/workflows/dpcpp-none.env
+      dpcpp-version: intel/2024.1.0
+      artifact-out: linux-DG2-JIT-PUBLIC-ICX-build
+      artifact-path: ./build/*.tar.gz
       cmd: |
-        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.18.1 isa:SSE2 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON sycl:none tasking:TBB2021.7.0 intensity:4
-        scripts/test.py build --debug
-        scripts/test.py build
-
-  linux-DG2-JIT-GPU-BVH-BUILDER:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
-    with:
-      image: embree/ubuntu:22.04
-      runs-on: '[ "Linux", "docker", "build" ]'
-      project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-embargo.env
-      artifact-out: linux-DG2-JIT-GPU-BVH-BUILDER
-      artifact-path: ./build .ctest_conf
-      cmd: |
-        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.18.1 isa:SSE2 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON EMBREE_SYCL_GPU_BVH_BUILDER:ON USERGEOM:CONTEXT FILTER_FUNCTION:CONTEXT gfx:EMBARGO sycl:none tasking:TBB2021.7.0 intensity:4
-        scripts/test.py build --debug
-        scripts/test.py build
-
-  linux-DG2-JIT-PUBLIC-GPU-BVH-BUILDER:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
-    with:
-      image: embree/ubuntu:22.04
-      runs-on: '[ "Linux", "docker", "build" ]'
-      project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-public.env
-      install-gfx-driver: false      
-      artifact-out: linux-DG2-JIT-PUBLIC-GPU-BVH-BUILDER
-      artifact-path: ./build .ctest_conf
-      cmd: |
-        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.18.1 isa:AVX512 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON EMBREE_SYCL_GPU_BVH_BUILDER:ON USERGEOM:CONTEXT FILTER_FUNCTION:CONTEXT gfx:PUBLIC sycl:none tasking:TBB2021.7.0 intensity:4
-        scripts/test.py build --debug
-        scripts/test.py build
-
-
- # linux-DG2-AOT-INTERNAL-test:
- #   secrets: inherit
- #   uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
- #   needs: ["linux-DG2-AOT-INTERNAL"]
- #   with:
- #     image: embree/ubuntu:22.04
- #     options: --device=/dev/dri:/dev/dri
- #     runs-on: '[ "Linux", "docker", "dg2" ]'
- #     project: embree
- #     env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
- #     artifact-in: linux-DG2-AOT-INTERNAL
- #     cmd: |
- #       scripts/test.py test --debug
- #       scripts/test.py test
-
-  linux-DG2-JIT-PUBLIC-ICX:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
-    with:
-      image: embree/ubuntu:22.04
-      runs-on: '[ "Linux", "docker", "build" ]'
-      project: embree
-      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
-      artifact-out: linux-DG2-JIT-PUBLIC-ICX
-      artifact-path: ./build .ctest_conf
-      cmd: |
-        scripts/test.py configure platform:x64 compiler:ICX2023.0.0 ispc:ispc1.18.1 isa:AVX512 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON sycl:none tasking:TBB2021.7.0 intensity:4
-        scripts/test.py build --debug
-        scripts/test.py build
+        module load cmake/3.25.3
+        cmake --preset linux-DG2-JIT-PUBLIC-ICX -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target build
 
   linux-DG2-JIT-PUBLIC-ICX-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: ["linux-DG2-JIT-PUBLIC-ICX"]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["linux-DG2-JIT-PUBLIC-ICX-build"]
     with:
       image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "dg2" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
-      artifact-in: linux-DG2-JIT-PUBLIC-ICX
+      env-from-files: ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: intel/2024.1.0
+      artifact-in: linux-DG2-JIT-PUBLIC-ICX-build
       cmd: |
-        scripts/test.py test --debug
-        scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset linux-DG2-JIT-PUBLIC-ICX -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-  ########################################
-  # Linux ATS-M
-  ########################################
+#   ########################################
+#   # Linux ATS-M
+#   ########################################
 
   linux-ATSM-JIT-PUBLIC-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: ["linux-DG2-JIT-PUBLIC"]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["linux-DG2-JIT-PUBLIC-build"]
     with:
       image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "atsm" ]'
       project: embree
       env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-public.env
-      artifact-in: linux-DG2-JIT-PUBLIC
+      artifact-in: linux-DG2-JIT-PUBLIC-build
       cmd: |
-        scripts/test.py test --debug
-        scripts/test.py test
-
+        module load cmake/3.25.3
+        cmake --preset linux-DG2-JIT-PUBLIC -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
   ########################################
   # Linux PVC
   ########################################
 
-  linux-PVC-JIT-INTERNAL:
+  linux-PVC-JIT-INTERNAL-build:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
       image: embree/ubuntu:22.04
       runs-on: '[ "Linux", "docker", "build" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-embargo.env
-      artifact-out: linux-PVC-JIT-INTERNAL
-      artifact-path: ./build .ctest_conf
+      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env
+      artifact-out: linux-PVC-JIT-INTERNAL-build
+      artifact-path: ./build/*.tar.gz
       cmd: |
-        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.16.1 isa:SSE2 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON sycl:none tasking:TBB intensity:4
-        scripts/test.py build --debug
-        scripts/test.py build
+        module load cmake/3.25.3
+        cmake --preset linux-PVC-JIT-INTERNAL-SYCL-RC -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target build
 
-  linux-PVC-1T-JIT-INTERNAL-test:
+  #  TODO: move to nightly
+  linux-PVC-JIT-INTERNAL-test-1T:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: ["linux-PVC-JIT-INTERNAL"]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["linux-PVC-JIT-INTERNAL-build"]
     with:
       image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "pvc" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-embargo.env
-      artifact-in: linux-PVC-JIT-INTERNAL
+      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
+      artifact-in: linux-PVC-JIT-INTERNAL-build
       cmd: |
+        export NEOReadDebugKeys=1
+        export UseKmdMigration=0
         export EnableImplicitScaling=0
-        scripts/test.py test --debug
-        scripts/test.py test
+        export ZE_FLAT_DEVICE_HIERARCHY=COMPOSITE
+        module load cmake/3.25.3
+        cmake --preset linux-PVC-JIT-INTERNAL-SYCL-RC -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-  linux-PVC-2T-JIT-INTERNAL-test:
+  linux-PVC-JIT-INTERNAL-test-2T:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
-    needs: ["linux-PVC-JIT-INTERNAL"]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["linux-PVC-JIT-INTERNAL-build"]
     with:
       image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "pvc" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-embargo.env
-      artifact-in: linux-PVC-JIT-INTERNAL
+      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
+      artifact-in: linux-PVC-JIT-INTERNAL-build
       cmd: |
-        export EnableImplicitScaling=1
-        scripts/test.py test --debug
-        scripts/test.py test
+        export NEOReadDebugKeys=1
+        export UseKmdMigration=0
+        export ZE_FLAT_DEVICE_HIERARCHY=COMPOSITE
+        module load cmake/3.25.3
+        cmake --preset linux-PVC-JIT-INTERNAL-SYCL-RC -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-#  linux-PVC-JIT-PUBLIC:
-#    secrets: inherit
-#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-#    needs: [setup_gfx, setup_dpcpp]
-#    with:
-#      image: embree/ubuntu:22.04
-#      runs-on: '[ "Linux", "docker", "build" ]'
-#      artifact-out: linux-PVC-JIT-PUBLIC
-#      artifact-path: ./build .ctest_conf
-#      cmd: |
-#        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.16.1 isa:SSE2 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON sycl:none tasking:TBB intensity:4
-#        scripts/test.py build --debug
-#        scripts/test.py build
-#
-#  linux-PVC-JIT-PUBLIC-test:
-#    secrets: inherit
-#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-#    needs: ["linux-PVC-JIT-PUBLIC"]
-#    with:
-#      image: embree/ubuntu:22.04
-#      options: --device=/dev/dri:/dev/dri
-#      runs-on: '[ "Linux", "docker", "pvc" ]'
-#      artifact-in: linux-PVC-JIT-PUBLIC
-#      cmd: |
-#        scripts/test.py test --debug
-#        scripts/test.py test
+  linux-PVC-JIT-INTERNAL-test-flat:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["linux-PVC-JIT-INTERNAL-build"]
+    with:
+      image: embree/ubuntu:22.04
+      options: --device=/dev/dri:/dev/dri
+      runs-on: '[ "Linux", "docker", "pvc" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-ubuntu22-internal.env
+      artifact-in: linux-PVC-JIT-INTERNAL-build
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset linux-PVC-JIT-INTERNAL-SYCL-RC -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
+
+
+  #  linux-PVC-JIT-PUBLIC:
+  #    secrets: inherit
+  #    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+  #    with:
+  #      image: embree/ubuntu:22.04
+  #      runs-on: '[ "Linux", "docker", "build" ]'
+  #      artifact-out: linux-PVC-JIT-PUBLIC
+  #      artifact-path: ./build .ctest_conf
+  #      cmd: |
+  #        scripts/test.py configure platform:x64 compiler:dpcpp ispc:ispc1.16.1 isa:SSE2 build:RelWithAssert EMBREE_SYCL_SUPPORT:ON sycl:none tasking:TBB intensity:4
+  #        scripts/test.py build --debug
+  #        scripts/test.py build
+  #
+  #  linux-PVC-JIT-PUBLIC-test:
+  #    secrets: inherit
+  #    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+  #    needs: ["linux-PVC-JIT-PUBLIC"]
+  #    with:
+  #      image: embree/ubuntu:22.04
+  #      options: --device=/dev/dri:/dev/dri
+  #      runs-on: '[ "Linux", "docker", "pvc" ]'
+  #      artifact-in: linux-PVC-JIT-PUBLIC
+  #      cmd: |
+  #        scripts/test.py test --debug
+  #        scripts/test.py test
 
 
   ########################################
   # Windows DG2
   ########################################
 
-  windows-DG2-JIT-INTERNAL:
+  windows-DG2-JIT-build:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+    with:
+      force-delete: true
+      runs-on: '[ "Windows", "NAS", "build" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env
+      artifact-out: windows-DG2-JIT-build
+      artifact-path: ./build/*.zip
+      cmd: |
+        cmake --preset windows-DG2-JIT-INTERNAL-L0RTAS -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target build
+
+  windows-DG2-JIT-INTERNAL-test:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+    needs: ["windows-DG2-JIT-build"]
     with:
       runs-on: '[ "Windows", "NAS", "dg2" ]'
       project: embree
       env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-windows-internal.env
+      artifact-in: windows-DG2-JIT-build
       cmd: |
-        python scripts/test.py configure platform:x64 compiler:dpcpp isa:SSE2 build:Release EMBREE_SYCL_SUPPORT:ON sycl:none intensity:3
-        python scripts/test.py build
-        python scripts/test.py test
+        cmake --preset windows-DG2-JIT-INTERNAL-L0RTAS -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-  windows-DG2-JIT-PUBLIC:
+  windows-DG2-JIT-PUBLIC-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+    needs: ["windows-DG2-JIT-build"]
     with:
       runs-on: '[ "Windows", "NAS", "dg2" ]'
       project: embree
       env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-windows-public.env
+      artifact-in: windows-DG2-JIT-build
       cmd: |
-        python scripts/test.py configure platform:x64 compiler:dpcpp isa:SSE2 build:Release EMBREE_SYCL_SUPPORT:ON sycl:none intensity:3
-        python scripts/test.py build
-        python scripts/test.py test
+        cmake --preset windows-DG2-JIT-INTERNAL-L0RTAS -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-  windows-DG2-JIT-PUBLIC-ICX:
+  windows-DG2-JIT-ICX-build:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
     with:
       runs-on: '[ "Windows", "NAS", "dg2" ]'
       project: embree
+      dpcpp-version: oneAPI/compiler/2024.1
       env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-windows-public.env
+      artifact-out: windows-DG2-JIT-ICX-build
+      artifact-path: ./build/*.zip
       cmd: |
-        python scripts/test.py configure platform:x64 compiler:ICX2023.0.0 ispc:ispc1.18.1 isa:AVX512 tasking:TBB2021.7.0 build:Release EMBREE_SYCL_SUPPORT:ON sycl:none intensity:2
-        python scripts/test.py build
-        python scripts/test.py test
+        cmake --preset nightly-windows-DG2-JIT-ICX -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target build
 
-  windows-DG2-JIT-INTERNAL-TBB2020_3:
+  windows-DG2-JIT-ICX-INTERNAL-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@devel
-    needs: [setup_gfx, setup_dpcpp]
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+    needs: ["windows-DG2-JIT-ICX-build"]
     with:
       runs-on: '[ "Windows", "NAS", "dg2" ]'
       project: embree
+      dpcpp-version: oneAPI/compiler/2024.1
       env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-windows-internal.env
+      artifact-in: windows-DG2-JIT-ICX-build
       cmd: |
-        python scripts/test.py configure platform:x64 compiler:dpcpp isa:SSE2 build:Release EMBREE_SYCL_SUPPORT:ON sycl:none intensity:3 tasking:TBB2020.3
-        python scripts/test.py build
-        python scripts/test.py test
+        cmake --preset nightly-windows-DG2-JIT-ICX -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
 
-  ################################################################################
-  ################################################################################
-  # CPU tests
-  ################################################################################
-  ################################################################################
-
-  ########################################
-  # Linux
-  ########################################
-
-  linux-fedora29_ICX:
+  windows-DG2-JIT-ICX-PUBLIC-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+    needs: ["windows-DG2-JIT-ICX-build"]
+    with:
+      runs-on: '[ "Windows", "NAS", "dg2" ]'
+      project: embree
+      dpcpp-version: oneAPI/compiler/2024.1
+      env-from-files: ./.github/workflows/dpcpp-sycl-nightly.env ./.github/workflows/gfx-windows-public.env
+      artifact-in: windows-DG2-JIT-ICX-build
+      cmd: |
+        cmake --preset nightly-windows-DG2-JIT-ICX -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTENSITY=3
+        cmake --build build --config Release --target test_package
+
+#   ################################################################################
+#   ################################################################################
+#   # CPU tests
+#   ################################################################################
+#   ################################################################################
+
+#   ########################################
+#   # Linux
+#   ########################################
+
+  linux-fedora29-CLANG4:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
     with:
       image: embree/fedora:29
       runs-on: '[ "Linux", "docker", "avx512" ]'
       cmd: |
-        python scripts/test.py configure platform:x64 build:Debug compiler:ICX2023.0.0 ispc:ispc1.12.0 isa:AVX512 frequency_level:simd256 tasking:TBB2021.2.0 intensity:2 maxinstancelevelcount:4
-        python scripts/test.py build
-        python scripts/test.py test
-#        python scripts/test.py test sde:hsw
+        module load cmake/3.25.3
+        cmake --preset linux-fedora29-CLANG4 -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
 
-  linux-fedora32-GCC:
+  linux-fedora29-CLANG5-INT-ADDRSANITIZER:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
     with:
-      image: embree/fedora:32
-      runs-on: '[ "Linux", "docker", "avx512" ]'
-      cmd: |
-        python scripts/test.py configure platform:x64 build:Debug compiler:GCC ispc:ispc isa:AVX512 tasking:TBB COMPACT_POLYS:ON MIN_WIDTH:ON intensity:2
-        python scripts/test.py build
-        python scripts/test.py test
-
-  linux-fedora32_ICX:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-    with:
-      image: embree/fedora:32
-      runs-on: '[ "Linux", "docker", "avx512" ]'
-      cmd: |
-        python scripts/test.py configure platform:x64 build:Debug compiler:ICX2023.0.0 ispc:ispc1.12.0 isa:AVX512 frequency_level:simd256 tasking:TBB2021.7.0 intensity:2 maxinstancelevelcount:4
-        python scripts/test.py build
-        python scripts/test.py test
-
-  linux-fedora26_CLANG4:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-    with:
-      image: embree/fedora:26
-      runs-on: '[ "Linux", "docker", "avx512" ]'
-      cmd: |
-        python scripts/test.py configure platform:x64 build:Debug compiler:CLANG4.0.0 ispc:ispc1.17.0 isa:AVX512 frequency_level:simd128 tasking:TBB2021.2.0 GLFW:OFF intensity:2
-        python scripts/test.py build
-        python scripts/test.py test
-
-  linux-fedora26_CLANG5-INT-ADDRSANITIZER:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-    with:
-      image: embree/fedora:26
+      image: embree/fedora:29
       runs-on: '[ "Linux", "docker", "build" ]'
       cmd: |
-        python scripts/test.py configure platform:x64 build:Debug compiler:CLANG5.0.0 ispc:ispc1.17.0 isa:AVX tasking:INT intensity:1 addrsanitizer:ON
-        python scripts/test.py build
-        python scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset linux-fedora29-CLANG5-INT-ADDRSANITIZER -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=1
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
+
+  linux-fedora29-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    with:
+      image: embree/fedora:29
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset linux-fedora29-ICX -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
+
+  linux-fedora32-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    with:
+      image: embree/fedora:32
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset linux-fedora32-ICX -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
 
   linux-ubuntu20_04-GCC-NAMESPACE:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
     with:
       image: embree/ubuntu:20.04
       runs-on: '[ "Linux", "docker", "build" ]'
       cmd: |
-        python scripts/test.py configure platform:x64 build:Debug compiler:GCC ispc:ispc1.12.0 isa:AVX2 frequency_level:simd256 tasking:TBB intensity:1 api_namespace:myembree
-        python scripts/test.py build
-        python scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset linux-ubuntu20_04-GCC-NAMESPACE -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=1
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
 
   linux-ubuntu20_04-ICX:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
       image: embree/ubuntu:20.04
       runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
       cmd: |
-        python scripts/test.py configure platform:x64 build:Release threads:4 compiler:ICX2023.0.0 ispc:ispc1.17.0 isa:AVX512 frequency_level:simd256 tasking:TBB2021.5.0 intensity:2
-        python scripts/test.py build
-        python scripts/test.py test
-
-  linux-centos7_4-GCC:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@devel
-    with:
-      image: embree/centos:7.4
-      runs-on: '[ "Linux", "docker", "build" ]'
-      cmd: |
-        python scripts/test.py configure platform:x64 build:Debug compiler:GCC ispc:ispc1.17.0 isa:AVX2 frequency_level:simd128 tasking:TBB intensity:1
-        python scripts/test.py build
-        python scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset linux-ubuntu20_04-ICX -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
 
 
   ########################################
   # Windows
   ########################################
-  windows-V140:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_windows.yml@devel
-    with:
-      runs-on: '[ "Windows", "NAS", "build" ]'
-      cmd: |
-        python scripts/test.py configure platform:x64 threads:4 build:RelWithDebInfo compiler:V140 ispc:ispc1.16.1 isa:SSE2 tasking:TBB2020.2 intensity:2
-        python scripts/test.py build
-        python scripts/test.py test
-
-  windows-V141:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_windows.yml@devel
-    with:
-      runs-on: '[ "Windows", "NAS", "build" ]'
-      cmd: |
-        python scripts/test.py configure platform:x64 build:RelWithDebInfo compiler:V141 ispc:ispc1.17.0 isa:SSE2 tasking:TBB2020.2 intensity:2 maxinstancelevelcount:8
-        python scripts/test.py build
-        python scripts/test.py test
-
   windows-V142:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_windows.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
     with:
+      force-delete: true
       runs-on: '[ "Windows", "NAS", "build" ]'
+      dpcpp-version: none
       cmd: |
-        python scripts/test.py configure platform:x64 build:RelWithDebInfo compiler:V142 ispc:ispc1.18.1 isa:SSE2 tasking:TBB2021.7.0 intensity:2 maxinstancelevelcount:8
-        python scripts/test.py build
-        python scripts/test.py test
+        cmake --preset windows-V142 -D CMAKE_BUILD_TYPE=RelWithDebInfo -D EMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --target test_package
 
   windows-ICX:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_windows.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
     with:
+      force-delete: true
       runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+      dpcpp-version: oneAPI/compiler/2024.1
       cmd: |
-        python scripts/test.py configure platform:x64 build:RelWithDebInfo compiler:ICX2023.0.0 ispc:ispc1.17.0 isa:AVX512 tasking:TBB2021.5.0 threads:16 intensity:2
-        python scripts/test.py build
-        python scripts/test.py test
-
-  ################################################################################
-  ################################################################################
-  # CPU tests
-  ################################################################################
-  ################################################################################
+        cmake --preset windows-ICX -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
 
   ########################################
   # MacOS
   ########################################
 
-  macosx-ARM-NEON-AVX2:
+  macosx-ARM-CLANG-NEON-AVX2:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_macos.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
     with:
-      runs-on: '[ "macOS", "arm", "build" ]'
+      runs-on: '[ "macOS", "build", "arm" ]'
       cmd: |
-        scripts/test.py configure platform:arm64 build:Debug compiler:CLANG ispc:ispc1.17.0 isa:NEON2X tasking:TBB_HOMEBREW intensity:2
-        scripts/test.py build
-        scripts/test.py test
+        cmake --preset macos-ARM-CLANG-NEON-AVX2 -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
+
+  macosx-ARM-CLANG-NEON:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    with:
+      runs-on: '[ "macOS", "build", "arm" ]'
+      cmd: |
+        cmake --preset macos-ARM-CLANG-NEON -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
 
   macosx-x64-CLANG:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_macos.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
     with:
-      runs-on: '[ "macOS", "embree", "build" ]'
+      runs-on: '[ "macOS", "build", "x86_64" ]'
       cmd: |
-        scripts/test.py configure platform:x64 build:Debug compiler:CLANG ispc:ispc1.18.1 isa:SSE2 tasking:TBB2021.1.1 intensity:2
-        scripts/test.py build
-        scripts/test.py test
-
-
-  ########################################
-  # Klockwork
-  ########################################
-
-  linux-GCC-KLOCWORK:
+        cmake --preset macos-x64-CLANG -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
+  
+  macosx-ARM-CLANG-NEON-AVX2-universal-binaries:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_static_analysis.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
     with:
-      image: embree/centos:7.4
-      runs-on: '[ "Linux", "docker", "build" ]'
+      runs-on: '[ "macOS", "build", "arm" ]'
       cmd: |
-        export CI_PROJECT_DIR=$GITHUB_WORKSPACE
-        export CI_JOB_ID=$GITHUB_RUN_ID
-        python scripts/test.py configure platform:x64 build:Debug compiler:GCC ispc:ispc1.17.0 isa:SSE2 tasking:TBB klocwork:ON intensity:0
-        python scripts/test.py build
-        python scripts/test.py test
-        scripts/klocwork_gen_report.sh
-        scripts/store_files.sh $GITHUB_REPOSITORY $GITHUB_RUN_ID klocwork "klocwork/report.log"
+        cmake --preset macos-ARM-CLANG-NEON-AVX2-UB -DCMAKE_BUILD_TYPE=RelWithDebInfo -DEMBREE_TESTING_INTENSITY=2 -DCMAKE_OSX_ARCHITECTURES="arm64;x86_64" -DEMBREE_BUILD_GLFW_FROM_SOURCE=ON -DEMBREE_ISPC_SUPPORT=OFF
+        cmake --build build --config RelWithDebInfo --target build
+        cmake --build build --config RelWithDebInfo --target test_package
+
+  ########################################
+  # Coverity
+  ########################################
+
+  static-analysis:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/static_analysis.yml@main
+    with:
+      project: Embree
+      submodules: true
+      coverity: true
+      prebuild: >
+        module load cmake/3.25.3 &&
+        export NAS_LINUX=$STORAGE_PATH/packages/apps &&
+        cmake --preset linux-coverity
+      build: sh scripts/coverity.sh
 
   success:
     runs-on: ubuntu-latest
-    steps:
-      - name: Success
-        run: echo Success
     needs:
-      - linux-ze_raytracing
-      - linux-ze_raytracing-test-DG2
-      - windows-ze_raytracing
       - linux-DG2-JIT-INTERNAL-test
       - linux-DG2-JIT-PUBLIC-test
       - linux-ATSM-JIT-PUBLIC-test
-      - linux-DG2-JIT-EMBARGO
       - linux-DG2-JIT-PUBLIC-ICX-test
-      - linux-PVC-1T-JIT-INTERNAL-test
-      - linux-PVC-2T-JIT-INTERNAL-test
-      - windows-DG2-JIT-INTERNAL
-      - windows-DG2-JIT-PUBLIC
-      - windows-DG2-JIT-PUBLIC-ICX
-      - windows-DG2-JIT-INTERNAL-TBB2020_3
-      - linux-fedora29_ICX
-      - linux-fedora32-GCC
-      - linux-fedora32_ICX
-      - linux-fedora26_CLANG4
-      - linux-fedora26_CLANG5-INT-ADDRSANITIZER
+      - linux-PVC-JIT-INTERNAL-test-1T
+      - linux-PVC-JIT-INTERNAL-test-2T
+      - linux-PVC-JIT-INTERNAL-test-flat
+      - windows-DG2-JIT-INTERNAL-test
+      - windows-DG2-JIT-PUBLIC-test
+      - windows-DG2-JIT-ICX-INTERNAL-test
+      - windows-DG2-JIT-ICX-PUBLIC-test
+      - linux-fedora29-CLANG4
+      - linux-fedora29-CLANG5-INT-ADDRSANITIZER
+      - linux-fedora29-ICX
+      - linux-fedora32-ICX
       - linux-ubuntu20_04-GCC-NAMESPACE
       - linux-ubuntu20_04-ICX
-      - linux-centos7_4-GCC
-      - windows-V140
-      - windows-V141
       - windows-V142
       - windows-ICX
-      - macosx-ARM-NEON-AVX2
+      - macosx-ARM-CLANG-NEON-AVX2
+      - macosx-ARM-CLANG-NEON
       - macosx-x64-CLANG
-      - linux-GCC-KLOCWORK
+      - macosx-ARM-CLANG-NEON-AVX2-universal-binaries
+      - static-analysis
+
+
+    if: failure() || cancelled()
+    steps:
+      - name: Failure
+        run: |
+          echo "::notice title=Success::Workflow failed"
+          exit 1

.github/workflows/dpcpp-sycl-nightly.env

@@ -1 +1 @@
-DPCPP_VERSION=sycl-nightly/20230304
+DPCPP_VERSION=intel-llvm/nightly-2024-03-17

.github/workflows/gfx-ubuntu22-embargo.env

@@ -1 +0,0 @@
-GFX_DRIVER_VERSION=gfx-driver-builds/ci/master/gfx-driver-ci-master-13324/artifacts/Linux/Ubuntu/22.04/Release

.github/workflows/gfx-ubuntu22-internal.env

@@ -1 +1 @@
-GFX_DRIVER_VERSION=neo-builds/ci/master/ci-neo-master-025813/artifacts/linux/ubuntu/22.04
+GFX_DRIVER_VERSION=neo-builds/ci/master/ci-neo-master-029525/artifacts/linux/ubuntu/22.04

.github/workflows/gfx-ubuntu22-public.env

@@ -1 +1 @@
-GFX_DRIVER_VERSION=neo-builds/releases/releases_23.05/neo-releases-23.05.25593.11/artifacts/linux/ubuntu/22.04
+GFX_DRIVER_VERSION=linux-latest

.github/workflows/gfx-ubuntu22-release.env

@@ -1 +0,0 @@
-GFX_DRIVER_VERSION=neo-builds/releases/releases_23.05/neo-releases-23.05.25593.11/artifacts/linux/ubuntu/20.04

.github/workflows/gfx-windows-internal.env

@@ -1 +1 @@
-GFX_DRIVER_VERSION=gfx-driver-builds/ci/master/gfx-driver-ci-master-13324/artifacts/Windows
+GFX_DRIVER_VERSION=gfx-driver-builds/ci/master/gfx-driver-ci-master-16477/artifacts/Windows

.github/workflows/gfx-windows-public.env

@@ -1 +1 @@
-GFX_DRIVER_VERSION=gfx-driver-builds/prod/releases_23ww08-ci-master-13370/prod-hini-releases_23ww08-ci-master-13370-revenue-pr-1014250/artifacts/Windows/
+GFX_DRIVER_VERSION=windows-101.5445

.github/workflows/gfx-windows-release.env

@@ -1 +0,0 @@
-GFX_DRIVER_VERSION=gfx-driver-builds/prod/releases_23ww08-ci-master-13370/prod-hini-releases_23ww08-ci-master-13370-revenue-pr-1014250/artifacts/Windows/

.github/workflows/integrate.yml

@@ -0,0 +1,688 @@
+## Copyright 2022 Intel Corporation
+## SPDX-License-Identifier: Apache-2.0
+
+name: integrate
+
+permissions: read-all
+
+# same as nightly...
+on:
+  schedule:
+    - cron:  '00 00 * * *'
+  workflow_dispatch:
+
+jobs:
+#####################################################################
+# Integration Tests
+####################################################################
+
+# TODO: investigate windows integration tests not working
+#####################################################################
+## Windows
+#####################################################################
+#  integrate-windows-ICX:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-out: integrate-windows-ICX
+#      artifact-path: ./build/embree-*.zip
+#      dpcpp-version: oneAPI/compiler/2024.1
+#      cmd: |
+#        cmake --preset integrate-windows-icx
+#        cmake --build build --config Release --target package -j8
+#        cmake --build build --config Release --target post_package
+#
+#  integrate-windows-ICX-test:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+#    needs: ["integrate-windows-ICX"]
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-in: integrate-windows-ICX
+#      dpcpp-version: oneAPI/compiler/2024.1
+#      cmd: |
+#        cd build
+#        . ..\tests\integration\test_embree_release\unpack-windows.ps1
+#        cd ..\tests\integration\test_embree_release
+#        cmake --preset test-windows-icx
+#        cmake --build build --config Release
+#        .\build\test.exe
+#        
+#
+#
+#
+#
+#  integrate-windows-v142:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows.yml@main
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-out: integrate-windows-v142
+#      artifact-path: ./build/embree-*.zip
+#      cmd: |
+#        cmake --preset integrate-windows-v142
+#        cmake --build build --config Release --target package -j8
+#        cmake --build build --config Release --target post_package
+#
+#  integrate-windows-v142-test-v142:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows.yml@main
+#    needs: ["integrate-windows-v142"]
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-in: integrate-windows-v142
+#      cmd: |
+#        cd build
+#        . ..\tests\integration\test_embree_release\unpack-windows.ps1
+#        cd ..\tests\integration\test_embree_release
+#        cmake --preset test-windows-v142
+#        cmake --build build --config Release
+#        .\build\Release\test.exe
+#
+#  integrate-windows-v142-test-ICX:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+#    needs: ["integrate-windows-v142"]
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-in: integrate-windows-v142
+#      dpcpp-version: oneAPI/compiler/2024.1
+#      cmd: |
+#        cd build
+#        . ..\tests\integration\test_embree_release\unpack-windows.ps1
+#        cd ..\tests\integration\test_embree_release
+#        cmake --preset test-windows-icx
+#        cmake --build build --config Release
+#        .\build\test.exe
+#        
+#
+#
+#
+#
+#####################################################################
+## Windows STATIC
+#####################################################################
+#  integrate-static-windows-ICX:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-out: integrate-static-windows-ICX
+#      artifact-path: ./build/embree-*.zip
+#      dpcpp-version: oneAPI/compiler/2024.1
+#      cmd: |
+#        cmake --preset integrate-static-windows-icx
+#        cmake --build build --config Release --target package -j8
+#        cmake --build build --config Release --target post_package
+#
+#  integrate-static-windows-ICX-test:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+#    needs: ["integrate-static-windows-ICX"]
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-in: integrate-static-windows-ICX
+#      dpcpp-version: oneAPI/compiler/2024.1
+#      cmd: |
+#        cd build
+#        . ..\tests\integration\test_embree_release\unpack-windows.ps1
+#        cd ..\tests\integration\test_embree_release
+#        cmake --preset test-windows-icx
+#        cmake --build build --config Release
+#        .\build\test.exe
+#        
+#
+#
+#
+#
+#  integrate-static-windows-v142:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows.yml@main
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-out: integrate-static-windows-v142
+#      artifact-path: ./build/embree-*.zip
+#      cmd: |
+#        cmake --preset integrate-windows-v142
+#        cmake --build build --config Release --target package -j8
+#        cmake --build build --config Release --target post_package
+#
+#  integrate-static-windows-v142-test-v142:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows.yml@main
+#    needs: ["integrate-static-windows-v142"]
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-in: integrate-static-windows-v142
+#      cmd: |
+#        cd build
+#        . ..\tests\integration\test_embree_release\unpack-windows.ps1
+#        cd ..\tests\integration\test_embree_release
+#        cmake --preset test-windows-v142
+#        cmake --build build --config Release
+#        .\build\Release\test.exe
+#
+#  integrate-static-windows-v142-test-ICX:
+#    secrets: inherit
+#    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
+#    needs: ["integrate-static-windows-v142"]
+#    with:
+#      runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
+#      artifact-in: integrate-static-windows-v142
+#      dpcpp-version: oneAPI/compiler/2024.1
+#      cmd: |
+#        cd build
+#        . ..\tests\integration\test_embree_release\unpack-windows.ps1
+#        cd ..\tests\integration\test_embree_release
+#        cmake --preset test-windows-icx
+#        cmake --build build --config Release
+#        .\build\test.exe
+
+
+
+
+
+####################################################################
+# Linux
+####################################################################
+  integrate-linux-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-out: integrate-linux-GCC
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-gcc -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  integrate-linux-GCC-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-linux-GCC" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-linux-GCC
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-linux-GCC-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-linux-GCC" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-linux-GCC
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-clang
+        cmake --build build --config Release --target test_integration
+
+  integrate-linux-GCC-test-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-linux-GCC" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-linux-GCC
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+        
+
+
+
+
+  integrate-linux-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-out: integrate-linux-ICX
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-icx -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  integrate-linux-ICX-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-linux-ICX" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-linux-ICX
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-linux-ICX-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-linux-ICX" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-linux-ICX
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-icx -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-clang
+        cmake --build build --config Release --target test_integration
+
+  integrate-linux-ICX-test-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-linux-ICX" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-linux-ICX
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-icx -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+        
+
+
+
+
+  integrate-static-linux-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-out: integrate-static-linux-CLANG
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-clang -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  integrate-static-linux-CLANG-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-static-linux-CLANG" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-static-linux-CLANG
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-static-linux-CLANG-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-static-linux-CLANG" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-static-linux-CLANG
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-clang
+        cmake --build build --config Release --target test_integration
+
+  integrate-static-linux-CLANG-test-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-static-linux-CLANG" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-static-linux-CLANG
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+        
+
+
+
+
+####################################################################
+# Linux STATIC
+####################################################################
+  integrate-static-linux-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-out: integrate-static-linux-GCC
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-gcc -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  integrate-static-linux-GCC-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-static-linux-GCC" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-static-linux-GCC
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-static-linux-GCC-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-static-linux-GCC" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-static-linux-GCC
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-clang
+        cmake --build build --config Release --target test_integration
+
+  integrate-static-linux-GCC-test-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-static-linux-GCC" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-static-linux-GCC
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+
+
+
+
+
+  integrate-static-linux-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-out: integrate-static-linux-ICX
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-icx -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  integrate-static-linux-ICX-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-static-linux-ICX" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-static-linux-ICX
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-icx -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-static-linux-ICX-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-static-linux-ICX" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-static-linux-ICX
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-icx -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-clang
+        cmake --build build --config Release --target test_integration
+
+  integrate-static-linux-ICX-test-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-static-linux-ICX" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-static-linux-ICX
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-static-linux-icx -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+        
+
+
+
+
+  integrate-linux-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-out: integrate-linux-CLANG
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-clang -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  integrate-linux-CLANG-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-linux-CLANG" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-linux-CLANG
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-linux-CLANG-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker.yml@main
+    needs: [ "integrate-linux-CLANG" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      artifact-in: integrate-linux-CLANG
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-clang
+        cmake --build build --config Release --target test_integration
+
+  integrate-linux-CLANG-test-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-linux-CLANG" ]
+    with:
+      image: embree/rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "avx512" ]'
+      dpcpp-version: intel/2024.1.0
+      artifact-in: integrate-linux-CLANG
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-linux-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+
+
+
+
+
+####################################################################
+# Linux SYCL
+####################################################################
+  integrate-sycl-linux-ICX:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    with:
+      image: embree/ubuntu:22.04
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: icx/compiler_ppkg-rel/20240110
+      artifact-out: integrate-sycl-linux-ICX
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-sycl-linux-icx -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  integrate-sycl-linux-ICX-test:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-sycl-linux-ICX" ]
+    with:
+      image: embree/ubuntu:22.04
+      options: --device=/dev/dri:/dev/dri
+      runs-on: '[ "Linux", "docker", "dg2" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: icx/compiler_ppkg-rel/20240110
+      artifact-in: integrate-sycl-linux-ICX
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-sycl-linux-icx -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+        
+
+
+
+
+  integrate-sycl-linux-DPCPP:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    with:
+      image: embree/ubuntu:22.04
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: intel-llvm/nightly-2023-09-22-rk
+      artifact-out: integrate-sycl-linux-DPCPP
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-sycl-linux-dpcpp -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  istegrate-sycl-linux-DPCPP-test:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [ "integrate-sycl-linux-DPCPP" ]
+    with:
+      image: embree/ubuntu:22.04
+      options: --device=/dev/dri:/dev/dri
+      runs-on: '[ "Linux", "docker", "dg2" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: icx/compiler_ppkg-rel/20240110
+      artifact-in: integrate-sycl-linux-DPCPP
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset integrate-sycl-linux-dpcpp -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --config Release --target test_integration
+        
+
+
+
+
+####################################################################
+# MacOS
+####################################################################
+  integrate-macos-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    with:
+      runs-on: '[ "macOS", "build", "x86_64" ]'
+      artifact-out: integrate-macos-GCC
+      artifact-path: ./build/*.zip
+      cmd: |
+        cmake --preset integrate-macos-gcc -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+#        cmake --build build --config Release --target post_package_notarize     TODO: reactivate this when notarization problems are solved
+
+  integrate-macos-GCC-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    needs: [ "integrate-macos-GCC" ]
+    with:
+      runs-on: '[ "macOS", "build", "x86_64" ]'
+      artifact-in: integrate-macos-GCC
+      cmd: |
+        cmake --preset integrate-macos-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-macos-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-macos-GCC-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    needs: [ "integrate-macos-GCC" ]
+    with:
+      runs-on: '[ "macOS", "build", "x86_64" ]'
+      artifact-in: integrate-macos-GCC
+      cmd: |
+        cmake --preset integrate-macos-gcc -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-macos-clang
+        cmake --build build --config Release --target test_integration
+      
+
+
+
+
+#  TODO: cant find tbb/tbb.h ???
+  integrate-macos-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    with:
+      runs-on: '[ "macOS", "build", "x86_64" ]'
+      artifact-out: integrate-macos-CLANG
+      artifact-path: ./build/*.zip
+      cmd: |
+        cmake --preset integrate-macos-clang -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+#        cmake --build build --config Release --target post_package_notarize     TODO: reactivate this when notarization problems are solved
+
+  integrate-macos-CLANG-test-GCC:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    needs: [ "integrate-macos-CLANG" ]
+    with:
+      runs-on: '[ "macOS", "build", "x86_64" ]'
+      artifact-in: integrate-macos-CLANG
+      cmd: |
+        cmake --preset integrate-macos-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-macos-gcc
+        cmake --build build --config Release --target test_integration
+
+  integrate-macos-CLANG-test-CLANG:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    needs: [ "integrate-macos-CLANG" ]
+    with:
+      runs-on: '[ "macOS", "build", "x86_64" ]'
+      artifact-in: integrate-macos-CLANG
+      cmd: |
+        cmake --preset integrate-macos-clang -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-macos-clang
+        cmake --build build --config Release --target test_integration

.github/workflows/performance.yml

@@ -3,137 +3,177 @@
 
 name: performance
 
-on: [workflow_dispatch]
+permissions: read-all
+
+on: 
+  schedule:
+    - cron:  '00 01 * * *'
+  workflow_dispatch:
 
 jobs:
-
-
-  ########################################
-  # Linux build
-  ########################################
-
-  linux-build-DG2-DPCPP:
+  perf-linux-SYCL-DPCPP-build:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@main
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
-      image: embree/ubuntu:20.04
+      image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "dg2" ]'
-      artifact-out: linux-DG2-DPCPP
-      artifact-path: ./build/*.tar.gz .ctest_conf
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: intel/2023.0
+      artifact-out: perf-linux-SYCL-DPCPP
+      artifact-path: ./build/*.tar.gz
       cmd: |
-        scripts/test.py configure platform:x64 build:Release compiler:ICX2023.0.0 EMBREE_SYCL_SUPPORT:ON sycl:none gfx:RELEASE ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2-AVX512 tasking:TBB2021.7.0 intensity:4 package:ZIP
-        scripts/test.py build
-        scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset perf-linux-icx-sycl -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
 
-
-  linux-build-x86-ICX-ISPC:
+  perf-linux-DG2-A770-test:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@main
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["perf-linux-SYCL-DPCPP-build"]
+    with:
+      image: embree/ubuntu:22.04
+      options: --device=/dev/dri:/dev/dri
+      runs-on: '[ "Linux", "docker", "dg2", "perf", "A770" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      artifact-in: perf-linux-SYCL-DPCPP
+      setup-benny: true
+      cmd: |
+        cd build
+        ls *.tar.gz
+        tar -xzf embree-?.?.?.sycl.x86_64.linux.tar.gz
+        cd bin
+        source ../embree-vars.sh
+        $GITHUB_WORKSPACE/scripts/run-perf.sh gpu
+
+  perf-linux-PVC-test:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["perf-linux-SYCL-DPCPP-build"]
+    with:
+      image: embree/ubuntu:22.04
+      options: --device=/dev/dri:/dev/dri
+      runs-on: '[ "Linux", "docker", "pvc", "perf" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      artifact-in: perf-linux-SYCL-DPCPP
+      setup-benny: true
+      cmd: |
+        # Run it on single tile (disable implicit scaling)
+        export EnableImplicitScaling=0
+        export NEOReadDebugKeys=1
+        export ZE_FLAT_DEVICE_HIERARCHY=COMPOSITE
+        cd build
+        ls *.tar.gz
+        tar -xzf embree-?.?.?.sycl.x86_64.linux.tar.gz
+        cd bin
+        source ../embree-vars.sh
+        $GITHUB_WORKSPACE/scripts/run-perf.sh gpu
+
+
+
+
+
+
+  perf-linux-x86-ICX-ISPC-build:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
       image: embree/rockylinux:8.7
       runs-on: '[ "Linux", "docker", "avx512" ]'
-      artifact-out: linux-build-x86-ICX-ISPC
-      artifact-path: ./build/*.tar.gz .ctest_conf
+      dpcpp-version: intel/2024.1.0
+      artifact-out: perf-linux-x86-ICX-ISPC-build
+      artifact-path: ./build/*.tar.gz
       cmd: |
-        scripts/test.py configure platform:x64 build:Release compiler:ICX2023.0.0 ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2-AVX512 tasking:TBB2021.7.0 intensity:4 package:ZIP
-        scripts/test.py build
-        scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset perf-linux-icx -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
 
-
-
-  linux-test-DG2-A750-not_perf:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@main
-    needs: ["linux-build-DG2-DPCPP"]
-    with:
-      image: embree/ubuntu:22.04
-      options: --device=/dev/dri:/dev/dri
-      runs-on: '[ "Linux", "docker", "dg2" ]'
-      artifact-in: linux-DG2-DPCPP
-      cmd: |
-        source <(scripts/test.py env --debug)
-        cd build
-        tar -xf embree-4.0.0.sycl.x86_64.linux.tar.gz
-        cd embree-4.0.0.sycl.x86_64.linux
-        chmod +x embree-vars.sh
-        ./embree-vars.sh
-        cd bin
-        $GITHUB_WORKSPACE/scripts/run-perf.sh gpu
-
-  linux-test-DG2-A770:
-    secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@main
-    needs: ["linux-build-DG2-DPCPP"]
-    with:
-      image: embree/ubuntu:22.04
-      options: --device=/dev/dri:/dev/dri
-      runs-on: '[ "Linux", "docker", "A770", "perf" ]'
-      artifact-in: linux-DG2-DPCPP
-      cmd: |
-        source <(scripts/test.py env --debug)
-        cd build
-        tar -xf embree-4.0.0.sycl.x86_64.linux.tar.gz
-        cd embree-4.0.0.sycl.x86_64.linux
-        chmod +x embree-vars.sh
-        ./embree-vars.sh
-        cd bin
-        $GITHUB_WORKSPACE/scripts/run-perf.sh gpu
-
-
-  linux-test-i9-12900k:
+  perf-linux-i9-12900k-test:
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/perf.yml@main
-    needs: ["linux-build-x86-ICX-ISPC"]
+    needs: ["perf-linux-x86-ICX-ISPC-build"]
     with:
       runs-on: '[ "Linux", "perf", "vis-perf-i9-12900k-1" ]'
-      artifact-in: linux-build-x86-ICX-ISPC
+      artifact-in: perf-linux-x86-ICX-ISPC-build
+      setup-benny: true
       cmd: |
         cd build
-        tar -xf embree-4.0.0.x86_64.linux.tar.gz
-        cd embree-4.0.0.x86_64.linux
+        tar -xzf embree-?.?.?.x86_64.linux.tar.gz
         cd bin
         $GITHUB_WORKSPACE/scripts/run-perf.sh cpu
 
-  linux-test-x8380-1:
+
+  perf-linux-x8380-1-test:
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/perf.yml@main
-    needs: ["linux-build-x86-ICX-ISPC"]
+    needs: ["perf-linux-x86-ICX-ISPC-build"]
     with:
       runs-on: '[ "Linux", "perf", "vis-perf-x8380-1" ]'
-      artifact-in: linux-build-x86-ICX-ISPC
+      artifact-in: perf-linux-x86-ICX-ISPC-build
+      setup-benny: true
       cmd: |
         cd build
-        tar -xf embree-4.0.0.x86_64.linux.tar.gz
-        cd embree-4.0.0.x86_64.linux
+        tar -xzf embree-?.?.?.x86_64.linux.tar.gz
         cd bin
         $GITHUB_WORKSPACE/scripts/run-perf.sh cpu
 
-  linux-test-x8280-1:
+
+  perf-linux-x8280-1-test:
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/perf.yml@main
-    needs: ["linux-build-x86-ICX-ISPC"]
+    needs: ["perf-linux-x86-ICX-ISPC-build"]
     with:
       runs-on: '[ "Linux", "perf", "vis-perf-x8280-1" ]'
-      artifact-in: linux-build-x86-ICX-ISPC
+      artifact-in: perf-linux-x86-ICX-ISPC-build
+      setup-benny: true
       cmd: |
         cd build
-        tar -xf embree-4.0.0.x86_64.linux.tar.gz
-        cd embree-4.0.0.x86_64.linux
+        tar -xzf embree-?.?.?.x86_64.linux.tar.gz
         cd bin
         $GITHUB_WORKSPACE/scripts/run-perf.sh cpu
 
-  linux-test-a3970x-1:
+
+  perf-linux-a3970x-1-test:
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/perf.yml@main
-    needs: ["linux-build-x86-ICX-ISPC"]
+    needs: ["perf-linux-x86-ICX-ISPC-build"]
     with:
       runs-on: '[ "Linux", "perf", "vis-perf-a3970x-1" ]'
-      artifact-in: linux-build-x86-ICX-ISPC
+      artifact-in: perf-linux-x86-ICX-ISPC-build
+      setup-benny: true
       cmd: |
         cd build
-        tar -xf embree-4.0.0.x86_64.linux.tar.gz
-        cd embree-4.0.0.x86_64.linux
+        tar -xzf embree-?.?.?.x86_64.linux.tar.gz
         cd bin
         $GITHUB_WORKSPACE/scripts/run-perf.sh cpu
 
+  linux-DG2-A770-buildbench:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["perf-linux-SYCL-DPCPP-build"]
+    with:
+      image: embree/ubuntu:22.04
+      options: --device=/dev/dri:/dev/dri
+      runs-on: '[ "Linux", "docker", "dg2", "perf", "A770" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      artifact-in: perf-linux-SYCL-DPCPP
+      setup-benny: true
+      cmd: |
+        cd build
+        ls *.tar.gz
+        tar -xzf embree-?.?.?.sycl.x86_64.linux.tar.gz
+        git clone https://github.com/intel-innersource/libraries.graphics.renderkit.embree.models.git embree-models
+        ls -al
+        source embree-vars.sh
+        echo "bin/embree_buildbench --legacy"
+        bin/embree_buildbench --legacy
+        echo "bin/embree_buildbench_sycl --legacy"
+        bin/embree_buildbench_sycl --legacy
+        echo "bin/embree_buildbench --legacy -c embree-models/crown/crown.ecs"
+        bin/embree_buildbench --legacy -c embree-models/crown/crown.ecs
+        echo "bin/embree_buildbench_sycl --legacy -c embree-models/crown/crown.ecs"
+        bin/embree_buildbench_sycl --legacy -c embree-models/crown/crown.ecs

.github/workflows/release.yml

@@ -3,6 +3,8 @@
 
 name: release
 
+permissions: read-all
+
 on:
   schedule:
     - cron:  '00 00 * * *'
@@ -13,146 +15,269 @@ jobs:
 #####################################################################
 # Release Builds
 ####################################################################
-
-  release-windows-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP:
+  release-windows:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_windows.yml@main
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
     with:
       runs-on: '[ "Windows", "NAS", "build", "avx512" ]'
-      artifact-out: release-windows-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-path: ./build/*.zip
+      artifact-out: release-windows
+      artifact-path: ./build/embree-*.zip
+      dpcpp-version: intel-llvm/nightly-2023-10-26-rk
+      shell: cmd
       cmd: |
-        python scripts/test.py configure platform:x64 build:Release compiler:ICX2023.0.0 ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2-AVX512 tasking:TBB2021.7.0 intensity:4 package:ZIP
-        python scripts/test.py build
-        python scripts/test.py test
+        cmake --preset release-windows -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+        cmake --build build --target test_package
 
-  release-macosx-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP:
+  release-macosx:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_macos.yml@main
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
     with:
-      runs-on: '[ "macOS", "embree", "sign", "avx2" ]'
-      artifact-out: release-macosx-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-path: ./build/*.zip
+      project: embree
+      runs-on: '[ "macOS", "build", "x86_64" ]'
       cmd: |
-        scripts/test.py configure platform:x64 build:Release compiler:ICX2023.0.0 ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2 tasking:TBB2021.7.0 intensity:4 package:ZIP
-        scripts/test.py build
-        scripts/test.py test
+        source scripts/cmake-presets/macos-env.sh
+        cmake --preset release-macos -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+        cmake --build build --target test_package
+        rm build/embree-*-testing.zip
+      notarize-package: build/embree-*.zip
+      artifact-out: release-macosx
+      artifact-path: ./build/*.zip
 
-  release-linux-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP:
+  verify-package-macosx:
+    needs: release-macosx
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_docker.yml@main
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
     with:
-      image: embree/rockylinux:8.7
+      project: embree
+      runs-on: '[ "macOS", "build", "x86_64" ]'
+      artifact-in: release-macosx
+      cmd: |
+        mkdir package
+        tar -xzvf build/embree-* -C ./package
+      sign-path-verify: ./package
+
+  release-macosx-arm:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    with:
+      project: embree
+      runs-on: '[ "macOS13", "test", "arm" ]'
+      cmd: |
+        source scripts/cmake-presets/macos-env.sh
+        cmake --preset release-macos-arm -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+        cmake --build build --target test_package
+        rm build/embree-*-testing.zip
+      notarize-package: build/embree-*.zip
+      artifact-out: release-macosx-arm
+      artifact-path: ./build/*.zip
+
+  verify-package-macosx-arm:
+    needs: release-macosx-arm
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/macos.yml@main
+    with:
+      project: embree
+      runs-on: '[ "macOS13", "test", "arm" ]'
+      artifact-in: release-macosx-arm
+      cmd: |
+        mkdir package
+        tar -xzvf build/embree-* -C ./package
+      sign-path-verify: ./package
+
+  release-linux:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    with:
+      image: rockylinux:8.7
       runs-on: '[ "Linux", "docker", "avx512" ]'
-      artifact-out: release-linux-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      artifact-out: release-linux
       artifact-path: ./build/*.tar.gz
+      dpcpp-version: intel-llvm/nightly-2023-10-26-rk
+      pre-cmd: |
+        yum -y install epel-release
+        dnf makecache --refresh
+        dnf config-manager --set-enabled powertools
+        dnf -y update
+        dnf -y install ninja-build
       cmd: |
-        scripts/test.py configure platform:x64 build:Release compiler:ICX2023.0.0 ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2-AVX512 tasking:TBB2021.7.0 intensity:4 package:ZIP
-        scripts/test.py build
-        scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset release-linux -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+        cmake --build build --target test_package
 
 
 #####################################################################
 # DPC++ Release Builds
 ####################################################################
-
-  release-windows-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP:
+  release-windows-sycl:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/windows_gpu.yml@main
     with:
       runs-on: '[ "Windows", "NAS", "dg2" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-windows-internal.env
-      artifact-out: release-windows-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      env-from-files: ./.github/workflows/gfx-windows-internal.env
+      dpcpp-version: intel-llvm/nightly-2023-10-26-rk
+      artifact-out: release-windows-sycl
       artifact-path: ./build/*.zip
+      shell: cmd
       cmd: |
-        python scripts/test.py configure platform:x64 build:Release compiler:ICX2023.0.0 EMBREE_SYCL_SUPPORT:ON sycl:none gfx:RELEASE ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2-AVX512 tasking:TBB2021.7.0 intensity:4 package:ZIP
-        python scripts/test.py build
-        python scripts/test.py test
+        cmake --preset release-windows-sycl -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+        cmake --build build --target test_package
 
-  # FIXME: remove this AOT release once JIT properly tested in CI
-  #release-windows-x64-Release-SYCL-ICX2023_0_0-AOT-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP:
-  #  secrets: inherit
-  #  uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/embree_windows.yml@main
-  #  with:
-  #    runs-on: '[ "Windows", "NAS", "dg2" ]'
-  #    artifact-out: release-windows-x64-Release-SYCL-ICX2023_0_0-AOT-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-  #    artifact-path: ./build/*.zip
-  #    cmd: |
-  #      python scripts/test.py configure platform:x64 build:Release compiler:dpcpp EMBREE_SYCL_SUPPORT:ON sycl:dg2 gfx:RELEASE ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2-AVX512 intensity:4 package:ZIP
-  #      python scripts/test.py build
-  #      python scripts/test.py test
-
-  release-linux-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP:
+  release-linux-sycl:
     secrets: inherit
-    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@devel
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
     with:
-      image: embree/ubuntu:20.04
+      image: rockylinux:8.7
+      runs-on: '[ "Linux", "docker", "build" ]'
+      project: embree
+      dpcpp-version: intel-llvm/nightly-2023-10-26-rk
+      artifact-out: release-linux-sycl
+      artifact-path: ./build/*.tar.gz
+      pre-cmd: |
+        yum -y install epel-release
+        dnf makecache --refresh
+        dnf config-manager --set-enabled powertools
+        dnf -y update
+        dnf -y install ninja-build
+      cmd: |
+        module load cmake/3.25.3
+        cmake --preset release-linux-sycl -DCMAKE_BUILD_TYPE=Release
+        cmake --build build --config Release --target build
+
+  release-linux-sycl-test:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["release-linux-sycl"]
+    with:
+      image: embree/ubuntu:22.04
       options: --device=/dev/dri:/dev/dri
       runs-on: '[ "Linux", "docker", "dg2" ]'
       project: embree
-      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-release.env
-      artifact-out: release-linux-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-path: ./build/*.tar.gz
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: intel-llvm/nightly-2023-10-26-rk
+      artifact-in: release-linux-sycl
       cmd: |
-        scripts/test.py configure platform:x64 build:Release compiler:ICX2023.0.0 EMBREE_SYCL_SUPPORT:ON sycl:none gfx:RELEASE ispc:ispc1.18.1 isas:SSE2-SSE42-AVX-AVX2-AVX512 tasking:TBB2021.7.0 intensity:4 package:ZIP
-        scripts/test.py build
-        scripts/test.py test
+        module load cmake/3.25.3
+        cmake --preset integrate-sycl-linux-dpcpp -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --target test_package
 
+  release-linux-sycl-pvc-test:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: ["release-linux-sycl"]
+    with:
+      image: embree/ubuntu:22.04
+      options: --device=/dev/dri:/dev/dri
+      runs-on: '[ "Linux", "docker", "pvc" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      dpcpp-version: intel-llvm/nightly-2023-10-26-rk
+      artifact-in: release-linux-sycl
+      cmd: |
+        module load cmake/3.25.3
+        export ZE_FLAT_DEVICE_HIERARCHY=COMPOSITE
+        cmake --preset integrate-sycl-linux-dpcpp -DCMAKE_BUILD_TYPE=Release -DEMBREE_TESTING_INTEGRATION_PRESET=test-linux-icx
+        cmake --build build --target test_package
 
 #####################################################################
 # Release binaries scans
 #####################################################################
   binary-analysis:
     needs:
-      - release-windows-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      - release-macosx-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      - release-linux-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      - release-windows
+      - release-macosx
+      - release-linux
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/binary_analysis.yml@main
     with:
       project: embree
-      artifact-in-windows: release-windows-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-in-macos: release-macosx-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-in-linux: release-linux-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      artifact-in-windows: release-windows
+      artifact-in-macos: release-macosx
+      artifact-in-linux: release-linux
       path: "build/*.zip build/*.gz"
 
   antivirus-scan:
     needs:
-      - release-windows-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      - release-macosx-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      - release-linux-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      - release-windows
+      - release-macosx
+      - release-linux
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/antivirus_scan.yml@main
     with:
       project: embree
-      artifact-in-windows: release-windows-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-in-macos: release-macosx-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-in-linux: release-linux-x64-Release-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      artifact-in-windows: release-windows
+      artifact-in-macos: release-macosx
+      artifact-in-linux: release-linux
       path: "build/*.zip build/*.gz"
 
   binary-analysis-dpcpp:
     needs:
-      - release-windows-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      - release-linux-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      - release-windows-sycl
+      - release-linux-sycl-test
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/binary_analysis.yml@main
     with:
       project: embree
-      artifact-in-windows: release-windows-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-in-linux: release-linux-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      artifact-in-windows: release-windows-sycl
+      artifact-in-linux: release-linux-sycl
       path: "build/*.zip build/*.gz"
       artifact-out-name: "binary-analysis-dpcpp"
 
   antivirus-scan-dpcpp:
     needs:
-      - release-windows-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      - release-linux-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      - release-windows-sycl
+      - release-linux-sycl-test
     secrets: inherit
     uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/antivirus_scan.yml@main
     with:
       project: embree
-      artifact-in-windows: release-windows-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
-      artifact-in-linux: release-linux-x64-Release-SYCL-ICX2023_0_0-ISPC1_18_1-TBB2021_7_0-PACKAGE-ZIP
+      artifact-in-windows: release-windows-sycl
+      artifact-in-linux: release-linux-sycl
       path: "build/*.zip build/*.gz"
       artifact-out-name: "antivirus-scan-dpcpp"
+
+  binary-analysis-arm:
+    needs:
+      - release-macosx-arm
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/binary_analysis.yml@main
+    with:
+      project: embree
+      artifact-in-macos: release-macosx-arm
+      path: "build/*.zip"
+      artifact-out-name: "binary-analysis-arm"
+
+  antivirus-scan-arm:
+    needs:
+      - release-macosx-arm
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/antivirus_scan.yml@main
+    with:
+      project: embree
+      artifact-in-macos: release-macosx-arm
+      path: "build/*.zip"
+      artifact-out-name: "antivirus-scan-arm"
+
+#####################################################################
+# Testing package
+#####################################################################
+  test-package:
+    secrets: inherit
+    uses: intel-innersource/libraries.devops.renderkit.workflows/.github/workflows/docker_gpu.yml@main
+    needs: [release-linux]
+    with:
+      image: embree/ubuntu:22.04
+      runs-on: '[ "Linux", "docker", "dg2" ]'
+      project: embree
+      env-from-files: ./.github/workflows/dpcpp-none.env ./.github/workflows/gfx-ubuntu22-public.env
+      artifact-in: release-linux
+      artifact-out: release-testing-PACKAGE-ZIP
+      artifact-path: ./build/*.tar.gz
+      cmd: |
+        rm build/embree-4.3.2.x86_64.linux.tar.gz
+        mv build/embree-4.3.2.x86_64.linux-testing.tar.gz build/embree-4.3.2-testing.tar.gz

.gitignore

@@ -9,3 +9,4 @@ kernels/config.h
 kernels/hash.h
 include/embree4/rtcore_config.h
 build
+out

CHANGELOG.md

@@ -1,6 +1,49 @@
 Version History
 ---------------
 
+### Embree 4.3.2
+-   MacOS universal binary compilation now works.
+-   Some bugfixes for AVX512 on MacOS x86 machines.
+-   Embree now uses level zero raytracing extension to build BVH which enables forward-compatibility.
+
+### Embree 4.3.1
+-   Add missing EMBREE_GEOMETRY types to embree-config.cmake
+-   User defined thread count now takes precedence for internal task scheduler
+-   Fixed static linking issue with ze_wrapper library
+-   Better error reporting for SYCL platform and driver problems in embree_info and tutorial apps.
+-   Patch to glfw source is not applied by default anymore.
+-   Known issue: Running Embree on Intel® Data Center GPU Max Series with 2 tiles (e.g. Intel® Data Center GPU Max 1550) requires setting the environment variable ZE_FLAT_DEVICE_HIERARCHY=COMPOSITE.
+-   Known issue: Embree build using Apple Clang 15 and ARM support (via the SEE2NEON library) may cause "EXEC_BAD_INSTRUCTION" runtime exceptions. Please use Apple Clang <= 14 on macOS.
+
+### Embree 4.3.0
+-   Added instance array primitive for reducing memony requirements in scenes
+    with large amounts of similar instances.
+-   Properly checks driver if L0 RTAS extension can get loaded.
+-   Added varying version of rtcGetGeometryTransform for ISPC.
+-   Fixed signature of RTCMemoryMonitorFunction for ISPC.
+-   Add support for ARM64 Windows platform in CMake.
+
+### Embree 4.2.0
+-   SYCL version of Embree with GPU support is no longer in beta phase.
+-   Improved BVH build performance on many core machines for applications that oversubscribe threads.
+-   Added rtcGetGeometryTransformFromScene API function that can get used inside SYCL kernels.
+-   No longer linking to ze_loader in SYCL mode to avoid Intel(R) oneAPI Level Zero dependency
+    for CPU rendering.
+-   Releasing test package to test Embree.
+
+### Embree 4.1.0
+-   Added support for Intel® Data Center GPU Max Series.
+-   Added ARM64 Linux support.
+-   Added EMBREE_BACKFACE_CULLING_SPHERES cmake option. The new cmake option defaults to OFF.
+
+### Embree 4.0.1
+-   Improved performance for Tiger Lake, Comet Lake, Cannon Lake, Kaby Lake,
+    and Skylake client CPUs by using 256 bit SIMD instructions by default.
+-   Fixed broken motion blur of RTC_GEOMETRY_TYPE_ROUND_LINEAR_CURVE geometry type.
+-   Fixed bvh build retry issue for TBB 2020.3
+-   Added support for Intel® Data Center GPU Flex Series
+-   Fixed issue on systems without a SYCL platform.
+
 ### Embree 4.0.0
 -   This Embree release adds support for Intel® Arc™ GPUs through SYCL.
 -   The SYCL support of Embree is in beta phase. Current functionality, quality,

CMakeLists.txt

@@ -1,16 +1,12 @@
 ## Copyright 2009-2021 Intel Corporation
 ## SPDX-License-Identifier: Apache-2.0
 
-CMAKE_MINIMUM_REQUIRED(VERSION 3.1.0)
+CMAKE_MINIMUM_REQUIRED(VERSION 3.5)
 
 SET(EMBREE_VERSION_MAJOR 4)
-SET(EMBREE_VERSION_MINOR 0)
-SET(EMBREE_VERSION_PATCH 1)
-IF (EMBREE_SYCL_SUPPORT)
-  SET(EMBREE_VERSION_NOTE "-beta") # Embree SYCL version is in beta phase
-ELSE()
-  SET(EMBREE_VERSION_NOTE "")
-ENDIF()
+SET(EMBREE_VERSION_MINOR 3)
+SET(EMBREE_VERSION_PATCH 2)
+SET(EMBREE_VERSION_NOTE "")
 
 SET(EMBREE_VERSION ${EMBREE_VERSION_MAJOR}.${EMBREE_VERSION_MINOR}.${EMBREE_VERSION_PATCH})
 MATH(EXPR EMBREE_VERSION_NUMBER "10000*${EMBREE_VERSION_MAJOR} + 100*${EMBREE_VERSION_MINOR} + ${EMBREE_VERSION_PATCH}")
@@ -76,7 +72,13 @@ IF(COMMAND cmake_policy)
   if(POLICY CMP0074)
     cmake_policy(SET CMP0074 NEW)
   endif()
-ENDIF(COMMAND cmake_policy)
+  if(POLICY CMP0135)
+    cmake_policy(SET CMP0135 NEW)
+  endif()
+  if(POLICY CMP0057)
+    cmake_policy(SET CMP0057 NEW)
+  endif()
+ENDIF()
 
 MARK_AS_ADVANCED(CMAKE_BACKWARDS_COMPATIBILITY)
 MARK_AS_ADVANCED(EXECUTABLE_OUTPUT_PATH)
@@ -89,9 +91,7 @@ MARK_AS_ADVANCED(CLEAR CMAKE_CXX_COMPILER)
 
 SET(CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/common/cmake" ${CMAKE_MODULE_PATH})
 
-IF (BUILD_TESTING)
-  INCLUDE(test)
-ENDIF()
+INCLUDE(test)
 
 set(BUILD_DOC OFF CACHE INTERNAL "build documentation (internal only)")
 IF (BUILD_DOC)
@@ -145,14 +145,6 @@ OPTION(EMBREE_STACK_PROTECTOR "When enabled Embree compiles with stack protectio
 
 IF (NOT APPLE)
   OPTION(EMBREE_SYCL_SUPPORT "Enables SYCL GPU support." OFF)
-  IF (EMBREE_SYCL_SUPPORT)
-    MESSAGE(WARNING
-       "==================================================================================\n"
-       "  The SYCL support of Embree is in beta phase. Current functionality, quality,    \n"
-       "  and GPU performance may not reflect that of the final product. Please read the  \n"
-       "  documentation section \"Embree SYCL Known Issues\" for known limitations.       \n"
-       "==================================================================================\n")
-  ENDIF()
 ENDIF()
 CMAKE_DEPENDENT_OPTION(EMBREE_SYCL_LARGEGRF "Enables SYCL large GRF support." OFF "EMBREE_SYCL_SUPPORT" OFF)
 CMAKE_DEPENDENT_OPTION(EMBREE_SYCL_DBG "Enables DPC++ debug info." OFF "EMBREE_SYCL_SUPPORT" OFF)
@@ -170,12 +162,6 @@ IF (EMBREE_SYCL_SUPPORT)
   ADD_DEFINITIONS("-DEMBREE_SYCL_SUPPORT")  # FIXME: only use define from rtcore_config.h
 ENDIF()
 
-OPTION(EMBREE_SYCL_GPU_BVH_BUILDER "Using fast GPU-based BVH builder" OFF)
-IF (EMBREE_SYCL_GPU_BVH_BUILDER)
-  SET(EMBREE_STATIC_LIB ON)  
-  ADD_DEFINITIONS("-DEMBREE_SYCL_GPU_BVH_BUILDER")
-ENDIF()
-
 CMAKE_DEPENDENT_OPTION(EMBREE_SYCL_RT_SIMULATION "Using hardware simulation" OFF "EMBREE_SYCL_SUPPORT" OFF)
 MARK_AS_ADVANCED(EMBREE_SYCL_RT_SIMULATION)
 IF (EMBREE_SYCL_RT_SIMULATION AND NOT EMBREE_SYCL_RT_VALIDATION_API AND EMBREE_SYCL_IMPLICIT_DISPATCH_GLOBALS)
@@ -186,13 +172,12 @@ IF (EMBREE_SYCL_RT_SIMULATION)
 ENDIF()
 
 CMAKE_DEPENDENT_OPTION(EMBREE_SYCL_RT_VALIDATION_API "Use rt_validation API instead of IGC provided rt_production API" OFF "EMBREE_SYCL_SUPPORT" OFF)
-
 IF (EMBREE_SYCL_RT_VALIDATION_API)
   ADD_DEFINITIONS("-DEMBREE_SYCL_RT_VALIDATION_API")
 ENDIF()
 
 CMAKE_DEPENDENT_OPTION(EMBREE_SYCL_IMPLICIT_DISPATCH_GLOBALS "Using L0 allocated Dispatch Globals" ON "EMBREE_SYCL_RT_VALIDATION_API" OFF)
-IF (NOT EMBREE_SYCL_IMPLICIT_DISPATCH_GLOBALS)
+IF (EMBREE_SYCL_RT_VALIDATION_API AND NOT EMBREE_SYCL_IMPLICIT_DISPATCH_GLOBALS)
   ADD_DEFINITIONS("-DEMBREE_SYCL_ALLOC_DISPATCH_GLOBALS")
 ENDIF()
 
@@ -209,8 +194,15 @@ IF (EMBREE_SYCL_SUPPORT)
   ENDIF()
 ENDIF()
 
+CMAKE_DEPENDENT_OPTION(EMBREE_SYCL_L0_RTAS_BUILDER "Enable Level Zero RTAS builder" ON "EMBREE_SYCL_SUPPORT" OFF)
+IF (EMBREE_SYCL_L0_RTAS_BUILDER)
+  ADD_DEFINITIONS("-DEMBREE_SYCL_L0_RTAS_BUILDER")
+ENDIF()
+
 OPTION(EMBREE_RAY_MASK "Enables ray mask support." ON)
 OPTION(EMBREE_BACKFACE_CULLING "Enables backface culling.")
+OPTION(EMBREE_BACKFACE_CULLING_CURVES "Enables backface culling for curve primitives." OFF)
+OPTION(EMBREE_BACKFACE_CULLING_SPHERES "Enables backface culling for sphere primitives." OFF)
 
 OPTION(EMBREE_FILTER_FUNCTION "Enables filter functions." ON)
 OPTION(EMBREE_IGNORE_INVALID_RAYS "Ignores invalid rays." OFF) # FIXME: enable by default?
@@ -221,10 +213,11 @@ OPTION(EMBREE_GEOMETRY_CURVE "Enables support for curve geometries." ON)
 OPTION(EMBREE_GEOMETRY_SUBDIVISION "Enables support for subdiv geometries." ON)
 OPTION(EMBREE_GEOMETRY_USER "Enables support for user geometries." ON)
 OPTION(EMBREE_GEOMETRY_INSTANCE "Enables support for instances." ON)
+OPTION(EMBREE_GEOMETRY_INSTANCE_ARRAY "Enables support for instance arrays." ON)
 SET(EMBREE_MAX_INSTANCE_LEVEL_COUNT 1 CACHE STRING "Maximum number of instance levels.")
 
-IF (NOT EMBREE_GEOMETRY_INSTANCE AND NOT EMBREE_MAX_INSTANCE_LEVEL_COUNT EQUAL 1)
-  message(FATAL_ERROR "EMBREE_MAX_INSTANCE_LEVEL_COUNT must be 1 when EMBREE_GEOMETRY_INSTANCE is disabled")
+IF (NOT EMBREE_GEOMETRY_INSTANCE AND NOT EMBREE_GEOMETRY_INSTANCE_ARRAY AND NOT EMBREE_MAX_INSTANCE_LEVEL_COUNT EQUAL 1)
+  message(FATAL_ERROR "EMBREE_MAX_INSTANCE_LEVEL_COUNT must be 1 when EMBREE_GEOMETRY_INSTANCE and EMBREE_GEOMETRY_INSTANCE_ARRAY are disabled")
 ENDIF()
 
 OPTION(EMBREE_GEOMETRY_GRID "Enables support for grid geometries." ON)
@@ -242,10 +235,10 @@ OPTION(EMBREE_MIN_WIDTH "Enables min-width feature to enlarge curve and point th
 ##############################################################
 
 # detect ARM compilation
-IF (APPLE AND CMAKE_SYSTEM_NAME STREQUAL "Darwin" AND (CMAKE_SYSTEM_PROCESSOR STREQUAL "arm64" OR CMAKE_OSX_ARCHITECTURES MATCHES "arm64"))
+IF (APPLE AND CMAKE_SYSTEM_NAME STREQUAL "Darwin" AND (CMAKE_SYSTEM_PROCESSOR STREQUAL "arm64" AND CMAKE_OSX_ARCHITECTURES STREQUAL "") OR ("arm64" IN_LIST CMAKE_OSX_ARCHITECTURES))
   MESSAGE(STATUS "Building for Apple silicon")
   SET(EMBREE_ARM ON)
-ELSEIF(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
+ELSEIF(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64" OR CMAKE_SYSTEM_PROCESSOR STREQUAL "ARM64")
   MESSAGE(STATUS "Building for AArch64")
   SET(EMBREE_ARM ON)
 ENDIF()
@@ -301,6 +294,9 @@ message(DEBUG "CMAKE_GENERATOR_TOOLSET: ${CMAKE_GENERATOR_TOOLSET}")
 message(DEBUG "CMAKE_CXX_COMPILER_ID: ${CMAKE_CXX_COMPILER_ID}")
 message(DEBUG "CMAKE_CXX_COMPILER: ${CMAKE_CXX_COMPILER}")
 
+include(CheckCXXCompilerFlag)
+check_cxx_compiler_flag(-fsycl COMPILER_HAS_SYCL_SUPPORT)
+
 IF (WIN32)
   IF (NOT DEFAULT_CMAKE_CONFIGURATION_TYPES_SET)
     SET(CMAKE_CONFIGURATION_TYPES "${CONFIGURATION_TYPES}" CACHE STRING "List of generated configurations." FORCE)
@@ -309,8 +305,7 @@ IF (WIN32)
   SET_PROPERTY(GLOBAL PROPERTY USE_FOLDERS ON)
 
   IF ((${CMAKE_CXX_COMPILER_ID} MATCHES "IntelLLVM") OR
-      (${CMAKE_CXX_COMPILER_ID} MATCHES "Clang" AND ${CMAKE_CXX_COMPILER} MATCHES "icx") OR
-      (${CMAKE_CXX_COMPILER} MATCHES "dpcpp"))
+      (${CMAKE_CXX_COMPILER_ID} MATCHES "Clang" AND COMPILER_HAS_SYCL_SUPPORT))
     MESSAGE("-- DPCPP compiler detected")
     INCLUDE(dpcpp)
   ELSEIF(${CMAKE_GENERATOR_TOOLSET} MATCHES "^LLVM")
@@ -418,6 +413,9 @@ IF (EMBREE_MAX_ISA STREQUAL "NONE")
     ELSE()
       OPTION(EMBREE_ISA_AVX512 "Enables AVX512 ISA." ${COMPILER_SUPPORTS_AVX512})
     ENDIF()
+    # Don't use OPTION, but still set them to OFF, so that embree-config.cmake is consisten with its definitions
+    SET(EMBREE_ISA_NEON   OFF)
+    SET(EMBREE_ISA_NEON2X OFF)
   ENDIF()
 
 ELSEIF (EMBREE_MAX_ISA STREQUAL "DEFAULT")
@@ -715,4 +713,3 @@ ENDIF()
 ##############################################################
 
 INCLUDE(CPack)
-

CMakePresets.json

@@ -0,0 +1,103 @@
+{
+  "version": 4,
+  "cmakeMinimumRequired": {
+    "major": 3,
+    "minor": 11,
+    "patch": 0
+  },
+  "include": [
+    "scripts/cmake-presets/os.json", 
+    "scripts/cmake-presets/package.json",
+    "scripts/cmake-presets/compiler.json",
+    "scripts/cmake-presets/tbb.json",
+    "scripts/cmake-presets/continuous.json",
+    "scripts/cmake-presets/nightly.json",
+    "scripts/cmake-presets/release.json",
+    "scripts/cmake-presets/integrate.json",
+    "scripts/cmake-presets/performance.json"
+  ],
+  "configurePresets": [
+
+    {
+      "name": "package-linux-icx-ispc1_19_0-tbb2021_9_0-perf",
+      "inherits": ["package-linux", "env", "icx", "ispc1_19_0", "tbb2021_9_0"],
+      "binaryDir": "${sourceDir}/build",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Release",
+        "EMBREE_MAX_ISA": "AVX512",
+        "EMBREE_USE_GOOGLE_BENCHMARK": "ON",
+        "EMBREE_BUILD_GOOGLE_BENCHMARK_FROM_SOURCE": "ON"
+      }
+    },
+
+
+
+    {
+      "name": "package-linux-icx-sycl-ispc1_19_0-tbb2021_9_0-perf",
+      "inherits": ["package-linux", "env", "icx", "ispc1_19_0", "tbb2021_9_0"],
+      "binaryDir": "${sourceDir}/build",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Release",
+        "EMBREE_SYCL_SUPPORT": "ON",
+        "EMBREE_SYCL_L0_RTAS_BUILDER" : "OFF",
+        "EMBREE_SYCL_AOT_DEVICES": "none",
+        "EMBREE_MAX_ISA": "AVX512",
+        "EMBREE_USE_GOOGLE_BENCHMARK": "ON",
+        "EMBREE_BUILD_GOOGLE_BENCHMARK_FROM_SOURCE": "ON"
+      }
+    },
+
+
+
+ 
+
+
+
+
+    {
+      "name": "package-windows-v141-ispc1_19_0-tbb2021_9_0",
+      "inherits": ["package-windows", "env", "v141", "ispc1_19_0", "tbb2021_9_0"],
+      "binaryDir": "${sourceDir}/build",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Release",
+        "EMBREE_MAX_ISA": "SSE2"
+      }
+    },
+
+    {
+      "name": "package-windows-v140-ispc1_19_0-tbb2021_9_0",
+      "inherits": ["package-windows", "env", "v140", "ispc1_19_0", "tbb2021_9_0"],
+      "binaryDir": "${sourceDir}/build",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Release",
+        "EMBREE_MAX_ISA": "SSE2"
+      }
+    },
+    
+
+    {
+      "name": "package-windows-icx-sycl-ispc1_19_0-tbb2021_9_0",
+      "inherits": ["package-windows", "env", "icx-windows", "ispc1_19_0", "tbb2021_9_0"],
+      "binaryDir": "${sourceDir}/build",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Release",
+        "EMBREE_SYCL_SUPPORT": "ON",
+        "EMBREE_SYCL_L0_RTAS_BUILDER" : "OFF",
+        "EMBREE_SYCL_AOT_DEVICES": "none",
+        "EMBREE_MAX_ISA": "AVX512"
+      }
+    },
+
+
+
+    {
+      "name": "linux-coverity",
+      "inherits": ["package-linux", "env", "gcc", "tasking_internal"],
+      "binaryDir": "${sourceDir}/build",
+      "cacheVariables": {
+        "CMAKE_BUILD_TYPE": "Release",
+        "EMBREE_MAX_ISA": "SSE2"
+      }
+    }
+  ]
+}

common/algorithms/parallel_any_of.h

@@ -12,7 +12,8 @@ namespace embree
   template<typename Index, class UnaryPredicate>
     __forceinline bool parallel_any_of (Index first, Index last, UnaryPredicate pred)
   {
-    bool ret = false;
+    std::atomic_bool ret;
+    ret = false;
     
 #if defined(TASKING_TBB)
 #if TBB_INTERFACE_VERSION >= 12002

common/cmake/FindTBB.cmake

@@ -37,7 +37,7 @@
 #===============================================================================
 
 # We use INTERFACE libraries, which are only supported in 3.x
-cmake_minimum_required(VERSION 3.1)
+cmake_minimum_required(VERSION 3.5)
 
 # These two are used to automatically find the root and include directories.
 set(_TBB_INCLUDE_SUBDIR "include")

common/cmake/clang.cmake

@@ -8,10 +8,19 @@ MACRO(_SET_IF_EMPTY VAR VALUE)
 ENDMACRO()
 
 IF (EMBREE_ARM)
-  SET(FLAGS_SSE2 "-D__SSE__ -D__SSE2__")
-  SET(FLAGS_SSE42 "-D__SSE4_2__  -D__SSE4_1__")
-  SET(FLAGS_AVX "-D__AVX__ -D__SSE4_2__  -D__SSE4_1__  -D__BMI__ -D__BMI2__ -D__LZCNT__")
-  SET(FLAGS_AVX2 "-D__AVX2__ -D__AVX__ -D__SSE4_2__  -D__SSE4_1__  -D__BMI__ -D__BMI2__ -D__LZCNT__")
+  IF ("x86_64" IN_LIST CMAKE_OSX_ARCHITECTURES)
+    # set ARM an x86 flags for macOS universal binary build
+    SET(FLAGS_SSE2 "-D__SSE__ -D__SSE2__ -msse -msse2 -mno-sse4.2")
+    SET(FLAGS_SSE42 "-D__SSE4_2__  -D__SSE4_1__ -msse4.2")
+    SET(FLAGS_AVX "-D__AVX__ -D__SSE4_2__  -D__SSE4_1__  -D__BMI__ -D__BMI2__ -D__LZCNT__ -mavx")
+    SET(FLAGS_AVX2 "-D__AVX2__ -D__AVX__ -D__SSE4_2__  -D__SSE4_1__  -D__BMI__ -D__BMI2__ -D__LZCNT__ -mf16c -mavx2 -mfma -mlzcnt -mbmi -mbmi2")
+    _SET_IF_EMPTY(FLAGS_AVX512 "-march=skx")
+  ELSE ()
+    SET(FLAGS_SSE2 "-D__SSE__ -D__SSE2__")
+    SET(FLAGS_SSE42 "-D__SSE4_2__  -D__SSE4_1__")
+    SET(FLAGS_AVX "-D__AVX__ -D__SSE4_2__  -D__SSE4_1__  -D__BMI__ -D__BMI2__ -D__LZCNT__")
+    SET(FLAGS_AVX2 "-D__AVX2__ -D__AVX__ -D__SSE4_2__  -D__SSE4_1__  -D__BMI__ -D__BMI2__ -D__LZCNT__")
+  ENDIF ()
 ELSE ()
   # for `thread` keyword
   _SET_IF_EMPTY(FLAGS_SSE2  "-msse -msse2 -mno-sse4.2")
@@ -109,7 +118,7 @@ ELSE()
     SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fexceptions")              # enable exceptions
     SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -pthread")                  # enable threads
     SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -msimd128")                 # enable SIMD intrinsics
-  ENDIF()  
+  ENDIF()
 
   SET(CMAKE_CXX_FLAGS_DEBUG "")
   SET(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -g")              # generate debug information

common/cmake/dpcpp.cmake

@@ -20,8 +20,6 @@ IF (NOT WIN32)
   ENDIF()
 ENDIF()
 
-OPTION(EMBREE_LEVEL_ZERO "Use level zero API" OFF)
-
 GET_FILENAME_COMPONENT(SYCL_COMPILER_DIR ${CMAKE_CXX_COMPILER} PATH)
 GET_FILENAME_COMPONENT(SYCL_COMPILER_NAME ${CMAKE_CXX_COMPILER} NAME_WE)
 IF (NOT SYCL_COMPILER_NAME STREQUAL "clang++")
@@ -74,6 +72,7 @@ IF (EMBREE_SYCL_SUPPORT)
            ${SYCL_COMPILER_LIB_DIR}/sycl[0-9].lib
            ${SYCL_COMPILER_LIB_DIR}/sycl[0-9][0-9].lib)
     ENDIF()
+    LIST(GET SYCL_LIB 0 SYCL_LIB)
     GET_FILENAME_COMPONENT(SYCL_LIB_NAME ${SYCL_LIB} NAME_WE)
   ELSE()
     SET(SYCL_LIB_NAME "sycl")
@@ -145,10 +144,16 @@ IF (EMBREE_SYCL_SUPPORT)
   SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-pessimizing-move") # disabled: warning: moving a temporary object prevents copy elision [-Wpessimizing-move]
 
   IF (SYCL_ONEAPI_ICX AND WIN32)
+    IF (${CMAKE_CXX_COMPILER_VERSION} VERSION_GREATER_EQUAL 2024.0)
+      SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -I\"${SYCL_COMPILER_DIR}/../opt/compiler/include/sycl\" -I\"${SYCL_COMPILER_DIR}/../opt/compiler/include/sycl/sycl\"")       # disable warning from SYCL header
+    ENDIF()
     SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -I\"${SYCL_COMPILER_DIR}/../include/sycl\" -I\"${SYCL_COMPILER_DIR}/../include/\"")       # disable warning from SYCL header
     SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Qstd=c++17")
   ELSE()
     SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++17")
+    IF (SYCL_ONEAPI_ICX AND ${CMAKE_CXX_COMPILER_VERSION} VERSION_GREATER_EQUAL 2024.0)
+      SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -isystem \"${SYCL_COMPILER_DIR}/../opt/compiler/include/sycl\" -isystem \"${SYCL_COMPILER_DIR}/../opt/compiler/include/sycl/sycl\"")       # disable warning from SYCL header
+    ENDIF()
     SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -isystem \"${SYCL_COMPILER_DIR}/../include/sycl\" -isystem \"${SYCL_COMPILER_DIR}/../include/\"")       # disable warning from SYCL header
   ENDIF()
 
@@ -168,6 +173,8 @@ IF(SYCL_ONEAPI_ICX)
     SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} /Qimf-use-svml:false")
     SET(CMAKE_CXX_LINK_FLAGS "${CMAKE_CXX_LINK_FLAGS} /Qno-intel-lib")
     SET(CMAKE_C_LINK_FLAGS "${CMAKE_C_LINK_FLAGS} /Qno-intel-lib")
+    SET(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} /Qoption,link,/DEPENDENTLOADFLAG:0x2000")
+    SET(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} /Qoption,link,/DEPENDENTLOADFLAG:0x2000")
   ELSE()
     SET(CMAKE_CXX_LINK_FLAGS "${CMAKE_CXX_LINK_FLAGS} -static-intel")
     SET(CMAKE_C_LINK_FLAGS "${CMAKE_C_LINK_FLAGS} -static-intel")

common/cmake/embree-config.cmake

@@ -22,6 +22,7 @@ SET(EMBREE_ISA_AVX2  @EMBREE_ISA_AVX2@)
 SET(EMBREE_ISA_AVX512 @EMBREE_ISA_AVX512@)
 SET(EMBREE_ISA_AVX512SKX @EMBREE_ISA_AVX512@) # just for compatibility
 SET(EMBREE_ISA_NEON @EMBREE_ISA_NEON@)
+SET(EMBREE_ISA_NEON2X @EMBREE_ISA_NEON2X@)
 
 SET(EMBREE_BUILD_TYPE @CMAKE_BUILD_TYPE@)
 SET(EMBREE_ISPC_SUPPORT @EMBREE_ISPC_SUPPORT@)
@@ -44,6 +45,9 @@ SET(EMBREE_GEOMETRY_QUAD @EMBREE_GEOMETRY_QUAD@)
 SET(EMBREE_GEOMETRY_CURVE @EMBREE_GEOMETRY_CURVE@)
 SET(EMBREE_GEOMETRY_SUBDIVISION @EMBREE_GEOMETRY_SUBDIVISION@)
 SET(EMBREE_GEOMETRY_USER @EMBREE_GEOMETRY_USER@)
+SET(EMBREE_GEOMETRY_INSTANCE @EMBREE_GEOMETRY_INSTANCE@)
+SET(EMBREE_GEOMETRY_INSTANCE_ARRAY @EMBREE_GEOMETRY_INSTANCE_ARRAY@)
+SET(EMBREE_GEOMETRY_GRID @EMBREE_GEOMETRY_GRID@)
 SET(EMBREE_GEOMETRY_POINT @EMBREE_GEOMETRY_POINT@)
 
 SET(EMBREE_RAY_PACKETS @EMBREE_RAY_PACKETS@)
@@ -52,8 +56,6 @@ SET(EMBREE_CURVE_SELF_INTERSECTION_AVOIDANCE_FACTOR @EMBREE_CURVE_SELF_INTERSECT
 SET(EMBREE_DISC_POINT_SELF_INTERSECTION_AVOIDANCE @EMBREE_DISC_POINT_SELF_INTERSECTION_AVOIDANCE@)
 SET(EMBREE_MIN_WIDTH @EMBREE_MIN_WIDTH@)
 
-SET(EMBREE_SYCL_GPU_BVH_BUILDER @EMBREE_SYCL_GPU_BVH_BUILDER@)
-
 IF (EMBREE_STATIC_LIB AND (EMBREE_TASKING_SYSTEM STREQUAL "TBB"))
   INCLUDE(CMakeFindDependencyMacro)
   FIND_DEPENDENCY(TBB)
@@ -95,9 +97,16 @@ IF (EMBREE_SYCL_SUPPORT)
   ENDIF()
   IF (EMBREE_STATIC_LIB)
     INCLUDE("${EMBREE_ROOT_DIR}/@EMBREE_CMAKEEXPORT_DIR@/embree_rthwif-targets.cmake")
+    INCLUDE("${EMBREE_ROOT_DIR}/@EMBREE_CMAKEEXPORT_DIR@/ze_wrapper-targets.cmake")
   ENDIF()
   INCLUDE("${EMBREE_ROOT_DIR}/@EMBREE_CMAKEEXPORT_DIR@/embree_sycl-targets.cmake")
 
 ENDIF()
 
 INCLUDE("${EMBREE_ROOT_DIR}/@EMBREE_CMAKEEXPORT_DIR@/embree-targets.cmake")
+
+
+SET(EMBREE_TESTING_ONLY_SYCL_TESTS @EMBREE_TESTING_ONLY_SYCL_TESTS@)
+SET(EMBREE_TESTING_INTENSITY @EMBREE_TESTING_INTENSITY@)
+SET(EMBREE_TESTING_MEMCHECK @EMBREE_TESTING_MEMCHECK@)
+SET(EMBREE_TESTING_BENCHMARK @EMBREE_TESTING_BENCHMARK@)

common/cmake/ispc.cmake

@@ -39,8 +39,6 @@ IF (NOT EMBREE_ISPC_EXECUTABLE)
   FIND_PROGRAM(EMBREE_ISPC_EXECUTABLE ispc PATHS ${ISPC_DIR_HINT} DOC "Path to the ISPC executable.")
   IF (NOT EMBREE_ISPC_EXECUTABLE)
     MESSAGE(FATAL_ERROR "Intel SPMD Compiler (ISPC) not found. Disable EMBREE_ISPC_SUPPORT or install ISPC.")
-  ELSE()
-    MESSAGE(STATUS "Found Intel SPMD Compiler (ISPC): ${EMBREE_ISPC_EXECUTABLE}")
   ENDIF()
 ENDIF()
 
@@ -51,6 +49,8 @@ EXECUTE_PROCESS(COMMAND ${EMBREE_ISPC_EXECUTABLE} --version
 
 IF (NOT ${ISPC_RESULT} STREQUAL "0")
   MESSAGE(FATAL_ERROR "Error executing ISPC executable '${EMBREE_ISPC_EXECUTABLE}': ${ISPC_RESULT}")
+ELSE()
+  MESSAGE(STATUS "Found working Intel SPMD Compiler (ISPC): ${EMBREE_ISPC_EXECUTABLE}")
 ENDIF()
 
 STRING(REGEX MATCH "([0-9]+[.][0-9]+[.][0-9]+)" DUMMY "${ISPC_OUTPUT}")

common/cmake/msvc.cmake

@@ -22,6 +22,9 @@ MACRO(DISABLE_STACK_PROTECTOR_FOR_FILE file)
   ENDIF()
 ENDMACRO()
 
+SET(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} /DEPENDENTLOADFLAG:0x2000")
+SET(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} /DEPENDENTLOADFLAG:0x2000")
+
 SET(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} ${COMMON_CXX_FLAGS}")
 STRING(REPLACE "/RTC1" "" CMAKE_CXX_FLAGS_DEBUG ${CMAKE_CXX_FLAGS_DEBUG})         # disable native runtime checks
 SET(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /DDEBUG")                     # enables assertions

common/cmake/package.cmake

@@ -39,7 +39,8 @@ IF (EMBREE_SYCL_SUPPORT AND EMBREE_INSTALL_DEPENDENCIES)
       "${DPCPP_COMPILER_DIR}/../bin/sycl?.dll")
     INSTALL(FILES ${LIB_SYCL_DLL_FILES} DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
     
-    INSTALL(FILES "${DPCPP_COMPILER_DIR}/../bin/pi_level_zero.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
+    INSTALL(FILES "${DPCPP_COMPILER_DIR}/pi_level_zero.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
+    INSTALL(FILES "${DPCPP_COMPILER_DIR}/win_proxy_loader.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib OPTIONAL)
 
     IF (EXISTS "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/svml_dispmd.dll")
       INSTALL(FILES "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/svml_dispmd.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
@@ -47,14 +48,21 @@ IF (EMBREE_SYCL_SUPPORT AND EMBREE_INSTALL_DEPENDENCIES)
     IF (EXISTS "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/libmmd.dll")
       INSTALL(FILES "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/libmmd.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
     ENDIF()
-    
+    IF (EXISTS "${DPCPP_COMPILER_DIR}/libmmd.dll")
+      INSTALL(FILES "${DPCPP_COMPILER_DIR}/libmmd.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
+    ENDIF()
+
+    IF (EXISTS "${DPCPP_COMPILER_DIR}/pi_win_proxy_loader.dll")
+      INSTALL(FILES "${DPCPP_COMPILER_DIR}/pi_win_proxy_loader.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
+    ENDIF()
   ELSE()
     
     FILE(GLOB LIB_SYCL_FILES LIST_DIRECTORIES FALSE
       "${DPCPP_COMPILER_DIR}/../lib/libsycl.so"
       "${DPCPP_COMPILER_DIR}/../lib/libsycl.so.?"
       "${DPCPP_COMPILER_DIR}/../lib/libsycl.so.?.?"
-      "${DPCPP_COMPILER_DIR}/../lib/libsycl.so.?.?.?")
+      "${DPCPP_COMPILER_DIR}/../lib/libsycl.so.?.?.?"
+      "${DPCPP_COMPILER_DIR}/../lib/libsycl.so.?.?.?-?")
     INSTALL(FILES ${LIB_SYCL_FILES} DESTINATION "${CMAKE_INSTALL_LIBDIR}" COMPONENT lib)
     
     INSTALL(FILES "${DPCPP_COMPILER_DIR}/../lib/libpi_level_zero.so" DESTINATION "${CMAKE_INSTALL_LIBDIR}" COMPONENT lib)
@@ -88,6 +96,9 @@ IF (WIN32)
     IF (EXISTS "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/libmmd.dll")
       INSTALL(FILES "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/libmmd.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
     ENDIF()
+    IF (EXISTS "${DPCPP_COMPILER_DIR}//libmmd.dll")
+      INSTALL(FILES "${DPCPP_COMPILER_DIR}/libmmd.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
+    ENDIF()
     IF (EXISTS "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/svml_dispmd.dll")
       INSTALL(FILES "${DPCPP_COMPILER_DIR}/../redist/intel64_win/compiler/svml_dispmd.dll" DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT lib)
     ENDIF()
@@ -233,6 +244,7 @@ SET(CPACK_PACKAGE_NAME "Intel(R) Embree Ray Tracing Kernels")
 SET(CPACK_PACKAGE_FILE_NAME "embree-${EMBREE_VERSION}${EMBREE_VERSION_NOTE}")
 IF (EMBREE_SYCL_SUPPORT)
   SET(CPACK_PACKAGE_FILE_NAME "${CPACK_PACKAGE_FILE_NAME}.sycl")
+  SET(EMBREE_VERSION_SUFFIX)
 ENDIF()
 #SET(CPACK_PACKAGE_ICON "${PROJECT_SOURCE_DIR}/embree-doc/images/icon.png")
 #SET(CPACK_PACKAGE_RELOCATABLE TRUE)
@@ -245,14 +257,25 @@ SET(CPACK_PACKAGE_DESCRIPTION_SUMMARY "Intel(R) Embree implements high performan
 SET(CPACK_PACKAGE_VENDOR "Intel Corporation")
 SET(CPACK_PACKAGE_CONTACT embree_support@intel.com)
 
+
+SET(CPACK_ARCHIVE_COMPONENT_INSTALL ON)
+SET(CPACK_COMPONENTS_GROUPING ONE_PER_GROUP)
+
 SET(CPACK_COMPONENT_LIB_DISPLAY_NAME "Library")
 SET(CPACK_COMPONENT_LIB_DESCRIPTION "The Embree library including documentation.")
+SET(CPACK_COMPONENT_LIB_GROUP "embree")
 
 SET(CPACK_COMPONENT_DEVEL_DISPLAY_NAME "Development")
 SET(CPACK_COMPONENT_DEVEL_DESCRIPTION "Header Files for C and ISPC required to develop applications with Embree.")
+SET(CPACK_COMPONENT_DEVEL_GROUP "embree")
 
 SET(CPACK_COMPONENT_EXAMPLES_DISPLAY_NAME "Examples")
 SET(CPACK_COMPONENT_EXAMPLES_DESCRIPTION "Tutorials demonstrating how to use Embree.")
+SET(CPACK_COMPONENT_EXAMPLES_GROUP "embree")
+
+SET(CPACK_COMPONENT_TESTING_DISPLAY_NAME "Testing")
+SET(CPACK_COMPONENT_TESTING_DESCRIPTION "Models and reference images for tests")
+SET(CPACK_COMPONENT_TESTING_GROUP "embree-testing")
 
 # dependencies between components
 #SET(CPACK_COMPONENT_DEVEL_DEPENDS lib)
@@ -265,7 +288,6 @@ SET(CPACK_RESOURCE_FILE_LICENSE "${PROJECT_SOURCE_DIR}/LICENSE.txt")
 
 # Windows specific settings
 IF(WIN32)
-
   IF (CMAKE_SIZEOF_VOID_P EQUAL 8)
     SET(ARCH x64)
     SET(CPACK_PACKAGE_NAME "${CPACK_PACKAGE_NAME} x64")
@@ -276,11 +298,8 @@ IF(WIN32)
 
   SET(CPACK_GENERATOR ZIP)
   SET(CPACK_PACKAGE_FILE_NAME "${CPACK_PACKAGE_FILE_NAME}.${ARCH}.windows")
-  SET(CPACK_MONOLITHIC_INSTALL 1)
-  IF (EMBREE_TESTING_PACKAGE)
-    SET(PACKAGE_SCRIPT "${PROJECT_SOURCE_DIR}/scripts/package_win.bat")
-    ADD_TEST(NAME "BuildPackage" WORKING_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}" COMMAND ${PACKAGE_SCRIPT} ${CMAKE_BUILD_TYPE} "${CPACK_PACKAGE_FILE_NAME}.zip")
-  ENDIF()
+  SET(PACKAGE_BASE_NAME "${CPACK_PACKAGE_FILE_NAME}")
+  SET(PACKAGE_EXT "zip")
 
 # MacOSX specific settings
 ELSEIF(APPLE)
@@ -289,30 +308,52 @@ ELSEIF(APPLE)
   SET(CPACK_RESOURCE_FILE_README "${PROJECT_BINARY_DIR}/README.txt")
 
   SET(CPACK_GENERATOR ZIP)
-  SET(CPACK_PACKAGE_FILE_NAME "${CPACK_PACKAGE_FILE_NAME}.x86_64.macosx")
-  SET(CPACK_MONOLITHIC_INSTALL 1)
-  IF (EMBREE_TESTING_PACKAGE)
-    ADD_TEST(NAME "BuildPackage" WORKING_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}" COMMAND "${PROJECT_SOURCE_DIR}/scripts/package_macosx.sh" ${CMAKE_BUILD_TYPE} "${CPACK_PACKAGE_FILE_NAME}.zip" "${EMBREE_SIGN_FILE}")
+  IF (EMBREE_ARM)
+    SET(CPACK_PACKAGE_FILE_NAME "${CPACK_PACKAGE_FILE_NAME}.arm64.macosx")
+  ELSE()
+    SET(CPACK_PACKAGE_FILE_NAME "${CPACK_PACKAGE_FILE_NAME}.x86_64.macosx")
   ENDIF()
+  SET(PACKAGE_BASE_NAME "${CPACK_PACKAGE_FILE_NAME}")
+  SET(PACKAGE_EXT "zip")
+
+  add_custom_target(
+    post_package_notarize "${PROJECT_SOURCE_DIR}/scripts/package_post_build_notarize_macosx.sh" ${PACKAGE_BASE_NAME} ${EMBREE_SIGN_FILE}
+  )
 
 # Linux specific settings
 ELSE()
-
   SET(CPACK_GENERATOR TGZ)
   SET(CPACK_PACKAGE_FILE_NAME "${CPACK_PACKAGE_FILE_NAME}.x86_64.linux")
-  SET(CPACK_MONOLITHIC_INSTALL 1)
-  IF (EMBREE_TESTING_PACKAGE)
-    IF (EMBREE_SYCL_SUPPORT)
-      SET(PACKAGE_SCRIPT "${PROJECT_SOURCE_DIR}/scripts/package_linux_dpcpp.sh")
-    ELSE()
-      SET(PACKAGE_SCRIPT "${PROJECT_SOURCE_DIR}/scripts/package_linux.sh")
-    ENDIF()
-    ADD_TEST(NAME "BuildPackage" WORKING_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}" COMMAND ${PACKAGE_SCRIPT}
-      ${EMBREE_ZIP_MODE} ${EMBREE_LIBRARY_NAME} ${EMBREE_VERSION} ${EMBREE_VERSION_MAJOR} "${EMBREE_SIGN_FILE}")
+  SET(PACKAGE_BASE_NAME "${CPACK_PACKAGE_FILE_NAME}")
+  SET(PACKAGE_EXT "tar.gz")
+  IF (EMBREE_SYCL_SUPPORT)
+    SET(EMBREE_VERSION_SYCL_SUFFIX ".sycl")
   ENDIF()
-  
 ENDIF()
 
-IF (EMBREE_TESTING_PACKAGE)
-  SET_TESTS_PROPERTIES(BuildPackage PROPERTIES TIMEOUT 1200)
-ENDIF()
+
+add_custom_target(
+  build ${CMAKE_COMMAND} --build . --config ${CMAKE_BUILD_TYPE} --target package -j8
+  COMMAND ${CMAKE_COMMAND} -DPACKAGE_BASENAME=${PACKAGE_BASE_NAME} -DPACKAGE_EXT=${PACKAGE_EXT} -P ${PROJECT_SOURCE_DIR}/scripts/package_build.cmake
+)
+
+add_custom_target(
+  test_package ${CMAKE_COMMAND} -DWHAT="UNPACK" -DPACKAGE_BASENAME=${PACKAGE_BASE_NAME} -DPACKAGE_EXT=${PACKAGE_EXT} -P ${PROJECT_SOURCE_DIR}/scripts/package_test.cmake
+  COMMAND cd embree_install/testing && ${CMAKE_COMMAND} -B build -DEMBREE_TESTING_INTENSITY=${EMBREE_TESTING_INTENSITY}
+  COMMAND ctest --test-dir ${CMAKE_CURRENT_BINARY_DIR}/embree_install/testing/build -VV -C ${CMAKE_BUILD_TYPE} --output-log ctest.output 
+  COMMAND ${CMAKE_COMMAND} -DWHAT="CHECK" -P ${PROJECT_SOURCE_DIR}/scripts/package_test.cmake
+)
+
+
+if(WIN32)
+  set(INTEGRATE_BINARY "./build/Release/test.exe")
+else()
+  set(INTEGRATE_BINARY "./build/test")
+endif()
+
+add_custom_target(
+  test_integration ${CMAKE_COMMAND} -DWHAT="UNPACK" -DPACKAGE_BASENAME=${PACKAGE_BASE_NAME} -DPACKAGE_EXT=${PACKAGE_EXT} -P ${PROJECT_SOURCE_DIR}/scripts/package_test.cmake 
+  COMMAND cd ${PROJECT_SOURCE_DIR}/tests/integration/test_embree_release && ${CMAKE_COMMAND} -B build --preset ${EMBREE_TESTING_INTEGRATION_PRESET} -Dembree_DIR="${CMAKE_CURRENT_BINARY_DIR}/embree_install/lib/cmake/embree-${EMBREE_VERSION}"
+  COMMAND cd ${PROJECT_SOURCE_DIR}/tests/integration/test_embree_release && ${CMAKE_COMMAND} --build build --config Release
+  COMMAND cd ${PROJECT_SOURCE_DIR}/tests/integration/test_embree_release && ${INTEGRATE_BINARY}
+)

common/cmake/test.cmake

@@ -1,267 +1,569 @@
 ## Copyright 2009-2021 Intel Corporation
 ## SPDX-License-Identifier: Apache-2.0
+if (BUILD_TESTING OR EMBREE_TESTING_INSTALL_TESTS)
 
-INCLUDE(CTest)
+  INCLUDE(CTest)
 
-IF (WIN32)
-    IF("${CMAKE_CXX_COMPILER_ID}" MATCHES "MSVC")
-      SET(MY_PROJECT_BINARY_DIR "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_BUILD_TYPE}")
+  if (EMBREE_TESTING_INSTALL_TESTS)
+    SET(EMBREE_INSTALL_CTESTTESTFILE "${CMAKE_CURRENT_BINARY_DIR}/embree-addtests.cmake")
+    file(WRITE "${EMBREE_INSTALL_CTESTTESTFILE}" "")
+  endif()
+
+  if (NOT EMBREE_TESTING_PACKAGE_TEST_PROJECT)
+    IF (WIN32)
+        IF("${CMAKE_CXX_COMPILER_ID}" MATCHES "MSVC")
+          SET(MY_PROJECT_BINARY_DIR "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_BUILD_TYPE}")
+        ELSE()
+          SET(MY_PROJECT_BINARY_DIR "${CMAKE_CURRENT_BINARY_DIR}")
+        ENDIF()
     ELSE()
-      SET(MY_PROJECT_BINARY_DIR "${CMAKE_CURRENT_BINARY_DIR}")
+        SET(MY_PROJECT_BINARY_DIR "${CMAKE_CURRENT_BINARY_DIR}")
     ENDIF()
-ELSE()
-    SET(MY_PROJECT_BINARY_DIR "${CMAKE_CURRENT_BINARY_DIR}")
-ENDIF()
+  else()
+    SET(MY_PROJECT_BINARY_DIR "${CMAKE_CURRENT_SOURCE_DIR}/../${EMBREE_INSTALL_BINDIR}")
+  endif()
 
-FIND_PATH(EMBREE_TESTING_MODEL_DIR
-  NAMES test-models-intensity2.txt
-  PATHS "${PROJECT_SOURCE_DIR}/models"
-  DOC "Path to the folder containing the Embree models for regression testing."
-  NO_DEFAULT_PATHS)
+  if (NOT EMBREE_TESTING_PACKAGE_TEST_PROJECT)
+    SET(EMBREE_TESTING_INTENSITY 1 CACHE STRING "Intensity of testing (0 = no testing, 1 = verify and tutorials, 2 = light testing, 3 = intensive testing, 4 = very intensive testing.")
+    SET_PROPERTY(CACHE EMBREE_TESTING_INTENSITY PROPERTY STRINGS 0 1 2 3 4)
+    SET(EMBREE_TESTING_ONLY_SYCL_TESTS OFF CACHE BOOL "Run only tests with the sycl support.")
+    SET(EMBREE_TESTING_MEMCHECK OFF CACHE BOOL "Turns on memory checking for some tests.")
+    SET(EMBREE_TESTING_BENCHMARK OFF CACHE BOOL "Turns benchmarking on.")
+    SET(EMBREE_TESTING_BENCHMARK_DATABASE "${PROJECT_BINARY_DIR}" CACHE PATH "Path to database for benchmarking.")
+    SET(EMBREE_TESTING_PACKAGE OFF CACHE BOOL "Packages release as test.")
+    SET(EMBREE_TESTING_KLOCWORK OFF CACHE BOOL "Runs Kocwork as test.")
+    SET(EMBREE_TESTING_SDE OFF CACHE STRING "Uses SDE to run tests for specified CPU.")
+    SET_PROPERTY(CACHE EMBREE_TESTING_SDE PROPERTY STRINGS OFF pnr nhm wsm snb ivb hsw bdw knl skl skx cnl)
+  endif()
 
-SET(EMBREE_TESTING_INTENSITY 1 CACHE STRING "Intensity of testing (0 = no testing, 1 = verify and tutorials, 2 = light testing, 3 = intensive testing, 4 = very intensive testing.")
-SET_PROPERTY(CACHE EMBREE_TESTING_INTENSITY PROPERTY STRINGS 0 1 2 3 4)
-SET(EMBREE_TESTING_ONLY_SYCL_TESTS OFF CACHE BOOL "Run only tests with the sycl support.")
-SET(EMBREE_TESTING_MEMCHECK OFF CACHE BOOL "Turns on memory checking for some tests.")
-SET(EMBREE_TESTING_BENCHMARK OFF CACHE BOOL "Turns benchmarking on.")
-SET(EMBREE_TESTING_BENCHMARK_DATABASE "${PROJECT_BINARY_DIR}" CACHE PATH "Path to database for benchmarking.")
-SET(EMBREE_TESTING_PACKAGE OFF CACHE BOOL "Packages release as test.")
-SET(EMBREE_TESTING_KLOCWORK OFF CACHE BOOL "Runs Kocwork as test.")
-SET(EMBREE_TESTING_SDE OFF CACHE STRING "Uses SDE to run tests for specified CPU.")
-SET_PROPERTY(CACHE EMBREE_TESTING_SDE PROPERTY STRINGS OFF pnr nhm wsm snb ivb hsw bdw knl skl skx cnl)
+  FUNCTION (SET_EMBREE_TEST_PROPERTIES testname)
+    SET(variants "_cpp;_ispc;_sycl")
+    foreach(v ${variants})
+      if (v STREQUAL "_cpp")
+        SET(v "")
+      endif()
+      set(testnamedef "EMBREE_TEST_${testname}${v}_DEFINED")
+      if(${testnamedef})
+        SET_TESTS_PROPERTIES(${testname}${v} ${ARGN})
+      endif()
+    endforeach()
 
-# checks if the current cmake configuration is compatible with the requirements declared in the optionsfile
-MACRO (EMBREE_TEST_CHECK_OPTIONS optionsfile out)  
+    if (EMBREE_TESTING_INSTALL_TESTS)
+      file(APPEND "${EMBREE_INSTALL_CTESTTESTFILE}" "SET_EMBREE_TEST_PROPERTIES(${testname} ${ARGN}) \n")
+    endif()
+  ENDFUNCTION()
 
-  # if no optionsfile is found return true (no requirements defined)
-  SET(${out} 1)
-
-  IF(EXISTS "${optionsfile}")
-    FILE(READ "${optionsfile}" lines)
-    STRING(REGEX REPLACE "\n" ";" lines "${lines}")
-
-    FOREACH (line ${lines})
-
-      # parse line into list
-      string(REGEX MATCHALL "([^\ ]+\ |[^\ ]+$)" tokens "${line}")
-      LIST(LENGTH tokens token_count)
-      IF (NOT ${token_count} EQUAL 3)
-        message(FATAL_ERROR "illegal .embree_opitons in ${optionsfile} at line ${line}")
-      ENDIF()
-
-      # we require every line to follow the scheme "option [==/!=] value"
-      LIST(GET tokens 0 option)
-      LIST(GET tokens 1 comp)
-      LIST(GET tokens 2 value)
-
-      STRING(STRIP ${option} option)
-      STRING(STRIP ${comp} comp)
-      STRING(STRIP ${value} value)
-
-      IF ("${comp}" STREQUAL "==")
-        IF (NOT "${${option}}" STREQUAL "${value}")
-          SET(${out} 0)
-        ENDIF()
-      ELSE()
-        IF ("${${option}}" STREQUAL "${value}")
-          SET(${out} 0)
-        ENDIF()
-      ENDIF()
-
-      IF (${out} EQUAL 0)
-        BREAK()
-      ENDIF()
-
-    ENDFOREACH()
-  ENDIF()
-
-ENDMACRO()
-
-MACRO (ADD_EMBREE_GENERIC_CPP_TEST testname executable args)  
-  IF (BUILD_TESTING)
+  MACRO (ADD_EMBREE_GENERIC_TEST testname executable)  
     ADD_TEST(NAME ${testname}
              WORKING_DIRECTORY "${MY_PROJECT_BINARY_DIR}"
-             COMMAND ${executable} ${args})
-  ENDIF()
-ENDMACRO()
+             COMMAND ${executable} ${ARGN})
+    SET(testnamedef EMBREE_TEST_${testname}_DEFINED)
+    SET(${testnamedef} "1" CACHE INTERNAL "")
+  ENDMACRO()
 
-MACRO (ADD_EMBREE_GENERIC_ISPC_TEST testname executable args)  
-  IF (BUILD_TESTING AND EMBREE_ISPC_SUPPORT AND EMBREE_RAY_PACKETS)
-    ADD_TEST(NAME ${testname}_ispc
-             WORKING_DIRECTORY "${MY_PROJECT_BINARY_DIR}"
-             COMMAND ${executable}_ispc ${args})
-  ENDIF()       
-ENDMACRO()
-
-MACRO (ADD_EMBREE_GENERIC_SYCL_TEST testname executable args)  
-  IF (BUILD_TESTING AND EMBREE_SYCL_SUPPORT)
-    ADD_TEST(NAME ${testname}_sycl
-             WORKING_DIRECTORY ${MY_PROJECT_BINARY_DIR}
-             COMMAND ${executable}_sycl ${args})
-    SET_TESTS_PROPERTIES(${testname}_sycl PROPERTIES TIMEOUT 50)
-  ENDIF()
-ENDMACRO()
-
-MACRO (ADD_EMBREE_TEST testname executable args)
-  # Check if we should dump the output of this tests
-  LIST(FIND EMBREE_TESTING_DUMPTESTS ${testname} dumptest)
-  if (NOT(EMBREE_TESTING_DUMPFORMAT))
-    SET(EMBREE_TESTING_DUMPFORMAT "tga")
-  endif()
-  set(DUMP_ARG "")
-  IF(NOT(${dumptest} EQUAL -1))
-    file(MAKE_DIRECTORY "${PROJECT_SOURCE_DIR}/tests/dump")
-    set(argsx "${args} -o ${PROJECT_SOURCE_DIR}/tests/dump/${testname}.${EMBREE_TESTING_DUMPFORMAT}")
-  ELSE()
-    set(argsx ${args})
-  ENDIF()
-
-  # By default add tests for all executables
-  SET (extra_args ${ARGN})
-  LIST(APPEND extra_args cpp)
-  LIST(APPEND extra_args ispc)
-  LIST(APPEND extra_args sycl)
-  LIST(REMOVE_DUPLICATES extra_args)
-
-  # Exclude exe with '!'
-  LIST(FIND extra_args "!cpp" p)
-  if (NOT (${p} EQUAL -1))
-    LIST(REMOVE_ITEM extra_args "!cpp")
-    LIST(REMOVE_ITEM extra_args "cpp")
-  endif()
-
-  LIST(FIND extra_args "!ispc" p)
-  if (NOT (${p} EQUAL -1))
-    LIST(REMOVE_ITEM extra_args "!ispc")
-    LIST(REMOVE_ITEM extra_args "ispc")
-  endif()
-
-  LIST(FIND extra_args "!sycl" p)
-  if (NOT (${p} EQUAL -1))
-    LIST(REMOVE_ITEM extra_args "!sycl")
-    LIST(REMOVE_ITEM extra_args "sycl")
-  endif()
-
-  # Add the tests
-  FOREACH(loopvar ${extra_args})
+  MACRO (ADD_EMBREE_GENERIC_CPP_TEST testname executable)  
     if((NOT ${EMBREE_SYCL_SUPPORT}) OR (NOT ${EMBREE_TESTING_ONLY_SYCL_TESTS}))
-      if(${loopvar} STREQUAL "cpp")
-        ADD_EMBREE_GENERIC_CPP_TEST(${testname} ${executable} ${argsx})
-      ENDIF()
-      if(${loopvar} STREQUAL "ispc")
-        ADD_EMBREE_GENERIC_ISPC_TEST(${testname} ${executable} ${argsx})
-      ENDIF()
+      ADD_TEST(NAME ${testname}
+               WORKING_DIRECTORY "${MY_PROJECT_BINARY_DIR}"
+               COMMAND ${executable} ${ARGN})
+      SET(testnamedef EMBREE_TEST_${testname}_DEFINED)
+      SET(${testnamedef} "1" CACHE INTERNAL "")
     endif()
-    if(${loopvar} STREQUAL "sycl")
-      ADD_EMBREE_GENERIC_SYCL_TEST(${testname} ${executable} ${argsx}) 
+  ENDMACRO()
+
+  MACRO (ADD_EMBREE_GENERIC_ISPC_TEST testname executable)  
+    if((NOT ${EMBREE_SYCL_SUPPORT}) OR (NOT ${EMBREE_TESTING_ONLY_SYCL_TESTS}))
+      IF (EMBREE_ISPC_SUPPORT AND EMBREE_RAY_PACKETS)
+        ADD_TEST(NAME ${testname}_ispc
+                 WORKING_DIRECTORY "${MY_PROJECT_BINARY_DIR}"
+                 COMMAND ${executable}_ispc ${ARGN})
+        SET(testnamedef EMBREE_TEST_${testname}_ispc_DEFINED)
+        SET(${testnamedef} "1" CACHE INTERNAL "")
+      ENDIF()       
+    endif()
+  ENDMACRO()
+
+  MACRO (ADD_EMBREE_GENERIC_SYCL_TEST testname executable)  
+    IF (EMBREE_SYCL_SUPPORT)
+      ADD_TEST(NAME ${testname}_sycl
+               WORKING_DIRECTORY ${MY_PROJECT_BINARY_DIR}
+               COMMAND ${executable}_sycl ${ARGN})
+      SET(testnamedef EMBREE_TEST_${testname}_sycl_DEFINED)
+      SET(${testnamedef} 1 CACHE INTERNAL "")
+      SET_TESTS_PROPERTIES(${testname}_sycl PROPERTIES TIMEOUT 50)
     ENDIF()
-  ENDFOREACH()
-ENDMACRO()
+  ENDMACRO()
 
-MACRO (ADD_EMBREE_TUTORIAL_TEST name args)
-  ADD_EMBREE_TEST(${name} embree_${name} "--compare ${PROJECT_SOURCE_DIR}/tests/models/reference/${name}.exr ${args}" ${ARGN})
-ENDMACRO()
+  
 
-MACRO (ADD_EMBREE_PRIMITIVE_TEST xml shader)
-  EMBREE_TEST_CHECK_OPTIONS("${xml}.embree_options" out)
-  IF (${out} EQUAL 1)
 
-    GET_FILENAME_COMPONENT(FN ${xml} NAME_WE)
-    SET(testname "prim_${FN}_${shader}")
-    SET(compare_threshold "55")
 
-    # TODO: APPLE with AVX2 enabled have a sphere cap intersection problem with linear curves
-    IF ((testname MATCHES "prim_curves_linear_flat.*") AND ((testname MATCHES ".*uv") OR testname MATCHES ".*Ng"))
-      IF (APPLE AND (EMBREE_ISA_AVX2 OR EMBREE_ISA_AVX512))
-        SET(compare_threshold "200")
-      ENDIF()
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+  # Checks if the current cmake configuration is compatible with <condition>
+  # condition may be a triple of CMAKE_VARIABLE_NAME op VALUE
+  # supported operators for op are: ==, !=, <, <=, >, >=
+  FUNCTION (EMBREE_TESTING_CHECK_OPTION out condition)  
+    # parse condition into list
+    string(REGEX MATCHALL "([^\ ]+\ |[^\ ]+$)" tokens "${condition}")
+    LIST(LENGTH tokens token_count)
+    IF (NOT ${token_count} EQUAL 3)
+      message(FATAL_ERROR "illegal embree_opitons condition: ${condition}")
     ENDIF()
 
-    SET(args "-i ${xml} --shader ${shader} --time 0.5 --compare ${PROJECT_SOURCE_DIR}/tests/primitives/reference/${testname}.exr --compare-threshold ${compare_threshold}")
-    ADD_EMBREE_TEST(${testname} embree_viewer ${args})
+    # we require every condition to follow the scheme "variable_name op value"
+    LIST(GET tokens 0 option)
+    LIST(GET tokens 1 comp)
+    LIST(GET tokens 2 value)
 
-  ENDIF()
-ENDMACRO()
+    STRING(STRIP ${option} option)
+    STRING(STRIP ${comp} comp)
+    STRING(STRIP ${value} value)
 
-MACRO (ADD_EMBREE_MODEL_TEST testname executable model reference args)
-  EMBREE_TEST_CHECK_OPTIONS("${PROJECT_SOURCE_DIR}/tests/models/${model}.embree_options" out)
-  IF (${out} EQUAL 1)
+    SET(${out} 0 PARENT_SCOPE)
+    if ("${comp}" STREQUAL "==")
+      if ("${${option}}" STREQUAL "${value}")
+        SET(${out} 1 PARENT_SCOPE)
+      endif()
+    elseif ("${comp}" STREQUAL "!=")
+      if (NOT ("${${option}}" STREQUAL "${value}"))
+        SET(${out} 1 PARENT_SCOPE)
+      endif()
+    elseif ("${comp}" STREQUAL ">")
+      if ("${${option}}" GREATER "${value}")
+        SET(${out} 1 PARENT_SCOPE)
+      endif()
+    elseif ("${comp}" STREQUAL ">=")
+      if ("${${option}}" GREATER_EQUAL "${value}")
+        SET(${out} 1 PARENT_SCOPE)
+      endif()
+    elseif ("${comp}" STREQUAL "<")
+      if ("${${option}}" LESS "${value}")
+        SET(${out} 1 PARENT_SCOPE)
+      endif()
+    elseif ("${comp}" STREQUAL "<=")
+      if ("${${option}}" LESS_EQUAL "${value}")
+        SET(${out} 1 PARENT_SCOPE)
+      endif()
+    else()
+      message(FATAL_ERROR "Could not parse embree_option condition: ${condition}")
+    endif()
 
-    SET(extra_args ${ARGN})
-    IF ("${testname}" MATCHES ".*subdiv.*")  # skip subdiv models for SYCL mode
-      LIST(APPEND extra_args !sycl)
-    ENDIF()
-
-    SET(argsx "-c ${PROJECT_SOURCE_DIR}/tests/models/${model} --compare ${PROJECT_SOURCE_DIR}/tests/models/reference/${reference}.exr ${args}")
-    ADD_EMBREE_TEST(${testname} ${executable} ${argsx} ${extra_args})
-
-  ENDIF()
-ENDMACRO()
-
-MACRO (ADD_EMBREE_MODELS_TEST model_list_file testname executable reference args)
-  IF (BUILD_TESTING)  
-
-    SET(full_model_list_file "${PROJECT_SOURCE_DIR}/tests/models/${model_list_file}")
-    
-    IF(NOT EXISTS "${full_model_list_file}")
-      MESSAGE(FATAL_ERROR "File ${PROJECT_SOURCE_DIR}/tests/models/${model_list_file} does not exist!")
-    ENDIF()
-
-    FILE(READ "${full_model_list_file}" models)
-    STRING(REGEX REPLACE "\n" ";" models "${models}")
-    
-    FOREACH (model ${models})
-      IF (model MATCHES "^#")
-        CONTINUE()
-      ENDIF()
-      IF (model MATCHES ".*mblur.*" OR model MATCHES ".*motion_blur.*")  # skip mblur models if motion blur is not enabled
-        IF (NOT EMBREE_SYCL_MBLUR)
-          CONTINUE()
-        ENDIF()
-      ENDIF()
-      STRING(REGEX REPLACE " .*" "" modelname "${model}")
-      STRING(REGEX REPLACE "/" "_" modelname "${modelname}")
-      STRING(REGEX REPLACE ".ecs" "" modelname "${modelname}")
-
-      ADD_EMBREE_MODEL_TEST(${testname}_${modelname} ${executable} ${model} ${reference}_${modelname} "${args}")
-    ENDFOREACH()
-  ENDIF()
-ENDMACRO()
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-# add klocwork test
-IF (EMBREE_TESTING_KLOCWORK)
-  ADD_TEST(NAME Klocwork-Build WORKING_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}" COMMAND "${PROJECT_SOURCE_DIR}/scripts/klocwork_build.sh")
-  ADD_TEST(NAME Klocwork-Check WORKING_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}" COMMAND "${PROJECT_SOURCE_DIR}/scripts/klocwork_check.sh")
-  SET_TESTS_PROPERTIES(Klocwork-Build PROPERTIES TIMEOUT 2400)
-  SET_TESTS_PROPERTIES(Klocwork-Check PROPERTIES TIMEOUT 300)
-ENDIF()
-
-# add valgrind test
-IF (EMBREE_TESTING_MEMCHECK)
-  find_program( EMBREE_MEMORYCHECK_COMMAND valgrind )
-  set( EMBREE_MEMORYCHECK_COMMAND_OPTIONS "--trace-children=yes --leak-check=full --show-leak-kinds=definite --errors-for-leak-kinds=definite --error-exitcode=1" )
-  IF (NOT EMBREE_MEMORYCHECK_COMMAND)
-    MESSAGE(FATAL_ERROR "valgrind not found")
-  ENDIF()
-  FUNCTION(ADD_MEMCHECK_TEST name binary)
-    set(memcheck_command "${EMBREE_MEMORYCHECK_COMMAND} ${EMBREE_MEMORYCHECK_COMMAND_OPTIONS}")
-    separate_arguments(memcheck_command)
-    add_test(NAME ${name} COMMAND ${memcheck_command} "${MY_PROJECT_BINARY_DIR}/${binary}" ${ARGN})
   ENDFUNCTION()
-ENDIF()
+
+  # Checks multiple options from a list with EMBREE_TESTING_CHECK_OPTION
+  FUNCTION (EMBREE_TESTING_CHECK_OPTIONS_LIST out conditions)  
+    SET(${out} 1 PARENT_SCOPE)
+    FOREACH (c ${conditions})
+      EMBREE_TESTING_CHECK_OPTION(myout ${c})
+
+      IF (myout EQUAL 0)
+        SET(${out} 0 PARENT_SCOPE)
+        BREAK()
+      ENDIF()
+    ENDFOREACH()
+  ENDFUNCTION()
+
+
+  # looks for ifile in multiple possible locations and outputs a file with absolute path in ofile
+  FUNCTION (EMBREE_FIND_TEST_FILE ifile ofile errmsgflavor)
+    if (EXISTS "${ifile}")                                  # abs path, use get_filename_component because it could also be relative to cwd
+      get_filename_component(absifile "${ifile}" ABSOLUTE)
+      SET(${ofile} ${absifile} PARENT_SCOPE)
+    elseif(EXISTS "${PROJECT_SOURCE_DIR}/tests/${ifile}")   # testing dir
+      set(${ofile} "${PROJECT_SOURCE_DIR}/tests/${ifile}" PARENT_SCOPE)
+    elseif(EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/${ifile}")   # relative to source folder
+      set(${ofile} "${CMAKE_CURRENT_SOURCE_DIR}/${ifile}" PARENT_SCOPE)
+    else()
+      set(${ofile} "" PARENT_SCOPE)
+      if (errmsgflavor)
+        message(FATAL_ERROR
+          "Could not find ${errmsgflavor} \"${ifile}\"\n"
+          "looked for:\n"
+          "  ${inputfile}\n"
+          "  ${PROJECT_SOURCE_DIR}/tests/${ifile}\n"
+          "  ${CMAKE_CURRENT_SOURCE_DIR}/${ifile}\n"
+        )
+      endif()
+    endif()
+  ENDFUNCTION()
+
+
+  FUNCTION (EMBREE_ADD_TEST_PARSE_SUBLIST args keyword sublist)
+    SET(myargs ${args})
+    SET(mysublist "")
+    SET(keywords "ECS;XML;NO_REFERENCE;REFERENCE;REFERENCE_SUFFIX;INTENSITY;CONDITION_FILE;CONDITION;ARGS;NO_CPP;NO_ISPC;NO_SYCL;GEN_REFERENCE;")
+
+    list(FIND nargs ${keyword} istart)
+    if (NOT(istart EQUAL -1))
+      list(LENGTH myargs iend)
+      foreach(k ${keywords})
+        list(FIND myargs ${k} i)
+        if (NOT(i EQUAL -1) AND (i GREATER istart) AND (i LESS iend))
+          SET(iend ${i})
+        endif()
+      endforeach()
+
+      MATH(EXPR i "${istart}+1")
+      while (i LESS iend)
+        list(GET myargs ${i} elem)
+        list(APPEND mysublist ${elem})
+        MATH(EXPR i "${i}+1")
+      endwhile()
+    endif()
+
+    SET(${sublist} ${mysublist} PARENT_SCOPE)
+  ENDFUNCTION()
+
+  FUNCTION (EMBREE_ADD_TEST_PARSE_FLAG args keyword value)
+    SET(myargs ${args})
+    SET(${value} OFF PARENT_SCOPE)
+    list(FIND nargs ${keyword} i)
+    if (NOT(i EQUAL -1))
+      SET(${value} ON PARENT_SCOPE)
+    endif()
+  ENDFUNCTION()
+
+    # ADD_EMBREE_TEST_ECS(testname exename [ECS <file> | XML <file>] [NO_REFERENCE | REFERENCE <path> | REFERENCE_SUFFIX <suffix>] [INTENSITY <i>] [CONDITION <conds>] [ARGS <args>] [GEN_REFERENCE])
+    # [ECS <file> | XML <inputfile> | OPTIONS <inputfile>]
+    #   - looks for file and calls the test command with either "-c <inputfile>.ecs" or "-i <inputfile>.xml"
+    #
+    # [NO_REFERENCE | REFERENCE <path> | REFERENCE_SUFFIX <suffix>]
+    #   - if not found a reference is will be expected in the same folder as *.ecs with name *.ecs.exename.exr
+    #   - NO_REFERENCE: don't look for a reference (no --compare in test command)
+    #   - REFERENCE <path>: use the reference located in <path>. Same rules aplly as for finding ecs files: absolute or relative to CMAKE_CURRENT_SOURCE_DIR path, must not be located outside the embree root dir.
+    #   - REFERENCE_SUFFIX <suffix>: use the default reference location and name with a suffix before the last file extension, e.g. *.ecs.exename<suffix>.exr
+    #   - if this argument is not specified, looks for 
+    #       1. <inputfile>.exename.exr, or
+    #       2. <testname>.exr, if no <inputfile> was given
+    #
+    # [INTENSITY <i>] 
+    #   - default i = 1
+    #   - sets the intensity level for the test, test is only run if ${EMBREE_TESTING_INTENSITY} GREATER_EQUAL i
+    #   - could be done with an *.embree_options file, but this is more flexible, e.g. you can easier share reference images without specifying an absolute path
+    #   - DOES NOT overrite EMBREE_TESTING_INTENSITY, if specified in *.embree_options
+    #
+    # [CONDITION_FILE <file>]
+    #   - file containing additional conditions
+    #   - conditions are specified linewise in form of: EMBREE_OPTION op VALUE, where EMBREE_OPTION is a cmake variable used during embree configuration and op is one of ==, !=, <, <=, >, >=
+    #   - if this argument is not specified, looks for 
+    #       1. <inputfile>.embree_options, or
+    #       2. <testname>.embree_options, if no <inputfile> was given
+    #
+    # [CONDITION <conds>]
+    #   - cmake list of additional conditions, specified the same way as in an embree_options file
+    #
+    # [ARGS <args>]
+    #   - additional arguments for the test command
+    #
+    # [GEN_REFERENCE]
+    #   - writes the reference to the expected location
+    #   - could also be done with the ARGS parameter, but this way we don't have to deal with paths
+    #
+    # EXAMPLES
+    #
+    # all optional arguments default, 
+    #  - looks for points.ecs.embree_options to filter out test by configured options, if not found no restrictions
+    #  - runs for EMBREE_TESTING_INTENSITY >= 1
+    #  - reference will be expected in the same folder as points.ecs with name points.ecs.embree_viewer.exr
+    #ADD_EMBREE_TEST_ECS(viewer_points embree_viewer "models/xxx/points.ecs")                                  # 1) 
+    #ADD_EMBREE_TEST_ECS(viewer_points embree_viewer "${PROJECT_SOURCE_DIR}/models/xxx/points.ecs")            # 2) same as 1) but with absolute path to ecs
+    #ADD_EMBREE_TEST_ECS(viewer_points embree_viewer "points.ecs")                                             # 3) same as 1) but looks for points.ecs in the ${CMAKE_CURRENT_SOURCE_DIR} i.e. <embree_root>/tutorials/viewer/points.ecs
+
+    #ADD_EMBREE_TEST_ECS(viewer_points embree_viewer "models/xxx/points.ecs" ARGS --coherent INTENSITY 2)      # 4) same as 1) but runs only at EMBREE_TESTING_INTENSITY >= 2
+    #ADD_EMBREE_TEST_ECS(viewer_points embree_viewer "models/xxx/points.ecs" REFERENCE_SUFFIX "_quads" ARGS 
+    #  --coherent 
+    #  --convert-triangles-to-quads INTENSITY 2)                                                               # 5) same as 4) but expects reference name points.ecr.embree_viewer_quads.ecs
+    #ADD_EMBREE_TEST_ECS(viewer_points embree_viewer "models/xxx/points.ecs" REFERENCE "special.exr" ARGS 
+    #  --coherent 
+    #  --convert-triangles-to-quads INTENSITY 2)                                                               # 5) same as 4) but expects reference <embree-root>/tutorials/viewer/special.exr
+    ## 1) - 5) all share the same base input arguments defined in points.ecs as well as the points.ecs.embree_options
+
+    #ADD_EMBREE_TEST_ECS(point_geometry embree_point_geometry)                                                 # 6) no ecs file, uses point_geometry.embree_options if exists, expects reference ${CMAKE_CURRENT_SOURCE_DIR}/point_geometry.exr
+    #ADD_EMBREE_TEST_ECS(verify verify NO_REFERENCE)                                                           # 7) no ecs file, uses verify.embree_options if exists, no reference
+    #ADD_EMBREE_TEST_ECS(verify_memcheck verify NO_REFERENCE CONDITIONS "EMBREE_TESTING_MEMCHECK == ON")       # 8) same as 7) but with extra condition
+
+  FUNCTION (ADD_EMBREE_TEST_ECS testname executable)
+    if (EMBREE_TESTING_FILTER_TESTNAMES)
+      list(FIND EMBREE_TESTING_FILTER_TESTNAMES "${testname}" i)
+      if (${i} EQUAL -1)
+        return()
+      endif()
+    endif()
+
+    SET(nargs ${ARGN})
+
+    # disable everything
+    SET(testnamedef EMBREE_TEST_${testname}_DEFINED)
+    SET(${testnamedef} "0" CACHE INTERNAL "")
+    SET(testnamedef EMBREE_TEST_${testname}_ispc_DEFINED)
+    SET(${testnamedef} "0" CACHE INTERNAL "")
+    SET(testnamedef EMBREE_TEST_${testname}_sycl_DEFINED)
+    SET(${testnamedef} "0" CACHE INTERNAL "")
+
+    # parsing input file
+    #   ECS mode -> single parameter with filename after ECS keyword
+    list(FIND nargs "ECS" i)
+    if (NOT(i EQUAL -1))
+      SET(inputtype "-c")
+      MATH(EXPR i "${i}+1")
+      list(GET nargs ${i} out)
+      EMBREE_FIND_TEST_FILE("${out}" inputfile "test file")
+    endif()
+    #   XML mode -> single parameter with filename after XML keyword
+    list(FIND nargs "XML" i)
+    if (NOT(i EQUAL -1))
+      SET(inputtype "-i")
+      MATH(EXPR i "${i}+1")
+      list(GET nargs ${i} out)
+      EMBREE_FIND_TEST_FILE("${out}" inputfile "test file")
+    endif()
+    #   no filetype keyword -> look for <testname>.ecs/xml
+    #   does not need to be specified
+    if (NOT inputtype)
+      EMBREE_FIND_TEST_FILE("${testname}.ecs" inputfile "")
+      if(inputfile)
+        SET(inputtype "-c")
+      endif()
+      EMBREE_FIND_TEST_FILE("${testname}.ecs" inputfile "")
+      if(inputfile)
+        SET(inputtype "-i")
+      endif()
+    endif()
+
+    # parsing refernce image
+    #   no reference image mode -> no additional parameter to parse
+    list(FIND nargs "NO_REFERENCE" i)
+    if (NOT(i EQUAL -1))
+      SET(no_reference ON)
+    endif()
+    #   reference suffix mode -> singe parameter with suffix to default reference file name
+    list(FIND nargs "REFERENCE_SUFFIX" i)
+    if (NOT(i EQUAL -1))
+      MATH(EXPR i "${i}+1")
+      list(GET nargs ${i} out)
+      EMBREE_FIND_TEST_FILE("${inputfile}.${executable}${out}.exr" referencefile "reference image")
+    endif()
+    #   reference mode -> singe parameter with absolute path to reference image
+    list(FIND nargs "REFERENCE" i)
+    if (NOT(i EQUAL -1))
+      MATH(EXPR i "${i}+1")
+      list(GET nargs ${i} out)
+      EMBREE_FIND_TEST_FILE("${out}" referencefile "reference image")
+    endif()
+    #   no reference keyword -> look for <testname>.exr and <inputfile>.<executable>.exr respectively
+    if ((NOT no_reference) AND (NOT referencefile))
+      if (NOT inputfile)
+        EMBREE_FIND_TEST_FILE("${testname}.exr" referencefile "reference image")
+      else()
+        EMBREE_FIND_TEST_FILE("${inputfile}.${executable}.exr" referencefile "reference image")
+      endif()
+    endif()
+
+    # parsing intensity
+    #   single integer parameter
+    SET(intensity 1)
+    list(FIND nargs "INTENSITY" i)
+    if (NOT(i EQUAL -1))
+      MATH(EXPR i "${i}+1")
+      list(GET nargs ${i} intensity)
+    endif()
+
+    # parsing condition
+    SET(conditions "")
+    SET(conditionsfile)
+    list(FIND nargs "CONDITION_FILE" i)
+    #   condition file -> single parameter to absolute path of conditions file
+    if (NOT(i EQUAL -1))
+      list(GET nargs ${i} conditionsfile)
+      EMBREE_FIND_TEST_FILE("${conditionsfile}" conditionsfile "")
+    #   no <inputfile> specified -> look for <testname>.embree_options
+    elseif (NOT inputtype)
+      EMBREE_FIND_TEST_FILE("${testname}.embree_options" conditionsfile "")
+    #   <inputfile> specified -> look for <inputfile>.embree_options
+    else()
+      EMBREE_FIND_TEST_FILE("${inputfile}.embree_options" conditionsfile "")
+    endif()
+    if (conditionsfile)
+      file(READ "${conditionsfile}" lines)
+      string(REGEX REPLACE "\n" ";" conditions "${lines}")
+    endif()
+
+    # parsing additional conditions
+    EMBREE_ADD_TEST_PARSE_SUBLIST("${nargs}" "CONDITION" extra_conditions)
+    list(APPEND conditions ${extra_conditions})
+
+    # parsing extra args
+    EMBREE_ADD_TEST_PARSE_SUBLIST("${nargs}" "ARGS" extraargs)
+
+    EMBREE_ADD_TEST_PARSE_FLAG("${nargs}" "MEMCHECK"      memcheck)
+    EMBREE_ADD_TEST_PARSE_FLAG("${nargs}" "NO_CPP"        no_cpp)
+    EMBREE_ADD_TEST_PARSE_FLAG("${nargs}" "NO_ISPC"       no_ispc)
+    EMBREE_ADD_TEST_PARSE_FLAG("${nargs}" "NO_SYCL"       no_sycl)
+    EMBREE_ADD_TEST_PARSE_FLAG("${nargs}" "NO_POSTFIX"    no_postfix)
+    EMBREE_ADD_TEST_PARSE_FLAG("${nargs}" "GEN_REFERENCE" gen_reference)
+
+    # add the test, if the conditions are met
+    SET(matchconditions ON)
+    if (conditions)
+      EMBREE_TESTING_CHECK_OPTIONS_LIST(out "${conditions}")
+      if (out EQUAL 0)
+        SET(matchconditions OFF)
+      endif()
+    endif()
+
+    # LESS_EQUAL not supported on all CI runners
+    if (matchconditions AND ((intensity LESS EMBREE_TESTING_INTENSITY) OR (intensity EQUAL EMBREE_TESTING_INTENSITY)))
+      SET(args "")
+      if (inputfile)
+        list(APPEND args "${inputtype} ${inputfile}")
+      endif()
+      if (NOT no_reference)
+        list(APPEND args "--compare ${referencefile}")
+      endif()
+      if (gen_reference)
+        list(APPEND args "-o ${referencefile}")
+      endif()
+      foreach(a ${extraargs})
+        list(APPEND args "${a}")
+      endforeach()
+
+
+
+      if (memcheck)
+        list(PREPEND args "${MY_PROJECT_BINARY_DIR}/${executable}")
+        list(PREPEND args "${EMBREE_MEMORYCHECK_COMMAND_OPTIONS}")
+        if (no_postfix)
+          ADD_EMBREE_GENERIC_TEST(${testname} ${EMBREE_MEMORYCHECK_COMMAND} ${args})
+        else()
+          if((NOT ${EMBREE_SYCL_SUPPORT}) OR (NOT ${EMBREE_TESTING_ONLY_SYCL_TESTS}))
+            if (NOT(no_cpp))
+              ADD_EMBREE_GENERIC_CPP_TEST(${testname} ${EMBREE_MEMORYCHECK_COMMAND} ${args})
+            endif()
+            if (NOT(no_ispc))
+              ADD_EMBREE_GENERIC_ISPC_TEST(${testname} ${EMBREE_MEMORYCHECK_COMMAND} ${args})
+            endif()
+          endif()
+          if (NOT(no_sycl))
+            ADD_EMBREE_GENERIC_SYCL_TEST(${testname} ${EMBREE_MEMORYCHECK_COMMAND} ${args}) 
+          endif()
+        endif()
+      else()
+        if (no_postfix)
+          ADD_EMBREE_GENERIC_TEST(${testname} ${executable} ${args})
+        else()
+          if (NOT(no_cpp))
+            ADD_EMBREE_GENERIC_CPP_TEST(${testname} ${executable} ${args})
+          endif()
+          if (NOT(no_ispc))
+            ADD_EMBREE_GENERIC_ISPC_TEST(${testname} ${executable} ${args})
+          endif()
+          if (NOT(no_sycl))
+            ADD_EMBREE_GENERIC_SYCL_TEST(${testname} ${executable} ${args}) 
+          endif()
+        endif()
+      endif()
+    endif()
+
+
+
+    if (EMBREE_TESTING_INSTALL_TESTS)
+      if (inputfile)
+        get_filename_component(inputpath ${inputfile} DIRECTORY)
+        STRING(REPLACE "${PROJECT_SOURCE_DIR}/" "" inputpath "${inputpath}")
+        INSTALL(FILES "${inputfile}"
+                DESTINATION "${CMAKE_INSTALL_TESTDIR}/${inputpath}"
+                COMPONENT testing)
+      endif()
+
+      if (conditionsfile)
+        get_filename_component(conditionspath ${conditionsfile} DIRECTORY)
+        STRING(REPLACE "${PROJECT_SOURCE_DIR}/" "" conditionspath "${conditionspath}")
+        INSTALL(FILES "${conditionsfile}"
+                DESTINATION "${CMAKE_INSTALL_TESTDIR}/${conditionspath}"
+                COMPONENT testing)
+      endif()
+
+      if (referencefile)
+        get_filename_component(referencepath ${referencefile} DIRECTORY)
+        STRING(REPLACE "${PROJECT_SOURCE_DIR}/" "" referencepath "${referencepath}")
+        INSTALL(FILES "${referencefile}"
+                DESTINATION "${CMAKE_INSTALL_TESTDIR}/${referencepath}"
+                COMPONENT testing)
+      endif()
+
+      SET(testcall "ADD_EMBREE_TEST_ECS(${testname} ${executable}")
+      if (inputfile)
+        STRING(REPLACE "${PROJECT_SOURCE_DIR}/" "" inputfile "${inputfile}")
+        if (inputtype STREQUAL "-c")
+          SET(testcall "${testcall} \n    ECS ${inputfile}")
+        elseif (inputtype STREQUAL "-i")
+          SET(testcall "${testcall} \n    XML ${inputfile}")
+        endif()
+      endif()
+
+      if (no_reference)
+        SET(testcall "${testcall} \n    NO_REFERENCE")
+      else()
+        STRING(REPLACE "${PROJECT_SOURCE_DIR}/" "" referencefile "${referencefile}")
+        SET(testcall "${testcall} \n    REFERENCE ${referencefile}")
+      endif()
+
+      SET(testcall "${testcall} \n    INTENSITY ${intensity}")
+      if (${memcheck})
+        SET(testcall "${testcall} \n    MEMCHECK")
+      endif()
+      if (${no_cpp})
+        SET(testcall "${testcall} \n    NO_CPP")
+      endif()
+      if (${no_ispc})
+        SET(testcall "${testcall} \n    NO_ISPC")
+      endif()
+      if (${no_sycl})
+        SET(testcall "${testcall} \n    NO_SYCL")
+      endif()
+      if (${no_postfix})
+        SET(testcall "${testcall} \n    NO_POSTFIX")
+      endif()
+      STRING(REPLACE "${PROJECT_SOURCE_DIR}/" "" conditionsfile "${conditionsfile}")
+      SET(testcall "${testcall} \n    CONDITION_FILE ${conditionsfile}")
+      SET(testcall "${testcall} \n    CONDITION ")
+      foreach(c ${conditions})
+        SET(testcall "${testcall} \"${c}\"")
+      endforeach()
+      SET(testcall "${testcall} \n    ARGS ")
+      foreach(a ${extraargs})
+        SET(testcall "${testcall} ${a}")
+      endforeach()
+
+      SET(testcall "${testcall})\n\n")
+      file(APPEND "${EMBREE_INSTALL_CTESTTESTFILE}" "${testcall}")
+    endif()
+  ENDFUNCTION()
+
+else()
+  FUNCTION(ADD_EMBREE_TEST_ECS testname executable)
+  ENDFUNCTION()
+  FUNCTION(ADD_EMBREE_GENERIC_TEST testname executable)
+  ENDFUNCTION()
+  FUNCTION(ADD_EMBREE_GENERIC_CPP_TEST testname executable)
+  ENDFUNCTION()
+  FUNCTION(ADD_EMBREE_GENERIC_SYCL_TEST testname executable)
+  ENDFUNCTION()
+  FUNCTION(ADD_EMBREE_GENERIC_ISPC_TEST testname executable)
+  ENDFUNCTION()
+  FUNCTION(SET_EMBREE_TEST_PROPERTIES testname)
+  ENDFUNCTION()
+endif()

common/lexers/stream.h

@@ -3,10 +3,6 @@
 
 #pragma once
 
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif
-
 #include "../sys/platform.h"
 #include "../sys/ref.h"
 #include "../sys/filename.h"
@@ -126,17 +122,16 @@ namespace embree
   class FileStream : public Stream<int>
   {
   public:
-
-    FileStream (FILE* file, const std::string& name = "file")
-      : file(file), lineNumber(1), colNumber(0), charNumber(0), name(std::shared_ptr<std::string>(new std::string(name))) {}
-
     FileStream (const FileName& fileName)
       : lineNumber(1), colNumber(0), charNumber(0), name(std::shared_ptr<std::string>(new std::string(fileName.str())))
     {
-      file = fopen(fileName.c_str(),"r");
-      if (file == nullptr) THROW_RUNTIME_ERROR("cannot open file " + fileName.str());
+      if (ifs) ifs.close();
+      ifs.open(fileName.str());
+      if (!ifs.is_open()) THROW_RUNTIME_ERROR("cannot open file " + fileName.str());
+    }
+    ~FileStream() { 
+      if (ifs) ifs.close();
     }
-    ~FileStream() { if (file) fclose(file); }
 
   public:
     ParseLocation location() {
@@ -144,14 +139,15 @@ namespace embree
     }
 
     int next() {
-      int c = fgetc(file);
+      int c = ifs.get();
       if (c == '\n') { lineNumber++; colNumber = 0; } else if (c != '\r') colNumber++;
       charNumber++;
       return c;
     }
 
+
   private:
-    FILE* file;
+    std::ifstream ifs;
     ssize_t lineNumber;           /// the line number the token is from
     ssize_t colNumber;            /// the character number in the current line
     ssize_t charNumber;           /// the character in the file

common/math/emath.h

@@ -48,6 +48,9 @@ namespace embree
   __forceinline int   toInt  (const float& a) { return int(a); }
   __forceinline float toFloat(const int&   a) { return float(a); }
 
+  __forceinline int   asInt  (const float& a) { return *((int*)&a); }
+  __forceinline float asFloat(const int&   a) { return *((float*)&a); }
+
 #if defined(__WIN32__)
   __forceinline bool finite ( const float x ) { return _finite(x) != 0; }
 #endif

common/math/math_sycl.h

@@ -75,8 +75,6 @@ namespace embree
   __forceinline float floor( const float x ) { return sycl::floor(x); }
   __forceinline float ceil ( const float x ) { return sycl::ceil(x); }
   __forceinline float frac ( const float x ) { return x-floor(x); }
-  __forceinline float sincos ( const float phi, float* sin, float* cos ) { return *sin = sycl::sincos(phi, cos); }
-  __forceinline float sincosf ( const float phi, float* sin, float* cos ) { return *sin = sycl::sincos(phi, cos); }
 
   //__forceinline double abs  ( const double x ) { return ::fabs(x); }
   //__forceinline double sign ( const double x ) { return x<0?-1.0:1.0; }

common/math/vec2fa_sycl.h

@@ -95,8 +95,8 @@ namespace embree
   __forceinline Vec2fa abs  ( const Vec2fa& a ) { return Vec2fa(sycl::fabs(a.x),sycl::fabs(a.y)); }
   __forceinline Vec2fa sign ( const Vec2fa& a ) { return Vec2fa(sycl::sign(a.x),sycl::sign(a.y)); }
 
-   //__forceinline Vec2fa rcp  ( const Vec2fa& a ) { return Vec2fa(sycl::recip(a.x),sycl::recip(a.y)); }
-  __forceinline Vec2fa rcp  ( const Vec2fa& a ) { return Vec2fa(__sycl_std::__invoke_native_recip<float>(a.x),__sycl_std::__invoke_native_recip<float>(a.y)); }
+   //__forceinline Vec2fa rcp  ( const Vec2fa& a ) { return Vec2fa(sycl::native::recip(a.x),sycl::native::recip(a.y)); }
+  __forceinline Vec2fa rcp  ( const Vec2fa& a ) { return Vec2fa(sycl::native::recip(a.x),sycl::native::recip(a.y)); }
   __forceinline Vec2fa sqrt ( const Vec2fa& a ) { return Vec2fa(sycl::sqrt(a.x),sycl::sqrt(a.y)); }
   __forceinline Vec2fa sqr  ( const Vec2fa& a ) { return Vec2fa(a.x*a.x,a.y*a.y); }
   

common/math/vec3fa_sycl.h

@@ -92,7 +92,7 @@ namespace embree
   __forceinline Vec3fa sign ( const Vec3fa& a ) { return Vec3fa(sycl::sign(a.x),sycl::sign(a.y),sycl::sign(a.z)); }
 
   //__forceinline Vec3fa rcp  ( const Vec3fa& a ) { return Vec3fa(sycl::recip(a.x),sycl::recip(a.y),sycl::recip(a.z)); }
-  __forceinline Vec3fa rcp  ( const Vec3fa& a ) { return Vec3fa(__sycl_std::__invoke_native_recip<float>(a.x),__sycl_std::__invoke_native_recip<float>(a.y),__sycl_std::__invoke_native_recip<float>(a.z)); }
+  __forceinline Vec3fa rcp  ( const Vec3fa& a ) { return Vec3fa(sycl::native::recip(a.x),sycl::native::recip(a.y),sycl::native::recip(a.z)); }
   __forceinline Vec3fa sqrt ( const Vec3fa& a ) { return Vec3fa(sycl::sqrt(a.x),sycl::sqrt(a.y),sycl::sqrt(a.z)); }
   __forceinline Vec3fa sqr  ( const Vec3fa& a ) { return Vec3fa(a.x*a.x,a.y*a.y,a.z*a.z); }
 
@@ -393,7 +393,7 @@ namespace embree
   __forceinline Vec3fx sign ( const Vec3fx& a ) { return Vec3fx(sycl::sign(a.x),sycl::sign(a.y),sycl::sign(a.z),sycl::sign(a.z)); }
 
   //__forceinline Vec3fx rcp  ( const Vec3fx& a ) { return Vec3fx(sycl::recip(a.x),sycl::recip(a.y),sycl::recip(a.z)); }
-  __forceinline Vec3fx rcp  ( const Vec3fx& a ) { return Vec3fx(__sycl_std::__invoke_native_recip<float>(a.x),__sycl_std::__invoke_native_recip<float>(a.y),__sycl_std::__invoke_native_recip<float>(a.z),__sycl_std::__invoke_native_recip<float>(a.w)); }
+  __forceinline Vec3fx rcp  ( const Vec3fx& a ) { return Vec3fx(sycl::native::recip(a.x),sycl::native::recip(a.y),sycl::native::recip(a.z),sycl::native::recip(a.w)); }
   __forceinline Vec3fx sqrt ( const Vec3fx& a ) { return Vec3fx(sycl::sqrt(a.x),sycl::sqrt(a.y),sycl::sqrt(a.z),sycl::sqrt(a.w)); }
   __forceinline Vec3fx sqr  ( const Vec3fx& a ) { return Vec3fx(a.x*a.x,a.y*a.y,a.z*a.z,a.w*a.w); }
 

common/simd/arm/emulation.h

@@ -34,6 +34,7 @@ __forceinline __m128 _mm_broadcast_ss (float const * mem_addr)
 #define _MM_SET_EXCEPTION_MASK(x)
 // #define _MM_SET_FLUSH_ZERO_MODE(x)
 
+/*
 __forceinline int _mm_getcsr()
 {
   return 0;
@@ -43,6 +44,7 @@ __forceinline void _mm_mfence()
 {
   __sync_synchronize();
 }
+*/
 
 __forceinline __m128i _mm_load4epu8_epi32(__m128i *ptr)
 {

common/simd/vboolf4_sse2.h

@@ -119,7 +119,7 @@ namespace embree
 #if defined(__aarch64__)
   template<int i0, int i1, int i2, int i3>
   __forceinline vboolf4 shuffle(const vboolf4& v) {
-    return vreinterpretq_f32_u8(vqtbl1q_u8( vreinterpretq_u8_s32(v), _MN_SHUFFLE(i0, i1, i2, i3)));
+    return vreinterpretq_f32_u8(vqtbl1q_u8( vreinterpretq_u8_s32((int32x4_t)v.v), _MN_SHUFFLE(i0, i1, i2, i3)));
   }
 
   template<int i0, int i1, int i2, int i3>

common/simd/vfloat16_avx512.h

@@ -316,6 +316,17 @@ namespace embree
     return madd(t,b-a,a);
   }
 
+  __forceinline bool isvalid (const vfloat16& v) {
+    return all((v > vfloat16(-FLT_LARGE)) & (v < vfloat16(+FLT_LARGE)));
+  }
+
+  __forceinline void xchg(vboolf16 m, vfloat16& a, vfloat16& b)
+  {
+    vfloat16 c = a;
+    a = select(m,b,a);
+    b = select(m,c,b); 
+  }
+
   ////////////////////////////////////////////////////////////////////////////////
   /// Rounding Functions
   ////////////////////////////////////////////////////////////////////////////////

common/sys/alloc.cpp

@@ -48,9 +48,8 @@ namespace embree
 
   void disableUSMAllocTutorial()
   {
-    if (tls_context_tutorial  == nullptr) throw std::runtime_error("USM allocation not enabled");
-    if (tls_device_tutorial  == nullptr) throw std::runtime_error("USM allocation not enabled");
-    
+    //if (tls_context_tutorial  == nullptr) throw std::runtime_error("USM allocation not enabled");
+    //if (tls_device_tutorial  == nullptr) throw std::runtime_error("USM allocation not enabled");
     tls_context_tutorial = nullptr;
     tls_device_tutorial = nullptr;
   }

common/sys/filename.cpp

@@ -35,18 +35,6 @@ namespace embree
       filename.resize(filename.size()-1);
   }
   
-  /*! returns path to home folder */
-  FileName FileName::homeFolder() 
-  {
-#ifdef __WIN32__
-    const char* home = getenv("UserProfile");
-#else
-    const char* home = getenv("HOME");
-#endif
-    if (home) return home;
-    return "";
-  }
-
   /*! returns path to executable */
   FileName FileName::executableFolder() {
     return FileName(getExecutableFileName()).path();

common/sys/filename.h

@@ -20,9 +20,6 @@ namespace embree
 
     /*! create a valid filename from a string */
     FileName (const std::string& filename);
-    
-    /*! returns path to home folder */
-    static FileName homeFolder();
 
     /*! returns path to executable */
     static FileName executableFolder();

common/sys/intrinsics.h

@@ -223,7 +223,7 @@ namespace embree
 
 #if defined(EMBREE_SYCL_SUPPORT) && defined(__SYCL_DEVICE_ONLY__)
   __forceinline unsigned int   bsf(unsigned v) {
-    return SYCL_CTZ::ctz(v);
+    return sycl::ctz(v);
   }
 
 #else
@@ -248,7 +248,7 @@ namespace embree
   
 #if defined(EMBREE_SYCL_SUPPORT) && defined(__SYCL_DEVICE_ONLY__)
   __forceinline size_t bsf(size_t v) {
-    return SYCL_CTZ::ctz(v);
+    return sycl::ctz(v);
   }
 #else
   

common/sys/platform.h

@@ -20,6 +20,7 @@
 #include <cstring>
 #include <stdint.h>
 #include <functional>
+#include <mutex>
 
 #if defined(EMBREE_SYCL_SUPPORT)
 
@@ -33,13 +34,6 @@
 
 #pragma clang diagnostic pop
 
-#define SYCL_ONEAPI sycl
-#define SYCL_EXT_ONEAPI sycl::ext::oneapi
-#define SYCL_SUBGROUP sycl
-#define SYCL_EXPERIMENTAL sycl::ext::oneapi::experimental
-#define SYCL_INTEL sycl::ext::intel
-#define SYCL_CTZ sycl
-
 #include "sycl.h"
 
 #if defined(EMBREE_SYCL_SUPPORT) && defined(__SYCL_DEVICE_ONLY__)
@@ -212,15 +206,11 @@
 #define PRINT2(x,y) embree_cout << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << embree_endl
 #define PRINT3(x,y,z) embree_cout << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << embree_endl
 #define PRINT4(x,y,z,w) embree_cout << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << ", " << STRING(w) << " = " << (w) << embree_endl
-#define PRINT5(x,y,z,w,a) embree_cout << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << ", " << STRING(w) << " = " << (w) << ", " << STRING(a) << " = " << (a) << embree_endl
-#define PRINT6(x,y,z,w,a,b) embree_cout << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << ", " << STRING(w) << " = " << (w) << ", " << STRING(a) << " = " << (a) << ", " << STRING(b) << " = " << (b) << embree_endl
 
 #define UPRINT(x) embree_cout_uniform << STRING(x) << " = " << (x) << embree_endl
 #define UPRINT2(x,y) embree_cout_uniform << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << embree_endl
 #define UPRINT3(x,y,z) embree_cout_uniform << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << embree_endl
 #define UPRINT4(x,y,z,w) embree_cout_uniform << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << ", " << STRING(w) << " = " << (w) << embree_endl
-#define UPRINT5(x,y,z,w,a) embree_cout_uniform << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << ", " << STRING(w) << " = " << (w) << ", " << STRING(a) << " = " << (a) << embree_endl
-#define UPRINT6(x,y,z,w,a,b) embree_cout_uniform << STRING(x) << " = " << (x) << ", " << STRING(y) << " = " << (y) << ", " << STRING(z) << " = " << (z) << ", " << STRING(w) << " = " << (w) << ", " << STRING(a) << " = " << (a) << ", " << STRING(b) << " = " << (b) << embree_endl
 
 #if defined(DEBUG) // only report file and line in debug mode
   #define THROW_RUNTIME_ERROR(str) \
@@ -365,49 +355,43 @@ __forceinline std::string toString(long long value) {
 /// SYCL specific
 ////////////////////////////////////////////////////////////////////////////////
 
-#if defined(__SYCL_DEVICE_ONLY__)
-#  define EMBREE_SYCL_SIMD(N) [[intel::reqd_sub_group_size(N)]]  
-#else
-#  define EMBREE_SYCL_SIMD(N)
-#endif
 
 #if defined(EMBREE_SYCL_SUPPORT) && defined(__SYCL_DEVICE_ONLY__)
 
-  
 #define sycl_printf0(format, ...) {               \
     static const CONSTANT char fmt[] = format;               \
-    if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))       \
-      SYCL_EXT_ONEAPI::experimental::printf(fmt, __VA_ARGS__ );  \
+    if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))       \
+      sycl::ext::oneapi::experimental::printf(fmt, __VA_ARGS__ );  \
   }
 
 #define sycl_printf0_(format) {               \
     static const CONSTANT char fmt[] = format;               \
-    if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))       \
-      SYCL_EXT_ONEAPI::experimental::printf(fmt);                \
+    if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))       \
+      sycl::ext::oneapi::experimental::printf(fmt);                \
   }
 
 #else
 
 #define sycl_printf0(format, ...) {                          \
     static const CONSTANT char fmt[] = format;               \
-    SYCL_EXT_ONEAPI::experimental::printf(fmt, __VA_ARGS__ );    \
+    sycl::ext::oneapi::experimental::printf(fmt, __VA_ARGS__ );    \
   }
 
 #define sycl_printf0_(format) {                              \
     static const CONSTANT char fmt[] = format;               \
-    SYCL_EXT_ONEAPI::experimental::printf(fmt);                  \
+    sycl::ext::oneapi::experimental::printf(fmt);                  \
   }
 
 #endif
 
 #define sycl_printf(format, ...) {               \
     static const CONSTANT char fmt[] = format;               \
-    SYCL_EXT_ONEAPI::experimental::printf(fmt, __VA_ARGS__ );    \
+    sycl::ext::oneapi::experimental::printf(fmt, __VA_ARGS__ );    \
   }
 
 #define sycl_printf_(format) {               \
     static const CONSTANT char fmt[] = format;               \
-    SYCL_EXT_ONEAPI::experimental::printf(fmt);                  \
+    sycl::ext::oneapi::experimental::printf(fmt);                  \
   }
 
 #if defined(EMBREE_SYCL_SUPPORT) && defined(__SYCL_DEVICE_ONLY__)
@@ -428,7 +412,7 @@ namespace embree
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, int   i)
   {
     if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+      if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
         sycl_printf("%i",i);
     }
     else
@@ -440,31 +424,18 @@ namespace embree
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, unsigned int i)
   {
     if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+      if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
         sycl_printf("%u",i);
     } else
       sycl_printf("%u ",i);
 
     return cout;
   }
-  
-#if defined(__WIN32__) and defined(__INTEL_LLVM_COMPILER)
-  inline sycl_ostream_ operator <<(sycl_ostream_ cout, size_t i)
-  {
-    if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
-        sycl_printf("%i",i);
-    } else {
-      sycl_printf("%i ",i);
-    }
-    return cout;
-  }
-#endif
-  
+
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, float f)
   {
     if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+      if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
         sycl_printf("%f",f);
     } else
       sycl_printf("%f ",f);
@@ -475,7 +446,7 @@ namespace embree
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, double d)
   {
     if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+      if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
         sycl_printf("%f",d);
     } else
       sycl_printf("%f ",d);
@@ -483,10 +454,10 @@ namespace embree
     return cout;
   }
 
-  inline sycl_ostream_ operator <<(sycl_ostream_ cout, ulong l)
+  inline sycl_ostream_ operator <<(sycl_ostream_ cout, uint64_t l)
   {
     if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+      if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
         sycl_printf("%lu",l);
     } else
       sycl_printf("%lu ",l);
@@ -497,7 +468,7 @@ namespace embree
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, long l)
   {
     if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+      if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
         sycl_printf("%l",l);
     } else
       sycl_printf("%l ",l);
@@ -509,7 +480,7 @@ namespace embree
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, void* p)
   {
     if (cout.uniform) {
-      if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+      if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
         sycl_printf("%p",p);
     } else
       sycl_printf("%p ",p);
@@ -519,14 +490,14 @@ namespace embree
 
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, const char* c)
   {
-     if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+     if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
        sycl_printf("%s",c);
      return cout;
   }
 
   inline sycl_ostream_ operator <<(sycl_ostream_ cout, sycl_endl_)
   {
-    if (get_sub_group_local_id() == SYCL_CTZ::ctz(intel_sub_group_ballot(true)))
+    if (get_sub_group_local_id() == sycl::ctz(intel_sub_group_ballot(true)))
       sycl_printf_("\n");
     return cout;
   }

common/sys/sycl.h

@@ -20,23 +20,16 @@ using sycl::uint8;
 using sycl::uint4;
 using sycl::uint3;
 using sycl::uint2;
-using sycl::uint;
-using sycl::ulong;
 using sycl::uchar16;
 using sycl::uchar8;
 using sycl::uchar4;
 using sycl::uchar3;
 using sycl::uchar2;
-using sycl::uchar;
 using sycl::ushort16;
 using sycl::ushort8;
 using sycl::ushort4;
 using sycl::ushort3;
 using sycl::ushort2;
-using sycl::ushort;
-
-using sycl::fmax;
-using sycl::fmin;
 
 #ifdef __SYCL_DEVICE_ONLY__
 #define GLOBAL __attribute__((opencl_global))
@@ -51,7 +44,7 @@ SYCL_EXTERNAL extern float atomic_min(volatile LOCAL  float *p, float val);
 SYCL_EXTERNAL extern float atomic_max(volatile GLOBAL float *p, float val);
 SYCL_EXTERNAL extern float atomic_max(volatile LOCAL  float *p, float val);
 
-SYCL_EXTERNAL extern "C" uint intel_sub_group_ballot(bool valid);
+SYCL_EXTERNAL extern "C" unsigned int intel_sub_group_ballot(bool valid);
 
 SYCL_EXTERNAL extern "C" void __builtin_IB_assume_uniform(void *p);
 
@@ -91,32 +84,32 @@ SYCL_EXTERNAL extern "C" void __builtin_IB_assume_uniform(void *p);
 
 ///////////////////////////////////////////////////////////////////////
 
-SYCL_EXTERNAL /* extern "C" */ uint          __builtin_IB_lsc_load_global_uchar_to_uint (const GLOBAL sycl::uchar  *base, int elemOff, enum LSC_LDCC cacheOpt);     //D8U32
-SYCL_EXTERNAL /* extern "C" */ uint          __builtin_IB_lsc_load_global_ushort_to_uint(const GLOBAL sycl::ushort *base, int elemOff, enum LSC_LDCC cacheOpt);   //D16U32
-SYCL_EXTERNAL /* extern "C" */ uint          __builtin_IB_lsc_load_global_uint  (const GLOBAL sycl::uint   *base, int elemOff, enum LSC_LDCC cacheOpt);       //D32V1
-SYCL_EXTERNAL /* extern "C" */ sycl::uint2   __builtin_IB_lsc_load_global_uint2 (const GLOBAL sycl::uint2  *base, int elemOff, enum LSC_LDCC cacheOpt);     //D32V2
-SYCL_EXTERNAL /* extern "C" */ sycl::uint3   __builtin_IB_lsc_load_global_uint3 (const GLOBAL sycl::uint3  *base, int elemOff, enum LSC_LDCC cacheOpt);     //D32V3
-SYCL_EXTERNAL /* extern "C" */ sycl::uint4   __builtin_IB_lsc_load_global_uint4 (const GLOBAL sycl::uint4  *base, int elemOff, enum LSC_LDCC cacheOpt);     //D32V4
-SYCL_EXTERNAL /* extern "C" */ sycl::uint8   __builtin_IB_lsc_load_global_uint8 (const GLOBAL sycl::uint8  *base, int elemOff, enum LSC_LDCC cacheOpt);     //D32V8
-SYCL_EXTERNAL /* extern "C" */ sycl::ulong   __builtin_IB_lsc_load_global_ulong (const GLOBAL sycl::ulong  *base, int elemOff, enum LSC_LDCC cacheOpt);    //D64V1
-SYCL_EXTERNAL /* extern "C" */ sycl::ulong2  __builtin_IB_lsc_load_global_ulong2(const GLOBAL sycl::ulong2 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V2
-SYCL_EXTERNAL /* extern "C" */ sycl::ulong3  __builtin_IB_lsc_load_global_ulong3(const GLOBAL sycl::ulong3 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V3
-SYCL_EXTERNAL /* extern "C" */ sycl::ulong4  __builtin_IB_lsc_load_global_ulong4(const GLOBAL sycl::ulong4 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V4
-SYCL_EXTERNAL /* extern "C" */ sycl::ulong8  __builtin_IB_lsc_load_global_ulong8(const GLOBAL sycl::ulong8 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V8
+SYCL_EXTERNAL /* extern "C" */ uint32_t     __builtin_IB_lsc_load_global_uchar_to_uint (const GLOBAL uint8_t *base,      int elemOff, enum LSC_LDCC cacheOpt);  //D8U32
+SYCL_EXTERNAL /* extern "C" */ uint32_t     __builtin_IB_lsc_load_global_ushort_to_uint(const GLOBAL uint16_t *base,     int elemOff, enum LSC_LDCC cacheOpt);  //D16U32
+SYCL_EXTERNAL /* extern "C" */ uint32_t     __builtin_IB_lsc_load_global_uint          (const GLOBAL uint32_t *base,     int elemOff, enum LSC_LDCC cacheOpt);  //D32V1
+SYCL_EXTERNAL /* extern "C" */ sycl::uint2  __builtin_IB_lsc_load_global_uint2         (const GLOBAL sycl::uint2  *base, int elemOff, enum LSC_LDCC cacheOpt);  //D32V2
+SYCL_EXTERNAL /* extern "C" */ sycl::uint3  __builtin_IB_lsc_load_global_uint3         (const GLOBAL sycl::uint3  *base, int elemOff, enum LSC_LDCC cacheOpt);  //D32V3
+SYCL_EXTERNAL /* extern "C" */ sycl::uint4  __builtin_IB_lsc_load_global_uint4         (const GLOBAL sycl::uint4  *base, int elemOff, enum LSC_LDCC cacheOpt);  //D32V4
+SYCL_EXTERNAL /* extern "C" */ sycl::uint8  __builtin_IB_lsc_load_global_uint8         (const GLOBAL sycl::uint8  *base, int elemOff, enum LSC_LDCC cacheOpt);  //D32V8
+SYCL_EXTERNAL /* extern "C" */ uint64_t     __builtin_IB_lsc_load_global_ulong         (const GLOBAL uint64_t     *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V1
+SYCL_EXTERNAL /* extern "C" */ sycl::ulong2 __builtin_IB_lsc_load_global_ulong2        (const GLOBAL sycl::ulong2 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V2
+SYCL_EXTERNAL /* extern "C" */ sycl::ulong3 __builtin_IB_lsc_load_global_ulong3        (const GLOBAL sycl::ulong3 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V3
+SYCL_EXTERNAL /* extern "C" */ sycl::ulong4 __builtin_IB_lsc_load_global_ulong4        (const GLOBAL sycl::ulong4 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V4
+SYCL_EXTERNAL /* extern "C" */ sycl::ulong8 __builtin_IB_lsc_load_global_ulong8        (const GLOBAL sycl::ulong8 *base, int elemOff, enum LSC_LDCC cacheOpt);  //D64V8
   
 //     global address space
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uchar_from_uint (GLOBAL sycl::uchar  *base, int immElemOff, uint val, enum LSC_STCC cacheOpt);     //D8U32
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ushort_from_uint(GLOBAL sycl::ushort *base, int immElemOff, uint val, enum LSC_STCC cacheOpt);  //D16U32
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint  (GLOBAL sycl::uint   *base, int immElemOff,       uint   val, enum LSC_STCC cacheOpt);        //D32V1
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint2 (GLOBAL sycl::uint2  *base, int immElemOff, sycl::uint2  val, enum LSC_STCC cacheOpt);     //D32V2
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint3 (GLOBAL sycl::uint3  *base, int immElemOff, sycl::uint3  val, enum LSC_STCC cacheOpt);     //D32V3
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint4 (GLOBAL sycl::uint4  *base, int immElemOff, sycl::uint4  val, enum LSC_STCC cacheOpt);     //D32V4
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint8 (GLOBAL sycl::uint8  *base, int immElemOff, sycl::uint8  val, enum LSC_STCC cacheOpt);     //D32V8
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong (GLOBAL sycl::ulong  *base, int immElemOff, sycl::ulong  val, enum LSC_STCC cacheOpt);     //D64V1
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong2(GLOBAL sycl::ulong2 *base, int immElemOff, sycl::ulong2 val, enum LSC_STCC cacheOpt);  //D64V2
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong3(GLOBAL sycl::ulong3 *base, int immElemOff, sycl::ulong3 val, enum LSC_STCC cacheOpt);  //D64V3
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong4(GLOBAL sycl::ulong4 *base, int immElemOff, sycl::ulong4 val, enum LSC_STCC cacheOpt);  //D64V4
-SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong8(GLOBAL sycl::ulong8 *base, int immElemOff, sycl::ulong8 val, enum LSC_STCC cacheOpt);  //D64V8
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uchar_from_uint (GLOBAL uint8_t  *base,     int immElemOff, uint32_t val,     enum LSC_STCC cacheOpt);  //D8U32
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ushort_from_uint(GLOBAL uint16_t *base,     int immElemOff, uint32_t val,     enum LSC_STCC cacheOpt);  //D16U32
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint            (GLOBAL uint32_t *base,     int immElemOff, uint32_t val,     enum LSC_STCC cacheOpt);  //D32V1
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint2           (GLOBAL sycl::uint2  *base, int immElemOff, sycl::uint2 val,  enum LSC_STCC cacheOpt);  //D32V2
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint3           (GLOBAL sycl::uint3  *base, int immElemOff, sycl::uint3 val,  enum LSC_STCC cacheOpt);  //D32V3
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint4           (GLOBAL sycl::uint4  *base, int immElemOff, sycl::uint4 val,  enum LSC_STCC cacheOpt);  //D32V4
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_uint8           (GLOBAL sycl::uint8  *base, int immElemOff, sycl::uint8 val,  enum LSC_STCC cacheOpt);  //D32V8
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong           (GLOBAL uint64_t *base,     int immElemOff, uint64_t val,     enum LSC_STCC cacheOpt);  //D64V1
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong2          (GLOBAL sycl::ulong2 *base, int immElemOff, sycl::ulong2 val, enum LSC_STCC cacheOpt);  //D64V2
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong3          (GLOBAL sycl::ulong3 *base, int immElemOff, sycl::ulong3 val, enum LSC_STCC cacheOpt);  //D64V3
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong4          (GLOBAL sycl::ulong4 *base, int immElemOff, sycl::ulong4 val, enum LSC_STCC cacheOpt);  //D64V4
+SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong8          (GLOBAL sycl::ulong8 *base, int immElemOff, sycl::ulong8 val, enum LSC_STCC cacheOpt);  //D64V8
 
 ///////////////////////////////////////////////////////////////////////
 // prefetching
@@ -124,14 +117,14 @@ SYCL_EXTERNAL extern "C"  void  __builtin_IB_lsc_store_global_ulong8(GLOBAL sycl
 //
 // LSC Pre-Fetch Load functions with CacheControls
 //     global address space
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uchar (const GLOBAL sycl::uchar  *base, int immElemOff, enum LSC_LDCC cacheOpt); //D8U32
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_ushort(const GLOBAL sycl::ushort *base, int immElemOff, enum LSC_LDCC cacheOpt); //D16U32
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint  (const GLOBAL sycl::uint   *base, int immElemOff, enum LSC_LDCC cacheOpt); //D32V1
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint2 (const GLOBAL sycl::uint2  *base, int immElemOff, enum LSC_LDCC cacheOpt); //D32V2
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint3 (const GLOBAL sycl::uint3  *base, int immElemOff, enum LSC_LDCC cacheOpt); //D32V3
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint4 (const GLOBAL sycl::uint4  *base, int immElemOff, enum LSC_LDCC cacheOpt); //D32V4
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint8 (const GLOBAL sycl::uint8  *base, int immElemOff, enum LSC_LDCC cacheOpt); //D32V8
-SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_ulong (const GLOBAL sycl::ulong  *base, int immElemOff, enum LSC_LDCC cacheOpt); //D64V1
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uchar (const GLOBAL uint8_t *base,      int immElemOff, enum LSC_LDCC cacheOpt); //D8U32
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_ushort(const GLOBAL uint16_t *base,     int immElemOff, enum LSC_LDCC cacheOpt); //D16U32
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint  (const GLOBAL uint32_t *base,     int immElemOff, enum LSC_LDCC cacheOpt); //D32V1
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint2 (const GLOBAL sycl::uint2 *base,  int immElemOff, enum LSC_LDCC cacheOpt); //D32V2
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint3 (const GLOBAL sycl::uint3 *base,  int immElemOff, enum LSC_LDCC cacheOpt); //D32V3
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint4 (const GLOBAL sycl::uint4 *base,  int immElemOff, enum LSC_LDCC cacheOpt); //D32V4
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_uint8 (const GLOBAL sycl::uint8 *base,  int immElemOff, enum LSC_LDCC cacheOpt); //D32V8
+SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_ulong (const GLOBAL uint64_t *base,     int immElemOff, enum LSC_LDCC cacheOpt); //D64V1
 SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_ulong2(const GLOBAL sycl::ulong2 *base, int immElemOff, enum LSC_LDCC cacheOpt); //D64V2
 SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_ulong3(const GLOBAL sycl::ulong3 *base, int immElemOff, enum LSC_LDCC cacheOpt); //D64V3
 SYCL_EXTERNAL extern "C"  void __builtin_IB_lsc_prefetch_global_ulong4(const GLOBAL sycl::ulong4 *base, int immElemOff, enum LSC_LDCC cacheOpt); //D64V4
@@ -150,14 +143,14 @@ inline float work_group_reduce_max(float x) { return x; }
 inline float atomic_min(volatile float *p, float val) { return val; };
 inline float atomic_max(volatile float *p, float val) { return val; };
 
-inline uint intel_sub_group_ballot(bool valid) { return 0; }
+inline uint32_t intel_sub_group_ballot(bool valid) { return 0; }
 
 #endif
 
 /* creates a temporary that is enforced to be uniform */
 #define SYCL_UNIFORM_VAR(Ty,tmp,k)					\
   Ty tmp##_data;							\
-  Ty* p##tmp##_data = (Ty*) sub_group_broadcast((ulong)&tmp##_data,k);	\
+  Ty* p##tmp##_data = (Ty*) sub_group_broadcast((uint64_t)&tmp##_data,k);	\
   Ty& tmp = *p##tmp##_data;
 
 #if !defined(__forceinline)
@@ -188,48 +181,55 @@ namespace embree
     return sycl::select(b,a,mask);
   }
   
-  __forceinline const SYCL_EXT_ONEAPI::sub_group this_sub_group() {
-    return SYCL_EXPERIMENTAL::this_sub_group(); 
+#define XSTR(x) STR(x)
+#define STR(x) #x
+
+  __forceinline const sycl::sub_group this_sub_group() {
+#if __LIBSYCL_MAJOR_VERSION >= 8
+    return sycl::ext::oneapi::this_work_item::get_sub_group();
+#else
+    return sycl::ext::oneapi::experimental::this_sub_group();
+#endif
   }
   
-  __forceinline const uint get_sub_group_local_id() {
+  __forceinline const uint32_t get_sub_group_local_id() {
     return this_sub_group().get_local_id()[0];
   }
 
-  __forceinline const uint get_sub_group_size() {
+  __forceinline const uint32_t get_sub_group_size() {
     return this_sub_group().get_max_local_range().size();
   }
 
-  __forceinline const uint get_sub_group_id() {
+  __forceinline const uint32_t get_sub_group_id() {
     return this_sub_group().get_group_id()[0];
   }
   
-  __forceinline const uint get_num_sub_groups() {
+  __forceinline const uint32_t get_num_sub_groups() {
     return this_sub_group().get_group_range().size();
   }
   
-  __forceinline uint sub_group_ballot(bool pred) {
+  __forceinline uint32_t sub_group_ballot(bool pred) {
     return intel_sub_group_ballot(pred);
   }
 
   __forceinline bool sub_group_all_of(bool pred) {
-    return SYCL_ONEAPI::all_of_group(this_sub_group(),pred);
+    return sycl::all_of_group(this_sub_group(),pred);
   }
 
   __forceinline bool sub_group_any_of(bool pred) {
-    return SYCL_ONEAPI::any_of_group(this_sub_group(),pred);
+    return sycl::any_of_group(this_sub_group(),pred);
   }
   
   __forceinline bool sub_group_none_of(bool pred) {
-    return SYCL_ONEAPI::none_of_group(this_sub_group(),pred);
+    return sycl::none_of_group(this_sub_group(),pred);
   }
 
   template <typename T> __forceinline T sub_group_broadcast(T x, sycl::id<1> local_id) {
-    return SYCL_SUBGROUP::group_broadcast<SYCL_EXT_ONEAPI::sub_group>(this_sub_group(),x,local_id);
+    return sycl::group_broadcast<sycl::sub_group>(this_sub_group(),x,local_id);
   }
   
   template <typename T> __forceinline T sub_group_make_uniform(T x) {
-    return sub_group_broadcast(x,SYCL_CTZ::ctz(intel_sub_group_ballot(true)));
+    return sub_group_broadcast(x,sycl::ctz(intel_sub_group_ballot(true)));
   }
 
   __forceinline void assume_uniform_array(void* ptr) {
@@ -239,47 +239,47 @@ namespace embree
   }
 
   template <typename T, class BinaryOperation> __forceinline T sub_group_reduce(T x, BinaryOperation binary_op) {
-    return SYCL_SUBGROUP::reduce_over_group<SYCL_EXT_ONEAPI::sub_group>(this_sub_group(),x,binary_op);
+    return sycl::reduce_over_group<sycl::sub_group>(this_sub_group(),x,binary_op);
   }
 
   template <typename T, class BinaryOperation> __forceinline T sub_group_reduce(T x, T init, BinaryOperation binary_op) {
-    return SYCL_SUBGROUP::reduce_over_group<SYCL_EXT_ONEAPI::sub_group>(this_sub_group(),x,init,binary_op);
+    return sycl::reduce_over_group<sycl::sub_group>(this_sub_group(),x,init,binary_op);
   }
   
   template <typename T> __forceinline T sub_group_reduce_min(T x, T init) {
-    return sub_group_reduce(x, init, SYCL_EXT_ONEAPI::minimum<T>());
+    return sub_group_reduce(x, init, sycl::ext::oneapi::minimum<T>());
   }
 
   template <typename T> __forceinline T sub_group_reduce_min(T x) {
-    return sub_group_reduce(x, SYCL_EXT_ONEAPI::minimum<T>());
+    return sub_group_reduce(x, sycl::ext::oneapi::minimum<T>());
   }
 
   template <typename T> __forceinline T sub_group_reduce_max(T x) {
-    return sub_group_reduce(x, SYCL_EXT_ONEAPI::maximum<T>());
+    return sub_group_reduce(x, sycl::ext::oneapi::maximum<T>());
   }
   
   template <typename T> __forceinline T sub_group_reduce_add(T x) {
-    return sub_group_reduce(x, SYCL_EXT_ONEAPI::plus<T>());
+    return sub_group_reduce(x, sycl::ext::oneapi::plus<T>());
   }
 
   template <typename T, class BinaryOperation> __forceinline T sub_group_exclusive_scan(T x, BinaryOperation binary_op) {
-    return SYCL_SUBGROUP::exclusive_scan_over_group(this_sub_group(),x,binary_op);
+    return sycl::exclusive_scan_over_group(this_sub_group(),x,binary_op);
   }
 
   template <typename T, class BinaryOperation> __forceinline T sub_group_exclusive_scan_min(T x) {
-    return sub_group_exclusive_scan(x,SYCL_EXT_ONEAPI::minimum<T>());
+    return sub_group_exclusive_scan(x,sycl::ext::oneapi::minimum<T>());
   }
 
   template <typename T, class BinaryOperation> __forceinline T sub_group_exclusive_scan(T x, T init, BinaryOperation binary_op) {
-    return SYCL_SUBGROUP::exclusive_scan_over_group(this_sub_group(),x,init,binary_op);
+    return sycl::exclusive_scan_over_group(this_sub_group(),x,init,binary_op);
   }
 
   template <typename T, class BinaryOperation> __forceinline T sub_group_inclusive_scan(T x, BinaryOperation binary_op) {
-    return SYCL_SUBGROUP::inclusive_scan_over_group(this_sub_group(),x,binary_op);
+    return sycl::inclusive_scan_over_group(this_sub_group(),x,binary_op);
   }
 
   template <typename T, class BinaryOperation> __forceinline T sub_group_inclusive_scan(T x, BinaryOperation binary_op, T init) {
-    return SYCL_SUBGROUP::inclusive_scan_over_group(this_sub_group(),x,binary_op,init);
+    return sycl::inclusive_scan_over_group(this_sub_group(),x,binary_op,init);
   }
 
   template <typename T> __forceinline T sub_group_shuffle(T x, sycl::id<1> local_id) {

common/sys/sysinfo.cpp

@@ -344,6 +344,19 @@ namespace embree
     if (cpuid_leaf_7[EBX] & CPU_FEATURE_BIT_AVX512VL  ) cpu_features |= CPU_FEATURE_AVX512VL;
     if (cpuid_leaf_7[ECX] & CPU_FEATURE_BIT_AVX512VBMI) cpu_features |= CPU_FEATURE_AVX512VBMI;
 
+#if defined(__MACOSX__)
+    if (   (cpu_features & CPU_FEATURE_AVX512F)
+        || (cpu_features & CPU_FEATURE_AVX512DQ)
+        || (cpu_features & CPU_FEATURE_AVX512CD)
+        || (cpu_features & CPU_FEATURE_AVX512BW)
+        || (cpu_features & CPU_FEATURE_AVX512VL) )
+      {
+        // on macOS AVX512 will be enabled automatically by the kernel when the first AVX512 instruction is called
+        // see https://github.com/apple/darwin-xnu/blob/0a798f6738bc1db01281fc08ae024145e84df927/osfmk/i386/fpu.c#L176
+        // therefore we ignore the state of XCR0
+        cpu_features |= CPU_FEATURE_ZMM_ENABLED;
+      }
+#endif
     return cpu_features;
 
 #elif defined(__ARM_NEON) || defined(__EMSCRIPTEN__)

common/tasking/taskschedulerinternal.cpp

@@ -150,7 +150,8 @@ namespace embree
   {
     Lock<MutexSys> lock(g_mutex);
     assert(newNumThreads);
-    newNumThreads = min(newNumThreads, (size_t) getNumberOfLogicalThreads());
+    if (newNumThreads == std::numeric_limits<size_t>::max())
+      newNumThreads = (size_t) getNumberOfLogicalThreads();
 
     numThreads = newNumThreads;
     if (!startThreads && !running) return;
@@ -230,7 +231,8 @@ namespace embree
   TaskScheduler::TaskScheduler()
     : threadCounter(0), anyTasksRunning(0), hasRootTask(false)
   {
-    threadLocal.resize(2*getNumberOfLogicalThreads()); // FIXME: this has to be 2x as in the compatibility join mode with rtcCommitScene the worker threads also join. When disallowing rtcCommitScene to join a build we can remove the 2x.
+    assert(threadPool);
+    threadLocal.resize(2 * TaskScheduler::threadCount()); // FIXME: this has to be 2x as in the compatibility join mode with rtcCommitScene the worker threads also join. When disallowing rtcCommitScene to join a build we can remove the 2x.
     for (size_t i=0; i<threadLocal.size(); i++)
       threadLocal[i].store(nullptr);
   }

doc/CMakeLists.txt

@@ -12,6 +12,7 @@ endif()
 
 string(REGEX REPLACE "\n$" "" intelstyle_url "${intelstyle_url}")
 string(REPLACE "embree" "intelstyle" intelstyle_url "${intelstyle_url}")
+string(REPLACE "libraries" "documentation" intelstyle_url "${intelstyle_url}")
 
 message(STATUS "Fetch intelstyle repository ${intelstyle_url}")
 include(FetchContent)

doc/src/api-ref.md

@@ -184,6 +184,11 @@
 ```
 \pagebreak
 
+## RTC_GEOMETRY_TYPE_INSTANCE_ARRAY
+``` {include=src/api/RTC_GEOMETRY_TYPE_INSTANCE_ARRAY.md}
+```
+\pagebreak
+
 ## RTCCurveFlags
 ``` {include=src/api/RTCCurveFlags.md}
 ```
@@ -355,6 +360,11 @@
 ```
 \pagebreak
 
+## rtcSetGeometryInstancedScenes
+``` {include=src/api/rtcSetGeometryInstancedScenes.md}
+```
+\pagebreak
+
 ## rtcSetGeometryTransform
 ``` {include=src/api/rtcSetGeometryTransform.md}
 ```
@@ -370,6 +380,15 @@
 ```
 \pagebreak
 
+## rtcGetGeometryTransformEx
+``` {include=src/api/rtcGetGeometryTransformEx.md}
+```
+\pagebreak
+
+## rtcGetGeometryTransformFromScene
+``` {include=src/api/rtcGetGeometryTransformFromScene.md}
+```
+\pagebreak
 
 ## rtcSetGeometryTessellationRate
 ``` {include=src/api/rtcSetGeometryTessellationRate.md}

doc/src/api.md

@@ -117,7 +117,7 @@ e.g. just triangles in this example.
 
 Inside the SYCL `parallel_for` loop you can use rendering related functions,
 such as the `rtcIntersect1` and `rtcOccluded1` functions to trace rays,
-`rtcForwardIntersect1` and `rtcForwardOccluded1` to continue object
+`rtcForwardIntersect1/Ex` and `rtcForwardOccluded1/Ex` to continue object
 traversal from inside a user geometry callback, 
 and `rtcGetGeometryUserDataFromScene` to get the user data pointer of some
 geometry.
@@ -224,21 +224,6 @@ Embree SYCL Known Issues
   quality, and GPU performance may not reflect that of the final
   product.
   
-- Currently only the following Intel® Arc™ GPUs are support:
-
-  - Intel® Arc™ A770 Graphics
-  - Intel® Arc™ A750 Graphics
-  - Intel® Arc™ A770M Graphics
-  - Intel® Arc™ A730M Graphics
-  - Intel® Arc™ A550M Graphics
-
-- Intel® Data Center GPU Max Series is currently not supported.
-
-- Ahead of time compilation is currently not working properly and you
-  will get this error during compilation:
-
-      llvm-foreach: Floating point exception (core dumped)
-
 - Compilation with build configuration "debug" is currently not working on
   Windows.
 

doc/src/api/RTCBufferType.md

@@ -26,6 +26,8 @@
       RTC_BUFFER_TYPE_VERTEX_CREASE_INDEX  = 20,
       RTC_BUFFER_TYPE_VERTEX_CREASE_WEIGHT = 21,
       RTC_BUFFER_TYPE_HOLE                 = 22,
+      
+      RTC_BUFFER_TYPE_TRANSFORM            = 23,
     
       RTC_BUFFER_TYPE_FLAGS = 32
     };
@@ -60,6 +62,10 @@ The `RTC_BUFFER_TYPE_FACE`, `RTC_BUFFER_TYPE_LEVEL`,
 special buffers required to create subdivision meshes (see
 [RTC_GEOMETRY_TYPE_SUBDIVISION]).
 
+The `RTC_BUFFER_TYPE_TRANSFORM` buffer is used to provide instance
+transformation information for instance array geometries (see
+[RTC_GEOMETRY_TYPE_INSTANCE_ARRAY]).
+
 The `RTC_BUFFER_TYPE_FLAGS` can get used to add additional flag per
 primitive of a geometry, and is currently only used for linear curves.
 

doc/src/api/RTCFormat.md

@@ -39,6 +39,8 @@
       RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR,
 
       RTC_FORMAT_GRID,
+
+      RTC_FORMAT_QUATERNION_DECOMPOSITION
     };
 
 #### DESCRIPTION
@@ -58,14 +60,16 @@ of (size 2,3,4, etc.). This format is typcally used to specify to
 format of vertex buffers, e.g. the `RTC_FORMAT_FLOAT3` type for vertex
 buffers of triangle meshes.
 
-The `RTC_FORMAT_FLOAT3X4_ROW_MAJOR` and
-`RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR` formats, specify a 3x4 floating
-point matrix layed out either row major or column major. The
-`RTC_FORMAT_FLOAT4X4_ROW_MAJOR` and `RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR`
-formats, specify a 4x4 floating point matrix layed out either row
-major or column major. These matrix formats are used in the
-[rtcSetGeometryTransform] function in order to set a transformation
-matrix for geometries.
+The `RTC_FORMAT_FLOAT3X4_ROW_MAJOR` and `RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR`
+formats, specify a 3x4 floating point matrix layed out either row major or
+column major. The `RTC_FORMAT_FLOAT4X4_ROW_MAJOR` and
+`RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR` formats, specify a 4x4 floating point matrix
+layed out either row major or column major. The
+`RTC_FORMAT_QUATERNION_DECOMPOSITION` format specifies a structure that
+represents a quaternion decomposition (see [RTCQuaternionDecomposition]) of an affine
+transformation. These formats are used in the [rtcSetGeometryTransform]
+function or in geometry buffers with type `RTC_BUFFER_TYPE_TRANSFORM` in order
+to set a transformation matrix for instance and instance array geometries.
 
 The `RTC_FORMAT_GRID` is a special data format used to specify grid
 primitives of layout RTCGrid when creating grid geometries
@@ -77,3 +81,4 @@ primitives of layout RTCGrid when creating grid geometries
 
 [rtcSetGeometryBuffer], [rtcSetSharedGeometryBuffer],
 [rtcSetNewGeometryBuffer], [rtcSetGeometryTransform]
+[RTCQuaternionDecomposition]

doc/src/api/RTC_GEOMETRY_TYPE_INSTANCE_ARRAY.md

@@ -0,0 +1,79 @@
+% RTC_GEOMETRY_TYPE_INSTANCE_ARRAY(3) | Embree Ray Tracing Kernels 4
+
+#### NAME
+
+    RTC_GEOMETRY_TYPE_INSTANCE_ARRAY - instance array geometry type
+
+#### SYNOPSIS
+
+    #include <embree4/rtcore.h>
+
+    RTCGeometry geometry =
+       rtcNewGeometry(device, RTC_GEOMETRY_TYPE_INSTANCE_ARRAY);
+
+#### DESCRIPTION
+
+Embree supports instance arrays, which is a more memory efficient way to
+represent large amounts of instances of the same or a small set of (sub)scenes.
+The main difference to regular Embree instances is that Embree instance arrays
+have a buffer of transformations (either affine transformations or quaternion
+decompositions [RTCQuaternionDecomposition]) that can be allocated by
+Embree or a shared buffer, similar to vertex buffers for triangle meshes.
+Optionally, instead of instancing only one scene, an instance array can instance
+multiple scenes by passing an array of scenes and a corresponding
+index buffer that specifies which instance of the instance array instances
+which of the scenes in the scenes array.
+
+Instance arrays are created by passing `RTC_GEOMETRY_TYPE_INSTANCE_ARRAY` to
+the `rtcNewGeometry` function call. The instanced scene can be either be set
+using the `rtcSetGeometryInstancedScene` call, or if multiple scenes should be
+instanced by passing an array of scenes using `rtcSetGeometryInstancedScenes`.
+The latter also requires to specify an index buffer using
+`rtcSetNewGeometryBuffer` or `rtcSetSharedGeometryBuffer` with
+`RTC_BUFFER_TYPE_INDEX` as the buffer type.
+
+Because the transformation information can become large for a large amount of
+instances, the instance array allows to share the transformation buffer between
+the user application and Embree. It can be either stored in a buffer created by
+Embree with `rtcSetNewGeometryBuffer` or an already existing buffer can be
+shared using `rtcSetSharedGeometryBuffer`. In either case, the buffer type has
+to be `RTC_BUFFER_TYPE_TRANSFORM` and the allowed formats are
+`RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR`, `RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR`,
+`RTC_FORMAT_FLOAT3X4_ROW_MAJOR`, and `RTC_FORMAT_QUATERNION_DECOMPOSITION`.
+Embree will not modify the data in the transformation buffer.
+
+Embree instance arrays support both single-level instancing and multi-level instancing.
+The maximum instance nesting depth is `RTC_MAX_INSTANCE_LEVEL_COUNT`; it
+can be configured at compile-time using the constant `EMBREE_MAX_INSTANCE_LEVEL_COUNT`.
+Users should adapt this constant to their needs: instances nested any deeper are silently
+ignored in release mode, and cause assertions in debug mode.
+
+Please note that `rtcCommitScene` on the instanced scene(s) should be
+called first, followed by `rtcCommitGeometry` on the instance array,
+followed by `rtcCommitScene` for the top-level scene containing the
+instance array.
+
+If a ray hits the instance, the `geomID` and `primID` members of the hit are
+set to the geometry ID and primitive ID of the hit primitive in the instanced
+scene. The `instID` member of the hit is set to the geometry ID of the instance
+array in the top-level scene and the `instPrimID` member is set to the index of
+the hit instance of the instance array.
+
+For multi-segment motion blur, the number of time steps must be first specified
+using the `rtcSetGeometryTimeStepCount` function. Then a transformation for
+each time step can be specified using the `rtcSetNewGeometryBuffer` or
+`rtcSetSharedGeometryBuffer` function and passing the time step as the `slot`
+parameter of these calls.
+
+See the [Instance Array Geometry] tutorial for an example of how to use instance arrays.
+
+#### EXIT STATUS
+
+On failure `NULL` is returned and an error code is set that can be
+queried using `rtcGetDeviceError`.
+
+#### SEE ALSO
+
+[rtcNewGeometry], [rtcSetGeometryInstancedScene],
+[rtcSetGeometryInstancedScenes], [rtcSetNewGeometryBuffer],
+[rtcSetSharedGeometryBuffer], [rtcGetGeometryTransformEx]

doc/src/api/rtcForwardIntersect1.md

@@ -1,8 +1,8 @@
-% rtcForwardIntersect1(3) | Embree Ray Tracing Kernels 4
+% rtcForwardIntersect1/Ex(3) | Embree Ray Tracing Kernels 4
 
 #### NAME
 
-    rtcForwardIntersect1 - forwards a single ray to new scene
+    rtcForwardIntersect1/Ex - forwards a single ray to new scene
       from user geometry callback
 
 #### SYNOPSIS
@@ -16,16 +16,23 @@
       unsigned int instID
     );
 
+    void rtcForwardIntersect1Ex(
+      const struct RTCIntersectFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay* ray,
+      unsigned int instID,
+      unsigned int instPrimID,
+    );
+
 #### DESCRIPTION
 
-The `rtcForwardIntersect1` function forwards the traversal of a
-transformed ray (`ray` argument) into a scene (`scene` argument) from
-a user geometry callback. The function can only get invoked from a
-user geometry callback for a ray traversal initiated with the
-`rtcIntersect1` function. The callback arguments structure of the
-callback invokation has to get passed to the ray forwarding (`args`
-argument). The user geometry callback should instantly terminate after
-invoking the `rtcForwardIntersect1` function.
+The `rtcForwardIntersect1` and `rtcForwardIntersect1Ex` functions forward the
+traversal of a transformed ray (`ray` argument) into a scene (`scene` argument)
+from a user geometry callback. The function can only get invoked from a user
+geometry callback for a ray traversal initiated with the `rtcIntersect1`
+function. The callback arguments structure of the callback invokation has to
+get passed to the ray forwarding (`args` argument). The user geometry callback
+should instantly terminate after invoking the `rtcForwardIntersect1/Ex` function.
 
 Only the ray origin and ray direction members of the ray
 argument are used for forwarding, all additional ray properties are
@@ -42,10 +49,15 @@ This function can get used to implement user defined instancing using
 user geometries, e.g. by transforming the ray in a special way, and/or
 selecting between different scenes to instantiate.
 
+For user defined instance arrays, the `rtcForwardIntersect1Ex` variant has an
+additional `instPrimID` argument which is pushed to the instance primitive ID
+stack. Instance primitive IDs identify which instance of an instance array was
+hit.
+
 When using Embree on the CPU it is possible to recursively invoke
 `rtcIntersect1` directly from a user geometry callback. However, when
 SYCL is used, recursively tracing rays is not directly supported, and
-the `rtcForwardIntersect1` function must be used.
+the `rtcForwardIntersect1/Ex` functions must be used.
 
 The ray structure must be aligned to 16 bytes.
 

doc/src/api/rtcForwardIntersect4.md

@@ -1,8 +1,8 @@
-% rtcForwardIntersect4/8/16(3) | Embree Ray Tracing Kernels 4
+% rtcForwardIntersect4/8/16/Ex(3) | Embree Ray Tracing Kernels 4
 
 #### NAME
 
-    rtcForwardIntersect4/8/16 - forwards a ray packet to new scene
+    rtcForwardIntersect4/8/16/Ex - forwards a ray packet to new scene
       from user geometry callback
 
 #### SYNOPSIS
@@ -17,11 +17,11 @@
       unsigned int instID
     );
 
-    void rtcForwardIntersect4(
+    void rtcForwardIntersect8(
       void int* valid,
       const struct RTCIntersectFunctionNArguments* args,
       RTCScene scene,
-      struct RTCRay4* ray,
+      struct RTCRay8* ray,
       unsigned int instID
     );
 
@@ -30,19 +30,47 @@
       const struct RTCIntersectFunctionNArguments* args,
       RTCScene scene,
       struct RTCRay16* ray,
-      unsigned int instID
+      unsigned int instID,
+      unsigned int instPrimID
+    );
+
+    void rtcForwardIntersect4Ex(
+      void int* valid,
+      const struct RTCIntersectFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay4* ray,
+      unsigned int instID,
+      unsigned int instPrimID
+    );
+
+    void rtcForwardIntersect8Ex(
+      void int* valid,
+      const struct RTCIntersectFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay8* ray,
+      unsigned int instID,
+      unsigned int instPrimID
+    );
+
+    void rtcForwardIntersect16Ex(
+      void int* valid,
+      const struct RTCIntersectFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay16* ray,
+      unsigned int instID,
+      unsigned int instPrimID
     );
 
 #### DESCRIPTION
 
-The `rtcForwardIntersect4/8/16` functions forward the traversal of a
-transformed ray packet (`ray` argument) into a scene (`scene` argument) from
-a user geometry callback. The function can only get invoked from a
-user geometry callback for a ray traversal initiated with the
-`rtcIntersect4/8/16` function. The callback arguments structure of the
-callback invokation has to get passed to the ray forwarding (`args`
-argument). The user geometry callback should instantly terminate after
-invoking the `rtcForwardIntersect4/8/16` function.
+The `rtcForwardIntersect4/8/16` and `rtcForwardIntersect4/8/16Ex` functions
+forward the traversal of a transformed ray packet (`ray` argument) into a scene
+(`scene` argument) from a user geometry callback. The function can only get
+invoked from a user geometry callback for a ray traversal initiated with the
+`rtcIntersect4/8/16` function. The callback arguments structure of the callback
+invokation has to get passed to the ray forwarding (`args` argument). The user
+geometry callback should instantly terminate after invoking the
+`rtcForwardIntersect4/8/16/Ex` function.
 
 Only the ray origin and ray direction members of the ray
 argument are used for forwarding, all additional ray properties are
@@ -59,6 +87,11 @@ This function can get used to implement user defined instancing using
 user geometries, e.g. by transforming the ray in a special way, and/or
 selecting between different scenes to instantiate.
 
+For user defined instance arrays, the `rtcForwardIntersect4/8/16Ex` variant has an
+additional `instPrimID` argument which is pushed to the instance primitive ID
+stack. Instance primitive IDs identify which instance of an instance array was
+hit.
+
 When using Embree on the CPU it is possible to recursively invoke
 `rtcIntersect4/8/16` directly from a user geometry callback. However, when
 SYCL is used, recursively tracing rays is not directly supported, and

doc/src/api/rtcForwardOccluded1.md

@@ -1,8 +1,8 @@
-% rtcForwardOccluded1(3) | Embree Ray Tracing Kernels 4
+% rtcForwardOccluded1/Ex(3) | Embree Ray Tracing Kernels 4
 
 #### NAME
 
-    rtcForwardOccluded1 - forwards a single ray to new scene
+    rtcForwardOccluded1/Ex - forwards a single ray to new scene
       from user geometry callback
 
 #### SYNOPSIS
@@ -16,16 +16,23 @@
       unsigned int instID
     );
 
+    void rtcForwardOccluded1(
+      const struct RTCOccludedFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay* ray,
+      unsigned int instID,
+      unsigned int instPrimID
+    );
+
 #### DESCRIPTION
 
-The `rtcForwardOccluded1` function forwards the traversal of a
-transformed ray (`ray` argument) into a scene (`scene` argument) from
-a user geometry callback. The function can only get invoked from a
-user geometry callback for a ray traversal initiated with the
-`rtcOccluded1` function. The callback arguments structure of the
-callback invokation has to get passed to the ray forwarding (`args`
-argument). The user geometry callback should instantly terminate after
-invoking the `rtcForwardOccluded1` function.
+The `rtcForwardOccluded1` and `rtcForwardOccluded1Ex` functions forward the
+traversal of a transformed ray (`ray` argument) into a scene (`scene` argument)
+from a user geometry callback. The function can only get invoked from a user
+geometry callback for a ray traversal initiated with the `rtcOccluded1`
+function. The callback arguments structure of the callback invokation has to
+get passed to the ray forwarding (`args` argument). The user geometry callback
+should instantly terminate after invoking the `rtcForwardOccluded1/Ex` function.
 
 Only the ray origin and ray direction members of the ray
 argument are used for forwarding, all additional ray properties are
@@ -42,10 +49,15 @@ This function can get used to implement user defined instancing using
 user geometries, e.g. by transforming the ray in a special way, and/or
 selecting between different scenes to instantiate.
 
+For user defined instance arrays, the `rtcForwardIntersect1Ex` variant has an
+additional `instPrimID` argument which is pushed to the instance primitive ID
+stack. Instance primitive IDs identify which instance of an instance array was
+hit.
+
 When using Embree on the CPU it is possible to recursively invoke
 `rtcOccluded1` directly from a user geometry callback. However, when
 SYCL is used, recursively tracing rays is not directly supported, and
-the `rtcForwardOccluded1` function must be used.
+the `rtcForwardOccluded1/Ex` function must be used.
 
 The ray structure must be aligned to 16 bytes.
 

doc/src/api/rtcForwardOccluded4.md

@@ -1,8 +1,8 @@
-% rtcForwardOccluded4/8/16(3) | Embree Ray Tracing Kernels 4
+% rtcForwardOccluded4/8/16/Ex(3) | Embree Ray Tracing Kernels 4
 
 #### NAME
 
-    rtcForwardOccluded4/8/16 - forwards a ray packet to new scene
+    rtcForwardOccluded4/8/16/Ex - forwards a ray packet to new scene
       from user geometry callback
 
 #### SYNOPSIS
@@ -17,11 +17,11 @@
       unsigned int instID
     );
 
-    void rtcForwardOccluded4(
+    void rtcForwardOccluded8(
       void int* valid,
       const struct RTCOccludedFunctionNArguments* args,
       RTCScene scene,
-      struct RTCRay4* ray,
+      struct RTCRay8* ray,
       unsigned int instID
     );
 
@@ -33,16 +33,43 @@
       unsigned int instID
     );
 
+    void rtcForwardOccluded4Ex(
+      void int* valid,
+      const struct RTCOccludedFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay4* ray,
+      unsigned int instID,
+      unsigned int instPrimID
+    );
+
+    void rtcForwardOccluded8Ex(
+      void int* valid,
+      const struct RTCOccludedFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay8* ray,
+      unsigned int instID,
+      unsigned int instPrimID
+    );
+
+    void rtcForwardOccluded16Ex(
+      void int* valid,
+      const struct RTCOccludedFunctionNArguments* args,
+      RTCScene scene,
+      struct RTCRay16* ray,
+      unsigned int instID,
+      unsigned int instPrimID
+    );
+
 #### DESCRIPTION
 
-The `rtcForwardOccluded4/8/16` functions forward the traversal of a
-transformed ray packet (`ray` argument) into a scene (`scene` argument) from
-a user geometry callback. The function can only get invoked from a
-user geometry callback for a ray traversal initiated with the
-`rtcOccluded4/8/16` function. The callback arguments structure of the
-callback invokation has to get passed to the ray forwarding (`args`
-argument). The user geometry callback should instantly terminate after
-invoking the `rtcForwardOccluded4/8/16` function.
+The `rtcForwardOccluded4/8/16` and `rtcForwardOccluded4/8/16Ex` functions
+forward the traversal of a transformed ray packet (`ray` argument) into a scene
+(`scene` argument) from a user geometry callback. The function can only get
+invoked from a user geometry callback for a ray traversal initiated with the
+`rtcOccluded4/8/16` function. The callback arguments structure of the callback
+invokation has to get passed to the ray forwarding (`args` argument). The user
+geometry callback should instantly terminate after invoking the
+`rtcForwardOccluded4/8/16/Ex` function.
 
 Only the ray origin and ray direction members of the ray
 argument are used for forwarding, all additional ray properties are
@@ -59,6 +86,11 @@ This function can get used to implement user defined instancing using
 user geometries, e.g. by transforming the ray in a special way, and/or
 selecting between different scenes to instantiate.
 
+For user defined instance arrays, the `rtcForwardIntersect4/8/16Ex` variant has
+an additional `instPrimID` argument which is pushed to the instance primitive
+ID stack. Instance primitive IDs identify which instance of an instance array
+was hit.
+
 When using Embree on the CPU it is possible to recursively invoke
 `rtcOccluded4/8/16` directly from a user geometry callback. However, when
 SYCL is used, recursively tracing rays is not directly supported, and

doc/src/api/rtcGetDeviceProperty.md

@@ -63,6 +63,14 @@ Possible properties to query are:
     back face culling is enabled. This is only the case if Embree is
     compiled with `EMBREE_BACKFACE_CULLING` enabled.
 
++   `RTC_DEVICE_PROPERTY_BACKFACE_CULLING_CURVES_ENABLED`: Queries whether
+    back face culling for curves is enabled. This is only the case if Embree is
+    compiled with `EMBREE_BACKFACE_CULLING_CURVES` enabled.
+
++   `RTC_DEVICE_PROPERTY_BACKFACE_CULLING_SPHERES_ENABLED`: Queries whether
+    back face culling for spheres is enabled. This is only the case if Embree is
+    compiled with `EMBREE_BACKFACE_CULLING_SPHERES` enabled.
+
 +   `RTC_DEVICE_PROPERTY_COMPACT_POLYS_ENABLED`: Queries whether
     compact polys is enabled. This is only the case if Embree is
     compiled with `EMBREE_COMPACT_POLYS` enabled.

doc/src/api/rtcGetGeometryTransform.md

@@ -35,6 +35,11 @@ Possible formats for the returned matrix are:
   in column-major form as a 4×4 homogeneous matrix with last row equal
   to (0, 0, 0, 1).
 
+This function is supposed to be used during rendering, but only
+supported on the CPU and not inside SYCL kernels on the GPU. Inside a
+SYCL kernel the `rtcGetGeometryTransformFromScene` function has to get
+used.
+
 #### EXIT STATUS
 
 On failure an error code is set that can be queried using
@@ -42,4 +47,4 @@ On failure an error code is set that can be queried using
 
 #### SEE ALSO
 
-[RTC_GEOMETRY_TYPE_INSTANCE], [rtcSetGeometryTransform]
+[RTC_GEOMETRY_TYPE_INSTANCE], [rtcSetGeometryTransform], [rtcGetGeometryTransformFromScene]

doc/src/api/rtcGetGeometryTransformEx.md

@@ -0,0 +1,49 @@
+% rtcGetGeometryTransformEx(3) | Embree Ray Tracing Kernels 4
+
+#### NAME
+
+    rtcGetGeometryTransformEx - returns the interpolated instance
+    transformation for an instance of an instance array geometry for the
+    specified time.
+
+#### SYNOPSIS
+
+    #include <embree4/rtcore.h>
+
+    void rtcGetGeometryTransformEx(
+      RTCGeometry geometry,
+      unsigned int instPrimID,
+      float time,
+      enum RTCFormat format,
+      void* xfm
+    );
+
+#### DESCRIPTION
+
+The `rtcGetGeometryTransformEx` function returns the interpolated local
+to world transformation (`xfm` parameter) of the `instPrimID`-th instance of an
+instance array geometry (`geometry` parameter) for a particular time (`time`
+parameter in range $[0,1]$) in the specified format (`format` parameter).
+The function can also be used when `geometry` refers to a regular instance, but
+then the `instPrimID` has to be $0$.
+
+Possible formats for the returned matrix are:
+
++ `RTC_FORMAT_FLOAT3X4_ROW_MAJOR`: The 3×4 float matrix is laid out
+  in row-major form.
+
++ `RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR`: The 3×4 float matrix is laid out
+  in column-major form.
+
++ `RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR`: The 3×4 float matrix is laid out
+  in column-major form as a 4×4 homogeneous matrix with last row equal
+  to (0, 0, 0, 1).
+
+#### EXIT STATUS
+
+On failure an error code is set that can be queried using
+`rtcGetDeviceError`.
+
+#### SEE ALSO
+
+[RTC_GEOMETRY_TYPE_INSTANCE_ARRAY]

doc/src/api/rtcGetGeometryTransformFromScene.md

@@ -0,0 +1,51 @@
+% rtcGetGeometryTransformFromScene(3) | Embree Ray Tracing Kernels 4
+
+#### NAME
+
+    rtcGetGeometryTransformFromScene - returns the interpolated instance
+      transformation for the specified time
+
+#### SYNOPSIS
+
+    #include <embree4/rtcore.h>
+
+    void rtcGetGeometryTransformFromScene(
+      RTCScene scene,
+      unsigned int geomID,
+      float time,
+      enum RTCFormat format,
+      void* xfm
+    );
+
+#### DESCRIPTION
+
+The `rtcGetGeometryTransformFromScene` function returns the
+interpolated local to world transformation (`xfm` output parameter) of
+an instance geometry specified by its geometry ID (`geomID` parameter)
+of a scene (`scene` parameter) for a particular time (`time` parameter
+in range $[0,1]$) in the specified format (`format` parameter).
+
+Possible formats for the returned matrix are:
+
++ `RTC_FORMAT_FLOAT3X4_ROW_MAJOR`: The 3×4 float matrix is laid out
+  in row-major form.
+
++ `RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR`: The 3×4 float matrix is laid out
+  in column-major form.
+
++ `RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR`: The 3×4 float matrix is laid out
+  in column-major form as a 4×4 homogeneous matrix with last row equal
+  to (0, 0, 0, 1).
+
+In contrast to the `rtcGetGeometryTransform` function, the
+`rtcGetGeometryTransformFromScene` function can get used during
+rendering inside a SYCL kernel.
+
+#### EXIT STATUS
+
+On failure an error code is set that can be queried using
+`rtcGetDeviceError`.
+
+#### SEE ALSO
+
+[RTC_GEOMETRY_TYPE_INSTANCE], [rtcSetGeometryTransform], [rtcGetGeometryTransform]

doc/src/api/rtcGetGeometryUserData.md

@@ -18,7 +18,9 @@ previously set with `rtcSetGeometryUserData`. When
 `rtcSetGeometryUserData` was not called yet, `NULL` is returned.
 
 This function is supposed to be used during rendering, but only
-supported on the CPU and in SYCL on the GPU.
+supported on the CPU and not inside SYCL kernels on the GPU. Inside a
+SYCL kernel the `rtcGetGeometryUserDataFromScene` function has to get
+used.
 
 #### EXIT STATUS
 
@@ -27,4 +29,4 @@ On failure an error code is set that can be queried using
 
 #### SEE ALSO
 
-[rtcSetGeometryUserData]
+[rtcSetGeometryUserData], [rtcGetGeometryUserDataFromScene]

doc/src/api/rtcGetGeometryUserDataFromScene.md

@@ -18,7 +18,9 @@ pointer previously set with `rtcSetGeometryUserData` from the geometry
 with index `geomID` from the specified scene `scene`. When
 `rtcSetGeometryUserData` was not called yet, `NULL` is returned.
 
-This function is supposed to be used during rendering.
+In contrast to the `rtcGetGeometryUserData` function, the
+`rtcGetGeometryUserDataFromScene` function an get used during
+rendering inside a SYCL kernel.
 
 #### EXIT STATUS
 

doc/src/api/rtcNewGeometry.md

@@ -33,7 +33,8 @@
      RTC_GEOMETRY_TYPE_DISC_POINT,
      RTC_GEOMETRY_TYPE_ORIENTED_DISC_POINT,
      RTC_GEOMETRY_TYPE_USER,
-     RTC_GEOMETRY_TYPE_INSTANCE
+     RTC_GEOMETRY_TYPE_INSTANCE,
+     RTC_GEOMETRY_TYPE_INSTANCE_ARRAY,
     };
 
     RTCGeometry rtcNewGeometry(
@@ -64,8 +65,9 @@ bases (`RTC_GEOMETRY_TYPE_FLAT_LINEAR_CURVE`, `RTC_GEOMETRY_TYPE_FLAT_BEZIER_CUR
 grid meshes (`RTC_GEOMETRY_TYPE_GRID`), point geometries
 (`RTC_GEOMETRY_TYPE_SPHERE_POINT`, `RTC_GEOMETRY_TYPE_DISC_POINT`,
 `RTC_TYPE_ORIENTED_DISC_POINT`),
-user-defined geometries (`RTC_GEOMETRY_TYPE_USER`), and instances
-(`RTC_GEOMETRY_TYPE_INSTANCE`).
+user-defined geometries (`RTC_GEOMETRY_TYPE_USER`), instances
+(`RTC_GEOMETRY_TYPE_INSTANCE`), and instance arrays
+(`RTC_GEOMETRY_TYPE_INSTANCE_ARRAY`).
 
 The types `RTC_GEOMETRY_TYPE_ROUND_BEZIER_CURVE`,
 `RTC_GEOMETRY_TYPE_ROUND_BSPLINE_CURVE`, and
@@ -123,4 +125,4 @@ queried using `rtcGetDeviceError`.
 [RTC_GEOMETRY_TYPE_TRIANGLE], [RTC_GEOMETRY_TYPE_QUAD],
 [RTC_GEOMETRY_TYPE_SUBDIVISION], [RTC_GEOMETRY_TYPE_CURVE],
 [RTC_GEOMETRY_TYPE_GRID], [RTC_GEOMETRY_TYPE_POINT],
-[RTC_GEOMETRY_TYPE_USER], [RTC_GEOMETRY_TYPE_INSTANCE]
+[RTC_GEOMETRY_TYPE_USER], [RTC_GEOMETRY_TYPE_INSTANCE], [RTC_GEOMETRY_TYPE_INSTANCE_ARRAY]

doc/src/api/rtcSetGeometryInstancedScenes.md

@@ -0,0 +1,37 @@
+% rtcSetGeometryInstancedScenes(3) | Embree Ray Tracing Kernels 4
+
+#### NAME
+
+    rtcSetGeometryInstancedScenes - sets an array of scenes that can be
+    instanced by an instance array geometry
+
+#### SYNOPSIS
+
+    #include <embree4/rtcore.h>
+
+    void rtcSetGeometryInstancedScenes(
+      RTCGeometry geometry,
+      RTCScene* scene,
+      size_t numScenes
+    );
+
+#### DESCRIPTION
+
+The `rtcSetGeometryInstancedScenes` function sets an array of type `RTCScene`
+with `numScenes` elements that the specified instance geometry (`geometry`
+argument) can instance. This call also requires setting an index buffer using
+either `rtcSetSharedGeometryBuffer` or `rtcSetNewGeometryBuffer` (similar to
+index buffers for triangle meshes), that specifies which instance of the
+instance array instances which scene in the scene array. If only one scene
+should be instanced the call `rtcSetGeometryInstancedScene` should be
+preferred.
+
+#### EXIT STATUS
+
+On failure an error code is set that can be queried using
+`rtcGetDeviceError`.
+
+#### SEE ALSO
+
+[RTC_GEOMETRY_TYPE_INSTANCE_ARRAY], [rtcSetNewGeometryBuffer],
+[rtcSetSharedGeometryBuffer], [rtcSetGeometryInstancedScene]

doc/src/api/rtcSetGeometryIntersectFilterFunction.md

@@ -70,11 +70,11 @@ ray and the potential hit are in object space.
 The filter callback function has the task to check for each valid ray
 whether it wants to accept or reject the corresponding hit. To reject
 a hit, the filter callback function just has to write `0` to the
-integer valid mask of the corresponding ray. To accept the hit, it just
-has to leave the valid mask set to `-1`. The filter function is further
-allowed to change the hit and decrease the `tfar` value of the ray but
-it should not modify other ray data nor any inactive components of the
-ray or hit.
+integer valid mask of the corresponding ray. To accept the hit, it
+just has to leave the valid mask set to `-1`. When accepting a hit,
+the filter function is further allowed to change the hit and decrease
+the `tfar` value of the ray but it should not modify other ray data
+nor any inactive components of the ray or hit.
 
 ``` {include=src/api/inc/reorder_callback_intersect.md}
 ```

doc/src/api/rtcSetGeometryOccludedFilterFunction.md

@@ -25,7 +25,7 @@ further invocations overwrite the previously set callback function.
 Passing `NULL` as function pointer disables the registered callback
 function.
 
-The registered intersection filter function is invoked for every hit
+The registered occlusion filter function is invoked for every hit
 encountered during the `rtcOccluded`-type ray queries and can accept
 or reject that hit. The feature can be used to define a silhouette for
 a primitive and reject hits that are outside the silhouette. E.g. a
@@ -36,6 +36,14 @@ Please see the description of the
 `rtcSetGeometryIntersectFilterFunction` for a description of the
 filter callback function.
 
+The `rtcOccluded`-type functions terminate traversal when a hit got
+committed. As filter functions can only set the `tfar` distance of the
+ray for a committed hit, the occlusion filter cannot influence the
+`tfar` value of subsequent traversal, as that directly ends. For that
+reason `rtcOccluded` and occlusion filters cannot get used to gather
+the next n-hits, and `rtcIntersect` and intersection filters should
+get used instead.
+
 #### EXIT STATUS
 
 On failure an error code is set that can be queried using

doc/src/changelog.md

@@ -1,6 +1,43 @@
 Version History
 ---------------
 
+### Embree 4.3.2
+-   Embree now uses level zero raytracing extension to build BVH which enables forward-compatibility. On Linux, the package intel-level-zero-gpu-raytracing has to be installed in addition to the other packages listed here [https://dgpu-docs.intel.com/](https://dgpu-docs.intel.com/).
+-   MacOS universal binary compilation now works.
+-   Some bugfixes for AVX512 on MacOS x86 machines.
+-   Known issue: It is recommended to run Embree on Intel® Data Center GPU Max Series (e.g. Intel® Data Center GPU Max 1550) with the following environment settings: NEOReadDebugKeys=1 UseKmdMigration=0
+-   Known issue: ISPC version of tutorials will not successfully build with MacOS universal binary compilation.
+
+### Embree 4.3.1
+-   Add missing EMBREE_GEOMETRY types to embree-config.cmake
+-   User defined thread count now takes precedence for internal task scheduler
+-   Fixed static linking issue with ze_wrapper library
+-   Better error reporting for SYCL platform and driver problems in embree_info and tutorial apps.
+-   Patch to glfw source is not applied by default anymore.
+-   Known issue: Running Embree on Intel® Data Center GPU Max Series with 2 tiles (e.g. Intel® Data Center GPU Max 1550) requires setting the environment variable ZE_FLAT_DEVICE_HIERARCHY=COMPOSITE.
+-   Known issue: Embree build using Apple Clang 15 and ARM support (via the SEE2NEON library) may cause "EXEC_BAD_INSTRUCTION" runtime exceptions. Please use Apple Clang <= 14 on macOS.
+
+### Embree 4.3.0
+-   Added instance array primitive for reducing memony requirements in scenes
+    with large amounts of similar instances.
+-   Properly checks driver if L0 RTAS extension can get loaded.
+-   Added varying version of rtcGetGeometryTransform for ISPC.
+-   Fixed signature of RTCMemoryMonitorFunction for ISPC.
+-   Add support for ARM64 Windows platform in CMake.
+
+### Embree 4.2.0
+-   SYCL version of Embree with GPU support is no longer in beta phase.
+-   Improved BVH build performance on many core machines for applications that oversubscribe threads.
+-   Added rtcGetGeometryTransformFromScene API function that can get used inside SYCL kernels.
+-   No longer linking to ze_loader in SYCL mode to avoid Intel(R) oneAPI Level Zero dependency
+    for CPU rendering.
+-   Releasing test package to test Embree.
+
+### Embree 4.1.0
+-   Added support for Intel® Data Center GPU Max Series.
+-   Added ARM64 Linux support.
+-   Added EMBREE_BACKFACE_CULLING_SPHERES cmake option. The new cmake option defaults to OFF.
+
 ### Embree 4.0.1
 -   Improved performance for Tiger Lake, Comet Lake, Cannon Lake, Kaby Lake,
     and Skylake client CPUs by using 256 bit SIMD instructions by default.

doc/src/compilation.md

@@ -17,31 +17,21 @@ C++11. Embree is tested with the following compilers:
 
 Linux
 
-  - Intel® oneAPI DPC++/C++ Compiler 2023.0.0
-  - oneAPI DPC++/C++ Compiler 2022-12-14
+  - Intel® oneAPI DPC++/C++ Compiler 2024.0.2
+  - oneAPI DPC++/C++ Compiler 2023-10-26
   - Clang 5.0.0
   - Clang 4.0.0
   - GCC 10.0.1 (Fedora 32) AVX512 support
-  - GCC  8.3.1 (Fedora 28) AVX512 support
-  - GCC  7.3.1 (Fedora 27) AVX2 support
-  - GCC  7.3.1 (Fedora 26) AVX2 support
-  - GCC  6.4.1 (Fedora 25) AVX2 support
-  - Intel® Implicit SPMD Program Compiler 1.18.1
-  - Intel® Implicit SPMD Program Compiler 1.17.0
-  - Intel® Implicit SPMD Program Compiler 1.16.1
-  - Intel® Implicit SPMD Program Compiler 1.15.0
-  - Intel® Implicit SPMD Program Compiler 1.14.1
-  - Intel® Implicit SPMD Program Compiler 1.13.0
-  - Intel® Implicit SPMD Program Compiler 1.12.0
-  
-macOS x86
+  - GCC  8.3.1 (Fedora 29) AVX512 support
+  - Intel® Implicit SPMD Program Compiler 1.22.0
 
-  - Intel® C++ Classic Compiler 2023.0.0
-  - Apple Clang 12.0.5 (macOS 11.7.1)
+macOS x86_64
 
-macOS M1
+  - Apple Clang 15
 
-  - Apple Clang 12.0.5 (macOS 11.7.1)
+macOS Arm64
+
+  - Apple Clang 14
 
 Embree supports using the Intel® Threading Building Blocks (TBB) as the
 tasking system. For performance and flexibility reasons we recommend
@@ -64,7 +54,7 @@ version of [GLFW](https://www.glfw.org/) version 3.
 Under macOS, all these dependencies can be installed
 using [MacPorts](http://www.macports.org/):
 
-    sudo port install cmake tbb-devel glfw-devel
+    sudo port install cmake tbb glfw-devel
 
 Depending on your Linux distribution you can install these dependencies
 using `yum` or `apt-get`.  Some of these packages might already be
@@ -141,18 +131,21 @@ The "oneAPI DPC++ Compiler" is more up-to-date than the "Intel(R) oneAPI
 DPC++/C++ Compiler" but less stable. The current tested version of the "oneAPI
 DPC++ compiler is
 
-  - [oneAPI DPC++ Compiler 2022-12-14](https://github.com/intel/llvm/releases/download/sycl-nightly%2F20221214/dpcpp-compiler.tar.gz)
+  - [oneAPI DPC++ Compiler 2023-10-26](https://github.com/intel/llvm/releases/tag/nightly-2023-10-26)
   
 The compiler can be downloaded and simply extracted. The oneAPI DPC++ compiler
-2022-12-14 can be set up executing the following command in a Linux (bash)
-shell:
+can be set up executing the following commands in a Linux (bash) shell:
 
-    wget https://github.com/intel/llvm/releases/download/sycl-nightly%2F20221214/dpcpp-compiler.tar.gz
-    tar xzf dpcpp-compiler.tar.gz
-    source ./dpcpp_compiler/startup.sh
+    export SYCL_BUNDLE_ROOT=path_to_dpcpp_compiler
+    export PATH=$SYCL_BUNDLE_ROOT/bin:$PATH
+    export CPATH=$SYCL_BUNDLE_ROOT/include:$CPATH
+    export LIBRARY_PATH=$SYCL_BUNDLE_ROOT/lib:$LIBRARY_PATH
+    export LD_LIBRARY_PATH=$SYCL_BUNDLE_ROOT/lib:$LD_LIBRARY_PATH
+    export LD_LIBRARY_PATH=$SYCL_BUNDLE_ROOT/linux/lib/x64:$LD_LIBRARY_PATH
 
-The `startup.sh` script will put `clang++` and `clang` from the
-oneAPI DPC++ Compiler into your path.
+where the `path_to_dpcpp_compiler` should point to the unpacked oneAPI DPC++
+compiler. This will put `clang++` and `clang` from the oneAPI DPC++ Compiler
+into your path.
 
 Please also install all Linux packages described in the previous
 section.
@@ -171,12 +164,12 @@ through `EMBREE_SYCL_SUPPORT=ON`.
 
 Alternatively, you can download and run the installer of the
 
- - [Intel(R) oneAPI DPC++/C++ Compiler 2023.0.0](https://www.intel.com/content/www/us/en/developer/articles/tool/oneapi-standalone-components.html#dpcpp-cpp).
+ - [Intel(R) oneAPI DPC++/C++ Compiler](https://www.intel.com/content/www/us/en/developer/articles/tool/oneapi-standalone-components.html#dpcpp-cpp).
 
 After installation, you can set up the compiler by sourcing the
 `vars.sh` script in the `env` directory of the compiler install directory, for example,
 
-    source /opt/intel/oneAPI/compiler/2023.0.0/env/vars.sh
+    source /opt/intel/oneAPI/compiler/latest/env/vars.sh
 
 This script will put the `icpx` and `icx` compiler executables from the
 Intel(R) oneAPI DPC++/C++ Compiler in your path.
@@ -210,36 +203,23 @@ your Intel Xe HPG/HPC GPUs from here
 the driver installation instructions for your graphics card and
 operating system.
 
-We tested Embree with the latest GPGPU driver Devel Release from
-20220809. The Intel(R) Graphics Compute Runtime for oneAPI Level Zero
-and OpenCL(TM) Driver from that release is too old for Embree to work
-properly. Thus if no newer version of the GPGPU driver is available,
-you need to additionally install the latest compute runtime from here
-[22.43.24595](https://github.com/intel/compute-runtime/releases/tag/22.43.24595.30).
+After installing the drivers you have to install an additional package
+manually using
 
-Unfortunately, these compute runtime packages are only available for
-Ubuntu 22.04. You can also install a newer version of the compute
-runtime if available.
+    sudo apt install intel-level-zero-gpu-raytracing
 
 
 Windows
 -------
-        
+
 Embree is tested using the following compilers under Windows:
 
-  - Intel® oneAPI DPC++/C++ Compiler 2023.0.0
-  - oneAPI DPC++/C++ Compiler 2022-12-14
+  - Intel® oneAPI DPC++/C++ Compiler 2024.0.2
+  - oneAPI DPC++/C++ Compiler 2023-10-26
+  - Visual Studio 2022
   - Visual Studio 2019
   - Visual Studio 2017
-  - Visual Studio 2015 (Update\ 1)
-  - Intel® Implicit SPMD Program Compiler 1.18.1
-  - Intel® Implicit SPMD Program Compiler 1.17.0
-  - Intel® Implicit SPMD Program Compiler 1.16.1
-  - Intel® Implicit SPMD Program Compiler 1.15.0
-  - Intel® Implicit SPMD Program Compiler 1.14.1
-  - Intel® Implicit SPMD Program Compiler 1.13.0
-  - Intel® Implicit SPMD Program Compiler 1.12.0
-    
+  - Intel® Implicit SPMD Program Compiler 1.22.0
 
 To compile Embree for AVX-512 you have to use the Intel® Compiler.
 
@@ -356,7 +336,7 @@ The "oneAPI DPC++ Compiler" is more up-to-date than the "Intel(R) oneAPI
 DPC++/C++ Compiler" but less stable. The current tested version of the oneAPI
 DPC++ compiler is
 
-  - [oneAPI DPC++ Compiler 2022-12-14](https://github.com/intel/llvm/releases/download/sycl-nightly%2F20221214/dpcpp-compiler-win.tar.gz)
+  - [oneAPI DPC++ Compiler 2023-10-26](https://github.com/intel/llvm/releases/tag/nightly-2023-10-26)
 
 Download and unpack the archive and open the "x64 Native Tools Command Prompt"
 of Visual Studio and execute the following lines to properly configure the
@@ -395,13 +375,13 @@ option.
 
 Alternatively, you can download and run the installer of the
 
- - [Intel(R) oneAPI DPC++/C++ Compiler 2023.0.0](https://www.intel.com/content/www/us/en/developer/articles/tool/oneapi-standalone-components.html#dpcpp-cpp).
+ - [Intel(R) oneAPI DPC++/C++ Compiler](https://www.intel.com/content/www/us/en/developer/articles/tool/oneapi-standalone-components.html#dpcpp-cpp).
 
 After installation, you can either open a regular `Command Prompt` and execute
 the `vars.bat` script in the `env` directory of the compiler install directory,
 for example
 
-    C:\Program Files (x86)\Intel\oneAPI\compiler\2023.0.0\env\vars.bat
+    C:\Program Files (x86)\Intel\oneAPI\compiler\latest\env\vars.bat
 
 or simply open the installed "Intel oneAPI command prompt for Intel 64 for Visual Studio".
 
@@ -442,9 +422,9 @@ your application with such an Embree package.
 ### Windows Graphics Driver Installation
 
 In order to run the SYCL tutorials on HPG hardware, you first need to
-install the proper graphics drivers for your graphics card from
-[https://www.intel.com](https://www.intel.com). Embree will work with
-graphics driver version 101.4027 or later.
+install the graphics drivers for your graphics card from
+[https://www.intel.com](https://www.intel.com). Please make sure to
+have installed version 31.0.101.4644 or newer.
 
 
 CMake Configuration
@@ -513,6 +493,14 @@ parameters that can be configured in CMake:
   surfaces facing a ray can be hit. This option is turned OFF by
   default.
 
++ `EMBREE_BACKFACE_CULLING_CURVES`: Enables backface culling for curves,
+  i.e. only surfaces facing a ray can be hit. This option is turned OFF
+  by default.
+
++ `EMBREE_BACKFACE_CULLING_SPHERES`: Enables backface culling for spheres,
+  i.e. only surfaces facing a ray can be hit. This option is turned OFF
+  by default.
+
 + `EMBREE_COMPACT_POLYS`: Enables compact tris/quads, i.e. only
   geomIDs and primIDs are stored inside the leaf nodes.  
 
@@ -590,6 +578,9 @@ parameters that can be configured in CMake:
 + `EMBREE_GEOMETRY_INSTANCE`: Enables support for instances (ON by
   default).
 
++ `EMBREE_GEOMETRY_INSTANCE_ARRAY`: Enables support for instance arrays (ON by
+  default).
+
 + `EMBREE_GEOMETRY_USER`: Enables support for user-defined geometries
   (ON by default).
 

doc/src/downloading.md

@@ -1,9 +1,9 @@
 Downloading Embree
 ------------------
 
-For Windows we provide Embree as ZIP files linked against the Visual Studio 2015/2017 (VC14) runtime:
+For Windows we provide Embree as a ZIP file:
 
-[embree-<EMBREE_VERSION>.x64.vc14.windows.zip](https://github.com/embree/embree/releases/download/v<EMBREE_VERSION>/embree-<EMBREE_VERSION>.x64.vc14.windows.zip)  
+[embree-<EMBREE_VERSION>.x64.windows.zip](https://github.com/embree/embree/releases/download/v<EMBREE_VERSION>/embree-<EMBREE_VERSION>.x64.windows.zip)
 
 For Linux we provide Embree as a `tar.gz` file:
 

doc/src/images_local.md

@@ -6,6 +6,7 @@
 [imgUserGeometry]: images/user_geometry.jpg
 [imgViewer]: images/viewer.jpg
 [imgInstancedGeometry]: images/instanced_geometry.jpg
+[imgForest]: images/forest.jpg
 [imgMultiLevelInstancing]: images/multi_level_instancing.jpg
 [imgIntersectionFilter]: images/intersection_filter.jpg
 [imgPathtracer]: images/pathtracer.jpg

doc/src/images_url.md

@@ -6,6 +6,7 @@
 [imgUserGeometry]: https://embree.github.io/images/user_geometry.jpg
 [imgViewer]: https://embree.github.io/images/viewer.jpg
 [imgInstancedGeometry]: https://embree.github.io/images/instanced_geometry.jpg
+[imgForest]: https://embree.github.io/images/forest.jpg
 [imgMultiLevelInstancing]: https://embree.github.io/images/multi_level_instancing.jpg
 [imgIntersectionFilter]: https://embree.github.io/images/intersection_filter.jpg
 [imgPathtracer]: https://embree.github.io/images/pathtracer.jpg

doc/src/install.md

@@ -5,8 +5,8 @@ Installation of Embree
 Windows Installation
 --------------------
 
-Embree linked against Visual Studio 2015 are provided as a ZIP file
-[embree-<EMBREE_VERSION>.x64.vc14.windows.zip](https://github.com/embree/embree/releases/download/v<EMBREE_VERSION>/embree-<EMBREE_VERSION>.x64.vc14.windows.zip). After
+A pre-built version of Embree for Windows is provided as a ZIP archive
+[embree-<EMBREE_VERSION>.x64.windows.zip](https://github.com/embree/embree/releases/download/v<EMBREE_VERSION>/embree-<EMBREE_VERSION>.x64.windows.zip). After
 unpacking this ZIP file, you should set the path to the `lib` folder
 manually to your `PATH` environment variable for applications to find
 Embree.
@@ -15,7 +15,7 @@ Embree.
 Linux Installation
 ------------------
 
-The Linux version of Embree is also delivered as a `tar.gz` file:
+A pre-built version of Embree for Linux is provided as a `tar.gz` archive:
 [embree-<EMBREE_VERSION>.x86_64.linux.tar.gz](https://github.com/embree/embree/releases/download/v<EMBREE_VERSION>/embree-<EMBREE_VERSION>.x86_64.linux.tar.gz). Unpack
 this file using `tar` and source the provided `embree-vars.sh` (if you
 are using the bash shell) or `embree-vars.csh` (if you are using the C
@@ -155,3 +155,20 @@ help of the device option in ocloc tool:
 
     ocloc compile --help
 
+
+Building Embree Tests
+---------------------
+
+Embree is released with a bundle of tests in an optional testing package.
+To run these tests extract the testing package in the same folder as your embree installation.
+e.g.:
+    
+    tar -xzf embree-<EMBREE_VERSION>-testing.zip -C /path/to/installed/embree
+
+The tests are extracted into a new folder inside you embree installation and can be run with:
+
+    cd /path/to/installed/embree/testing
+    cmake -B build
+    cmake --build build target=tests
+
+

doc/src/links_local.md

@@ -7,6 +7,7 @@
 [Stream Viewer]: #stream-viewer
 [User Geometry]: #user-geometry
 [Instanced Geometry]: #instanced-geometry
+[Instance Array Geometry]: #instance-array-geometry
 [Multi Level Instancing]: #multi-level-instancing
 [Intersection Filter]: #intersection-filter
 [Hair]: #hair

doc/src/links_web.md

@@ -7,6 +7,7 @@
 [Stream Viewer]: tutorials.html#stream-viewer
 [User Geometry]: tutorials.html#user-geometry
 [Instanced Geometry]: tutorials.html#instanced-geometry
+[Instance Array Geometry]: tutorials.html#instance-array-geometry
 [Multi Level Instancing]: tutorials.html#multi-level-instancing
 [Intersection Filter]: tutorials.html#intersection-filter
 [Hair]: tutorials.html#hair

doc/src/overview.md

@@ -4,8 +4,8 @@ Intel® Embree Overview
 Intel® Embree is a high-performance ray tracing library developed at
 Intel, which is released as open source under the [Apache 2.0
 license](http://www.apache.org/licenses/LICENSE-2.0). Intel® Embree
-supports x86 CPUs under Linux, macOS, and Windows; ARM CPUs on macOS;
-as well as Intel® GPUs under Linux and Windows.
+supports x86 CPUs under Linux, macOS, and Windows; ARM CPUs on Linux
+and macOS; as well as Intel® GPUs under Linux and Windows.
 
 Intel® Embree targets graphics application developers to improve the
 performance of photo-realistic rendering applications. Embree is

doc/src/platforms.md

@@ -3,24 +3,14 @@ Supported Platforms
 
 Embree supports Windows (32-bit and 64-bit), Linux (64-bit), and macOS
 (64-bit). Under Windows, Linux and macOS x86 based CPUs are supported,
-while ARM CPUs are currently only supported under macOS (e.g. Apple
-M1). ARM support for Windows and Linux is experimental.
+while ARM CPUs are currently only supported under Linux and macOS (e.g. 
+Apple M1). ARM support for Windows experimental.
 
 Embree supports Intel GPUs based on the Xe HPG microarchitecture
 (Intel® Arc™ GPU) under Linux and Windows and Xe HPC microarchitecture
 (Intel® Data Center GPU Flex Series and Intel® Data Center GPU Max
 Series) under Linux.
 
-Currently the following products are supported and further products
-will get enabled soon:
-
-- Intel® Arc™ A770 Graphics
-- Intel® Arc™ A750 Graphics
-- Intel® Arc™ A770M Graphics
-- Intel® Arc™ A730M Graphics
-- Intel® Arc™ A550M Graphics
-- Intel® Data Center GPU Flex 170
-
 The code compiles with the Intel® Compiler, Intel® oneAPI DPC++
 Compiler, GCC, Clang, and the Microsoft Compiler. To use Embree on the
 GPU the Intel® oneAPI DPC++ Compiler must be used. Please see section

doc/src/tutorials.md

@@ -199,6 +199,25 @@ ways can be distinguished.
 
 [Source Code](https://github.com/embree/embree/blob/master/tutorials/instanced_geometry/instanced_geometry_device.cpp)
 
+Instance Array Geometry
+----------------------
+
+[![][imgForest]](https://github.com/embree/embree/blob/master/tutorials/forest/forest_device.cpp)
+
+This tutorial demonstrates the usage of instance arrays in Embree. Instance arrays
+are large collections of similar objects. Examples are sand dunes that consist
+of millions of instances of a few grain models or, like here, a forest consisting of
+many instances of a few tree models.
+
+In this application can switch between representing the scene with regular
+instances or (one!) instance array. It also prints several stats, that
+demonstrate the memory savings and faster BVH build times when using instance
+arrays for such scenes. Instance arrays come with a small overhead on CPU and
+should be preferred if memory consumption is more important than raytracing
+performance.
+
+[Source Code](https://github.com/embree/embree/blob/master/tutorials/forest/forest_device.cpp)
+
 Multi Level Instancing
 ----------------------
 

include/embree4/rtcore_buffer.h

@@ -27,6 +27,8 @@ enum RTCBufferType
   RTC_BUFFER_TYPE_VERTEX_CREASE_WEIGHT = 21,
   RTC_BUFFER_TYPE_HOLE                 = 22,
 
+  RTC_BUFFER_TYPE_TRANSFORM            = 23,
+
   RTC_BUFFER_TYPE_FLAGS = 32
 };
 

include/embree4/rtcore_buffer.isph

@@ -26,6 +26,8 @@ enum RTCBufferType
   RTC_BUFFER_TYPE_VERTEX_CREASE_WEIGHT = 21,
   RTC_BUFFER_TYPE_HOLE                 = 22,
 
+  RTC_BUFFER_TYPE_TRANSFORM            = 23,
+
   RTC_BUFFER_TYPE_FLAGS = 32
 };
 

include/embree4/rtcore_common.h

@@ -147,7 +147,9 @@ enum RTCFormat
   RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR = 0x9244,
 
   /* special 12-byte format for grids */
-  RTC_FORMAT_GRID = 0xA001
+  RTC_FORMAT_GRID = 0xA001,
+
+  RTC_FORMAT_QUATERNION_DECOMPOSITION = 0xB001,
 };
 
 /* Build quality levels */
@@ -290,6 +292,8 @@ enum RTCFeatureFlags
 
   RTC_FEATURE_FLAG_32_BIT_RAY_MASK = 1 << 28,
 
+  RTC_FEATURE_FLAG_INSTANCE_ARRAY = 1 << 29,
+
   RTC_FEATURE_FLAG_ALL = 0xffffffff,
 };
 
@@ -334,19 +338,26 @@ struct RTCRayQueryContext
   unsigned int instStackSize;                        // Number of instances currently on the stack.
 #endif
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // The current stack of instance ids.
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // The current stack of instance primitive ids.
+#endif
 };
 
 /* Initializes an ray query context. */
 RTC_FORCEINLINE void rtcInitRayQueryContext(struct RTCRayQueryContext* context)
 {
   unsigned l = 0;
-  
+
 #if RTC_MAX_INSTANCE_LEVEL_COUNT > 1
   context->instStackSize = 0;
 #endif
 
-  for (; l < RTC_MAX_INSTANCE_LEVEL_COUNT; ++l)
+  for (; l < RTC_MAX_INSTANCE_LEVEL_COUNT; ++l) {
     context->instID[l] = RTC_INVALID_GEOMETRY_ID;
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    context->instPrimID[l] = RTC_INVALID_GEOMETRY_ID;
+#endif
+  }
 }
 
 /* Point query structure for closest point query */
@@ -404,6 +415,11 @@ struct RTC_ALIGN(16) RTCPointQueryContext
   // instance ids.
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT];
 
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  // instance prim ids.
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT];
+#endif
+
   // number of instances currently on the stack.
   unsigned int instStackSize;
 };
@@ -411,8 +427,16 @@ struct RTC_ALIGN(16) RTCPointQueryContext
 /* Initializes an ray query context. */
 RTC_FORCEINLINE void rtcInitPointQueryContext(struct RTCPointQueryContext* context)
 {
+  unsigned l = 0;
+
   context->instStackSize = 0;
-  context->instID[0] = RTC_INVALID_GEOMETRY_ID;
+
+  for (; l < RTC_MAX_INSTANCE_LEVEL_COUNT; ++l) {
+    context->instID[l] = RTC_INVALID_GEOMETRY_ID;
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    context->instPrimID[l] = RTC_INVALID_GEOMETRY_ID;
+#endif
+  }
 }
 
 struct RTC_ALIGN(16) RTCPointQueryFunctionArguments

include/embree4/rtcore_common.isph

@@ -265,6 +265,8 @@ enum RTCFeatureFlags
 
   RTC_FEATURE_FLAG_32_BIT_RAY_MASK = 1 << 28,
 
+  RTC_FEATURE_FLAG_INSTANCE_ARRAY = 1 << 29,
+
   RTC_FEATURE_FLAG_ALL = 0xffffffff,
 };
 
@@ -287,6 +289,10 @@ struct RTCRayQueryContext
   unsigned int instStackSize;                        // Number of instances currently on the stack.
 #endif
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // The current stack of instance ids.
+
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // The current stack of instance primitive ids.
+#endif
 };
 
 /* Initializes an ray query context. */
@@ -298,8 +304,12 @@ RTC_FORCEINLINE void rtcInitRayQueryContext(uniform RTCRayQueryContext* uniform
   context->instStackSize = 0;
 #endif
   
-  for (; l < RTC_MAX_INSTANCE_LEVEL_COUNT; ++l)
+  for (; l < RTC_MAX_INSTANCE_LEVEL_COUNT; ++l) {
     context->instID[l] = RTC_INVALID_GEOMETRY_ID;
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    context->instPrimID[l] = RTC_INVALID_GEOMETRY_ID;
+#endif
+  }
 }
 
 /* Arguments for RTCFilterFunctionN */
@@ -362,17 +372,22 @@ struct RTCPointQueryContext
 {
   // accumulated 4x4 column major matrices from world space to instance space.
   // undefined if size == 0.
-  float world2inst[RTC_MAX_INSTANCE_LEVEL_COUNT][16]; 
+  float world2inst[RTC_MAX_INSTANCE_LEVEL_COUNT][16];
 
   // accumulated 4x4 column major matrices from instance space to world space.
   // undefined if size == 0.
-  float inst2world[RTC_MAX_INSTANCE_LEVEL_COUNT][16]; 
+  float inst2world[RTC_MAX_INSTANCE_LEVEL_COUNT][16];
 
   // instance ids.
-  unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT]; 
+  unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT];
+
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  // instance prim ids.
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT];
+#endif
 
   // number of instances currently on the stack.
-  unsigned int instStackSize;                                 
+  unsigned int instStackSize;
 };
 
 /* Initializes an ray query context. */
@@ -380,6 +395,9 @@ RTC_FORCEINLINE void rtcInitPointQueryContext(uniform RTCPointQueryContext* unif
 {
   context->instStackSize = 0;
   context->instID[0] = RTC_INVALID_GEOMETRY_ID;
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  context->instPrimID[0] = RTC_INVALID_GEOMETRY_ID;
+#endif
 }
 
 struct RTCPointQueryFunctionArguments

include/embree4/rtcore_device.h

@@ -49,6 +49,7 @@ enum RTCDeviceProperty
   RTC_DEVICE_PROPERTY_NATIVE_RAY8_SUPPORTED  = 33,
   RTC_DEVICE_PROPERTY_NATIVE_RAY16_SUPPORTED = 34,
 
+  RTC_DEVICE_PROPERTY_BACKFACE_CULLING_SPHERES_ENABLED = 62,
   RTC_DEVICE_PROPERTY_BACKFACE_CULLING_CURVES_ENABLED = 63,
   RTC_DEVICE_PROPERTY_RAY_MASK_SUPPORTED          = 64,
   RTC_DEVICE_PROPERTY_BACKFACE_CULLING_ENABLED    = 65,
@@ -65,7 +66,10 @@ enum RTCDeviceProperty
 
   RTC_DEVICE_PROPERTY_TASKING_SYSTEM        = 128,
   RTC_DEVICE_PROPERTY_JOIN_COMMIT_SUPPORTED = 129,
-  RTC_DEVICE_PROPERTY_PARALLEL_COMMIT_SUPPORTED = 130
+  RTC_DEVICE_PROPERTY_PARALLEL_COMMIT_SUPPORTED = 130,
+
+  RTC_DEVICE_PROPERTY_CPU_DEVICE  = 140,
+  RTC_DEVICE_PROPERTY_SYCL_DEVICE = 141
 };
 
 /* Gets a device property. */

include/embree4/rtcore_device.isph

@@ -76,7 +76,7 @@ typedef unmasked void (*uniform RTCErrorFunction)(void* uniform userPtr, uniform
 RTC_API void rtcSetDeviceErrorFunction(RTCDevice device, uniform RTCErrorFunction error, void* uniform userPtr);
 
 /* Memory monitor callback function */
-typedef uniform bool (*uniform RTCMemoryMonitorFunction)(uniform intptr_t bytes, uniform bool post);
+typedef unmasked uniform bool (*uniform RTCMemoryMonitorFunction)(void* uniform ptr, uniform intptr_t bytes, uniform bool post);
 
 /* Sets the memory monitor callback function. */
 RTC_API void rtcSetDeviceMemoryMonitorFunction(RTCDevice device, RTCMemoryMonitorFunction memoryMonitor, void* uniform userPtr);

include/embree4/rtcore_geometry.h

@@ -48,7 +48,8 @@ enum RTCGeometryType
   RTC_GEOMETRY_TYPE_NORMAL_ORIENTED_CATMULL_ROM_CURVE  = 60, // flat normal-oriented Catmull-Rom curves
 
   RTC_GEOMETRY_TYPE_USER     = 120, // user-defined geometry
-  RTC_GEOMETRY_TYPE_INSTANCE = 121  // scene instance
+  RTC_GEOMETRY_TYPE_INSTANCE = 121,  // scene instance
+  RTC_GEOMETRY_TYPE_INSTANCE_ARRAY = 122,  // scene instance array
 };
 
 /* Interpolation modes for subdivision surfaces */
@@ -219,6 +220,9 @@ RTC_SYCL_API void rtcInvokeOccludedFilterFromGeometry(const struct RTCOccludedFu
 /* Sets the instanced scene of an instance geometry. */
 RTC_API void rtcSetGeometryInstancedScene(RTCGeometry geometry, RTCScene scene);
 
+/* Sets the instanced scenes of an instance array geometry. */
+RTC_API void rtcSetGeometryInstancedScenes(RTCGeometry geometry, RTCScene* scenes, size_t numScenes);
+
 /* Sets the transformation of an instance for the specified time step. */
 RTC_API void rtcSetGeometryTransform(RTCGeometry geometry, unsigned int timeStep, enum RTCFormat format, const void* xfm);
 
@@ -228,6 +232,12 @@ RTC_API void rtcSetGeometryTransformQuaternion(RTCGeometry geometry, unsigned in
 /* Returns the interpolated transformation of an instance for the specified time. */
 RTC_API void rtcGetGeometryTransform(RTCGeometry geometry, float time, enum RTCFormat format, void* xfm);
 
+/*
+ * Returns the interpolated transformation of the instPrimID'th instance of an
+ * instance array for the specified time. If geometry is an regular instance,
+ * instPrimID must be 0.
+ */
+RTC_API void rtcGetGeometryTransformEx(RTCGeometry geometry, unsigned int instPrimID, float time, enum RTCFormat format, void* xfm);
 
 /* Sets the uniform tessellation rate of the geometry. */
 RTC_API void rtcSetGeometryTessellationRate(RTCGeometry geometry, float tessellationRate);

include/embree4/rtcore_geometry.isph

@@ -47,7 +47,8 @@ enum RTCGeometryType
   RTC_GEOMETRY_TYPE_NORMAL_ORIENTED_CATMULL_ROM_CURVE  = 60, // flat normal-oriented Catmull-Rom curves  
 
   RTC_GEOMETRY_TYPE_USER     = 120, // user-defined geometry
-  RTC_GEOMETRY_TYPE_INSTANCE = 121  // scene instance
+  RTC_GEOMETRY_TYPE_INSTANCE = 121,  // scene instance
+  RTC_GEOMETRY_TYPE_INSTANCE_ARRAY = 122  // scene instance
 };
 
 /* Interpolation modes for subdivision surfaces */
@@ -225,6 +226,9 @@ RTC_API void rtcInvokeOccludedFilterFromGeometry(const uniform struct RTCOcclude
 /* Sets the instanced scene of an instance geometry. */
 RTC_API void rtcSetGeometryInstancedScene(RTCGeometry geometry, RTCScene scene);
 
+/* Sets the instanced scene of an instance array geometry. */
+RTC_API void rtcSetGeometryInstancedScenes(RTCGeometry geometry, uniform RTCScene* uniform scenes, uniform size_t numScenes);
+
 /* Sets the transformation of an instance for the specified time step. */
 RTC_API void rtcSetGeometryTransform(RTCGeometry geometry, uniform unsigned int timeStep, uniform RTCFormat format, const void* uniform xfm);
 
@@ -234,6 +238,53 @@ RTC_API void rtcSetGeometryTransformQuaternion(RTCGeometry geometry, uniform uns
 /* Returns the interpolated transformation of an instance for the specified time. */
 RTC_API void rtcGetGeometryTransform(RTCGeometry geometry, uniform float time, uniform RTCFormat format, void* uniform xfm);
 
+/* Returns the interpolated transformation of an instance for the specified time. Varying version. */
+inline void rtcGetGeometryTransform(RTCGeometry geometry, varying float time, uniform RTCFormat format, void* uniform xfm)
+{
+  varying float vmatrix[12];
+  foreach_unique(utime in time)
+  {
+    uniform float umatrix[12];
+    rtcGetGeometryTransform(geometry,utime,RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR,umatrix);
+    for (int uniform j=0; j<12; j++) vmatrix[j] = umatrix[j];
+  }
+
+  /* store to desired layout */
+  varying float* vxfm = (varying float*) xfm;
+  
+  switch (format)
+  {
+  case RTC_FORMAT_FLOAT3X4_ROW_MAJOR:
+    vxfm[ 0] = vmatrix[0];  vxfm[ 1] = vmatrix[3];  vxfm[ 2] = vmatrix[6];  vxfm[ 3] = vmatrix[9];
+    vxfm[ 4] = vmatrix[1];  vxfm[ 5] = vmatrix[4];  vxfm[ 6] = vmatrix[7];  vxfm[ 7] = vmatrix[10];
+    vxfm[ 8] = vmatrix[2];  vxfm[ 9] = vmatrix[5];  vxfm[10] = vmatrix[8];  vxfm[11] = vmatrix[11];
+    break;
+    
+  case RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR:
+    vxfm[ 0] = vmatrix[0];  vxfm[ 1] = vmatrix[1];  vxfm[ 2] = vmatrix[2];
+    vxfm[ 3] = vmatrix[3];  vxfm[ 4] = vmatrix[4];  vxfm[ 5] = vmatrix[5];
+    vxfm[ 6] = vmatrix[6];  vxfm[ 7] = vmatrix[7];  vxfm[ 8] = vmatrix[8];
+    vxfm[ 9] = vmatrix[9];  vxfm[10] = vmatrix[10]; vxfm[11] = vmatrix[11];
+    break;
+    
+  case RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR:
+    vxfm[ 0] = vmatrix[0];  vxfm[ 1] = vmatrix[1];  vxfm[ 2] = vmatrix[2];  vxfm[ 3] = 0.f;
+    vxfm[ 4] = vmatrix[3];  vxfm[ 5] = vmatrix[4];  vxfm[ 6] = vmatrix[5];  vxfm[ 7] = 0.f;
+    vxfm[ 8] = vmatrix[6];  vxfm[ 9] = vmatrix[7];  vxfm[10] = vmatrix[8];  vxfm[11] = 0.f;
+    vxfm[12] = vmatrix[9];  vxfm[13] = vmatrix[10]; vxfm[14] = vmatrix[11]; vxfm[15] = 1.f;
+    break;
+    
+  default:
+    break;
+  }
+}
+
+/*
+ * Returns the interpolated transformation of the instPrimID'th instance of an
+ * instance array for the specified time. If geometry is an regular instance,
+ * instPrimID must be 0.
+ */
+RTC_API void rtcGetGeometryTransformEx(RTCGeometry geometry, uniform unsigned int instPrimID, uniform float time, uniform RTCFormat format, void* uniform xfm);
 
 /* Sets the uniform tessellation rate of the geometry. */
 RTC_API void rtcSetGeometryTessellationRate(RTCGeometry geometry, uniform float tessellationRate);

include/embree4/rtcore_ray.h

@@ -39,6 +39,9 @@ struct RTC_ALIGN(16) RTCHit
   unsigned int primID; // primitive ID
   unsigned int geomID; // geometry ID
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // instance ID
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // instance primitive ID
+#endif
 };
 
 /* Combined ray/hit structure for a single ray */
@@ -80,6 +83,9 @@ struct RTC_ALIGN(16) RTCHit4
   unsigned int primID[4];
   unsigned int geomID[4];
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT][4];
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT][4];
+#endif
 };
 
 /* Combined ray/hit structure for a packet of 4 rays */
@@ -121,6 +127,9 @@ struct RTC_ALIGN(32) RTCHit8
   unsigned int primID[8];
   unsigned int geomID[8];
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT][8];
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT][8];
+#endif
 };
 
 /* Combined ray/hit structure for a packet of 8 rays */
@@ -162,6 +171,9 @@ struct RTC_ALIGN(64) RTCHit16
   unsigned int primID[16];
   unsigned int geomID[16];
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT][16];
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT][16];
+#endif
 };
 
 /* Combined ray/hit structure for a packet of 16 rays */
@@ -201,17 +213,20 @@ RTC_FORCEINLINE float& RTCHitN_Ng_z(RTCHitN* hit, unsigned int N, unsigned int i
 RTC_FORCEINLINE float& RTCHitN_u(RTCHitN* hit, unsigned int N, unsigned int i) { return ((float*)hit)[3*N+i]; }
 RTC_FORCEINLINE float& RTCHitN_v(RTCHitN* hit, unsigned int N, unsigned int i) { return ((float*)hit)[4*N+i]; }
 
-RTC_FORCEINLINE unsigned int& RTCHitN_primID(RTCHitN* hit, unsigned int N, unsigned int i) { return ((unsigned*)hit)[5*N+i]; }
-RTC_FORCEINLINE unsigned int& RTCHitN_geomID(RTCHitN* hit, unsigned int N, unsigned int i) { return ((unsigned*)hit)[6*N+i]; }
-RTC_FORCEINLINE unsigned int& RTCHitN_instID(RTCHitN* hit, unsigned int N, unsigned int i, unsigned int l) { return ((unsigned*)hit)[7*N+i+N*l]; }
+RTC_FORCEINLINE unsigned int& RTCHitN_primID    (RTCHitN* hit, unsigned int N, unsigned int i) { return ((unsigned*)hit)[5*N+i]; }
+RTC_FORCEINLINE unsigned int& RTCHitN_geomID    (RTCHitN* hit, unsigned int N, unsigned int i) { return ((unsigned*)hit)[6*N+i]; }
+RTC_FORCEINLINE unsigned int& RTCHitN_instID    (RTCHitN* hit, unsigned int N, unsigned int i, unsigned int l) { return ((unsigned*)hit)[7*N +                                  N*l + i]; }
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+RTC_FORCEINLINE unsigned int& RTCHitN_instPrimID(RTCHitN* hit, unsigned int N, unsigned int i, unsigned int l) { return ((unsigned*)hit)[7*N + N*RTC_MAX_INSTANCE_LEVEL_COUNT + N*l + i]; }
+#endif
 
 /* Helper functions to extract RTCRayN and RTCHitN from RTCRayHitN */
 RTC_FORCEINLINE RTCRayN* RTCRayHitN_RayN(RTCRayHitN* rayhit, unsigned int N) { return (RTCRayN*)&((float*)rayhit)[0*N]; }
 RTC_FORCEINLINE RTCHitN* RTCRayHitN_HitN(RTCRayHitN* rayhit, unsigned int N) { return (RTCHitN*)&((float*)rayhit)[12*N]; }
 
 /* Helper structure for a ray packet of compile-time size N */
-template<int N>
-struct RTCRayNt
+template<unsigned int N>
+struct RTC_ALIGN((N && !(N & (N - 1)) ? (N * 4 > 16 ? N * 4 : 16) : 16)) RTCRayNt
 {
   float org_x[N];
   float org_y[N];
@@ -230,8 +245,8 @@ struct RTCRayNt
 };
 
 /* Helper structure for a hit packet of compile-time size N */
-template<int N>
-struct RTCHitNt
+template<unsigned int N>
+struct RTC_ALIGN((N && !(N & (N - 1)) ? (N * 4 > 16 ? N * 4 : 16) : 16)) RTCHitNt
 {
   float Ng_x[N];
   float Ng_y[N];
@@ -243,6 +258,9 @@ struct RTCHitNt
   unsigned int primID[N];
   unsigned int geomID[N];
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT][N];
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT][N];
+#endif
 };
 
 /* Helper structure for a combined ray/hit packet of compile-time size N */
@@ -281,8 +299,12 @@ RTC_FORCEINLINE RTCHit rtcGetHitFromHitN(RTCHitN* hitN, unsigned int N, unsigned
   hit.v      = RTCHitN_v(hitN,N,i);
   hit.primID = RTCHitN_primID(hitN,N,i);
   hit.geomID = RTCHitN_geomID(hitN,N,i);
-  for (unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++)
+  for (unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++) {
     hit.instID[l] = RTCHitN_instID(hitN,N,i,l);
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    hit.instPrimID[l] = RTCHitN_instPrimID(hitN,N,i,l);
+#endif
+  }
   return hit;
 }
 
@@ -295,8 +317,12 @@ RTC_FORCEINLINE void rtcCopyHitToHitN(RTCHitN* hitN, const RTCHit* hit, unsigned
   RTCHitN_v(hitN,N,i)      = hit->v;
   RTCHitN_primID(hitN,N,i) = hit->primID;
   RTCHitN_geomID(hitN,N,i) = hit->geomID;
-  for (unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++)
+  for (unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++) {
     RTCHitN_instID(hitN,N,i,l) = hit->instID[l];
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    RTCHitN_instPrimID(hitN,N,i,l) = hit->instPrimID[l];
+#endif
+  }
 }
 
 RTC_FORCEINLINE RTCRayHit rtcGetRayHitFromRayHitN(RTCRayHitN* rayhitN, unsigned int N, unsigned int i)
@@ -325,8 +351,12 @@ RTC_FORCEINLINE RTCRayHit rtcGetRayHitFromRayHitN(RTCRayHitN* rayhitN, unsigned
   rh.hit.v      = RTCHitN_v(hit,N,i);
   rh.hit.primID = RTCHitN_primID(hit,N,i);
   rh.hit.geomID = RTCHitN_geomID(hit,N,i);
-  for (unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++)
+  for (unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++) {
     rh.hit.instID[l] = RTCHitN_instID(hit,N,i,l);
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    rh.hit.instPrimID[l] = RTCHitN_instPrimID(hit,N,i,l);
+#endif
+  }
 
   return rh;
 }

include/embree4/rtcore_ray.isph

@@ -38,6 +38,9 @@ struct RTC_ALIGN(16) RTCHit
   unsigned int primID; // primitive ID
   unsigned int geomID; // geometry ID
   unsigned int instID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // instance ID
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+  unsigned int instPrimID[RTC_MAX_INSTANCE_LEVEL_COUNT]; // instance prim ID
+#endif
 };
 
 /* Combined ray/hit structure */
@@ -76,6 +79,9 @@ RTC_FORCEINLINE varying float& RTCHitN_v   (const RTCHitN* uniform hit, uniform
 RTC_FORCEINLINE varying unsigned int& RTCHitN_primID(const RTCHitN* uniform hit, uniform unsigned int N, uniform unsigned int i) { return *((varying unsigned int* uniform) &((unsigned int* uniform  )hit)[5*N+i]); }
 RTC_FORCEINLINE varying unsigned int& RTCHitN_geomID(const RTCHitN* uniform hit, uniform unsigned int N, uniform unsigned int i) { return *((varying unsigned int* uniform) &((unsigned int* uniform  )hit)[6*N+i]); }
 RTC_FORCEINLINE varying unsigned int& RTCHitN_instID(const RTCHitN* uniform hit, uniform unsigned int N, uniform unsigned int i, uniform unsigned int l) { return *((varying unsigned int* uniform) &((unsigned int* uniform  )hit)[7*N+i+l*N]); }
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+RTC_FORCEINLINE varying unsigned int& RTCHitN_instPrimID(const RTCHitN* uniform hit, uniform unsigned int N, uniform unsigned int i, uniform unsigned int l) { return *((varying unsigned int* uniform) &((unsigned int* uniform  )hit)[7*N + N*RTC_MAX_INSTANCE_LEVEL_COUNT + N*l + i]); }
+#endif
 
 /* Helper functions to extract RTCRayN and RTCHitN from RTCRayHitN */
 RTC_FORCEINLINE RTCRayN* uniform RTCRayHitN_RayN(RTCRayHitN* uniform rayhit, uniform unsigned int N) { return (RTCRayN* uniform)&((uniform float* uniform)rayhit)[0*N]; }
@@ -109,8 +115,12 @@ RTC_FORCEINLINE RTCHit rtcGetHitFromHitN(RTCHitN* uniform hitN, uniform unsigned
   hit.v      = RTCHitN_v(hitN,N,i);
   hit.primID = RTCHitN_primID(hitN,N,i);
   hit.geomID = RTCHitN_geomID(hitN,N,i);
-  for (uniform unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++)
+  for (uniform unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++) {
     hit.instID[l] = RTCHitN_instID(hitN,N,i,l);
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    hit.instPrimID[l] = RTCHitN_instPrimID(hitN,N,i,l);
+#endif
+  }
   return hit;
 }
 
@@ -123,8 +133,12 @@ RTC_FORCEINLINE void rtcCopyHitToHitN(RTCHitN* uniform hitN, const varying RTCHi
   RTCHitN_v(hitN,N,i)      = hit->v;
   RTCHitN_primID(hitN,N,i) = hit->primID;
   RTCHitN_geomID(hitN,N,i) = hit->geomID;
-  for (uniform unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++)
+  for (uniform unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++) {
     RTCHitN_instID(hitN,N,i,l) = hit->instID[l];
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    RTCHitN_instPrimID(hitN,N,i,l) = hit->instPrimID[l];
+#endif
+  }
 }
 
 RTC_FORCEINLINE RTCRayHit rtcGetRayHitFromRayHitN(RTCRayHitN* uniform rayhitN, uniform unsigned int N, uniform unsigned int i)
@@ -153,8 +167,12 @@ RTC_FORCEINLINE RTCRayHit rtcGetRayHitFromRayHitN(RTCRayHitN* uniform rayhitN, u
   rh.hit.v      = RTCHitN_v(hit,N,i);
   rh.hit.primID = RTCHitN_primID(hit,N,i);
   rh.hit.geomID = RTCHitN_geomID(hit,N,i);
-  for (uniform unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++)
+  for (uniform unsigned int l = 0; l < RTC_MAX_INSTANCE_LEVEL_COUNT; l++) {
     rh.hit.instID[l] = RTCHitN_instID(hit,N,i,l);
+#if defined(RTC_GEOMETRY_INSTANCE_ARRAY)
+    rh.hit.instPrimID[l] = RTCHitN_instPrimID(hit,N,i,l);
+#endif
+  }
 
   return rh;
 }

include/embree4/rtcore_scene.h

@@ -16,11 +16,12 @@ struct RTCRayHit16;
 /* Scene flags */
 enum RTCSceneFlags
 {
-  RTC_SCENE_FLAG_NONE                    = 0,
-  RTC_SCENE_FLAG_DYNAMIC                 = (1 << 0),
-  RTC_SCENE_FLAG_COMPACT                 = (1 << 1),
-  RTC_SCENE_FLAG_ROBUST                  = (1 << 2),
-  RTC_SCENE_FLAG_FILTER_FUNCTION_IN_ARGUMENTS = (1 << 3)
+  RTC_SCENE_FLAG_NONE                         = 0,
+  RTC_SCENE_FLAG_DYNAMIC                      = (1 << 0),
+  RTC_SCENE_FLAG_COMPACT                      = (1 << 1),
+  RTC_SCENE_FLAG_ROBUST                       = (1 << 2),
+  RTC_SCENE_FLAG_FILTER_FUNCTION_IN_ARGUMENTS = (1 << 3),
+  RTC_SCENE_FLAG_PREFETCH_USM_SHARED_ON_GPU   = (1 << 4),
 };
 
 /* Additional arguments for rtcIntersect1/4/8/16 calls */
@@ -110,6 +111,9 @@ RTC_API RTCGeometry rtcGetGeometryThreadSafe(RTCScene scene, unsigned int geomID
 /* Gets the user-defined data pointer of the geometry. This function is not thread safe and should get used during rendering. */
 RTC_SYCL_API void* rtcGetGeometryUserDataFromScene(RTCScene scene, unsigned int geomID);
 
+/* Returns the interpolated transformation of an instance for the specified time. */
+RTC_SYCL_API void rtcGetGeometryTransformFromScene(RTCScene scene, unsigned int geomID, float time, enum RTCFormat format, void* xfm);
+
 
 /* Commits the scene. */
 RTC_API void rtcCommitScene(RTCScene scene);
@@ -169,15 +173,27 @@ RTC_API void rtcIntersect16(const int* valid, RTCScene scene, struct RTCRayHit16
 /* Forwards ray inside user geometry callback. */
 RTC_SYCL_API void rtcForwardIntersect1(const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay* ray, unsigned int instID);
 
+/* Forwards ray inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_SYCL_API void rtcForwardIntersect1Ex(const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay* ray, unsigned int instID, unsigned int instPrimID);
+
 /* Forwards ray packet of size 4 inside user geometry callback. */
 RTC_API void rtcForwardIntersect4(const int* valid, const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay4* ray, unsigned int instID);
 
+/* Forwards ray packet of size 4 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardIntersect4Ex(const int* valid, const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay4* ray, unsigned int instID, unsigned int primInstID);
+
 /* Forwards ray packet of size 8 inside user geometry callback. */
 RTC_API void rtcForwardIntersect8(const int* valid, const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay8* ray, unsigned int instID);
 
+/* Forwards ray packet of size 4 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardIntersect8Ex(const int* valid, const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay8* ray, unsigned int instID, unsigned int primInstID);
+
 /* Forwards ray packet of size 16 inside user geometry callback. */
 RTC_API void rtcForwardIntersect16(const int* valid, const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay16* ray, unsigned int instID);
 
+/* Forwards ray packet of size 4 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardIntersect16Ex(const int* valid, const struct RTCIntersectFunctionNArguments* args, RTCScene scene, struct RTCRay16* ray, unsigned int instID, unsigned int primInstID);
+
 
 /* Tests a single ray for occlusion with the scene. */
 RTC_SYCL_API void rtcOccluded1(RTCScene scene, struct RTCRay* ray, struct RTCOccludedArguments* args RTC_OPTIONAL_ARGUMENT);
@@ -195,15 +211,27 @@ RTC_API void rtcOccluded16(const int* valid, RTCScene scene, struct RTCRay16* ra
 /* Forwards single occlusion ray inside user geometry callback. */
 RTC_SYCL_API void rtcForwardOccluded1(const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay* ray, unsigned int instID);
 
+/* Forwards single occlusion ray inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_SYCL_API void rtcForwardOccluded1Ex(const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay* ray, unsigned int instID, unsigned int instPrimID);
+
 /* Forwards occlusion ray packet of size 4 inside user geometry callback. */
 RTC_API void rtcForwardOccluded4(const int* valid, const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay4* ray, unsigned int instID);
 
+/* Forwards occlusion ray packet of size 4 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardOccluded4Ex(const int* valid, const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay4* ray, unsigned int instID, unsigned int instPrimID);
+
 /* Forwards occlusion ray packet of size 8 inside user geometry callback. */
 RTC_API void rtcForwardOccluded8(const int* valid, const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay8* ray, unsigned int instID);
 
+/* Forwards occlusion ray packet of size 8 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardOccluded8Ex(const int* valid, const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay8* ray, unsigned int instID, unsigned int instPrimID);
+
 /* Forwards occlusion ray packet of size 16 inside user geometry callback. */
 RTC_API void rtcForwardOccluded16(const int* valid, const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay16* ray, unsigned int instID);
 
+/* Forwards occlusion ray packet of size 16 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardOccluded16Ex(const int* valid, const struct RTCOccludedFunctionNArguments* args, RTCScene scene, struct RTCRay16* ray, unsigned int instID, unsigned int instPrimID);
+
 
 /*! collision callback */
 struct RTCCollision { unsigned int geomID0; unsigned int primID0; unsigned int geomID1; unsigned int primID1; };

include/embree4/rtcore_scene.isph

@@ -101,6 +101,51 @@ RTC_API RTCGeometry rtcGetGeometryThreadSafe(RTCScene scene, uniform unsigned in
 /* Gets the user-defined data pointer of the geometry. This function is not thread safe and should get used during rendering. */
 RTC_API void* rtcGetGeometryUserDataFromScene(RTCScene scene, uniform unsigned int geomID);
 
+/* Returns the interpolated transformation of an instance for the specified time. */
+RTC_API void rtcGetGeometryTransformFromScene(RTCScene scene, uniform unsigned int geomID, uniform float time, uniform RTCFormat format, void* uniform xfm);
+
+/* Returns the interpolated transformation of an instance for the specified time. Varying version. */
+inline void rtcGetGeometryTransformFromScene(RTCScene scene, varying unsigned int geomID, varying float time, uniform RTCFormat format, void* uniform xfm)
+{
+  varying float vmatrix[12];
+
+  foreach_unique(ugeomID in geomID)
+    foreach_unique(utime in time)
+  {
+    uniform float umatrix[12];
+    rtcGetGeometryTransformFromScene(scene,ugeomID,utime,RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR,umatrix);
+    for (int uniform j=0; j<12; j++) vmatrix[j] = umatrix[j];
+  }
+
+  /* store to desired layout */
+  varying float* vxfm = (varying float*) xfm;
+  
+  switch (format)
+  {
+  case RTC_FORMAT_FLOAT3X4_ROW_MAJOR:
+    vxfm[ 0] = vmatrix[0];  vxfm[ 1] = vmatrix[3];  vxfm[ 2] = vmatrix[6];  vxfm[ 3] = vmatrix[9];
+    vxfm[ 4] = vmatrix[1];  vxfm[ 5] = vmatrix[4];  vxfm[ 6] = vmatrix[7];  vxfm[ 7] = vmatrix[10];
+    vxfm[ 8] = vmatrix[2];  vxfm[ 9] = vmatrix[5];  vxfm[10] = vmatrix[8];  vxfm[11] = vmatrix[11];
+    break;
+    
+  case RTC_FORMAT_FLOAT3X4_COLUMN_MAJOR:
+    vxfm[ 0] = vmatrix[0];  vxfm[ 1] = vmatrix[1];  vxfm[ 2] = vmatrix[2];
+    vxfm[ 3] = vmatrix[3];  vxfm[ 4] = vmatrix[4];  vxfm[ 5] = vmatrix[5];
+    vxfm[ 6] = vmatrix[6];  vxfm[ 7] = vmatrix[7];  vxfm[ 8] = vmatrix[8];
+    vxfm[ 9] = vmatrix[9];  vxfm[10] = vmatrix[10]; vxfm[11] = vmatrix[11];
+    break;
+    
+  case RTC_FORMAT_FLOAT4X4_COLUMN_MAJOR:
+    vxfm[ 0] = vmatrix[0];  vxfm[ 1] = vmatrix[1];  vxfm[ 2] = vmatrix[2];  vxfm[ 3] = 0.f;
+    vxfm[ 4] = vmatrix[3];  vxfm[ 5] = vmatrix[4];  vxfm[ 6] = vmatrix[5];  vxfm[ 7] = 0.f;
+    vxfm[ 8] = vmatrix[6];  vxfm[ 9] = vmatrix[7];  vxfm[10] = vmatrix[8];  vxfm[11] = 0.f;
+    vxfm[12] = vmatrix[9];  vxfm[13] = vmatrix[10]; vxfm[14] = vmatrix[11]; vxfm[15] = 1.f;
+    break;
+    
+  default:
+    break;
+  }
+}
 
 /* Commits the scene. */
 RTC_API void rtcCommitScene(RTCScene scene);
@@ -191,15 +236,26 @@ RTC_FORCEINLINE void rtcIntersectV(RTCScene scene, varying RTCRayHit* uniform ra
 /* Forwards ray inside user geometry callback. */
 RTC_API void rtcForwardIntersect1(const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, uniform RTCRay* uniform ray, uniform unsigned int instID);
 
+/* Forwards ray inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardIntersect1Ex(const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, uniform RTCRay* uniform ray, uniform unsigned int instID, uniform unsigned int instPrimID);
+
 /* Forwards ray packet of size 4 inside user geometry callback. */
 RTC_API void rtcForwardIntersect4(const uniform int* uniform valid, const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, void* uniform ray4, uniform unsigned int instID);
 
+/* Forwards ray packet of size 4 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardIntersect4Ex(const uniform int* uniform valid, const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, void* uniform ray4, uniform unsigned int instID, uniform unsigned int instPrimID);
+
 /* Forwards ray packet of size 8 inside user geometry callback. */
 RTC_API void rtcForwardIntersect8(const uniform int* uniform valid, const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, void* uniform ray8, uniform unsigned int instID);
 
+/* Forwards ray packet of size 8 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardIntersect8Ex(const uniform int* uniform valid, const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, void* uniform ray8, uniform unsigned int instID, uniform unsigned int instPrimID);
+
 /* Forwards ray packet of size 16 inside user geometry callback. */
 RTC_API void rtcForwardIntersect16(const uniform int* uniform valid, const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, void* uniform ray16, uniform unsigned int instID);
 
+/* Forwards ray packet of size 16 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardIntersect16Ex(const uniform int* uniform valid, const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, void* uniform ray16, uniform unsigned int instID, uniform unsigned int instPrimID);
 
 /* Forwards ray intersection query inside user geometry callback. */
 RTC_FORCEINLINE void rtcForwardIntersectV(const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, varying RTCRay* uniform iray, uniform unsigned int instID)
@@ -216,6 +272,21 @@ RTC_FORCEINLINE void rtcForwardIntersectV(const uniform RTCIntersectFunctionNArg
     rtcForwardIntersect16((uniform int* uniform)&imask, args, scene, iray, instID);
 }
 
+/* Forwards ray intersection query inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_FORCEINLINE void rtcForwardIntersectVEx(const uniform RTCIntersectFunctionNArguments* uniform args, RTCScene scene, varying RTCRay* uniform iray, uniform unsigned int instID, uniform unsigned int instPrimID)
+{
+  varying bool mask = __mask;
+  unmasked {
+    varying int imask = mask ? -1 : 0;
+  }
+  if (sizeof(varying float) == 16)
+    rtcForwardIntersect4Ex((uniform int* uniform)&imask, args, scene, iray, instID, instPrimID);
+  else if (sizeof(varying float) == 32)
+    rtcForwardIntersect8Ex((uniform int* uniform)&imask, args, scene, iray, instID, instPrimID);
+  else if (sizeof(varying float) == 64)
+    rtcForwardIntersect16Ex((uniform int* uniform)&imask, args, scene, iray, instID, instPrimID);
+}
+
 
 /* Tests a single ray for occlusion with the scene. */
 RTC_API void rtcOccluded1(RTCScene scene, uniform RTCRay* uniform ray, uniform RTCOccludedArguments* uniform args = NULL);
@@ -249,15 +320,27 @@ RTC_FORCEINLINE void rtcOccludedV(RTCScene scene, varying RTCRay* uniform ray, u
 /* Forwards single occlusion ray inside user geometry callback. */
 RTC_API void rtcForwardOccluded1(const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, uniform RTCRay* uniform ray, uniform unsigned int instID);
 
+/* Forwards single occlusion ray inside user geometry callback.  Extended to handle instance arrays using instPrimID parameter.*/
+RTC_API void rtcForwardOccluded1Ex(const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, uniform RTCRay* uniform ray, uniform unsigned int instID, uniform unsigned int instPrimID);
+
 /* Forwards occlusion ray packet of size 4 inside user geometry callback. */
 RTC_API void rtcForwardOccluded4(const uniform int* uniform valid, const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, void* uniform ray4, uniform unsigned int instID);
 
+/* Forwards occlusion ray packet of size 4 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardOccluded4Ex(const uniform int* uniform valid, const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, void* uniform ray4, uniform unsigned int instID, uniform unsigned int instPrimID);
+
 /* Forwards occlusion ray packet of size 8 inside user geometry callback. */
 RTC_API void rtcForwardOccluded8(const uniform int* uniform valid, const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, void* uniform ray8, uniform unsigned int instID);
 
+/* Forwards occlusion ray packet of size 8 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardOccluded8Ex(const uniform int* uniform valid, const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, void* uniform ray8, uniform unsigned int instID, uniform unsigned int instPrimID);
+
 /* Forwards occlusion ray packet of size 16 inside user geometry callback. */
 RTC_API void rtcForwardOccluded16(const uniform int* uniform valid, const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, void* uniform ray16, uniform unsigned int instID);
 
+/* Forwards occlusion ray packet of size 16 inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_API void rtcForwardOccluded16Ex(const uniform int* uniform valid, const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, void* uniform ray16, uniform unsigned int instID, uniform unsigned int instPrimID);
+
 /* Forwards ray occlusion query inside user geometry callback. */
 RTC_FORCEINLINE void rtcForwardOccludedV(const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, varying RTCRay* uniform iray, uniform unsigned int instID)
 {
@@ -274,6 +357,22 @@ RTC_FORCEINLINE void rtcForwardOccludedV(const uniform RTCOccludedFunctionNArgum
     rtcForwardOccluded16((uniform int* uniform)&imask, args, scene, iray, instID);
 }
 
+/* Forwards ray occlusion query inside user geometry callback. Extended to handle instance arrays using instPrimID parameter. */
+RTC_FORCEINLINE void rtcForwardOccludedVEx(const uniform RTCOccludedFunctionNArguments* uniform args, RTCScene scene, varying RTCRay* uniform iray, uniform unsigned int instID, uniform unsigned int instPrimID)
+{
+  varying bool mask = __mask;
+  unmasked {
+    varying int imask = mask ? -1 : 0;
+  }
+
+  if (sizeof(varying float) == 16)
+    rtcForwardOccluded4Ex((uniform int* uniform)&imask, args, scene, iray, instID, instPrimID);
+  else if (sizeof(varying float) == 32)
+    rtcForwardOccluded8Ex((uniform int* uniform)&imask, args, scene, iray, instID, instPrimID);
+  else if (sizeof(varying float) == 64)
+    rtcForwardOccluded16Ex((uniform int* uniform)&imask, args, scene, iray, instID, instPrimID);
+}
+
 
 /*! collision callback */
 struct RTCCollision { unsigned int geomID0; unsigned int primID0; unsigned int geomID1; unsigned int primID1; };

kernels/CMakeLists.txt

@@ -6,6 +6,7 @@ IF (EMBREE_SYCL_RT_SIMULATION)
 ENDIF()
 
 IF (EMBREE_SYCL_SUPPORT)
+  ADD_SUBDIRECTORY(level_zero)
   ADD_SUBDIRECTORY(rthwif)
 ENDIF()
 
@@ -47,6 +48,7 @@ SET(EMBREE_LIBRARY_FILES
   common/geometry.cpp
   common/scene_user_geometry.cpp
   common/scene_instance.cpp
+  common/scene_instance_array.cpp
   common/scene_triangle_mesh.cpp
   common/scene_quad_mesh.cpp
   common/scene_curves.cpp
@@ -61,6 +63,7 @@ SET(EMBREE_LIBRARY_FILES
 
   geometry/primitive4.cpp
   geometry/instance_intersector.cpp
+  geometry/instance_array_intersector.cpp
   geometry/curve_intersector_virtual_4v.cpp
   geometry/curve_intersector_virtual_4i.cpp
   geometry/curve_intersector_virtual_4i_mb.cpp
@@ -88,9 +91,13 @@ SET(EMBREE_LIBRARY_FILES
   bvh/bvh_intersector1_bvh4.cpp
   )
 
+
 IF (EMBREE_SYCL_SUPPORT)
   SET(EMBREE_LIBRARY_FILES ${EMBREE_LIBRARY_FILES}
-  sycl/rthwif_embree_builder.cpp)
+  sycl/rthwif_embree_builder.cpp
+  sycl/scene_sycl.cpp)
+
+  set_source_files_properties(sycl/scene_sycl.cpp PROPERTIES COMPILE_FLAGS ${CMAKE_CXX_FLAGS_SYCL})
 ENDIF()
 
 IF (EMBREE_GEOMETRY_SUBDIVISION)
@@ -113,6 +120,7 @@ MACRO(embree_files TARGET ISA)
       
   SET(${TARGET}
     geometry/instance_intersector.cpp
+    geometry/instance_array_intersector.cpp
     geometry/curve_intersector_virtual_4v.cpp
     geometry/curve_intersector_virtual_4i.cpp
     geometry/curve_intersector_virtual_4i_mb.cpp
@@ -130,6 +138,7 @@ MACRO(embree_files TARGET ISA)
     LIST(APPEND ${TARGET}
       common/scene_user_geometry.cpp
       common/scene_instance.cpp
+      common/scene_instance_array.cpp
       common/scene_triangle_mesh.cpp
       common/scene_quad_mesh.cpp 
       common/scene_curves.cpp
@@ -240,10 +249,9 @@ DISABLE_STACK_PROTECTOR_FOR_INTERSECTORS(${EMBREE_LIBRARY_FILES})
 ADD_LIBRARY(embree ${EMBREE_LIB_TYPE} ${EMBREE_LIBRARY_FILES})
 
 IF (EMBREE_SYCL_SUPPORT)
-
-  IF (EMBREE_LEVEL_ZERO)
-    TARGET_COMPILE_DEFINITIONS(embree PUBLIC EMBREE_LEVEL_ZERO)
-  ENDIF()
+  # add link flag '-fsycl' here to include the kernel for USM prefetch from
+  # scene_sycl.cpp in embree4.dll/embree4.so
+  TARGET_LINK_OPTIONS(embree PRIVATE "-fsycl")
 
   ADD_LIBRARY(embree_sycl STATIC
     sycl/rtcore_sycl.cpp
@@ -341,10 +349,7 @@ ENDIF()
 
 TARGET_LINK_LIBRARIES(embree PRIVATE ${EMBREE_LIBRARIES} sys math simd lexers tasking)
 IF (EMBREE_SYCL_SUPPORT)
-  TARGET_LINK_LIBRARIES(embree PRIVATE embree_rthwif ${SYCL_LIB_NAME} PUBLIC embree_sycl)
-  IF (EMBREE_LEVL_ZERO)
-    TARGET_LINK_LIBRARIES(embree PRIVATE ze_loader)
-  ENDIF()
+  TARGET_LINK_LIBRARIES(embree PRIVATE ${SYCL_LIB_NAME} ze_wrapper PUBLIC embree_sycl)
 ENDIF()
 
 #TARGET_LINK_LIBRARIES(embree PRIVATE ${EMBREE_LIBRARIES} sys math simd lexers tasking ${CMAKE_LINK_FLAGS_SYCL})  # FIXME: enable this line
@@ -357,7 +362,7 @@ target_include_directories(embree PUBLIC
 # libtbb is located in same install folder as libembree
 IF(WIN32)
 ELSEIF(APPLE)
-  SET_TARGET_PROPERTIES(embree PROPERTIES INSTALL_RPATH "@rpath") # On MacOSX the app is supposed to already have an rpath set to that folder
+  SET_TARGET_PROPERTIES(embree PROPERTIES INSTALL_RPATH "@loader_path") # On MacOSX we tell dyld to find libtbb in the folder libembree is placed
 ELSE()
   SET_TARGET_PROPERTIES(embree PROPERTIES INSTALL_RPATH "$ORIGIN")  # Linux we tell ld to find libtbb in the folder libembree is placed.
 ENDIF()

kernels/builders/bvh_builder_sah.h

@@ -48,10 +48,9 @@ namespace embree
         }
 
         Settings (size_t sahBlockSize, size_t minLeafSize, size_t maxLeafSize, float travCost, float intCost, size_t singleThreadThreshold, size_t primrefarrayalloc = inf)
-        : branchingFactor(2), maxDepth(32), logBlockSize(bsr(sahBlockSize)), minLeafSize(minLeafSize), maxLeafSize(maxLeafSize),
+        : branchingFactor(2), maxDepth(32), logBlockSize(bsr(sahBlockSize)), minLeafSize(min(minLeafSize,maxLeafSize)), maxLeafSize(maxLeafSize),
           travCost(travCost), intCost(intCost), singleThreadThreshold(singleThreadThreshold), primrefarrayalloc(primrefarrayalloc)
         {
-          minLeafSize = min(minLeafSize,maxLeafSize);
         }
 
       public:

kernels/builders/heuristic_spatial_array.h

@@ -283,11 +283,9 @@ namespace embree
 
                 if (likely(splits <= 1)) continue; /* todo: does this ever happen ? */
 
-                //int bin0 = split.mapping.bin(prims0[i].lower)[split.dim];
-                //int bin1 = split.mapping.bin(prims0[i].upper)[split.dim];
-                //if (unlikely(bin0 < split.pos && bin1 >= split.pos))
-
-                if (unlikely(prims0[i].lower[split.dim] < fpos && prims0[i].upper[split.dim] > fpos))
+                const int bin0 = split.mapping.bin(prims0[i].lower)[split.dim];
+                const int bin1 = split.mapping.bin(prims0[i].upper)[split.dim];
+                if (unlikely(bin0 < split.pos && bin1 >= split.pos))
                 {
                   assert(splits > 1);
 

kernels/builders/primrefgen.cpp

@@ -354,5 +354,6 @@ namespace embree
     IF_ENABLED_QUADS(template size_t createMortonCodeArray<QuadMesh>(QuadMesh* mesh COMMA mvector<BVHBuilderMorton::BuildPrim>& morton COMMA BuildProgressMonitor& progressMonitor));
     IF_ENABLED_USER (template size_t createMortonCodeArray<UserGeometry>(UserGeometry* mesh COMMA mvector<BVHBuilderMorton::BuildPrim>& morton COMMA BuildProgressMonitor& progressMonitor));
     IF_ENABLED_INSTANCE (template size_t createMortonCodeArray<Instance>(Instance* mesh COMMA mvector<BVHBuilderMorton::BuildPrim>& morton COMMA BuildProgressMonitor& progressMonitor));
+    IF_ENABLED_INSTANCE_ARRAY (template size_t createMortonCodeArray<InstanceArray>(InstanceArray* mesh COMMA mvector<BVHBuilderMorton::BuildPrim>& morton COMMA BuildProgressMonitor& progressMonitor));
   }
 }

kernels/builders/primrefgen_presplit.h

@@ -38,7 +38,6 @@ namespace embree
 
       __forceinline bool split_pos(const PrimRef& prim, unsigned int& dim_o, float& fsplit_o) const
       {
-#if !defined(__SYCL_DEVICE_ONLY__)        
         /* compute morton code */
         const Vec3fa lower = prim.lower;
         const Vec3fa upper = prim.upper;
@@ -73,7 +72,6 @@ namespace embree
 
         dim_o = dim;
         fsplit_o = fsplit;
-#endif        
         return true;
       }
 
@@ -85,11 +83,10 @@ namespace embree
         const Vec3fa gupper = (upper-base)*Vec3fa(scale)-Vec3fa(0.2f);
         Vec3ia ilower(floor(glower));
         Vec3ia iupper(floor(gupper));
-
-#if !defined(__SYCL_DEVICE_ONLY__)                
+        
         /* this ignores dimensions that are empty */
         iupper = (Vec3ia)select(vint4(glower) >= vint4(gupper),vint4(ilower),vint4(iupper));
-#endif        
+        
         /* compute a morton code for the lower and upper grid coordinates. */
         const unsigned int lower_code = bitInterleave(ilower.x,ilower.y,ilower.z);
         const unsigned int upper_code = bitInterleave(iupper.x,iupper.y,iupper.z);

kernels/bvh/bvh4_factory.cpp

@@ -17,6 +17,7 @@
 #include "../geometry/subdivpatch1.h"
 #include "../geometry/object.h"
 #include "../geometry/instance.h"
+#include "../geometry/instance_array.h"
 #include "../geometry/subgrid.h"
 #include "../common/accelinstance.h"
 
@@ -66,6 +67,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector1,BVH4InstanceIntersector1);
   DECLARE_SYMBOL2(Accel::Intersector1,BVH4InstanceMBIntersector1);
 
+  DECLARE_SYMBOL2(Accel::Intersector1,BVH4InstanceArrayIntersector1);
+  DECLARE_SYMBOL2(Accel::Intersector1,BVH4InstanceArrayMBIntersector1);
+
   DECLARE_SYMBOL2(Accel::Intersector1,BVH4GridIntersector1Moeller);
   DECLARE_SYMBOL2(Accel::Intersector1,BVH4GridMBIntersector1Moeller);
   DECLARE_SYMBOL2(Accel::Intersector1,BVH4GridIntersector1Pluecker);
@@ -104,6 +108,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector4,BVH4InstanceIntersector4Chunk);
   DECLARE_SYMBOL2(Accel::Intersector4,BVH4InstanceMBIntersector4Chunk);
 
+  DECLARE_SYMBOL2(Accel::Intersector4,BVH4InstanceArrayIntersector4Chunk);
+  DECLARE_SYMBOL2(Accel::Intersector4,BVH4InstanceArrayMBIntersector4Chunk);
+
   DECLARE_SYMBOL2(Accel::Intersector4,BVH4GridIntersector4HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector4,BVH4GridMBIntersector4HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector4,BVH4GridIntersector4HybridPluecker);
@@ -142,6 +149,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector8,BVH4InstanceIntersector8Chunk);
   DECLARE_SYMBOL2(Accel::Intersector8,BVH4InstanceMBIntersector8Chunk);
 
+  DECLARE_SYMBOL2(Accel::Intersector8,BVH4InstanceArrayIntersector8Chunk);
+  DECLARE_SYMBOL2(Accel::Intersector8,BVH4InstanceArrayMBIntersector8Chunk);
+
   DECLARE_SYMBOL2(Accel::Intersector8,BVH4GridIntersector8HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector8,BVH4GridMBIntersector8HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector8,BVH4GridIntersector8HybridPluecker);
@@ -180,6 +190,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector16,BVH4InstanceIntersector16Chunk);
   DECLARE_SYMBOL2(Accel::Intersector16,BVH4InstanceMBIntersector16Chunk);
 
+  DECLARE_SYMBOL2(Accel::Intersector16,BVH4InstanceArrayIntersector16Chunk);
+  DECLARE_SYMBOL2(Accel::Intersector16,BVH4InstanceArrayMBIntersector16Chunk);
+
   DECLARE_SYMBOL2(Accel::Intersector16,BVH4GridIntersector16HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector16,BVH4GridMBIntersector16HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector16,BVH4GridIntersector16HybridPluecker);
@@ -190,6 +203,7 @@ namespace embree
   DECLARE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelQuadMeshSAH,void* COMMA Scene* COMMA bool);
   DECLARE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelVirtualSAH,void* COMMA Scene* COMMA bool);
   DECLARE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelInstanceSAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
+  DECLARE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelInstanceArraySAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
 
   DECLARE_ISA_FUNCTION(Builder*,BVH4Curve4vBuilder_OBB_New,void* COMMA Scene* COMMA size_t);
   DECLARE_ISA_FUNCTION(Builder*,BVH4Curve4iBuilder_OBB_New,void* COMMA Scene* COMMA size_t);
@@ -220,7 +234,10 @@ namespace embree
 
   DECLARE_ISA_FUNCTION(Builder*,BVH4InstanceSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
   DECLARE_ISA_FUNCTION(Builder*,BVH4InstanceMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
-  
+
+  DECLARE_ISA_FUNCTION(Builder*,BVH4InstanceArraySceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+  DECLARE_ISA_FUNCTION(Builder*,BVH4InstanceArrayMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+
   DECLARE_ISA_FUNCTION(Builder*,BVH4GridSceneBuilderSAH,void* COMMA Scene* COMMA size_t);
   DECLARE_ISA_FUNCTION(Builder*,BVH4GridMBSceneBuilderSAH,void* COMMA Scene* COMMA size_t);
 
@@ -249,6 +266,7 @@ namespace embree
     IF_ENABLED_QUADS (SELECT_SYMBOL_DEFAULT_AVX(features,BVH4BuilderTwoLevelQuadMeshSAH));
     IF_ENABLED_USER (SELECT_SYMBOL_DEFAULT_AVX(features,BVH4BuilderTwoLevelVirtualSAH));
     IF_ENABLED_INSTANCE (SELECT_SYMBOL_DEFAULT_AVX(features,BVH4BuilderTwoLevelInstanceSAH));
+    IF_ENABLED_INSTANCE_ARRAY (SELECT_SYMBOL_DEFAULT_AVX(features,BVH4BuilderTwoLevelInstanceArraySAH));
 
     IF_ENABLED_CURVES_OR_POINTS(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4Curve4vBuilder_OBB_New));
     IF_ENABLED_CURVES_OR_POINTS(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4Curve4iBuilder_OBB_New));
@@ -279,7 +297,10 @@ namespace embree
 
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4InstanceSceneBuilderSAH));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4InstanceMBSceneBuilderSAH));
-    
+
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4InstanceArraySceneBuilderSAH));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4InstanceArrayMBSceneBuilderSAH));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4GridSceneBuilderSAH));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_DEFAULT_AVX(features,BVH4GridMBSceneBuilderSAH));
 
@@ -332,6 +353,9 @@ namespace embree
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceIntersector1));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceMBIntersector1));
 
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceArrayIntersector1));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceArrayMBIntersector1));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4GridIntersector1Moeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4GridMBIntersector1Moeller))
     IF_ENABLED_GRIDS(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4GridIntersector1Pluecker));
@@ -372,7 +396,10 @@ namespace embree
 
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceIntersector4Chunk));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceMBIntersector4Chunk));
-    
+
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceArrayIntersector4Chunk));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4InstanceArrayMBIntersector4Chunk));
+
     IF_ENABLED_QUADS(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4Quad4vIntersector4HybridMoeller));
 
     IF_ENABLED_GRIDS(SELECT_SYMBOL_DEFAULT_SSE42_AVX_AVX2_AVX512(features,BVH4GridIntersector4HybridMoeller));
@@ -407,13 +434,16 @@ namespace embree
 
     IF_ENABLED_SUBDIV(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4SubdivPatch1Intersector8));
     IF_ENABLED_SUBDIV(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4SubdivPatch1MBIntersector8));
-    
+
     IF_ENABLED_USER(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4VirtualIntersector8Chunk));
     IF_ENABLED_USER(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4VirtualMBIntersector8Chunk));
 
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4InstanceIntersector8Chunk));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4InstanceMBIntersector8Chunk));
 
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4InstanceArrayIntersector8Chunk));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4InstanceArrayMBIntersector8Chunk));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4GridIntersector8HybridMoeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4GridMBIntersector8HybridMoeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH4GridIntersector8HybridPluecker));
@@ -453,6 +483,9 @@ namespace embree
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX512(features,BVH4InstanceIntersector16Chunk));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX512(features,BVH4InstanceMBIntersector16Chunk));
 
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX512(features,BVH4InstanceArrayIntersector16Chunk));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX512(features,BVH4InstanceArrayMBIntersector16Chunk));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX512(features,BVH4GridIntersector16HybridMoeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX512(features,BVH4GridMBIntersector16HybridMoeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX512(features,BVH4GridIntersector16HybridPluecker));
@@ -816,6 +849,32 @@ namespace embree
 #endif
     return intersectors;
   }
+
+  Accel::Intersectors BVH4Factory::BVH4InstanceArrayIntersectors(BVH4* bvh)
+  {
+    Accel::Intersectors intersectors;
+    intersectors.ptr = bvh;
+    intersectors.intersector1  = BVH4InstanceArrayIntersector1();
+#if defined (EMBREE_RAY_PACKETS)
+    intersectors.intersector4  = BVH4InstanceArrayIntersector4Chunk();
+    intersectors.intersector8  = BVH4InstanceArrayIntersector8Chunk();
+    intersectors.intersector16 = BVH4InstanceArrayIntersector16Chunk();
+#endif
+    return intersectors;
+  }
+
+  Accel::Intersectors BVH4Factory::BVH4InstanceArrayMBIntersectors(BVH4* bvh)
+  {
+    Accel::Intersectors intersectors;
+    intersectors.ptr = bvh;
+    intersectors.intersector1  = BVH4InstanceArrayMBIntersector1();
+#if defined (EMBREE_RAY_PACKETS)
+    intersectors.intersector4  = BVH4InstanceArrayMBIntersector4Chunk();
+    intersectors.intersector8  = BVH4InstanceArrayMBIntersector8Chunk();
+    intersectors.intersector16 = BVH4InstanceArrayMBIntersector16Chunk();
+#endif
+    return intersectors;
+  }
   
   Accel::Intersectors BVH4Factory::BVH4SubdivPatch1Intersectors(BVH4* bvh)
   {
@@ -1193,6 +1252,35 @@ namespace embree
     return new AccelInstance(accel,builder,intersectors);
   }
 
+  Accel* BVH4Factory::BVH4InstanceArray(Scene* scene, BuildVariant bvariant)
+  {
+    BVH4* accel = new BVH4(InstanceArrayPrimitive::type,scene);
+    Accel::Intersectors intersectors = BVH4InstanceArrayIntersectors(accel);
+    auto gtype = Geometry::MTY_INSTANCE_ARRAY;
+
+    Builder* builder = nullptr;
+    if (scene->device->object_builder == "default") {
+      switch (bvariant) {
+      case BuildVariant::STATIC      : builder = BVH4InstanceArraySceneBuilderSAH(accel,scene,gtype); break;
+      case BuildVariant::DYNAMIC     : builder = BVH4BuilderTwoLevelInstanceArraySAH(accel,scene,gtype,false); break;
+      case BuildVariant::HIGH_QUALITY: assert(false); break;
+      }
+    }
+    else if (scene->device->object_builder == "sah") { builder = BVH4InstanceArraySceneBuilderSAH(accel,scene,gtype); }
+    else if (scene->device->object_builder == "dynamic") { builder = BVH4BuilderTwoLevelInstanceArraySAH(accel,scene,gtype,false); }
+    else throw_RTCError(RTC_ERROR_INVALID_ARGUMENT,"unknown builder "+scene->device->object_builder+" for BVH4<Object>");
+
+    return new AccelInstance(accel,builder,intersectors);
+  }
+
+  Accel* BVH4Factory::BVH4InstanceArrayMB(Scene* scene)
+  {
+    BVH4* accel = new BVH4(InstanceArrayPrimitive::type,scene);
+    Accel::Intersectors intersectors = BVH4InstanceArrayMBIntersectors(accel);
+    Builder* builder = BVH4InstanceArrayMBSceneBuilderSAH(accel,scene,Geometry::MTY_INSTANCE_ARRAY);
+    return new AccelInstance(accel,builder,intersectors);
+  }
+
   Accel::Intersectors BVH4Factory::BVH4GridIntersectors(BVH4* bvh, IntersectVariant ivariant)
   {
     Accel::Intersectors intersectors;

kernels/bvh/bvh4_factory.h

@@ -48,6 +48,9 @@ namespace embree
     Accel* BVH4Instance(Scene* scene, bool isExpensive, BuildVariant bvariant = BuildVariant::STATIC);
     Accel* BVH4InstanceMB(Scene* scene, bool isExpensive);
 
+    Accel* BVH4InstanceArray(Scene* scene, BuildVariant bvariant = BuildVariant::STATIC);
+    Accel* BVH4InstanceArrayMB(Scene* scene);
+
     Accel* BVH4Grid(Scene* scene, BuildVariant bvariant = BuildVariant::STATIC, IntersectVariant ivariant = IntersectVariant::FAST);
     Accel* BVH4GridMB(Scene* scene, BuildVariant bvariant = BuildVariant::STATIC, IntersectVariant ivariant = IntersectVariant::FAST);
 
@@ -77,7 +80,10 @@ namespace embree
 
     Accel::Intersectors BVH4InstanceIntersectors(BVH4* bvh);
     Accel::Intersectors BVH4InstanceMBIntersectors(BVH4* bvh);
-    
+
+    Accel::Intersectors BVH4InstanceArrayIntersectors(BVH4* bvh);
+    Accel::Intersectors BVH4InstanceArrayMBIntersectors(BVH4* bvh);
+
     Accel::Intersectors BVH4SubdivPatch1Intersectors(BVH4* bvh);
     Accel::Intersectors BVH4SubdivPatch1MBIntersectors(BVH4* bvh);
 
@@ -122,7 +128,10 @@ namespace embree
 
     DEFINE_SYMBOL2(Accel::Intersector1,BVH4InstanceIntersector1);
     DEFINE_SYMBOL2(Accel::Intersector1,BVH4InstanceMBIntersector1);
-        
+
+    DEFINE_SYMBOL2(Accel::Intersector1,BVH4InstanceArrayIntersector1);
+    DEFINE_SYMBOL2(Accel::Intersector1,BVH4InstanceArrayMBIntersector1);
+
     DEFINE_SYMBOL2(Accel::Intersector1,BVH4GridIntersector1Moeller);
     DEFINE_SYMBOL2(Accel::Intersector1,BVH4GridMBIntersector1Moeller);
     DEFINE_SYMBOL2(Accel::Intersector1,BVH4GridIntersector1Pluecker);
@@ -161,6 +170,9 @@ namespace embree
     DEFINE_SYMBOL2(Accel::Intersector4,BVH4InstanceIntersector4Chunk);
     DEFINE_SYMBOL2(Accel::Intersector4,BVH4InstanceMBIntersector4Chunk);
 
+    DEFINE_SYMBOL2(Accel::Intersector4,BVH4InstanceArrayIntersector4Chunk);
+    DEFINE_SYMBOL2(Accel::Intersector4,BVH4InstanceArrayMBIntersector4Chunk);
+
     DEFINE_SYMBOL2(Accel::Intersector4,BVH4GridIntersector4HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector4,BVH4GridMBIntersector4HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector4,BVH4GridIntersector4HybridPluecker);
@@ -201,6 +213,9 @@ namespace embree
     DEFINE_SYMBOL2(Accel::Intersector8,BVH4InstanceIntersector8Chunk);
     DEFINE_SYMBOL2(Accel::Intersector8,BVH4InstanceMBIntersector8Chunk);
 
+    DEFINE_SYMBOL2(Accel::Intersector8,BVH4InstanceArrayIntersector8Chunk);
+    DEFINE_SYMBOL2(Accel::Intersector8,BVH4InstanceArrayMBIntersector8Chunk);
+
     DEFINE_SYMBOL2(Accel::Intersector8,BVH4GridIntersector8HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector8,BVH4GridMBIntersector8HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector8,BVH4GridIntersector8HybridPluecker);
@@ -241,6 +256,9 @@ namespace embree
     DEFINE_SYMBOL2(Accel::Intersector16,BVH4InstanceIntersector16Chunk);
     DEFINE_SYMBOL2(Accel::Intersector16,BVH4InstanceMBIntersector16Chunk);
 
+    DEFINE_SYMBOL2(Accel::Intersector16,BVH4InstanceArrayIntersector16Chunk);
+    DEFINE_SYMBOL2(Accel::Intersector16,BVH4InstanceArrayMBIntersector16Chunk);
+
     DEFINE_SYMBOL2(Accel::Intersector16,BVH4GridIntersector16HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector16,BVH4GridMBIntersector16HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector16,BVH4GridIntersector16HybridPluecker);
@@ -274,6 +292,9 @@ namespace embree
     DEFINE_ISA_FUNCTION(Builder*,BVH4InstanceSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
     DEFINE_ISA_FUNCTION(Builder*,BVH4InstanceMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
 
+    DEFINE_ISA_FUNCTION(Builder*,BVH4InstanceArraySceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+    DEFINE_ISA_FUNCTION(Builder*,BVH4InstanceArrayMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+
     DEFINE_ISA_FUNCTION(Builder*,BVH4GridSceneBuilderSAH,void* COMMA Scene* COMMA size_t);
     DEFINE_ISA_FUNCTION(Builder*,BVH4GridMBSceneBuilderSAH,void* COMMA Scene* COMMA size_t);
 
@@ -292,5 +313,6 @@ namespace embree
     DEFINE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelQuadMeshSAH,void* COMMA Scene* COMMA bool);
     DEFINE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelVirtualSAH,void* COMMA Scene* COMMA bool);
     DEFINE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelInstanceSAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
+    DEFINE_ISA_FUNCTION(Builder*,BVH4BuilderTwoLevelInstanceArraySAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
   };
 }

kernels/bvh/bvh8_factory.cpp

@@ -21,6 +21,7 @@
 #include "../geometry/subdivpatch1.h"
 #include "../geometry/object.h"
 #include "../geometry/instance.h"
+#include "../geometry/instance_array.h"
 #include "../geometry/subgrid.h"
 #include "../common/accelinstance.h"
 
@@ -66,6 +67,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector1,BVH8InstanceIntersector1);
   DECLARE_SYMBOL2(Accel::Intersector1,BVH8InstanceMBIntersector1);
 
+  DECLARE_SYMBOL2(Accel::Intersector1,BVH8InstanceArrayIntersector1);
+  DECLARE_SYMBOL2(Accel::Intersector1,BVH8InstanceArrayMBIntersector1);
+
   DECLARE_SYMBOL2(Accel::Intersector1,BVH8GridIntersector1Moeller);
   DECLARE_SYMBOL2(Accel::Intersector1,BVH8GridMBIntersector1Moeller);
   DECLARE_SYMBOL2(Accel::Intersector1,BVH8GridIntersector1Pluecker);
@@ -101,6 +105,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector4,BVH8InstanceIntersector4Chunk);
   DECLARE_SYMBOL2(Accel::Intersector4,BVH8InstanceMBIntersector4Chunk);
 
+  DECLARE_SYMBOL2(Accel::Intersector4,BVH8InstanceArrayIntersector4Chunk);
+  DECLARE_SYMBOL2(Accel::Intersector4,BVH8InstanceArrayMBIntersector4Chunk);
+
   DECLARE_SYMBOL2(Accel::Intersector4,BVH8GridIntersector4HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector4,BVH8GridIntersector4HybridPluecker);
 
@@ -135,6 +142,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector8,BVH8InstanceIntersector8Chunk);
   DECLARE_SYMBOL2(Accel::Intersector8,BVH8InstanceMBIntersector8Chunk);
 
+  DECLARE_SYMBOL2(Accel::Intersector8,BVH8InstanceArrayIntersector8Chunk);
+  DECLARE_SYMBOL2(Accel::Intersector8,BVH8InstanceArrayMBIntersector8Chunk);
+
   DECLARE_SYMBOL2(Accel::Intersector8,BVH8GridIntersector8HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector8,BVH8GridIntersector8HybridPluecker);
 
@@ -169,6 +179,9 @@ namespace embree
   DECLARE_SYMBOL2(Accel::Intersector16,BVH8InstanceIntersector16Chunk);
   DECLARE_SYMBOL2(Accel::Intersector16,BVH8InstanceMBIntersector16Chunk);
 
+  DECLARE_SYMBOL2(Accel::Intersector16,BVH8InstanceArrayIntersector16Chunk);
+  DECLARE_SYMBOL2(Accel::Intersector16,BVH8InstanceArrayMBIntersector16Chunk);
+
   DECLARE_SYMBOL2(Accel::Intersector16,BVH8GridIntersector16HybridMoeller);
   DECLARE_SYMBOL2(Accel::Intersector16,BVH8GridIntersector16HybridPluecker);
 
@@ -194,6 +207,9 @@ namespace embree
   DECLARE_ISA_FUNCTION(Builder*,BVH8InstanceSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
   DECLARE_ISA_FUNCTION(Builder*,BVH8InstanceMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
 
+  DECLARE_ISA_FUNCTION(Builder*,BVH8InstanceArraySceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+  DECLARE_ISA_FUNCTION(Builder*,BVH8InstanceArrayMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+
   DECLARE_ISA_FUNCTION(Builder*,BVH8Triangle4SceneBuilderFastSpatialSAH,void* COMMA Scene* COMMA size_t);
   DECLARE_ISA_FUNCTION(Builder*,BVH8Triangle4vSceneBuilderFastSpatialSAH,void* COMMA Scene* COMMA size_t);
   DECLARE_ISA_FUNCTION(Builder*,BVH8Quad4vSceneBuilderFastSpatialSAH,void* COMMA Scene* COMMA size_t);
@@ -206,6 +222,7 @@ namespace embree
   DECLARE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelQuadMeshSAH,void* COMMA Scene* COMMA bool);
   DECLARE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelVirtualSAH,void* COMMA Scene* COMMA bool);
   DECLARE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelInstanceSAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
+  DECLARE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelInstanceArraySAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
 
   BVH8Factory::BVH8Factory(int bfeatures, int ifeatures)
   {
@@ -238,6 +255,9 @@ namespace embree
 
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX(features,BVH8InstanceSceneBuilderSAH));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX(features,BVH8InstanceMBSceneBuilderSAH));
+
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX(features,BVH8InstanceArraySceneBuilderSAH));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX(features,BVH8InstanceArrayMBSceneBuilderSAH));
     
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX(features,BVH8GridSceneBuilderSAH));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX(features,BVH8GridMBSceneBuilderSAH));
@@ -252,6 +272,7 @@ namespace embree
     IF_ENABLED_QUADS (SELECT_SYMBOL_INIT_AVX(features,BVH8BuilderTwoLevelQuadMeshSAH));
     IF_ENABLED_USER  (SELECT_SYMBOL_INIT_AVX(features,BVH8BuilderTwoLevelVirtualSAH));
     IF_ENABLED_INSTANCE (SELECT_SYMBOL_INIT_AVX(features,BVH8BuilderTwoLevelInstanceSAH));
+    IF_ENABLED_INSTANCE_ARRAY (SELECT_SYMBOL_INIT_AVX(features,BVH8BuilderTwoLevelInstanceArraySAH));
   }
 
   void BVH8Factory::selectIntersectors(int features)
@@ -295,6 +316,9 @@ namespace embree
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceIntersector1));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceMBIntersector1));
 
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceArrayIntersector1));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceArrayMBIntersector1));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8GridIntersector1Moeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8GridMBIntersector1Moeller))
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8GridIntersector1Pluecker));
@@ -333,6 +357,9 @@ namespace embree
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceIntersector4Chunk));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceMBIntersector4Chunk));
 
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceArrayIntersector4Chunk));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceArrayMBIntersector4Chunk));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8GridIntersector4HybridMoeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8GridIntersector4HybridPluecker));
 
@@ -368,6 +395,9 @@ namespace embree
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceIntersector8Chunk));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceMBIntersector8Chunk));
 
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceArrayIntersector8Chunk));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8InstanceArrayMBIntersector8Chunk));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8GridIntersector8HybridMoeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX_AVX2_AVX512(features,BVH8GridIntersector8HybridPluecker));
 
@@ -403,6 +433,9 @@ namespace embree
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX512(features,BVH8InstanceIntersector16Chunk));
     IF_ENABLED_INSTANCE(SELECT_SYMBOL_INIT_AVX512(features,BVH8InstanceMBIntersector16Chunk));
 
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX512(features,BVH8InstanceArrayIntersector16Chunk));
+    IF_ENABLED_INSTANCE_ARRAY(SELECT_SYMBOL_INIT_AVX512(features,BVH8InstanceArrayMBIntersector16Chunk));
+
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX512(features,BVH8GridIntersector16HybridMoeller));
     IF_ENABLED_GRIDS(SELECT_SYMBOL_INIT_AVX512(features,BVH8GridIntersector16HybridPluecker));
 
@@ -767,6 +800,19 @@ namespace embree
     return intersectors;
   }
 
+  Accel::Intersectors BVH8Factory::BVH8InstanceArrayIntersectors(BVH8* bvh)
+  {
+    Accel::Intersectors intersectors;
+    intersectors.ptr = bvh;
+    intersectors.intersector1  = BVH8InstanceArrayIntersector1();
+#if defined (EMBREE_RAY_PACKETS)
+    intersectors.intersector4  = BVH8InstanceArrayIntersector4Chunk();
+    intersectors.intersector8  = BVH8InstanceArrayIntersector8Chunk();
+    intersectors.intersector16 = BVH8InstanceArrayIntersector16Chunk();
+#endif
+    return intersectors;
+  }
+
   Accel::Intersectors BVH8Factory::BVH8InstanceMBIntersectors(BVH8* bvh)
   {
     Accel::Intersectors intersectors;
@@ -780,6 +826,19 @@ namespace embree
     return intersectors;
   }
 
+  Accel::Intersectors BVH8Factory::BVH8InstanceArrayMBIntersectors(BVH8* bvh)
+  {
+    Accel::Intersectors intersectors;
+    intersectors.ptr = bvh;
+    intersectors.intersector1  = BVH8InstanceArrayMBIntersector1();
+#if defined (EMBREE_RAY_PACKETS)
+    intersectors.intersector4  = BVH8InstanceArrayMBIntersector4Chunk();
+    intersectors.intersector8  = BVH8InstanceArrayMBIntersector8Chunk();
+    intersectors.intersector16 = BVH8InstanceArrayMBIntersector16Chunk();
+#endif
+    return intersectors;
+  }
+
   Accel* BVH8Factory::BVH8OBBVirtualCurve8v(Scene* scene, IntersectVariant ivariant)
   {
     BVH8* accel = new BVH8(Curve8v::type,scene);
@@ -1024,6 +1083,28 @@ namespace embree
     return new AccelInstance(accel,builder,intersectors);
   }
 
+  Accel* BVH8Factory::BVH8InstanceArray(Scene* scene, BuildVariant bvariant)
+  {
+    BVH8* accel = new BVH8(InstanceArrayPrimitive::type,scene);
+    Accel::Intersectors intersectors = BVH8InstanceArrayIntersectors(accel);
+    auto gtype = Geometry::MTY_INSTANCE_ARRAY;
+    // Builder* builder = BVH8InstanceSceneBuilderSAH(accel,scene,gtype);
+
+    Builder* builder = nullptr;
+    if (scene->device->object_builder == "default") {
+      switch (bvariant) {
+      case BuildVariant::STATIC      : builder = BVH8InstanceArraySceneBuilderSAH(accel,scene,gtype); break;
+      case BuildVariant::DYNAMIC     : builder = BVH8BuilderTwoLevelInstanceArraySAH(accel,scene,gtype,false); break;
+      case BuildVariant::HIGH_QUALITY: assert(false); break;
+      }
+    }
+    else if (scene->device->object_builder == "sah") builder = BVH8InstanceArraySceneBuilderSAH(accel,scene,gtype);
+    else if (scene->device->object_builder == "dynamic") builder = BVH8BuilderTwoLevelInstanceArraySAH(accel,scene,gtype,false);
+    else throw_RTCError(RTC_ERROR_INVALID_ARGUMENT,"unknown builder "+scene->device->object_builder+" for BVH8<Object>");
+
+    return new AccelInstance(accel,builder,intersectors);
+  }
+
   Accel* BVH8Factory::BVH8InstanceMB(Scene* scene, bool isExpensive)
   {
     BVH8* accel = new BVH8(InstancePrimitive::type,scene);
@@ -1033,6 +1114,15 @@ namespace embree
     return new AccelInstance(accel,builder,intersectors);
   }
 
+  Accel* BVH8Factory::BVH8InstanceArrayMB(Scene* scene)
+  {
+    BVH8* accel = new BVH8(InstanceArrayPrimitive::type,scene);
+    Accel::Intersectors intersectors = BVH8InstanceArrayMBIntersectors(accel);
+    auto gtype = Geometry::MTY_INSTANCE_ARRAY;
+    Builder* builder = BVH8InstanceArrayMBSceneBuilderSAH(accel,scene,gtype);
+    return new AccelInstance(accel,builder,intersectors);
+  }
+
   Accel::Intersectors BVH8Factory::BVH8GridIntersectors(BVH8* bvh, IntersectVariant ivariant)
   {
     Accel::Intersectors intersectors;

kernels/bvh/bvh8_factory.h

@@ -39,6 +39,9 @@ namespace embree
     Accel* BVH8Instance(Scene* scene, bool isExpensive, BuildVariant bvariant = BuildVariant::STATIC);
     Accel* BVH8InstanceMB(Scene* scene, bool isExpensive);
 
+    Accel* BVH8InstanceArray(Scene* scene, BuildVariant bvariant = BuildVariant::STATIC);
+    Accel* BVH8InstanceArrayMB(Scene* scene);
+
     Accel* BVH8Grid(Scene* scene, BuildVariant bvariant = BuildVariant::STATIC, IntersectVariant ivariant = IntersectVariant::FAST);
     Accel* BVH8GridMB(Scene* scene, BuildVariant bvariant = BuildVariant::STATIC, IntersectVariant ivariant = IntersectVariant::FAST);
 
@@ -70,6 +73,9 @@ namespace embree
     Accel::Intersectors BVH8InstanceIntersectors(BVH8* bvh);
     Accel::Intersectors BVH8InstanceMBIntersectors(BVH8* bvh);
 
+    Accel::Intersectors BVH8InstanceArrayIntersectors(BVH8* bvh);
+    Accel::Intersectors BVH8InstanceArrayMBIntersectors(BVH8* bvh);
+
     Accel::Intersectors BVH8GridIntersectors(BVH8* bvh, IntersectVariant ivariant);
     Accel::Intersectors BVH8GridMBIntersectors(BVH8* bvh, IntersectVariant ivariant);
 
@@ -111,6 +117,9 @@ namespace embree
     DEFINE_SYMBOL2(Accel::Intersector1,BVH8InstanceIntersector1);
     DEFINE_SYMBOL2(Accel::Intersector1,BVH8InstanceMBIntersector1);
 
+    DEFINE_SYMBOL2(Accel::Intersector1,BVH8InstanceArrayIntersector1);
+    DEFINE_SYMBOL2(Accel::Intersector1,BVH8InstanceArrayMBIntersector1);
+
     DEFINE_SYMBOL2(Accel::Intersector1,BVH8GridIntersector1Moeller);
     DEFINE_SYMBOL2(Accel::Intersector1,BVH8GridMBIntersector1Moeller);
     DEFINE_SYMBOL2(Accel::Intersector1,BVH8GridIntersector1Pluecker);
@@ -145,7 +154,10 @@ namespace embree
 
     DEFINE_SYMBOL2(Accel::Intersector4,BVH8InstanceIntersector4Chunk);
     DEFINE_SYMBOL2(Accel::Intersector4,BVH8InstanceMBIntersector4Chunk);
-    
+
+    DEFINE_SYMBOL2(Accel::Intersector4,BVH8InstanceArrayIntersector4Chunk);
+    DEFINE_SYMBOL2(Accel::Intersector4,BVH8InstanceArrayMBIntersector4Chunk);
+
     DEFINE_SYMBOL2(Accel::Intersector4,BVH8GridIntersector4HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector4,BVH8GridIntersector4HybridPluecker);
 
@@ -180,9 +192,12 @@ namespace embree
     DEFINE_SYMBOL2(Accel::Intersector8,BVH8InstanceIntersector8Chunk);
     DEFINE_SYMBOL2(Accel::Intersector8,BVH8InstanceMBIntersector8Chunk);
 
+    DEFINE_SYMBOL2(Accel::Intersector8,BVH8InstanceArrayIntersector8Chunk);
+    DEFINE_SYMBOL2(Accel::Intersector8,BVH8InstanceArrayMBIntersector8Chunk);
+
     DEFINE_SYMBOL2(Accel::Intersector8,BVH8GridIntersector8HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector8,BVH8GridIntersector8HybridPluecker);
-   
+
     DEFINE_SYMBOL2(Accel::Intersector16,BVH8OBBVirtualCurveIntersector16Hybrid);
     DEFINE_SYMBOL2(Accel::Intersector16,BVH8OBBVirtualCurveIntersector16HybridMB);
     DEFINE_SYMBOL2(Accel::Intersector16,BVH8OBBVirtualCurveIntersectorRobust16Hybrid);
@@ -213,7 +228,10 @@ namespace embree
 
     DEFINE_SYMBOL2(Accel::Intersector16,BVH8InstanceIntersector16Chunk);
     DEFINE_SYMBOL2(Accel::Intersector16,BVH8InstanceMBIntersector16Chunk);
-   
+
+    DEFINE_SYMBOL2(Accel::Intersector16,BVH8InstanceArrayIntersector16Chunk);
+    DEFINE_SYMBOL2(Accel::Intersector16,BVH8InstanceArrayMBIntersector16Chunk);
+
     DEFINE_SYMBOL2(Accel::Intersector16,BVH8GridIntersector16HybridMoeller);
     DEFINE_SYMBOL2(Accel::Intersector16,BVH8GridIntersector16HybridPluecker);
 
@@ -240,6 +258,9 @@ namespace embree
 
     DEFINE_ISA_FUNCTION(Builder*,BVH8InstanceSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
     DEFINE_ISA_FUNCTION(Builder*,BVH8InstanceMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+    
+    DEFINE_ISA_FUNCTION(Builder*,BVH8InstanceArraySceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
+    DEFINE_ISA_FUNCTION(Builder*,BVH8InstanceArrayMBSceneBuilderSAH,void* COMMA Scene* COMMA Geometry::GTypeMask);
 
     DEFINE_ISA_FUNCTION(Builder*,BVH8GridSceneBuilderSAH,void* COMMA Scene* COMMA size_t);
     DEFINE_ISA_FUNCTION(Builder*,BVH8GridMBSceneBuilderSAH,void* COMMA Scene* COMMA size_t);
@@ -258,5 +279,6 @@ namespace embree
     DEFINE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelQuadMeshSAH,void* COMMA Scene* COMMA bool);
     DEFINE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelVirtualSAH,void* COMMA Scene* COMMA bool);
     DEFINE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelInstanceSAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
+    DEFINE_ISA_FUNCTION(Builder*,BVH8BuilderTwoLevelInstanceArraySAH,void* COMMA Scene* COMMA Geometry::GTypeMask COMMA bool);
   };
 }

kernels/bvh/bvh_builder_morton.cpp

@@ -17,6 +17,7 @@
 #include "../geometry/quadi.h"
 #include "../geometry/object.h"
 #include "../geometry/instance.h"
+#include "../geometry/instance_array.h"
 
 #if defined(__64BIT__)
 #  define ROTATE_TREE 1 // specifies number of tree rotation rounds to perform
@@ -399,6 +400,50 @@ namespace embree
       unsigned int geomID_ = std::numeric_limits<unsigned int>::max();
     };
 
+    template<int N>
+    struct CreateMortonLeaf<N,InstanceArrayPrimitive>
+    {
+      typedef BVHN<N> BVH;
+      typedef typename BVH::NodeRef NodeRef;
+      typedef typename BVH::NodeRecord NodeRecord;
+
+      __forceinline CreateMortonLeaf (InstanceArray* mesh, unsigned int geomID, BVHBuilderMorton::BuildPrim* morton)
+        : mesh(mesh), morton(morton), geomID_(geomID) {}
+
+      __noinline NodeRecord operator() (const range<unsigned>& current, const FastAllocator::CachedAllocator& alloc)
+      {
+        vfloat4 lower(pos_inf);
+        vfloat4 upper(neg_inf);
+        size_t items = current.size();
+        size_t start = current.begin();
+        assert(items <= 1);
+
+        /* allocate leaf node */
+        InstanceArrayPrimitive* accel = (InstanceArrayPrimitive*) alloc.malloc1(items*sizeof(InstanceArrayPrimitive),BVH::byteAlignment);
+        NodeRef ref = BVH::encodeLeaf((char*)accel,items);
+        const InstanceArray* instance = this->mesh;
+
+        BBox3fa bounds = empty;
+        for (size_t i=0; i<items; i++)
+        {
+          const unsigned int primID = morton[start+i].index;
+          bounds.extend(instance->bounds(primID));
+          new (&accel[i]) InstanceArrayPrimitive(geomID_, primID);
+        }
+
+        BBox3fx box_o = (BBox3fx&)bounds;
+#if ROTATE_TREE
+        if (N == 4)
+          box_o.lower.a = current.size();
+#endif
+        return NodeRecord(ref,box_o);
+      }
+    private:
+      InstanceArray* mesh;
+      BVHBuilderMorton::BuildPrim* morton;
+      unsigned int geomID_ = std::numeric_limits<unsigned int>::max();
+    };
+
     template<typename Mesh>
     struct CalculateMeshBounds
     {
@@ -523,7 +568,14 @@ namespace embree
 #if defined(EMBREE_GEOMETRY_INSTANCE)
     Builder* BVH4InstanceMeshBuilderMortonGeneral (void* bvh, Instance* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) { return new class BVHNMeshBuilderMorton<4,Instance,InstancePrimitive>((BVH4*)bvh,mesh,gtype,geomID,1,BVH4::maxLeafBlocks); }
 #if defined(__AVX__)
-    Builder* BVH8InstanceMeshBuilderMortonGeneral (void* bvh, Instance* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) { return new class BVHNMeshBuilderMorton<8,Instance,InstancePrimitive>((BVH8*)bvh,mesh,gtype,geomID,1,BVH4::maxLeafBlocks); }    
+    Builder* BVH8InstanceMeshBuilderMortonGeneral (void* bvh, Instance* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) { return new class BVHNMeshBuilderMorton<8,Instance,InstancePrimitive>((BVH8*)bvh,mesh,gtype,geomID,1,BVH4::maxLeafBlocks); }
+#endif
+#endif
+
+#if defined(EMBREE_GEOMETRY_INSTANCE_ARRAY)
+    Builder* BVH4InstanceArrayMeshBuilderMortonGeneral (void* bvh, InstanceArray* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) { return new class BVHNMeshBuilderMorton<4,InstanceArray,InstanceArrayPrimitive>((BVH4*)bvh,mesh,gtype,geomID,1,BVH4::maxLeafBlocks); }
+#if defined(__AVX__)
+    Builder* BVH8InstanceArrayMeshBuilderMortonGeneral (void* bvh, InstanceArray* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) { return new class BVHNMeshBuilderMorton<8,InstanceArray,InstanceArrayPrimitive>((BVH8*)bvh,mesh,gtype,geomID,1,BVH4::maxLeafBlocks); }
 #endif
 #endif
 

kernels/bvh/bvh_builder_sah.cpp

@@ -15,6 +15,7 @@
 #include "../geometry/quadi.h"
 #include "../geometry/object.h"
 #include "../geometry/instance.h"
+#include "../geometry/instance_array.h"
 #include "../geometry/subgrid.h"
 
 #include "../common/state.h"
@@ -150,7 +151,7 @@ namespace embree
             const size_t leaf_bytes = size_t(1.2*Primitive::blocks(numPrimitives)*sizeof(Primitive));
             bvh->alloc.init_estimate(node_bytes+leaf_bytes);
             settings.singleThreadThreshold = bvh->alloc.fixSingleThreadThreshold(N,DEFAULT_SINGLE_THREAD_THRESHOLD,numPrimitives,node_bytes+leaf_bytes);
-            prims.resize(numPrimitives); 
+            prims.resize(numPrimitives);
 
             PrimInfo pinfo = mesh ?
               createPrimRefArray(mesh,geomID_,numPrimitives,prims,bvh->scene->progressInterface) :
@@ -518,14 +519,35 @@ namespace embree
 #endif
 
 #if defined(EMBREE_GEOMETRY_INSTANCE)
-    Builder* BVH4InstanceSceneBuilderSAH (void* bvh, Scene* scene, Geometry::GTypeMask gtype) { return new BVHNBuilderSAH<4,InstancePrimitive>((BVH4*)bvh,scene,4,1.0f,1,1,gtype); }
+    Builder* BVH4InstanceSceneBuilderSAH (void* bvh, Scene* scene, Geometry::GTypeMask gtype) {
+      return new BVHNBuilderSAH<4,InstancePrimitive>((BVH4*)bvh,scene,4,1.0f,1,1,gtype);
+    }
     Builder* BVH4InstanceMeshBuilderSAH (void* bvh, Instance* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) {
       return new BVHNBuilderSAH<4,InstancePrimitive>((BVH4*)bvh,mesh,geomID,4,1.0f,1,inf,gtype);
     }
 #if defined(__AVX__)
-    Builder* BVH8InstanceSceneBuilderSAH (void* bvh, Scene* scene, Geometry::GTypeMask gtype) { return new BVHNBuilderSAH<8,InstancePrimitive>((BVH8*)bvh,scene,8,1.0f,1,1,gtype); }
+    Builder* BVH8InstanceSceneBuilderSAH (void* bvh, Scene* scene, Geometry::GTypeMask gtype) {
+      return new BVHNBuilderSAH<8,InstancePrimitive>((BVH8*)bvh,scene,8,1.0f,1,1,gtype);
+    }
     Builder* BVH8InstanceMeshBuilderSAH (void* bvh, Instance* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) {
-      return new BVHNBuilderSAH<8,InstancePrimitive>((BVH8*)bvh,mesh,geomID,8,1.0f,1,inf,gtype);
+      return new BVHNBuilderSAH<8,InstancePrimitive>((BVH8*)bvh,mesh,geomID,8,1.0f,1,1,gtype);
+    }
+#endif
+#endif
+
+#if defined(EMBREE_GEOMETRY_INSTANCE_ARRAY)
+    Builder* BVH4InstanceArraySceneBuilderSAH (void* bvh, Scene* scene, Geometry::GTypeMask gtype) {
+      return new BVHNBuilderSAH<4,InstanceArrayPrimitive>((BVH4*)bvh,scene,4,1.0f,1,1,gtype);
+    }
+    Builder* BVH4InstanceArrayMeshBuilderSAH (void* bvh, InstanceArray* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) {
+      return new BVHNBuilderSAH<4,InstanceArrayPrimitive>((BVH4*)bvh,mesh,geomID,4,1.0f,1,1,gtype);
+    }
+#if defined(__AVX__)
+    Builder* BVH8InstanceArraySceneBuilderSAH (void* bvh, Scene* scene, Geometry::GTypeMask gtype) {
+      return new BVHNBuilderSAH<8,InstanceArrayPrimitive>((BVH8*)bvh,scene,8,1.0f,1,1,gtype);
+    }
+    Builder* BVH8InstanceArrayMeshBuilderSAH (void* bvh, InstanceArray* mesh, Geometry::GTypeMask gtype, unsigned int geomID, size_t mode) {
+      return new BVHNBuilderSAH<8,InstanceArrayPrimitive>((BVH8*)bvh,mesh,geomID,8,1.0f,1,1,gtype);
     }
 #endif
 #endif