pytorch · charan-ponnada · May 16, 2025
@@ -596,7 +596,9 @@ test_perf_for_dashboard() {
 
   local device=cuda
   if [[ "${TEST_CONFIG}" == *cpu* ]]; then
-    if [[ "${TEST_CONFIG}" == *cpu_x86* ]]; then
+    if [[ "${TEST_CONFIG}" == *zen_cpu_x86* ]]; then
+      device=zen_cpu_x86
+    elif [[ "${TEST_CONFIG}" == *cpu_x86* ]]; then
       device=cpu_x86
     elif [[ "${TEST_CONFIG}" == *cpu_aarch64* ]]; then
       device=cpu_aarch64

@@ -14,6 +14,7 @@ self-hosted-runner:
     - linux.12xlarge
     - linux.24xlarge
     - linux.24xlarge.ephemeral
+    - linux.24xlarge.amd
     - linux.arm64.2xlarge
     - linux.arm64.2xlarge.ephemeral
     - linux.arm64.m7g.4xlarge

@@ -11,6 +11,7 @@ ciflow_push_tags:
 - ciflow/inductor-perf-compare
 - ciflow/inductor-micro-benchmark
 - ciflow/inductor-micro-benchmark-cpu-x86
+- ciflow/inductor-perf-test-nightly-x86-zen
 - ciflow/inductor-cu126
 - ciflow/linux-aarch64
 - ciflow/mps

@@ -0,0 +1,129 @@
+name: inductor-perf-nightly-x86-zen
+
+on:
+  push:
+    tags:
+      - ciflow/inductor-perf-test-nightly-x86-zen/*
+  schedule:
+    # - cron: 0 7 * * 1-6
+    # - cron: 0 7 * * 0
+    # Does not perform max_autotune on CPU, so skip the weekly run setup
+    - cron: 0 7 * * *
+  # NB: GitHub has an upper limit of 10 inputs here
+  workflow_dispatch:
+    inputs:
+      training:
+        # CPU for training is not typical, but leave the option open here
+        description: Run training (off by default)?
+        required: false
+        type: boolean
+        default: false
+      inference:
+        description: Run inference (on by default)?
+        required: false
+        type: boolean
+        default: true
+      default:
+        description: Run inductor_default?
+        required: false
+        type: boolean
+        default: true
+      dynamic:
+        description: Run inductor_dynamic_shapes?
+        required: false
+        type: boolean
+        default: false
+      cppwrapper:
+        description: Run inductor_cpp_wrapper?
+        required: false
+        type: boolean
+        default: false
+      aotinductor:
+        description: Run aot_inductor for inference?
+        required: false
+        type: boolean
+        default: false
+      benchmark_configs:
+        description: The list of configs used the benchmark
+        required: false
+        type: string
+        default: inductor_huggingface_perf_zen_cpu_x86,inductor_timm_perf_zen_cpu_x86,inductor_torchbench_perf_zen_cpu_x86
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref_name }}-${{ github.ref_type == 'branch' && github.sha }}-${{ github.event_name == 'workflow_dispatch' }}-${{ github.event_name == 'schedule' }}
+  cancel-in-progress: true
+
+permissions: read-all
+
+jobs:
+  get-label-type:
+    name: get-label-type
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    if: ${{ (github.event_name != 'schedule' || github.repository == 'pytorch/pytorch') && github.repository_owner == 'pytorch' }}
+    with:
+      triggering_actor: ${{ github.triggering_actor }}
+      issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
+      curr_branch: ${{ github.head_ref || github.ref_name }}
+      curr_ref_type: ${{ github.ref_type }}
+      opt_out_experiments: lf
+
+  linux-jammy-zen-cpu-py3_9-gcc11-inductor-build:
+    name: linux-jammy-zen-cpu-py3.9-gcc11-inductor
+    uses: ./.github/workflows/_linux-build.yml
+    needs: get-label-type
+    with:
+      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
+      build-environment: linux-jammy-py3.9-gcc11-build
+      docker-image-name: ci-image:pytorch-linux-jammy-py3.9-gcc11-inductor-benchmarks
+      test-matrix: |
+        { include: [
+          { config: "inductor_huggingface_perf_zen_cpu_x86", shard: 1, num_shards: 3, runner: "linux.24xlarge.amd" },
+          { config: "inductor_huggingface_perf_zen_cpu_x86", shard: 2, num_shards: 3, runner: "linux.24xlarge.amd" },
+          { config: "inductor_huggingface_perf_zen_cpu_x86", shard: 3, num_shards: 3, runner: "linux.24xlarge.amd" },
+          { config: "inductor_timm_perf_zen_cpu_x86", shard: 1, num_shards: 5, runner: "linux.24xlarge.amd" },
+          { config: "inductor_timm_perf_zen_cpu_x86", shard: 2, num_shards: 5, runner: "linux.24xlarge.amd" },
+          { config: "inductor_timm_perf_zen_cpu_x86", shard: 3, num_shards: 5, runner: "linux.24xlarge.amd" },
+          { config: "inductor_timm_perf_zen_cpu_x86", shard: 4, num_shards: 5, runner: "linux.24xlarge.amd" },
+          { config: "inductor_timm_perf_zen_cpu_x86", shard: 5, num_shards: 5, runner: "linux.24xlarge.amd" },
+          { config: "inductor_torchbench_perf_zen_cpu_x86", shard: 1, num_shards: 4, runner: "linux.24xlarge.amd" },
+          { config: "inductor_torchbench_perf_zen_cpu_x86", shard: 2, num_shards: 4, runner: "linux.24xlarge.amd" },
+          { config: "inductor_torchbench_perf_zen_cpu_x86", shard: 3, num_shards: 4, runner: "linux.24xlarge.amd" },
+          { config: "inductor_torchbench_perf_zen_cpu_x86", shard: 4, num_shards: 4, runner: "linux.24xlarge.amd" },
+        ]}
+      selected-test-configs: ${{ inputs.benchmark_configs }}
+    secrets: inherit
+
+  linux-jammy-zen-cpu-py3_9-gcc11-inductor-test-nightly:
+    name: linux-jammy-zen-cpu-py3.9-gcc11-inductor
+    uses: ./.github/workflows/_linux-test.yml
+    needs: linux-jammy-zen-cpu-py3_9-gcc11-inductor-build
+    if: github.event.schedule == '0 7 * * *'
+    with:
+      build-environment: linux-jammy-py3.9-gcc11-build
+      dashboard-tag: training-false-inference-true-default-true-dynamic-true-cppwrapper-true-aotinductor-true
+      docker-image: ${{ needs.linux-jammy-zen-cpu-py3_9-gcc11-inductor-build.outputs.docker-image }}
+      test-matrix: ${{ needs.linux-jammy-zen-cpu-py3_9-gcc11-inductor-build.outputs.test-matrix }}
+      timeout-minutes: 720
+      # disable monitor in perf tests
+      disable-monitor: false
+      monitor-log-interval: 15
+      monitor-data-collect-interval: 4
+    secrets: inherit
+
+
+  linux-jammy-zen-cpu-py3_9-gcc11-inductor-test:
+    name: linux-jammy-zen-cpu-py3.9-gcc11-inductor
+    uses: ./.github/workflows/_linux-test.yml
+    needs: linux-jammy-zen-cpu-py3_9-gcc11-inductor-build
+    if: github.event_name == 'workflow_dispatch'
+    with:
+      build-environment: linux-jammy-py3.9-gcc11-build
+      dashboard-tag: training-${{ inputs.training }}-inference-${{ inputs.inference }}-default-${{ inputs.default }}-dynamic-${{ inputs.dynamic }}-cppwrapper-${{ inputs.cppwrapper }}-aotinductor-${{ inputs.aotinductor }}
+      docker-image: ${{ needs.linux-jammy-zen-cpu-py3_9-gcc11-inductor-build.outputs.docker-image }}
+      test-matrix: ${{ needs.linux-jammy-zen-cpu-py3_9-gcc11-inductor-build.outputs.test-matrix }}
+      timeout-minutes: 720
+      # disable monitor in perf tests
+      disable-monitor: false
+      monitor-log-interval: 15
+      monitor-data-collect-interval: 4
+    secrets: inherit