Skip to content

attention_op_microbenchmark #24

attention_op_microbenchmark

attention_op_microbenchmark #24

name: attention_op_microbenchmark
on:
push:
tags:
- ciflow/op-benchmark/*
workflow_dispatch:
schedule:
# Run at 06:00 UTC everyday
- cron: 0 7 * * *
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:
id-token: write
contents: read
jobs:
attn-microbenchmark-build:
if: github.repository_owner == 'pytorch'
uses: ./.github/workflows/_linux-build.yml
with:
runner: linux.12xlarge.memory
build-environment: linux-jammy-cuda12.8-py3.10-gcc9-sm80
docker-image-name: ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11
cuda-arch-list: '8.0 9.0'
test-matrix: |
{ include: [
{ config: "attention_microbenchmark_test", shard: 1, num_shards: 1, runner: "linux.aws.a100" },
{ config: "attention_microbenchmark_test", shard: 1, num_shards: 1, runner: "linux.aws.h100" },
]}
secrets: inherit
attn-microbenchmark-test:
name: attn-microbenchmark-test
uses: ./.github/workflows/_linux-test.yml
needs: attn-microbenchmark-build
with:
timeout-minutes: 500
build-environment: linux-jammy-cuda12.8-py3.10-gcc9-sm80
docker-image: ${{ needs.attn-microbenchmark-build.outputs.docker-image }}
test-matrix: ${{ needs.attn-microbenchmark-build.outputs.test-matrix }}
secrets: inherit
# B200 runner
opmicrobenchmark-build-b200:
if: github.repository_owner == 'pytorch'
name: opmicrobenchmark-build-b200
uses: ./.github/workflows/_linux-build.yml
with:
runner: linux.12xlarge.memory
build-environment: linux-jammy-cuda12.8-py3.10-gcc9-sm100
docker-image-name: ci-image:pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc11
cuda-arch-list: '10.0'
test-matrix: |
{ include: [
{ config: "operator_microbenchmark_test", shard: 1, num_shards: 1, runner: "linux.dgx.b200" },
]}
secrets: inherit
opmicrobenchmark-test-b200:
name: opmicrobenchmark-test-b200
uses: ./.github/workflows/_linux-test.yml
needs: opmicrobenchmark-build-b200
with:
timeout-minutes: 500
build-environment: linux-jammy-cuda12.8-py3.10-gcc9-sm100
docker-image: ${{ needs.opmicrobenchmark-build-b200.outputs.docker-image }}
test-matrix: ${{ needs.opmicrobenchmark-build-b200.outputs.test-matrix }}
aws-role-to-assume: arn:aws:iam::308535385114:role/gha_workflow_s3_and_ecr_read_only
secrets: inherit