[https://nvbugs/6185212][fix] Fix B300 MoE test list ids - #14401
Conversation
|
/bot run --extra-stage "DGX_B300-4_GPUs-PyTorch-1" |
📝 WalkthroughWalkthroughThis PR promotes MoE multi-GPU test execution from post-merge to pre-merge stage by relocating test blocks in the L0 DGX B300 test configuration, updating parameter sizing for specific backends, and removing corresponding test waivers to align with the earlier execution tier. ChangesMoE Test Promotion
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Possibly related PRs
Suggested reviewers
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
tests/integration/test_lists/test-db/l0_dgx_b300.yml (1)
87-114: QA list follow-up is likely unnecessary for this layer.This change updates authoritative
test-dbpre-merge selection and does not add newtests/integration/defs/cases or new perf definitions. A QA scheduled list update is optional here unless you want QA cadence to mirror this exact stage/ID migration immediately (thentests/integration/test_lists/qa/llm_function_core.txtis the likely target).As per coding guidelines, "If the PR only touches unittest/ or narrow unit scope, say explicitly whether QA list updates are unnecessary or optional."
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@tests/integration/test_lists/test-db/l0_dgx_b300.yml` around lines 87 - 114, The QA list update mention is unnecessary or optional for this PR because it only modifies authoritative test-db pre-merge selections under tests/integration/test_lists and does not add new test definitions; explicitly state in the PR description that QA list updates are optional and not required (or that no changes to tests/integration/defs/ or perf definitions were made) so reviewers know QA cadence need not change; update the PR description and any changelog notes to include this explicit statement and reference the modified test list file name (tests/integration/test_lists/test-db/l0_dgx_b300.yml) and the set of removed MoE test cases to make the intent clear.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@tests/integration/test_lists/test-db/l0_dgx_b300.yml`:
- Line 109: The pre-merge list still contains the local-only MoE shape ID
"e256_k8_h7168_i2048" and the routing tag "DeepSeekV3"; replace that stale shape
ID with the CI-sanctioned DeepSeek-V4-Flash shape ID used elsewhere in the suite
and update the routing tag to "DeepSeekV4-Flash" so the test entry (the string
containing
e256_k8_h7168_i2048-seq=1-dtype=torch.bfloat16-backend=MEGAMOE_DEEPGEMM-quant=W4A8_MXFP4_MXFP8-routing=DeepSeekV3)
matches CI IDs and avoids mismatches.
---
Nitpick comments:
In `@tests/integration/test_lists/test-db/l0_dgx_b300.yml`:
- Around line 87-114: The QA list update mention is unnecessary or optional for
this PR because it only modifies authoritative test-db pre-merge selections
under tests/integration/test_lists and does not add new test definitions;
explicitly state in the PR description that QA list updates are optional and not
required (or that no changes to tests/integration/defs/ or perf definitions were
made) so reviewers know QA cadence need not change; update the PR description
and any changelog notes to include this explicit statement and reference the
modified test list file name
(tests/integration/test_lists/test-db/l0_dgx_b300.yml) and the set of removed
MoE test cases to make the intent clear.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: f1410bf1-365d-4bb6-9641-43131950b801
📒 Files selected for processing (2)
tests/integration/test_lists/test-db/l0_dgx_b300.ymltests/integration/test_lists/waives.txt
💤 Files with no reviewable changes (1)
- tests/integration/test_lists/waives.txt
|
PR_Github #49667 [ run ] triggered by Bot. Commit: |
|
PR_Github #49667 [ run ] completed with state
|
Signed-off-by: xxi <xxi@nvidia.com>
c028186 to
c5b63c3
Compare
|
/bot run --extra-stage "DGX_B300-4_GPUs-PyTorch-1" |
|
PR_Github #49787 [ run ] triggered by Bot. Commit: |
|
PR_Github #49787 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #49851 [ run ] triggered by Bot. Commit: |
|
PR_Github #49851 [ run ] completed with state
|
|
/bot run --disable-fail-fast |
|
PR_Github #49904 [ run ] triggered by Bot. Commit: |
|
PR_Github #49904 [ run ] completed with state |
Signed-off-by: xxi <xxi@nvidia.com>
Signed-off-by: xxi <xxi@nvidia.com>
3 participants
Summary
e256_k8_h7168_i2048IDs by switching the affected coverage to CI DeepSeek-V4-Flashe256_k6_h4096_i2048.post_mergetopre_mergeinl0_dgx_b300.yml.Validation
post_merge: 0 explicit MoE unit cases,pre_merge: 19)git diff --check: PASS./scripts/check_test_list.py --validate: PASS (2925 unique entries)Notes
Summary by CodeRabbit