frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 12:54:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

[ROCm][CI] add mi355 to inductor perf test nightly (#165326)

Browse Source 
Fixes #ISSUE_NUMBER

Pull Request resolved: https://github.com/pytorch/pytorch/pull/165326
Approved by: https://github.com/jeffdaily

Co-authored-by: Jeff Daily <jeff.daily@amd.com>
This commit is contained in:
Jeff Daily
2025-10-14 20:03:21 +00:00
committed by PyTorch MergeBot
parent 08f09d9543
commit d7e3f493d9
3 changed files with 159 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
.github/pytorch-probot.yml vendored
View File

@ -15,7 +15,8 @@ ciflow_push_tags:
- ciflow/inductor-micro-benchmark
- ciflow/inductor-micro-benchmark-cpu-x86
- ciflow/inductor-perf-compare
- ciflow/inductor-perf-test-nightly-rocm
- ciflow/inductor-perf-test-nightly-rocm-mi300
- ciflow/inductor-perf-test-nightly-rocm-mi355
- ciflow/inductor-perf-test-nightly-x86-zen
- ciflow/inductor-periodic
- ciflow/inductor-rocm

132
.github/workflows/inductor-perf-test-nightly-rocm-mi300.yml vendored Normal file
View File

@ -0,0 +1,132 @@
name: inductor-perf-nightly-rocm-mi300
on:
push:
tags:
- ciflow/inductor-perf-test-nightly-rocm-mi300/*
schedule:
- cron: 15 0 * * *
# NB: GitHub has an upper limit of 10 inputs here, so before we can sort it
# out, let try to run torchao cudagraphs_low_precision as part of cudagraphs
workflow_dispatch:
inputs:
training:
description: Run training (on by default)?
required: false
type: boolean
default: true
inference:
description: Run inference (on by default)?
required: false
type: boolean
default: true
default:
description: Run inductor_default?
required: false
type: boolean
default: false
dynamic:
description: Run inductor_dynamic_shapes?
required: false
type: boolean
default: false
cppwrapper:
description: Run inductor_cpp_wrapper?
required: false
type: boolean
default: false
cudagraphs:
description: Run inductor_cudagraphs?
required: false
type: boolean
default: true
freezing_cudagraphs:
description: Run inductor_cudagraphs with freezing for inference?
required: false
type: boolean
default: false
aotinductor:
description: Run aot_inductor for inference?
required: false
type: boolean
default: false
maxautotune:
description: Run inductor_max_autotune?
required: false
type: boolean
default: false
benchmark_configs:
description: The list of configs used the benchmark
required: false
type: string
default: inductor_huggingface_perf_rocm_mi300,inductor_timm_perf_rocm_mi300,inductor_torchbench_perf_rocm_mi300
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-rocm-py3_10-inductor-benchmark-build:
if: github.repository_owner == 'pytorch'
name: rocm-py3_10-inductor-benchmark-build
uses: ./.github/workflows/_linux-build.yml
with:
build-environment: linux-jammy-rocm-py3_10
docker-image-name: ci-image:pytorch-linux-jammy-rocm-n-py3-benchmarks
test-matrix: |
{ include: [
{ config: "inductor_huggingface_perf_rocm_mi300", shard: 1, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm_mi300", shard: 2, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm_mi300", shard: 3, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm_mi300", shard: 4, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm_mi300", shard: 5, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm_mi300", shard: 1, num_shards: 7, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm_mi300", shard: 2, num_shards: 7, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm_mi300", shard: 3, num_shards: 7, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm_mi300", shard: 4, num_shards: 7, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm_mi300", shard: 5, num_shards: 7, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm_mi300", shard: 6, num_shards: 7, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm_mi300", shard: 7, num_shards: 7, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 1, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 2, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 3, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 4, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 5, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 6, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 7, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 8, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm_mi300", shard: 9, num_shards: 9, runner: "linux.rocm.gpu.gfx942.1" },
]}
secrets: inherit
linux-jammy-rocm-py3_10-inductor-benchmark-test:
permissions:
id-token: write
contents: read
name: rocm-py3_10-inductor-benchmark-test
uses: ./.github/workflows/_rocm-test.yml
needs: linux-jammy-rocm-py3_10-inductor-benchmark-build
with:
build-environment: linux-jammy-rocm-py3_10
dashboard-tag: training-true-inference-true-default-true-dynamic-true-cudagraphs-true-cppwrapper-true-aotinductor-true-freezing_cudagraphs-true-cudagraphs_low_precision-true
docker-image: ${{ needs.linux-jammy-rocm-py3_10-inductor-benchmark-build.outputs.docker-image }}
test-matrix: ${{ needs.linux-jammy-rocm-py3_10-inductor-benchmark-build.outputs.test-matrix }}
timeout-minutes: 720
# Disable monitor in perf tests for more investigation
disable-monitor: true
monitor-log-interval: 10
monitor-data-collect-interval: 2
secrets: inherit

46
.github/workflows/inductor-perf-test-nightly-rocm.yml → .github/workflows/inductor-perf-test-nightly-rocm-mi355.yml vendored
View File

@ -1,11 +1,11 @@
name: inductor-perf-nightly-rocm
name: inductor-perf-nightly-rocm-mi355
on:
push:
tags:
- ciflow/inductor-perf-test-nightly-rocm/*
- ciflow/inductor-perf-test-nightly-rocm-mi355/*
schedule:
- cron: 0 7 * * 0,3
- cron: 15 0 * * *
# NB: GitHub has an upper limit of 10 inputs here, so before we can sort it
# out, let try to run torchao cudagraphs_low_precision as part of cudagraphs
workflow_dispatch:
@ -59,7 +59,7 @@ on:
description: The list of configs used the benchmark
required: false
type: string
default: inductor_huggingface_perf_rocm,inductor_timm_perf_rocm,inductor_torchbench_perf_rocm
default: inductor_huggingface_perf_rocm_mi355,inductor_timm_perf_rocm_mi355,inductor_torchbench_perf_rocm_mi355
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' }}
@ -88,23 +88,27 @@ jobs:
docker-image-name: ci-image:pytorch-linux-jammy-rocm-n-py3-benchmarks
test-matrix: |
{ include: [
{ config: "inductor_huggingface_perf_rocm", shard: 1, num_shards: 4, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm", shard: 2, num_shards: 4, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm", shard: 3, num_shards: 4, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm", shard: 4, num_shards: 4, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm", shard: 1, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm", shard: 2, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm", shard: 3, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm", shard: 4, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_timm_perf_rocm", shard: 5, num_shards: 5, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 1, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 2, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 3, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 4, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 5, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 6, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 7, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_torchbench_perf_rocm", shard: 8, num_shards: 8, runner: "linux.rocm.gpu.gfx942.1" },
{ config: "inductor_huggingface_perf_rocm_mi355", shard: 1, num_shards: 5, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_huggingface_perf_rocm_mi355", shard: 2, num_shards: 5, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_huggingface_perf_rocm_mi355", shard: 3, num_shards: 5, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_huggingface_perf_rocm_mi355", shard: 4, num_shards: 5, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_huggingface_perf_rocm_mi355", shard: 5, num_shards: 5, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_timm_perf_rocm_mi355", shard: 1, num_shards: 7, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_timm_perf_rocm_mi355", shard: 2, num_shards: 7, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_timm_perf_rocm_mi355", shard: 3, num_shards: 7, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_timm_perf_rocm_mi355", shard: 4, num_shards: 7, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_timm_perf_rocm_mi355", shard: 5, num_shards: 7, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_timm_perf_rocm_mi355", shard: 6, num_shards: 7, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_timm_perf_rocm_mi355", shard: 7, num_shards: 7, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 1, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 2, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 3, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 4, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 5, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 6, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 7, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 8, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
{ config: "inductor_torchbench_perf_rocm_mi355", shard: 9, num_shards: 9, runner: "linux.rocm.gpu.mi355.2" },
]}
secrets: inherit