[https://nvbugs/6274934][fix] Skip DeepSeek-V3-Lite dis-agg test on non-Hopper#15577
[https://nvbugs/6274934][fix] Skip DeepSeek-V3-Lite dis-agg test on non-Hopper#15577Shixiaowei02 wants to merge 1 commit into
Conversation
e6c42c1 to
d3d6119
Compare
|
/bot run --add-multi-gpu-test --disable-fail-fast |
📝 WalkthroughWalkthroughThe ChangesTest Hardware Gate
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes Suggested reviewers
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
|
/bot run --add-multi-gpu-test --disable-fail-fast |
|
PR_Github #55642 [ run ] triggered by Bot. Commit: |
|
PR_Github #55642 [ run ] completed with state |
d3d6119 to
a9454a5
Compare
|
/bot run --add-multi-gpu-test --disable-fail-fast |
a9454a5 to
c56defe
Compare
|
/bot run --add-multi-gpu-test --disable-fail-fast |
|
PR_Github #56871 [ run ] triggered by Bot. Commit: |
|
PR_Github #56871 [ run ] completed with state
|
…disagg worker test on non-Hopper test_workers_conditional_disaggregation_deepseek_v3_lite_bf16 runs DeepSeek-V3-Lite, which uses MLA. The context-phase MLA FMHA kernels (headSize=192, headSizeV=128) are only available on Hopper (SM90), so on pre-Hopper GPUs AttentionOp::initialize() aborts with "Deepseek should be supported by fmha in context part." Add @skip_no_hopper, matching the sibling DeepSeek-V3-Lite disaggregation tests (test_workers_kv_cache_aware_router_deepseek_v3_lite_bf16 and test_disaggregated_deepseek_v3_lite_bf16_conditional). Signed-off-by: Shixiaowei02 <39303645+Shixiaowei02@users.noreply.github.com>
c56defe to
d554c77
Compare
|
/bot run --stage-list "DGX_H100-4_GPUs-PyTorch-DeepSeek-1" |
|
PR_Github #59443 [ run ] triggered by Bot. Commit: |
|
PR_Github #59443 [ run ] completed with state
|
Summary by CodeRabbit
Description
Test Coverage
PR Checklist
Please review the following before submitting your PR:
PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
If PR introduces API changes, an appropriate PR label is added - either
api-compatibleorapi-breaking. Forapi-breaking, includeBREAKINGin the PR title.Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
Update tava architecture diagram if there is a significant design change in PR.
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.
GitHub Bot Help
To see a list of available CI bot commands, please comment
/bot help.