Skip to content

[WIP] Make the nightly job use TheRock instead of ROCm #305

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
charleshofer wants to merge 1 commit into
base: master
Choose a base branch
from
Draft

[WIP] Make the nightly job use TheRock instead of ROCm #305

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 8 additions & 1 deletion .github/workflows/build-base-docker.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,13 +17,16 @@ jobs:
strategy:
fail-fast: false
matrix:
rocm-version: ["7.1.1", "7.2.0"]
rocm-version: ["7.1.1", "7.2.0", "7.12.0"]
Copy link
Contributor

@alekstheod alekstheod Feb 12, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not sure if you are interested but we could do a hermetic build with therock. Check openxla rocm_ci_hermetic config for more details.

install-llvm: [true, false]
include:
- rocm-version: "7.1.1"
runner-label: "linux-x86-64-1gpu-amd"
- rocm-version: "7.2.0"
runner-label: "linux-x86-64-1gpu-amd"
- rocm-version: "7.12.0"
therock-path: "https://rocm.nightlies.amd.com/tarball/therock-dist-linux-gfx94X-dcgpu-7.12.0a20260210.tar.gz"
runner-label: "linux-x86-64-1gpu-amd"
steps:
- name: Clean up old runs
run: |
Expand Down Expand Up @@ -51,6 +54,7 @@ jobs:
env:
ROCM_BUILD_JOB: ${{ matrix.rocm-build-job }}
ROCM_BUILD_NUM: ${{ matrix.rocm-build-num }}
THEROCK_PATH: ${{ matrix.therock-path }}
run: |
BUILD_ARGS=""
if [ -n "$ROCM_BUILD_JOB" ]; then
Expand All @@ -59,6 +63,9 @@ jobs:
if [ -n "$ROCM_BUILD_NUM" ]; then
BUILD_ARGS="$BUILD_ARGS --rocm-build-num=$ROCM_BUILD_NUM"
fi
if [ -n "$THEROCK_PATH" ]; then
BUILD_ARGS="$BUILD_ARGS --therock-path=$THEROCK_PATH"
fi
python3 build/ci_build \
--rocm-version="${{ matrix.rocm-version }}" \
$BUILD_ARGS \
Expand Down
4 changes: 4 additions & 0 deletions .github/workflows/build-docker.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,9 @@ on:
rocm-build-num:
required: false
type: string
therock-path:
required: false
type: string
runner-label:
required: false
type: string
Expand Down Expand Up @@ -67,6 +70,7 @@ jobs:
--rocm-version="${{ inputs.rocm-version }}" \
--rocm-build-job="${{ inputs.rocm-build-job }}" \
--rocm-build-num="${{ inputs.rocm-build-num }}" \
--therock-path="${{ inputs.therock-path }}" \
build_dockers
- name: Push docker images
env:
Expand Down
4 changes: 4 additions & 0 deletions .github/workflows/build-wheels.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,9 @@ on:
rocm-build-num:
required: false
type: string
therock-path:
required: false
type: string
runner-label:
required: false
type: string
Expand Down Expand Up @@ -86,6 +89,7 @@ jobs:
--rocm-version="${{ inputs.rocm-version }}" \
--rocm-build-job="${{ inputs.rocm-build-job }}" \
--rocm-build-num="${{ inputs.rocm-build-num }}" \
--therock-path="${{ inputs.therock-path }}" \
--jax-source-dir="./jax" \
dist_wheels \
--rbe
Expand Down
22 changes: 9 additions & 13 deletions .github/workflows/nightly.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,18 +21,16 @@ jobs:
strategy:
fail-fast: false
matrix:
rocm-version: ["7.1.1", "7.2.0"]
rocm-version: ["7.12.0"]
include:
- rocm-version: "7.1.1"
runner-label: '["linux-x86-64-1gpu-amd"]'
- rocm-version: "7.2.0"
- rocm-version: "7.12.0"
therock-path: "https://rocm.nightlies.amd.com/tarball/therock-dist-linux-gfx94X-dcgpu-7.12.0a20260210.tar.gz"
runner-label: '["linux-x86-64-1gpu-amd"]'
uses: ./.github/workflows/build-wheels.yml
with:
python-versions: "3.11,3.12,3.13,3.14"
rocm-version: ${{ matrix.rocm-version }}
rocm-build-job: ${{ matrix.rocm-build-job }}
rocm-build-num: ${{ matrix.rocm-build-num }}
therock-path: ${{ matrix.therock-path }}
runner-label: ${{ matrix.runner-label }}
secrets:
rbe_ci_cert: ${{ secrets.RBE_CI_CERT }}
Expand All @@ -42,17 +40,15 @@ jobs:
strategy:
fail-fast: false
matrix:
rocm-version: ["7.1.1", "7.2.0"]
rocm-version: ["7.12.0"]
include:
- rocm-version: "7.1.1"
runner-label: '["linux-x86-64-1gpu-amd"]'
- rocm-version: "7.2.0"
- rocm-version: "7.12.0"
therock-path: "https://rocm.nightlies.amd.com/tarball/therock-dist-linux-gfx94X-dcgpu-7.12.0a20260210.tar.gz"
runner-label: '["linux-x86-64-1gpu-amd"]'
uses: ./.github/workflows/build-docker.yml
with:
rocm-version: ${{ matrix.rocm-version }}
rocm-build-job: ${{ matrix.rocm-build-job }}
rocm-build-num: ${{ matrix.rocm-build-num }}
therock-path: ${{ matrix.therock-path }}
runner-label: ${{ matrix.runner-label }}
extra-cr-tag: "nightly"
call-test-and-upload:
Expand All @@ -63,7 +59,7 @@ jobs:
test-command:
- "python jax_rocm_plugin/build/rocm/run_single_gpu.py -c -s"
- "python jax_rocm_plugin/build/rocm/run_multi_gpu.py -c -s"
rocm-version: ["7.1.1", "7.2.0"]
rocm-version: ["7.12.0"]
include:
- test-command: "python jax_rocm_plugin/build/rocm/run_single_gpu.py -c -s"
runner-label: '["linux-x86-64-1gpu-amd"]'
Expand Down
3 changes: 2 additions & 1 deletion docker/Dockerfile.base-ubu24
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,7 +76,8 @@ RUN --mount=type=cache,target=/var/cache/apt \
apt-transport-https \
ca-certificates \
gnupg \
curl && \
curl \
wget && \
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg \
| gpg --dearmor -o /usr/share/keyrings/cloud.google.gpg && \
echo "deb [signed-by=/usr/share/keyrings/cloud.google.gpg] https://packages.cloud.google.com/apt cloud-sdk main" \
Expand Down
16 changes: 11 additions & 5 deletions jax_rocm_plugin/build/rocm/tools/get_rocm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,11 +233,17 @@ def _install_therock(rocm_version, therock_path):
os.symlink(rocm_real_path, rocm_sym_path, target_is_directory=True)

# Make a symlink to amdgcn to fix LLVM not being able to find binaries
os.symlink(
rocm_real_path + "/lib/llvm/amdgcn/",
rocm_real_path + "/amdgcn",
target_is_directory=True,
)
try:
os.symlink(
rocm_real_path + "/lib/llvm/amdgcn/",
rocm_real_path + "/amdgcn",
target_is_directory=True,
)
except FileExistsError:
LOG.info(
"%s/amdgc already exists. Not creating symlink",
rocm_real_path
)


def _setup_internal_repo(system, rocm_version, job_name, build_num):
Expand Down
Loading