Merge branch 'main' into cuda-malloc-managed-optimize

Author: JackAKirk

.github/workflows/benchmarks_core.yml renamed to .github/workflows/benchmarks_compute.yml

@@ -1,60 +1,60 @@
 name: Compute Benchmarks
 
 on:
-  # this workflow can by only triggered by other workflows
-  # for example by: e2e_cuda.yml or e2e_opencl.yml
-  workflow_call:
-    # acceptable input from adapter-specific workflows
+  # Can be triggered via manual "dispatch" (from workflow view in GitHub Actions tab)
+  workflow_dispatch:
+    # acceptable input for adapter-specific runs
     inputs:
-      name:
-        description: Adapter name
-        type: string
-        required: true
       str_name:
         description: Formatted adapter name
-        type: string
-        required: true
-      config:
-        description: Params for sycl configuration
-        type: string
+        type: choice
         required: true
+        default: 'level_zero'
+        options:
+          - level_zero
       unit:
         description: Test unit (cpu/gpu)
-        type: string
+        type: choice
         required: true
-      runner_tag:
-        description: Tag defined for the runner
-        type: string
+        default: 'gpu'
+        options:
+          - cpu
+          - gpu
+      pr_no:
+        description: PR number (if 0, it'll run on the main)
+        type: number
         required: true
-      trigger:
-        description: Type of workflow trigger
+      bench_script_params:
+        description: Parameters passed to script executing benchmark
         type: string
-        required: true
-      comment:
-        description: Text if triggered by a comment
+        required: false
+        default: ''
+      sycl_config_params:
+        description: Extra params for SYCL configuration
         type: string
         required: false
+        default: ''
 
 permissions:
   contents: read
   pull-requests: write
 
 jobs:
   e2e-build-hw:
-    if: github.repository == 'oneapi-src/unified-runtime'  # run only on upstream; forks will not have the HW
+    # Run only on upstream; forks will not have the HW
+    if: github.repository == 'oneapi-src/unified-runtime'
     name: Build SYCL, UR, run Compute Benchmarks
     strategy:
       matrix:
         adapter: [
-          {name: "${{inputs.name}}",
-          str_name: "${{inputs.str_name}}",
-          config: "${{inputs.config}}",
+          {str_name: "${{inputs.str_name}}",
+          sycl_config: "${{inputs.sycl_config_params}}",
           unit: "${{inputs.unit}}"}
         ]
         build_type: [Release]
         compiler: [{c: clang, cxx: clang++}]
 
-    runs-on: ${{inputs.runner_tag}}
+    runs-on: "${{inputs.str_name}}_PERF"
 
     steps:
     # Workspace on self-hosted runners is not cleaned automatically.
@@ -67,15 +67,17 @@ jobs:
 
     - name: Add comment to PR
       uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
-      if: ${{ always() && inputs.trigger != 'schedule' }}
+      if: ${{ always() && inputs.pr_no != 0 }}
       with:
         script: |
-          const adapter = '${{ matrix.adapter.name }}';
+          const pr_no = '${{ inputs.pr_no }}';
+          const adapter = '${{ matrix.adapter.str_name }}';
           const url = '${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}';
-          const body = `Compute Benchmarks ${adapter} run: \n${url}`;
+          const params = '${{ inputs.bench_script_params }}';
+          const body = `Compute Benchmarks ${adapter} run (with params: ${params}):\n${url}`;
 
           github.rest.issues.createComment({
-            issue_number: context.issue.number,
+            issue_number: pr_no,
             owner: context.repo.owner,
             repo: context.repo.repo,
             body: body
@@ -86,13 +88,12 @@ jobs:
       with:
         path: ur-repo
 
-    # On issue_comment trigger (for PRs) we need to fetch special ref for
-    # proper PR's merge commit. Note, this ref may be absent if the PR is already merged.
+    # We need to fetch special ref for proper PR's merge commit. Note, this ref may be absent if the PR is already merged.
     - name: Fetch PR's merge commit
-      if: ${{ inputs.trigger != 'schedule' }}
+      if: ${{ inputs.pr_no != 0 }}
       working-directory: ${{github.workspace}}/ur-repo
       env:
-        PR_NO: ${{github.event.issue.number}}
+        PR_NO: ${{ inputs.pr_no }}
       run: |
         git fetch -- https://github.com/${{github.repository}} +refs/pull/${PR_NO}/*:refs/remotes/origin/pr/${PR_NO}/*
         git checkout origin/pr/${PR_NO}/merge
@@ -108,7 +109,7 @@ jobs:
         fetch-tags: false
 
     - name: Set CUDA env vars
-      if: matrix.adapter.name == 'CUDA'
+      if: matrix.adapter.str_name == 'cuda'
       run: |
         echo "CUDA_LIB_PATH=/usr/local/cuda/lib64/stubs" >> $GITHUB_ENV
         echo "LD_LIBRARY_PATH=/usr/local/cuda/compat/:/usr/local/cuda/lib64:$LD_LIBRARY_PATH" >> $GITHUB_ENV
@@ -119,7 +120,7 @@ jobs:
         -t ${{matrix.build_type}}
         -o ${{github.workspace}}/sycl_build
         --cmake-gen "Ninja"
-        --ci-defaults ${{matrix.adapter.config}}
+        --ci-defaults ${{matrix.adapter.sycl_config}}
         --cmake-opt="-DLLVM_INSTALL_UTILS=ON"
         --cmake-opt="-DSYCL_PI_TESTS=OFF"
         --cmake-opt="-DSYCL_PI_UR_USE_FETCH_CONTENT=OFF"
@@ -165,17 +166,13 @@ jobs:
       run: |
         echo "ONEAPI_DEVICE_SELECTOR=${{ matrix.adapter.str_name }}:${{ matrix.adapter.unit }}" >> $GITHUB_ENV
 
-    - name: Extract arguments from comment
-      id: args
-      run: echo "ARGS=$(echo '${{ inputs.comment }}' | sed -n 's/.*\/benchmarks-[^ ]* \(.*\)/\1/p')" >> $GITHUB_ENV
-
     - name: Run SYCL API Overhead benchmark
       id: benchmarks
-      run: ${{github.workspace}}/ur-repo/.github/scripts/compute_benchmarks.py ${{github.workspace}}/compute-benchmarks-build/bin/ $ARGS
+      run: ${{ github.workspace }}/ur-repo/.github/scripts/compute_benchmarks.py ${{ github.workspace }}/compute-benchmarks-build/bin/ ${{ inputs.bench_script_params }}
 
     - name: Add comment to PR
       uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
-      if: ${{ always() && inputs.trigger != 'schedule' }}
+      if: ${{ always() && inputs.pr_no != 0 }}
       with:
         script: |
           let markdown = ""
@@ -185,14 +182,16 @@ jobs:
           } catch(err) {
           }
 
-          const adapter = '${{ matrix.adapter.name }}';
+          const pr_no = '${{ inputs.pr_no }}';
+          const adapter = '${{ matrix.adapter.str_name }}';
           const url = '${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}';
           const test_status = '${{ steps.benchmarks.outcome }}';
           const job_status = '${{ job.status }}';
-          const body = `Compute Benchmarks ${adapter} run:\n${url}\nJob status: ${job_status}. Test status: ${test_status}.\n ${markdown}`;
+          const params = '${{ inputs.bench_script_params }}';
+          const body = `Compute Benchmarks ${adapter} run (with params: ${params}):\n${url}\nJob status: ${job_status}. Test status: ${test_status}.\n ${markdown}`;
 
           github.rest.issues.createComment({
-            issue_number: context.issue.number,
+            issue_number: pr_no,
             owner: context.repo.owner,
             repo: context.repo.repo,
             body: body

.github/workflows/benchmarks_level_zero.yml

@@ -8,6 +8,7 @@ concurrency:
 
 permissions:
   contents: read
+  pull-requests: write
 
 jobs:
   ubuntu-build:
@@ -230,6 +231,9 @@ jobs:
            adapter: {name: L0, var: '-DUR_BUILD_ADAPTER_L0=ON'}
          - adapter: {name: L0, var: '-DUR_BUILD_ADAPTER_L0=ON'}
            compiler: {c: clang-cl, cxx: clang-cl}
+         # TODO: testing is flaky on windows-2022 in Release mode
+         - os: 'windows-2022'
+           build_type: Release
 
         build_type: [Debug, Release]
         compiler: [{c: cl, cxx: cl}, {c: clang-cl, cxx: clang-cl}]

.github/workflows/cmake.yml

@@ -49,6 +49,7 @@ on:
 
 permissions:
   contents: read
+  pull-requests: write
 
 jobs:
   changed-files:
@@ -177,20 +178,23 @@ jobs:
       id: tests
       run: ninja -C build-e2e check-sycl-e2e
 
-    - name: Add comment to PR
-      uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
-      if: ${{ always() }}
-      with:
-        script: |
-          const adapter = '${{ matrix.adapter.name }}';
-          const url = '${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}';
-          const test_status = '${{ steps.tests.outcome }}';
-          const job_status = '${{ job.status }}';
-          const body = `E2E ${adapter} build:\n${url}\nJob status: ${job_status}. Test status: ${test_status}`;
-
-          github.rest.issues.createComment({
-            issue_number: context.issue.number,
-            owner: context.repo.owner,
-            repo: context.repo.repo,
-            body: body
-          })
+    # FIXME: Requires pull-request: write permissions but this is only granted
+    # on pull requests from forks if using pull_request_target workflow
+    # trigger but not the pull_request trigger..
+    # - name: Add comment to PR
+    #   uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+    #   if: ${{ always() }}
+    #   with:
+    #     script: |
+    #       const adapter = '${{ matrix.adapter.name }}';
+    #       const url = '${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}';
+    #       const test_status = '${{ steps.tests.outcome }}';
+    #       const job_status = '${{ job.status }}';
+    #       const body = `E2E ${adapter} build:\n${url}\nJob status: ${job_status}. Test status: ${test_status}`;
+
+    #       github.rest.issues.createComment({
+    #         issue_number: context.issue.number,
+    #         owner: context.repo.owner,
+    #         repo: context.repo.repo,
+    #         body: body
+    #       })

.github/workflows/e2e_core.yml

@@ -0,0 +1,26 @@
+# Copyright (C) 2024 Intel Corporation
+# Part of the Unified-Runtime Project, under the Apache License v2.0 with LLVM Exceptions.
+# See LICENSE.TXT
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+# This is lifted from intel-llvm's FindCUDACupti implementation
+# https://github.com/intel/llvm/blob/0cd04144d9ca83371c212e8e4709a59c968291b9/sycl/cmake/modules/FindCUDACupti.cmake
+
+macro(find_cuda_cupti_library)
+  find_library(CUDA_cupti_LIBRARY
+    NAMES cupti
+    HINTS ${CUDA_TOOLKIT_ROOT_DIR}
+          ENV CUDA_PATH
+    PATH_SUFFIXES nvidia/current lib64 lib/x64 lib
+                  ../extras/CUPTI/lib64/
+                  ../extras/CUPTI/lib/
+  )
+endmacro()
+
+macro(find_cuda_cupti_include_dir)
+  find_path(CUDA_CUPTI_INCLUDE_DIR cupti.h PATHS
+      "${CUDA_TOOLKIT_ROOT_DIR}/extras/CUPTI/include"
+      "${CUDA_INCLUDE_DIRS}/../extras/CUPTI/include"
+      "${CUDA_INCLUDE_DIRS}"
+      NO_DEFAULT_PATH)
+endmacro()