Merge remote-tracking branch 'origin/sycl' into llvmspirv_pulldown

Conflicts: sycl/test/check_device_code/vector/vector_math_ops.cpp
intel · Dec 17, 2024 · 5ca9829 · 5ca9829
2 parents eb615f5 + 839f0af
commit 5ca9829
Show file tree

Hide file tree

Showing 85 changed files with 1,003 additions and 556 deletions.
diff --git a/.github/workflows/sycl-containers-igc-dev.yaml b/.github/workflows/sycl-containers-igc-dev.yaml
@@ -6,11 +6,13 @@ on:
       - sycl
     paths:
       - 'devops/actions/build_container/**'
+      - 'devops/scripts/**'
       - 'devops/dependencies-igc-dev.json'
       - '.github/workflows/sycl-containers-igc-dev.yaml'
   pull_request:
     paths:
       - 'devops/actions/build_container/**'
+      - 'devops/scripts/**'
       - 'devops/dependencies-igc-dev.json'
       - '.github/workflows/sycl-containers-igc-dev.yaml'
 

diff --git a/.github/workflows/sycl-containers.yaml b/.github/workflows/sycl-containers.yaml
@@ -47,6 +47,10 @@ jobs:
             file: ubuntu2204_build
             tag: latest
             build_args: ""
+          - name: Build Ubuntu 24.04 oneAPI Docker image
+            file: ubuntu2404_build_oneapi
+            tag: latest
+            build_args: ""
           - name: Intel Drivers Ubuntu 22.04 Docker image
             file: ubuntu2204_intel_drivers
             tag: latest
@@ -74,6 +78,7 @@ jobs:
           file: ${{ matrix.file }}
           username: ${{ github.repository_owner }}
           password: ${{ secrets.GITHUB_TOKEN }}
+          sycl_ci_passwd: ${{ secrets.DOCKER_SUDO_PASSWORD }}
           tags: |
             ghcr.io/${{ github.repository }}/${{ matrix.file }}:${{ matrix.tag }}-${{ github.sha }}
             ghcr.io/${{ github.repository }}/${{ matrix.file }}:${{ matrix.tag }}

diff --git a/.github/workflows/sycl-linux-precommit.yml b/.github/workflows/sycl-linux-precommit.yml
@@ -134,7 +134,8 @@ jobs:
       install_dev_igc_driver: >-
         ${{ !contains(matrix.target_devices, 'ext_oneapi_cuda') &&
         !contains(matrix.target_devices, 'ext_oneapi_hip') &&
-        matrix.use_igc_dev && contains(needs.detect_changes.outputs.filters, 'devigccfg') ||
+        matrix.use_igc_dev &&
+        (contains(needs.detect_changes.outputs.filters, 'devigccfg') || contains(needs.detect_changes.outputs.filters, 'drivers')) ||
         'false' }}
       # Run only if the PR does not have the 'ci-no-devigc' label.
       skip_run: ${{matrix.use_igc_dev && contains(github.event.pull_request.labels.*.name, 'ci-no-devigc') || 'false'}}

diff --git a/.github/workflows/sycl-rel-nightly.yml b/.github/workflows/sycl-rel-nightly.yml
@@ -7,12 +7,6 @@ on:
 
 permissions: read-all
 
-env:
-  # NOTE: so that GitHub Actions can trigger the scheduled workflow run - the
-  # workflow file should be on the default branch. Therefore every job should
-  # checkout the release branch.
-  SYCL_REL_BRANCH: "draft-sycl-rel-6_0_0"
-
 jobs:
   # To avoid unnecessary scheduled runs this job checks if there are new commits
   # since the last run. More precisely, it checks if the last commit is older
@@ -25,7 +19,7 @@ jobs:
     steps:
     - uses: actions/checkout@v4
       with:
-        ref: ${{ env.SYCL_REL_BRANCH }}
+        ref: draft-sycl-rel-6_0_0
     - run: git show --quiet | tee -a $GITHUB_STEP_SUMMARY
 
     - id: is_new_commit
@@ -45,7 +39,7 @@ jobs:
       build_artifact_suffix: v6
       build_configure_extra_args: '--hip --cuda'
       merge_ref: ''
-      ref: ${{ env.SYCL_REL_BRANCH }}
+      ref: draft-sycl-rel-6_0_0
 
       # We upload the build for people to download/use, override its name and
       # prefer widespread gzip compression.
@@ -113,7 +107,7 @@ jobs:
       tests_selector: ${{ matrix.tests_selector }}
       extra_lit_opts: ${{ matrix.extra_lit_opts }}
       reset_intel_gpu: ${{ matrix.reset_intel_gpu }}
-      ref: ${{ env.SYCL_REL_BRANCH }}
+      ref: draft-sycl-rel-6_0_0
       merge_ref: ''
       sycl_toolchain_artifact: sycl_linux_default
       sycl_toolchain_archive: ${{ needs.ubuntu2204_build.outputs.artifact_archive_name }}
@@ -127,7 +121,7 @@ jobs:
       # We upload both Linux/Windows build via Github's "Releases"
       # functionality, make sure Linux/Windows names follow the same pattern.
       artifact_archive_name: sycl_windows.tar.gz
-      build_ref: ${{ env.SYCL_REL_BRANCH }}
+      build_ref: draft-sycl-rel-6_0_0
 
   e2e-win:
     needs: build-win
@@ -142,7 +136,7 @@ jobs:
       runner: '["Windows","gen12"]'
       sycl_toolchain_archive: ${{ needs.build-win.outputs.artifact_archive_name }}
       extra_lit_opts: --param gpu-intel-gen12=True
-      ref: ${{ env.SYCL_REL_BRANCH }}
+      ref: draft-sycl-rel-6_0_0
 
   cuda-aws-start:
     needs: [ubuntu2204_build]
@@ -151,7 +145,7 @@ jobs:
     secrets: inherit
     with:
       mode: start
-      ref: ${{ env.SYCL_REL_BRANCH }}
+      ref: draft-sycl-rel-6_0_0
 
   cuda-run-tests:
     needs: [ubuntu2204_build, cuda-aws-start]
@@ -163,7 +157,7 @@ jobs:
       image: ghcr.io/intel/llvm/ubuntu2204_build:latest-0300ac924620a51f76c4929794637b82790f12ab
       image_options: -u 1001 --gpus all --cap-add SYS_ADMIN --env NVIDIA_DISABLE_REQUIRE=1
       target_devices: ext_oneapi_cuda:gpu
-      ref: ${{ env.SYCL_REL_BRANCH }}
+      ref: draft-sycl-rel-6_0_0
       merge_ref: ''
 
       sycl_toolchain_artifact: sycl_linux_default
@@ -177,4 +171,4 @@ jobs:
     secrets: inherit
     with:
       mode: stop
-      ref: ${{ env.SYCL_REL_BRANCH }}
+      ref: draft-sycl-rel-6_0_0
diff --git a/.github/workflows/sycl-windows-build.yml b/.github/workflows/sycl-windows-build.yml
@@ -73,20 +73,19 @@ jobs:
     name: Build + LIT
     runs-on: [Windows, build]
     environment: WindowsCILock
-    # TODO use cached checkout
     outputs:
       build_conclusion: ${{ steps.build.conclusion }}
     steps:
     - uses: actions/checkout@v4
       with:
-        path: src
-        ref: ${{ inputs.build_ref || github.sha }}
-        fetch-depth: 1
+        sparse-checkout: |
+          devops/actions
+        ref: ${{ inputs.ref || github.sha }} 
     - uses: ilammy/msvc-dev-cmd@0b201ec74fa43914dc39ae48a89fd1d8cb592756
       with:
         arch: amd64
     - name: Setup oneAPI env
-      uses: ./src/devops/actions/setup_windows_oneapi_env
+      uses: ./devops/actions/setup_windows_oneapi_env
       if: ${{ always() && !cancelled() && inputs.compiler == 'icx' }}
     - name: Set env
       run: |
@@ -96,7 +95,12 @@ jobs:
         echo "CCACHE_DIR=D:\github\_work\cache\${{ inputs.build_cache_suffix }}" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append
         echo "CCACHE_MAXSIZE=10G" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append
     - name: Register cleanup after job is finished
-      uses: ./src/devops/actions/cleanup
+      uses: ./devops/actions/cleanup
+    - uses: ./devops/actions/cached_checkout
+      with:
+        path: src
+        ref: ${{ inputs.build_ref || github.sha }}
+        cache_path: "D:\\\\github\\\\_work\\\\repo_cache\\\\"
     - name: Configure
       shell: cmd
       env:

diff --git a/.github/workflows/sycl-windows-run-tests.yml b/.github/workflows/sycl-windows-run-tests.yml
@@ -47,25 +47,29 @@ jobs:
     environment: WindowsCILock
     env: ${{ fromJSON(inputs.env) }}
     steps:
-    # TODO: use cached_checkout
     - uses: actions/checkout@v4
       with:
-        persist-credentials: false
+        sparse-checkout: |
+          devops/actions
         ref: ${{ inputs.ref || github.sha }}
-        path: llvm
     - uses: ilammy/msvc-dev-cmd@0b201ec74fa43914dc39ae48a89fd1d8cb592756
       with:
         arch: amd64
     - name: Setup oneAPI env
-      uses: ./llvm/devops/actions/setup_windows_oneapi_env
+      uses: ./devops/actions/setup_windows_oneapi_env
       if: ${{ always() && !cancelled() && inputs.compiler == 'icx' }}
     - name: Set env
       run: |
         git config --system core.longpaths true
         git config --global core.autocrlf false
         echo "C:\Program Files\Git\usr\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
     - name: Register cleanup after job is finished
-      uses: ./llvm/devops/actions/cleanup
+      uses: ./devops/actions/cleanup
+    - uses: ./devops/actions/cached_checkout
+      with:
+        path: llvm
+        ref: ${{ inputs.build_ref || github.sha }}
+        cache_path: "D:\\\\github\\\\_work\\\\repo_cache\\\\"
     - name: Download compiler toolchain
       uses: actions/download-artifact@v4
       with:

diff --git a/devops/actions/build_container/action.yml b/devops/actions/build_container/action.yml
@@ -21,6 +21,9 @@ inputs:
   file:
     description: "Dockerfile"
     required: true
+  sycl_ci_passwd:
+    description: "Password to assign to sycl_ci user within a container"
+    required: true
 
 runs:
   using: "composite"
@@ -41,3 +44,4 @@ runs:
       file: ${{ github.workspace }}/devops/containers/${{ inputs.file }}.Dockerfile
       secrets: |
         github_token=${{ github.token }}
+        sycl_ci_passwd=${{ inputs.sycl_ci_passwd }}
diff --git a/devops/containers/ubuntu2404_build_oneapi.Dockerfile b/devops/containers/ubuntu2404_build_oneapi.Dockerfile
@@ -0,0 +1,53 @@
+FROM nvidia/cuda:12.6.3-devel-ubuntu24.04
+
+ENV DEBIAN_FRONTEND=noninteractive
+
+USER root
+
+# Install SYCL prerequisites
+COPY scripts/install_build_tools.sh /install.sh
+RUN /install.sh
+
+SHELL ["/bin/bash", "-ec"]
+
+# Install oneAPI
+
+# Make the directory if it doesn't exist yet.
+# This location is recommended by the distribution maintainers.
+RUN mkdir --parents --mode=0755 /etc/apt/keyrings
+# Download the key, convert the signing-key to a full
+# keyring required by apt and store in the keyring directory
+RUN wget https://repo.radeon.com/rocm/rocm.gpg.key -O - | \
+gpg --dearmor | tee /etc/apt/keyrings/rocm.gpg > /dev/null && \
+# Add rocm repo
+echo "deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/amdgpu/6.3/ubuntu noble main" \
+    | tee /etc/apt/sources.list.d/amdgpu.list && \
+echo "deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/rocm/apt/6.3 noble main" \
+    |  tee --append /etc/apt/sources.list.d/rocm.list && \
+echo -e 'Package: *\nPin: release o=repo.radeon.com\nPin-Priority: 600' \
+    | tee /etc/apt/preferences.d/rocm-pin-600 && \
+echo -e 'Package: *\nPin: release o=repo.radeon.com\nPin-Priority: 600' \
+    | tee /etc/apt/preferences.d/rocm-pin-600 && \
+wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor \
+| tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null && \
+echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" \
+| tee /etc/apt/sources.list.d/oneAPI.list && \
+apt update
+# Install the ROCM kernel driver and oneAPI
+RUN apt install -yqq rocm-dev intel-oneapi-compiler-dpcpp-cpp && \
+  apt-get clean && \
+  rm -rf /var/lib/apt/lists/*
+
+# By default Ubuntu sets an arbitrary UID value, that is different from host
+# system. When CI passes default UID value of 1001, some of LLVM tools fail to
+# discover user home directory and fail a few LIT tests. Fixes UID and GID to
+# 1001, that is used as default by GitHub Actions.
+RUN groupadd -g 1001 sycl && useradd sycl -u 1001 -g 1001 -m -s /bin/bash
+# Add sycl user to video/irc groups so that it can access GPU
+RUN usermod -aG video sycl
+RUN usermod -aG irc sycl
+
+COPY scripts/docker_entrypoint.sh /docker_entrypoint.sh
+
+ENTRYPOINT ["/docker_entrypoint.sh"]
+
diff --git a/devops/dependencies-igc-dev.json b/devops/dependencies-igc-dev.json
@@ -1,10 +1,10 @@
 {
   "linux": {
     "igc_dev": {
-      "github_tag": "igc-dev-3db59df",
-      "version": "3db59df",
-      "updated_at": "2024-12-03T20:43:00Z",
-      "url": "https://api.github.com/repos/intel/intel-graphics-compiler/actions/artifacts/2248119261/zip",
+      "github_tag": "igc-dev-e0d826a",
+      "version": "e0d826a",
+      "updated_at": "2024-12-17T21:18:30Z",
+      "url": "https://api.github.com/repos/intel/intel-graphics-compiler/actions/artifacts/2327583926/zip",
       "root": "{DEPS_ROOT}/opencl/runtime/linux/oclgpu"
     }
   }

diff --git a/devops/scripts/install_drivers.sh b/devops/scripts/install_drivers.sh
@@ -162,6 +162,9 @@ InstallIGFX () {
     echo "Install libopencl-clang"
     # Workaround only, will download deb and install with dpkg once fixed.
     cp -d libopencl-clang.so.14*  /usr/local/lib/
+    rm /usr/local/lib/libigc.so /usr/local/lib/libigc.so.1* && \
+       ln -s /usr/local/lib/libigc.so.2 /usr/local/lib/libigc.so && \
+       ln -s /usr/local/lib/libigc.so.2 /usr/local/lib/libigc.so.1
     echo "Clean up"
     rm *.deb libopencl-clang.so.14*
     echo "$IGC_DEV_TAG" > /usr/local/lib/igc/IGCTAG.txt

diff --git a/llvm/include/llvm/SYCLLowerIR/DeviceConfigFile.td b/llvm/include/llvm/SYCLLowerIR/DeviceConfigFile.td
@@ -189,11 +189,26 @@ class IntelTargetInfo<string Name, list<Aspect> Aspects, list<int> subGroupSizes
 // Note: only the "canonical" target names are listed here - see
 // SYCL::gen::resolveGenDevice().
 //
-// TODO: instructions on how to get "???" values below are required. Currently
-// device architectures below which have "???" in it are not fully supported.
-//
-// def : IntelTargetInfo<"intel_gpu_ptl_u", ???, ???>;
-// def : IntelTargetInfo<"intel_gpu_ptl_h", ???, ???>;
+// When adding the new Intel architectures, the ideal way to get these 
+// values is to run sycl-ls --verbose on the target machine. However,
+// if you don't have access to the target machine, then you can look in
+// the intel/compute-runtime source code to get the values. For 
+// the fp64 and atomic64 aspects, look for the definition 
+// of specific RuntimeCapabilityTable's. These will be defined in 
+// a file called hw_info_<device>.cpp. For example, for PVC, it 
+// would be hw_info_pvc.cpp. For fp16, this is supported by all 
+// Intel GPUs currently. (see definition of getDeviceExtensions in 
+// compiler_product_helper_base.inl). For the supported sub-group sizes,
+// it is not as straightforward. They are defined per family in 
+// GfxCoreHelperHw<Family>::getDeviceSubGroupSizes and by default
+// the default sub-group sizes are 8, 16, and 32. (see gfx_core_helper_base.inl)
+// However, this can be overriden by specific families. For example,
+// gfx_core_helper_xe_hpc_core.cpp defines the supported sub-group 
+// sizes to be 16 and 32, which corresponds to PVC having 16 and 32 
+// as sub-group sizes the supported sub-group sizes. 
+
+def : IntelTargetInfo<"intel_gpu_ptl_u", Fp16Fp64Atomic64, Sg16_32>;
+def : IntelTargetInfo<"intel_gpu_ptl_h", Fp16Fp64Atomic64, Sg16_32>;
 def : IntelTargetInfo<"intel_gpu_lnl_m", Fp16Fp64Atomic64, Sg16_32>;
 def : IntelTargetInfo<"intel_gpu_bmg_g21", Fp16Fp64Atomic64, Sg16_32>;
 def : IntelTargetInfo<"intel_gpu_arl_h", Fp16Fp64Atomic64, Sg8_16_32>;

diff --git a/sycl/cmake/modules/UnifiedRuntimeTag.cmake b/sycl/cmake/modules/UnifiedRuntimeTag.cmake
@@ -1,7 +1,7 @@
-# commit b7047f6c36ec17b8560c2f1cd9ac9521715a9127
-# Merge: 73e5f3c6ff2d fcddf077c290
-# Author: Martin Grant <martin.morrisongrant@codeplay.com>
-# Date:   Fri Dec 13 14:20:15 2024 +0000
-#     Merge pull request #2454 from Bensuo/l0_cmd-buf_multi-device
-#     Fix L0 command-buffer consumption of multi-device kernels
-set(UNIFIED_RUNTIME_TAG b7047f6c36ec17b8560c2f1cd9ac9521715a9127)
+# commit 39df0317814c164f5242eda8d6f08550f6268492
+# Merge: 68d93efd be27d8f0
+# Author: Kenneth Benzie (Benie) <k.benzie@codeplay.com>
+# Date:   Mon Dec 16 13:53:13 2024 +0000
+#     Merge pull request #2467 from nrspruit/fix_external_import_function_call
+#     [L0] Fix external semaphore import function calls to match the header
+set(UNIFIED_RUNTIME_TAG 39df0317814c164f5242eda8d6f08550f6268492)
diff --git a/sycl/doc/EnvironmentVariables.md b/sycl/doc/EnvironmentVariables.md
@@ -196,7 +196,6 @@ variables in production code.</span>
 
 | Environment variable | Values | Description |
 | -------------------- | ------ | ----------- |
-| `SYCL_PREFER_UR` | Integer | If non-0 then run through Unified Runtime if desired backend is supported there. Default is 0.  |
 | `SYCL_UR_TRACE` | Integer | Described [below](#sycl_ur_trace-options)  | Enable specified level of tracing for UR. |
 | `SYCL_QUEUE_THREAD_POOL_SIZE` | Positive integer | Number of threads in thread pool of queue. |
 | `SYCL_DEVICELIB_NO_FALLBACK` | Any(\*) | Disable loading and linking of device library images |

diff --git a/sycl/include/sycl/detail/type_traits.hpp b/sycl/include/sycl/detail/type_traits.hpp
@@ -384,6 +384,22 @@ struct map_type<T, From, To, Rest...> {
 template <typename T, typename... Ts>
 constexpr bool check_type_in_v = ((std::is_same_v<T, Ts> || ...));
 
+#if __has_builtin(__type_pack_element)
+template <int N, typename... Ts>
+using nth_type_t = __type_pack_element<N, Ts...>;
+#else
+template <int N, typename T, typename... Ts> struct nth_type {
+  using type = typename nth_type<N - 1, Ts...>::type;
+};
+
+template <typename T, typename... Ts> struct nth_type<0, T, Ts...> {
+  using type = T;
+};
+
+template <int N, typename... Ts>
+using nth_type_t = typename nth_type<N, Ts...>::type;
+#endif
+
 } // namespace detail
 } // namespace _V1
 } // namespace sycl