[None][fix] Update CUTLASS C++ to 4.4.2#12897
[None][fix] Update CUTLASS C++ to 4.4.2#12897depaulmillz wants to merge 1 commit intoNVIDIA:mainfrom
Conversation
📝 WalkthroughWalkthroughUpdated the cutlass dependency version from v4.4.1 to v4.4.2 in the fetch content configuration file. All other repository settings and fetch parameters remain unchanged. Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes 🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
There was a problem hiding this comment.
🧹 Nitpick comments (1)
3rdparty/fetch_content.json (1)
14-14: Pin to commit SHA for reproducibility and verify alignment with SM12x carveout fix.The tag
v4.4.2resolves to commitda5e086dab31d63815acafdac9a9c5893b1c69e2. Pinning to the commit SHA instead of the tag hardens reproducibility and supply-chain resilience. Note that the release notes do not explicitly mention the SM12x shared-memory carveout group GEMM fix stated in the PR objective—confirm this tag addresses the intended issue.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@3rdparty/fetch_content.json` at line 14, Replace the "git_tag" pin with the exact commit SHA da5e086dab31d63815acafdac9a9c5893b1c69e2 in fetch_content.json (replace the "git_tag" value with this SHA) to ensure reproducible fetches, and then verify the checked-out commit contains the SM12x shared-memory carveout GEMM fix by inspecting the commit diff/PR referenced in that commit to confirm it matches the intended fix before merging.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@3rdparty/fetch_content.json`:
- Line 14: Replace the "git_tag" pin with the exact commit SHA
da5e086dab31d63815acafdac9a9c5893b1c69e2 in fetch_content.json (replace the
"git_tag" value with this SHA) to ensure reproducible fetches, and then verify
the checked-out commit contains the SM12x shared-memory carveout GEMM fix by
inspecting the commit diff/PR referenced in that commit to confirm it matches
the intended fix before merging.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: f752def2-bb46-4a8e-b180-7cada48f7a25
📒 Files selected for processing (1)
3rdparty/fetch_content.json
|
/bot run |
|
PR_Github #42564 [ run ] triggered by Bot. Commit: |
Summary by CodeRabbit
Description
CUTLASS 4.4.2 fixed an auto smem carveout issue for SM12x which resulted in some group GEMM launch failures. This MR updates CUTLASS C++ from 4.4.1 to 4.4.2 in order to get the fix. This should ensure that prior kernel launches that were failing on SM12x with 4.4.1 will be fixed.
Test Coverage
N/A
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)
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.
[ x] 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.