[geomin12]

The test_type_for_family flag (e.g., "quick" for MI455/gfx125x) was not being strictly honored by fetch_test_configurations.py. Now it overrides the TEST_TYPE env var regardless of what's passed, ensuring families with limited hardware always run the appropriate test type.

follow-up PR: fetch_test_configurations.py is abstracting a lot from configure_multi_arch_ci.py. I'm planning to combine them so configure does everything and we don't miss cases between

[therock-pr-bot]

❌ PR Check — Action Required

Check	Status	Details
📝 PR Description	❌ Fail	Error: PR description must reference a JIRA ID, ISSUE ID, or a GitHub closing keyword. Expected: include a JIRA ID / ISSUE ID line (separator : or -, or omitted; value may be a JIRA key, a number with/without #, or a link), OR a closing keyword + issue reference. Accepted examples: • JIRA ID : TESTAUTO-6039 • JIRA ID - #330 • JIRA ID #330 • JIRA ID (on separate line) ROCM-25757 • ISSUE ID : TESTUTO-3334 • ISSUE ID #3334 • ISSUE ID - TESTAUTO-3433 • ISSUE ID (on separate line) AIRUNTIME-2352 • ISSUE ID : https://github.com/<org_name>/<repo_name>/issues/1234 • Closes #10 • Fixes octo-org/octo-repo#100 • Resolves: #123 • #123 • https://github.com/<org_name>/<repo_name>/issues/123 Current: no valid JIRA/ISSUE/closing-keyword reference found
⛔ Forbidden Files	✅ Pass	—
🧪 Unit Test	✅ Pass	—
🔎 pre-commit	✅ Pass	—
🚫 Draft PR	🔜 To Be Enabled	—
🚩 Feature Flag	🔜 To Be Enabled	—
📊 Code Coverage	🔜 To Be Enabled	—

⚠️ 1 policy check(s) failed. Please address the issues above before this PR can be Reviewed.

🚫 Please fix the failed policies

• ❌ PR Description

The Not ready to Review label was added to this PR. Once all policies pass, the label is removed automatically.

📖 Need help? See the Policy FAQ for details on every check and how to fix failures.

🙋 Wish to Override Policy?

[therock-pr-bot]

🚫 Please fix the failed policies before requesting reviews.

The following policy checks failed:

• ❌ PR Description

The Not ready to Review label has been added to this PR.
Once all policies pass, the label will be removed automatically.

[dysu-amd]

From my understanding of the context around this:

• We set gfx1250 GPU tests to only run for nightlies
  • feat(ci): enable nightly tests for gfx125x #8772
  • https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/amdgpu_family_matrix.py#L425
• gfx1250 CPU tests are still enabled for other run types, eg. submodule bumps
  • Bump rocm-systems from 105dd4f to 8a343ff #8831
  • https://github.com/ROCm/TheRock/actions/runs/37655107204/job/112956067247?pr=8831#step:7:129
• test_type_for_family was introduced to always run quick tests for gfx1250, to further reduce test pressure
  • feat(ci): adding trigger_test_label_only and test_type_for_family flags for amdgpu_family_matrix.py #8169
  • https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/amdgpu_family_matrix.py#L427

The issue:

• configure_multi_arch_ci.py only considers test_type_for_family when test_runs_on is nonempty
  • https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/configure_multi_arch_ci.py#L1596
• When GPU tests are disabled, eg. submodule bumps, test_runs_on is empty
• Therefore test_type_for_family is not set when GPU tests are disabled, leading the issue that this PR addresses
  • In the above submodule PR run, the standard filter is selected instead
  • There's not yet been a nightly run for gfx1250 tests yet, but it's expected to correctly use the quick filter since test_runs_on should be nonempty

Instead of having fetch_test_configurations.py also need to check if a gfx has an override, could we just always have configure_multi_arch_ci.py set test_type_for_family regardless of whether or not test_runs_on is empty? Would that address the issue? The specific line in question: https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/configure_multi_arch_ci.py#L1596

follow-up PR: fetch_test_configurations.py is abstracting a lot from configure_multi_arch_ci.py. I'm planning to combine them so configure does everything and we don't miss cases between

100% agree :D

[geomin12]

From my understanding of the context around this:

• We set gfx1250 GPU tests to only run for nightlies

  • feat(ci): enable nightly tests for gfx125x #8772
  • https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/amdgpu_family_matrix.py#L425

• gfx1250 CPU tests are still enabled for other run types, eg. submodule bumps

  • Bump rocm-systems from 105dd4f to 8a343ff #8831
  • https://github.com/ROCm/TheRock/actions/runs/37655107204/job/112956067247?pr=8831#step:7:129

• test_type_for_family was introduced to always run quick tests for gfx1250, to further reduce test pressure

  • feat(ci): adding trigger_test_label_only and test_type_for_family flags for amdgpu_family_matrix.py #8169
  • https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/amdgpu_family_matrix.py#L427

The issue:

• configure_multi_arch_ci.py only considers test_type_for_family when test_runs_on is nonempty

  • https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/configure_multi_arch_ci.py#L1596

• When GPU tests are disabled, eg. submodule bumps, test_runs_on is empty

• Therefore test_type_for_family is not set when GPU tests are disabled, leading the issue that this PR addresses

  • In the above submodule PR run, the standard filter is selected instead
  • There's not yet been a nightly run for gfx1250 tests yet, but it's expected to correctly use the quick filter since test_runs_on should be nonempty

Instead of having fetch_test_configurations.py also need to check if a gfx has an override, could we just always have configure_multi_arch_ci.py set test_type_for_family regardless of whether or not test_runs_on is empty? Would that address the issue? The specific line in question: https://github.com/ROCm/TheRock/blob/main/build_tools/github_actions/configure_multi_arch_ci.py#L1596

follow-up PR: fetch_test_configurations.py is abstracting a lot from configure_multi_arch_ci.py. I'm planning to combine them so configure does everything and we don't miss cases between

100% agree :D

this is true! however, it requires a bit more effort which I will be doing in a follow-up PR

this helps resolve immediate testing needed for mi455

[dysu-amd]

I suppose this is ok if a follow up is coming. Tbh I did not know of this script's existence until now, I'm far from familiar with all the many ci scripts and the interactions between them