CI: parallel download steps, bump {up,down}load-artifacts versions

[matthiasdiener]

Description

• Parallelize the two download steps (docker image & artifact) which can take 45+ minutes when run sequentially (e.g., https://github.com/ROCm/TransformerEngine/actions/runs/33089752428/job/98588663888)
  • On the CI run for this PR (https://github.com/ROCm/TransformerEngine/actions/runs/33101164648/job/98630586120?pr=719), parallelization reduced total download step time between 12%-42% (mm:ss - 1:25-7:40)
• Bump {up,down}load-artifacts versions to avoid node.js deprecation warnings (e.g. at the bottom here: https://github.com/ROCm/TransformerEngine/actions/runs/33067995737)

Type of change

• Documentation change (change only to the documentation, either a fix or a new content)
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Infra/Build change
• Code refactoring

Changes

Please list the changes introduced in this PR:

• Change A
• Change B

Checklist:

• I have read and followed the contributing guidelines
• The functionality is complete
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[Copilot]

Pull request overview

Parallelizes artifact downloads and Docker image pulls to reduce ROCm CI startup time while updating artifact actions to Node.js-compatible releases.

Changes:

• Run artifact download and Docker pull steps concurrently.
• Upgrade upload/download artifact actions to v7/v8.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
.github/workflows/rocm-wheels-build.yml	Upgrades artifact uploads to v7.
.github/workflows/rocm-ci.yml	Parallelizes setup operations and upgrades artifact actions.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[ipanfilo]

It is good change. Will it be also beneficial to put other downloads to this parallel step: repo checkout + submodules init?

[matthiasdiener]

From what I have seen, these steps are quite fast already (<10s each):

Checkout probably can't share the group with the download, since it runs git clean and that might race with the artifact download to dist/.

We could:

• run docker pull as the first step, in the background (it needs nothing from the repo)
• run checkout, submodule init, artifact download sequentially (but in parallel to the backgrounded docker pull)

I'm doubtful that the added complexity would be worth it.