Full CI support for public builds + switch to use cibuildwheel by leofang · Pull Request #267 · NVIDIA/cuda-python

leofang · 2024-12-05T05:04:25Z

Close#227. Part of #81.

The earlier build pipeline was deficient because

it only covered a single point in our large support matrix (ex: no aarch64/win-64 coverage)
it used a miniforge container to build wheels, which does not guarantee compatibility with the manylinux/glibc requirement
since a container is used, it is impossible to reuse many widely used/validated, OSS GHA workflows, and everything has to be homegrown scripts, which is an unnecessary burden to both developer and devops teams
- one addition downside of this is it's really hard to understand the overall CI design, since codes are scattered into many files that are hard to follow
the old CI had too many unnecessary pieces not suitable for this project (ex: CMake, Legate, ...)

This PR switches to use public GHA workflows. Most steps are meant to be easily reusable (after some tweaking/refactoring) by other Python/CUDA projects, either first-party or third-party. For example, cibuildwheels has been used to offer guarantee to manylinux. It also lays the ground work for later bring-up of the CI test stage. (This PR only covers the build stage.)

As of commit 0d554f0, we can build all needed wheels using GHA now! 🎉

cuda.bindings
cuda.core
Python: 3.9-3.12
OS: linux-64, linux-aarch64, win-64
CUDA 12.6 (we can evaluate backporting to the cuda-11 branch)

The artifacts are available here for inspection: https://github.com/NVIDIA/cuda-python/actions/runs/12192916020

copy-pr-bot · 2024-12-05T05:04:30Z

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

leofang · 2024-12-05T05:04:47Z

/ok to test

leofang · 2024-12-05T05:23:47Z

/ok to test

leofang · 2024-12-05T05:29:30Z

/ok to test

leofang · 2024-12-05T05:31:52Z

/ok to test

leofang · 2024-12-05T05:42:30Z

/ok to test

leofang · 2024-12-05T05:44:59Z

/ok to test

leofang · 2024-12-05T05:54:22Z

/ok to test

leofang · 2024-12-05T06:01:26Z

/ok to test

leofang · 2024-12-05T06:05:11Z

/ok to test

leofang · 2024-12-05T06:24:25Z

/ok to test

leofang · 2024-12-05T06:34:34Z

/ok to test

leofang · 2024-12-05T06:43:26Z

/ok to test

leofang · 2024-12-05T06:46:13Z

/ok to test

leofang · 2024-12-06T06:00:22Z

+ PARALLEL_LEVEL=${{ env.PARALLEL_LEVEL }}
+ CIBW_ENVIRONMENT_WINDOWS: >
+ CUDA_HOME="$(cygpath -w $(realpath ./cuda_toolkit))"
+ # PARALLEL_LEVEL=${{ env.PARALLEL_LEVEL }}


Note: This was commented out because of #271.

leofang · 2024-12-06T19:00:10Z

+ # WAR: setup-python is not relocatable...
+ # see https://github.com/actions/setup-python/issues/871
+ - name: Set up Python ${{ inputs.python-version }}
+ if: ${{ startsWith(inputs.host-platform, 'linux') }}
+ id: setup-python
+ uses: actions/setup-python@v5
+ with:
+ python-version: "3.12"


Technically we don't need this step. However, because actions/setup-python is not relocatable (its RPATH is hard-wired, see actions/setup-python#871), on a self-hosted runner cibuildwheels cannot properly load libpython3.12. This is a WAR by forcing the same Python version installed on the host (and mounted into the container).

keenan-simpson · 2024-12-06T20:43:03Z

Last comment I think: Perhaps we should squash these commits. Not hellbent on this

leofang · 2024-12-06T21:25:38Z

/ok to test

leofang · 2024-12-06T21:26:47Z

Perhaps we should squash these commits.

Yes when we merge there's a drop-down menu to select merge or squash.

keenan-simpson · 2024-12-06T21:28:09Z

ahh, that option is grayed out for me on this project

leofang · 2024-12-06T21:37:13Z

@ksimpson-work is it still grey after you approved it?

keenan-simpson · 2024-12-06T21:39:50Z

No, I noticed it when I went to merge my own PRs. I'll take a screenshot next time, unless I was hallucinating

leofang · 2024-12-06T21:43:15Z

Let's merge! Thanks for quick reviews @ksimpson-work!

switch to cibuildwheel + some cleanups
0141cbf

leofang self-assigned this Dec 5, 2024

leofang added CI/CD CI/CD infrastructure P0 High priority - Must do! labels Dec 5, 2024

try setting up Python manually
79fd67a

comment out un-needed code + propagate python-version
92283dc

fix: need to check out first
6666e90

only build natively; add -v; ensure targeting manylinux
c96fef3

single quotes
ea77572

restore env setup to get artifact dir
508a126

fix artifact dir
beff50d

fix artifact name
a935da2

restore & fix artifact name again
df982e5

build on all platforms!
426cd58

fix CIBW_BUILD for windows
d167a55

fix typo
9cd0803

leofang marked this pull request as ready for review December 6, 2024 05:19

leofang added cuda.bindings Everything related to the cuda.bindings module cuda.core Everything related to the cuda.core module labels Dec 6, 2024

leofang mentioned this pull request Dec 6, 2024
Building cuda.bindings in parallel might fail non-deterministically #271
Closed

leofang commented Dec 6, 2024
View reviewed changes

leofang mentioned this pull request Dec 6, 2024
Add py313 builds #272
Merged

leofang commented Dec 6, 2024
View reviewed changes

Comment thread.github/actions/setup/action.yml Outdated

leofang commented Dec 6, 2024
View reviewed changes

Comment thread.github/actions/setup/action.yml Outdated

keenan-simpson reviewed Dec 6, 2024
View reviewed changes

Comment thread.github/actions/setup/action.yml

keenan-simpson reviewed Dec 6, 2024
View reviewed changes

Comment thread.github/workflows/gh-build-and-test.yml

keenan-simpson reviewed Dec 6, 2024
View reviewed changes

Comment thread.github/actions/setup/action.yml Outdated

leofang added 2 commits December 6, 2024 15:41

clean up unused (but still relevant) code
19445d5

clean up unused (but still relevant) code - cont'd
15626a8

keenan-simpson reviewed Dec 6, 2024
View reviewed changes

Comment thread.github/actions/setup/action.yml

consolidate with PYTHON_VERSION_FORMATTED
07b8d1d

keenan-simpson self-requested a review December 6, 2024 21:27

keenan-simpson approved these changes Dec 6, 2024
View reviewed changes

leofang merged commit 1c86afa into NVIDIA:mainDec 6, 2024

leofang deleted the more_ci_tweak branch December 6, 2024 21:43

This was referenced Dec 7, 2024
Publish standalone cuda-core packages #147
Closed
Set up a public, GitHub-Action-based CI infrastructure #81
Closed

leofang mentioned this pull request Feb 24, 2025
CI: Enable Cython tests on Windows runners #466
Closed

Uh oh!

Conversation

leofang commented Dec 5, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

copy-pr-botBot commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofang commented Dec 5, 2024

Uh oh!

leofangDec 6, 2024

Choose a reason for hiding this comment

Uh oh!

leofangDec 6, 2024

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

keenan-simpson commented Dec 6, 2024

Uh oh!

Uh oh!

leofang commented Dec 6, 2024

Uh oh!

leofang commented Dec 6, 2024

Uh oh!

keenan-simpson commented Dec 6, 2024

Uh oh!

leofang commented Dec 6, 2024

Uh oh!

keenan-simpson commented Dec 6, 2024

Uh oh!

leofang commented Dec 6, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

leofang commented Dec 5, 2024•
edited
Loading