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/CD] upgrade nightly wheels to ROCm 7.0 (#163937)

Browse Source 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/163937
Approved by: https://github.com/jeffdaily

Co-authored-by: Jeff Daily <jeff.daily@amd.com>
This commit is contained in:
Jeff Daily
2025-09-26 21:42:09 +00:00
committed by PyTorch MergeBot
parent 28c7d11428
commit f1260c9b9a
9 changed files with 907 additions and 909 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
.ci/docker/common/install_rocm_magma.sh
View File

@ -12,8 +12,8 @@ function do_install() {
rocm_version_nodot=${rocm_version//./}
# Version 2.7.2 + ROCm related updates
MAGMA_VERSION=a1625ff4d9bc362906bd01f805dbbe12612953f6
# https://github.com/icl-utk-edu/magma/pull/65
MAGMA_VERSION=d6e4117bc88e73f06d26c6c2e14f064e8fc3d1ec
magma_archive="magma-rocm${rocm_version_nodot}-${MAGMA_VERSION}-1.tar.bz2"
rocm_dir="/opt/rocm"

2
.github/scripts/generate_binary_build_matrix.py vendored
View File

@ -30,7 +30,7 @@ CUDA_ARCHES_CUDNN_VERSION = {
}
# NOTE: Please also update the ROCm sources in `PIP_SOURCES` in tools/nightly.py when changing this
ROCM_ARCHES = ["6.3", "6.4"]
ROCM_ARCHES = ["6.4", "7.0"]
XPU_ARCHES = ["xpu"]

1
.github/workflows/build-libtorch-images.yml vendored
View File

@ -52,7 +52,6 @@ jobs:
{ tag: "cuda12.9" },
{ tag: "cuda12.8" },
{ tag: "cuda12.6" },
{ tag: "rocm6.3" },
{ tag: "rocm6.4" },
{ tag: "rocm7.0" },
{ tag: "cpu" },

2
.github/workflows/build-magma-rocm-linux.yml vendored
View File

@ -34,7 +34,7 @@ jobs:
id-token: write
strategy:
matrix:
rocm_version: ["70", "64", "63"]
rocm_version: ["70", "64"]
steps:
- name: Checkout PyTorch
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2

1
.github/workflows/build-manywheel-images.yml vendored
View File

@ -52,7 +52,6 @@ jobs:
{ name: "manylinuxaarch64-builder", tag: "cuda13.0", runner: "linux.arm64.2xlarge.ephemeral" },
{ name: "manylinuxaarch64-builder", tag: "cuda12.8", runner: "linux.arm64.2xlarge.ephemeral" },
{ name: "manylinuxaarch64-builder", tag: "cuda12.6", runner: "linux.arm64.2xlarge.ephemeral" },
{ name: "manylinux2_28-builder", tag: "rocm6.3", runner: "linux.9xlarge.ephemeral" },
{ name: "manylinux2_28-builder", tag: "rocm6.4", runner: "linux.9xlarge.ephemeral" },
{ name: "manylinux2_28-builder", tag: "rocm7.0", runner: "linux.9xlarge.ephemeral" },
{ name: "manylinux2_28-builder", tag: "cpu", runner: "linux.9xlarge.ephemeral" },

2
.github/workflows/build-triton-wheel.yml vendored
View File

@ -55,7 +55,7 @@ jobs:
docker-image: ["pytorch/manylinux2_28-builder:cpu"]
include:
- device: "rocm"
rocm_version: "6.4"
rocm_version: "7.0"
runs_on: "${{ needs.get-label-type.outputs.label-type }}linux.4xlarge"
- device: "cuda"
rocm_version: ""

230
.github/workflows/generated-linux-binary-libtorch-nightly.yml generated vendored
View File

@ -316,121 +316,6 @@ jobs:
github-token: ${{ secrets.GITHUB_TOKEN }}
uses: ./.github/workflows/_binary-upload.yml
libtorch-rocm6_3-shared-with-deps-release-build:
if: ${{ github.repository_owner == 'pytorch' }}
uses: ./.github/workflows/_binary-build-linux.yml
needs: get-label-type
with:
PYTORCH_ROOT: /pytorch
PACKAGE_TYPE: libtorch
# TODO: This is a legacy variable that we eventually want to get rid of in
# favor of GPU_ARCH_VERSION
DESIRED_CUDA: rocm6.3
GPU_ARCH_VERSION: "6.3"
GPU_ARCH_TYPE: rocm
DOCKER_IMAGE: libtorch-cxx11-builder
DOCKER_IMAGE_TAG_PREFIX: rocm6.3
LIBTORCH_CONFIG: release
LIBTORCH_VARIANT: shared-with-deps
runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
timeout-minutes: 300
build_name: libtorch-rocm6_3-shared-with-deps-release
build_environment: linux-binary-libtorch
secrets:
github-token: ${{ secrets.GITHUB_TOKEN }}
libtorch-rocm6_3-shared-with-deps-release-test: # Testing
if: ${{ github.repository_owner == 'pytorch' }}
needs:
- libtorch-rocm6_3-shared-with-deps-release-build
- get-label-type
runs-on: linux.rocm.gpu.mi250
timeout-minutes: 240
env:
PYTORCH_ROOT: /pytorch
PACKAGE_TYPE: libtorch
# TODO: This is a legacy variable that we eventually want to get rid of in
# favor of GPU_ARCH_VERSION
DESIRED_CUDA: rocm6.3
GPU_ARCH_VERSION: "6.3"
GPU_ARCH_TYPE: rocm
SKIP_ALL_TESTS: 1
DOCKER_IMAGE: libtorch-cxx11-builder
DOCKER_IMAGE_TAG_PREFIX: rocm6.3
LIBTORCH_CONFIG: release
LIBTORCH_VARIANT: shared-with-deps
steps:
- name: Setup ROCm
uses: ./.github/actions/setup-rocm
- uses: actions/download-artifact@v4.1.7
name: Download Build Artifacts
with:
name: libtorch-rocm6_3-shared-with-deps-release
path: "${{ runner.temp }}/artifacts/"
- name: Checkout PyTorch
uses: actions/checkout@v4
with:
ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
submodules: recursive
path: pytorch
show-progress: false
- name: Clean PyTorch checkout
run: |
# Remove any artifacts from the previous checkouts
git clean -fxd
working-directory: pytorch
- name: ROCm set GPU_FLAG
run: |
echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
- name: configure aws credentials
id: aws_creds
if: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') }}
uses: aws-actions/configure-aws-credentials@v4
with:
role-to-assume: arn:aws:iam::308535385114:role/gha_workflow_s3_and_ecr_read_only
aws-region: us-east-1
role-duration-seconds: 18000
- name: Calculate docker image
id: calculate-docker-image
uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
with:
docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
docker-image-name: libtorch-cxx11-builder
custom-tag-prefix: rocm6.3
docker-build-dir: .ci/docker
working-directory: pytorch
- name: Pull Docker image
uses: pytorch/test-infra/.github/actions/pull-docker-image@main
with:
docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
- name: Test Pytorch binary
uses: ./pytorch/.github/actions/test-pytorch-binary
env:
DOCKER_IMAGE: ${{ steps.calculate-docker-image.outputs.docker-image }}
- name: Teardown ROCm
uses: ./.github/actions/teardown-rocm
libtorch-rocm6_3-shared-with-deps-release-upload: # Uploading
if: ${{ github.repository_owner == 'pytorch' }}
permissions:
id-token: write
contents: read
needs: libtorch-rocm6_3-shared-with-deps-release-test
with:
PYTORCH_ROOT: /pytorch
PACKAGE_TYPE: libtorch
# TODO: This is a legacy variable that we eventually want to get rid of in
# favor of GPU_ARCH_VERSION
DESIRED_CUDA: rocm6.3
GPU_ARCH_VERSION: "6.3"
GPU_ARCH_TYPE: rocm
DOCKER_IMAGE: libtorch-cxx11-builder
DOCKER_IMAGE_TAG_PREFIX: rocm6.3
LIBTORCH_CONFIG: release
LIBTORCH_VARIANT: shared-with-deps
build_name: libtorch-rocm6_3-shared-with-deps-release
secrets:
github-token: ${{ secrets.GITHUB_TOKEN }}
uses: ./.github/workflows/_binary-upload.yml
libtorch-rocm6_4-shared-with-deps-release-build:
if: ${{ github.repository_owner == 'pytorch' }}
uses: ./.github/workflows/_binary-build-linux.yml
@ -545,3 +430,118 @@ jobs:
secrets:
github-token: ${{ secrets.GITHUB_TOKEN }}
uses: ./.github/workflows/_binary-upload.yml
libtorch-rocm7_0-shared-with-deps-release-build:
if: ${{ github.repository_owner == 'pytorch' }}
uses: ./.github/workflows/_binary-build-linux.yml
needs: get-label-type
with:
PYTORCH_ROOT: /pytorch
PACKAGE_TYPE: libtorch
# TODO: This is a legacy variable that we eventually want to get rid of in
# favor of GPU_ARCH_VERSION
DESIRED_CUDA: rocm7.0
GPU_ARCH_VERSION: "7.0"
GPU_ARCH_TYPE: rocm
DOCKER_IMAGE: libtorch-cxx11-builder
DOCKER_IMAGE_TAG_PREFIX: rocm7.0
LIBTORCH_CONFIG: release
LIBTORCH_VARIANT: shared-with-deps
runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
timeout-minutes: 300
build_name: libtorch-rocm7_0-shared-with-deps-release
build_environment: linux-binary-libtorch
secrets:
github-token: ${{ secrets.GITHUB_TOKEN }}
libtorch-rocm7_0-shared-with-deps-release-test: # Testing
if: ${{ github.repository_owner == 'pytorch' }}
needs:
- libtorch-rocm7_0-shared-with-deps-release-build
- get-label-type
runs-on: linux.rocm.gpu.mi250
timeout-minutes: 240
env:
PYTORCH_ROOT: /pytorch
PACKAGE_TYPE: libtorch
# TODO: This is a legacy variable that we eventually want to get rid of in
# favor of GPU_ARCH_VERSION
DESIRED_CUDA: rocm7.0
GPU_ARCH_VERSION: "7.0"
GPU_ARCH_TYPE: rocm
SKIP_ALL_TESTS: 1
DOCKER_IMAGE: libtorch-cxx11-builder
DOCKER_IMAGE_TAG_PREFIX: rocm7.0
LIBTORCH_CONFIG: release
LIBTORCH_VARIANT: shared-with-deps
steps:
- name: Setup ROCm
uses: ./.github/actions/setup-rocm
- uses: actions/download-artifact@v4.1.7
name: Download Build Artifacts
with:
name: libtorch-rocm7_0-shared-with-deps-release
path: "${{ runner.temp }}/artifacts/"
- name: Checkout PyTorch
uses: actions/checkout@v4
with:
ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
submodules: recursive
path: pytorch
show-progress: false
- name: Clean PyTorch checkout
run: |
# Remove any artifacts from the previous checkouts
git clean -fxd
working-directory: pytorch
- name: ROCm set GPU_FLAG
run: |
echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
- name: configure aws credentials
id: aws_creds
if: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') }}
uses: aws-actions/configure-aws-credentials@v4
with:
role-to-assume: arn:aws:iam::308535385114:role/gha_workflow_s3_and_ecr_read_only
aws-region: us-east-1
role-duration-seconds: 18000
- name: Calculate docker image
id: calculate-docker-image
uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
with:
docker-registry: ${{ startsWith(github.event.ref, 'refs/tags/ciflow/') && '308535385114.dkr.ecr.us-east-1.amazonaws.com' || 'docker.io' }}
docker-image-name: libtorch-cxx11-builder
custom-tag-prefix: rocm7.0
docker-build-dir: .ci/docker
working-directory: pytorch
- name: Pull Docker image
uses: pytorch/test-infra/.github/actions/pull-docker-image@main
with:
docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
- name: Test Pytorch binary
uses: ./pytorch/.github/actions/test-pytorch-binary
env:
DOCKER_IMAGE: ${{ steps.calculate-docker-image.outputs.docker-image }}
- name: Teardown ROCm
uses: ./.github/actions/teardown-rocm
libtorch-rocm7_0-shared-with-deps-release-upload: # Uploading
if: ${{ github.repository_owner == 'pytorch' }}
permissions:
id-token: write
contents: read
needs: libtorch-rocm7_0-shared-with-deps-release-test
with:
PYTORCH_ROOT: /pytorch
PACKAGE_TYPE: libtorch
# TODO: This is a legacy variable that we eventually want to get rid of in
# favor of GPU_ARCH_VERSION
DESIRED_CUDA: rocm7.0
GPU_ARCH_VERSION: "7.0"
GPU_ARCH_TYPE: rocm
DOCKER_IMAGE: libtorch-cxx11-builder
DOCKER_IMAGE_TAG_PREFIX: rocm7.0
LIBTORCH_CONFIG: release
LIBTORCH_VARIANT: shared-with-deps
build_name: libtorch-rocm7_0-shared-with-deps-release
secrets:
github-token: ${{ secrets.GITHUB_TOKEN }}
uses: ./.github/workflows/_binary-upload.yml

1568
.github/workflows/generated-linux-binary-manywheel-nightly.yml generated vendored
View File

File diff suppressed because it is too large Load Diff

6
tools/nightly.py
View File

@ -149,9 +149,9 @@ PIP_SOURCES = {
supported_platforms={"Linux"},
accelerator="rocm",
),
"rocm-6.4": PipSource(
name="rocm-6.4",
index_url=f"{PYTORCH_NIGHTLY_PIP_INDEX_URL}/rocm6.4",
"rocm-7.0": PipSource(
name="rocm-7.0",
index_url=f"{PYTORCH_NIGHTLY_PIP_INDEX_URL}/rocm7.0",
supported_platforms={"Linux"},
accelerator="rocm",
),