feat: bump nvidia-cutlass-dsl to >=4.4.2#2833
feat: bump nvidia-cutlass-dsl to >=4.4.2#2833limin2021 wants to merge 5 commits intoflashinfer-ai:mainfrom
Conversation
Update requirements.txt to require >=4.4.2 and add CI override via setup_python.env so tests use the new version immediately without waiting for Docker image rebuild. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
limin2021
requested review from
aleozlx,
bkryu,
cyx-6,
jimmyzho,
kahyunnam,
nv-yunzheq,
saltyminty,
sricketts,
yongwww,
yyihuang and
yzh119
as code owners
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request aims to upgrade the Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. Footnotes
|
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughThis PR bumps the nvidia-cutlass-dsl dependency from version 4.3.4 to 4.4.2 across the project. It introduces a configurable CUTLASS_DSL_VERSION environment variable in CI and updates the setup script to handle version overrides with CUDA-specific package variants. Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Possibly related PRs
Suggested labels
Suggested reviewers
Poem
🚥 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)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request correctly bumps the nvidia-cutlass-dsl dependency version to >=4.4.2. The changes include updating requirements.txt and adding a CI override mechanism to immediately use the new version. The implementation is solid. I have one minor suggestion to improve a comment's clarity in the CI environment file to prevent future confusion.
![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
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@ci/setup_python.env`:
- Around line 19-20: The active override CUTLASS_DSL_VERSION=4.4.2 is
inconsistent with the header guidance that overrides should not be merged;
update the file to either (a) comment the line back out if this was temporary,
or (b) keep it but add a one-line explanatory comment above CUTLASS_DSL_VERSION
noting why this is an intentional exception (e.g., temporary bridge until Docker
image rebuilt) and include a TODO with the intended remediation (e.g., "TODO:
remove when image rebuilt"), so reviewers know this is deliberate and tracked.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 3fdc60bd-51f4-48b3-83e5-1fcc619f6881
📒 Files selected for processing (3)
ci/setup_python.envrequirements.txtscripts/setup_test_env.sh
|
/bot run |
| # TVM_FFI_REF= | ||
|
|
||
| # Uncomment to override nvidia-cutlass-dsl version: | ||
| CUTLASS_DSL_VERSION=4.4.2 |
There was a problem hiding this comment.
Please remove it after the CI passed.
From 4.4 onwards, nvidia-cutlass-dsl requires the [cu13] extra to install nvidia-cutlass-dsl-libs-cu13. Without it, CI fails. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…install - requirements.txt: remove [cu13] extra (keep generic for Docker builds) - setup_test_env.sh: detect CUDA major version from torch to select [cu12] or [cu13] extra, and clean uninstall old packages before installing (per NVIDIA docs recommendation) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
/bot run |
|
@limin2021 is not authorized to trigger this CI job. cc: @yzh119, @sricketts, @yongwww |
|
/bot run |
|
/bot run |
|
@limin2021 is not authorized to trigger this CI job. cc: @yzh119, @sricketts, @yongwww |
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.claude/skills/update-cutlass-dsl/SKILL.md:
- Around line 53-59: The unlabeled fenced code blocks in SKILL.md (the example
blocks around the "Before/After" snippet and the PR flow diagram near lines 53
and 137) trigger markdownlint MD040; update each triple-backtick fence to
include a language (e.g., change ``` to ```text) for both the small Before/After
snippet and the larger PR flow block so the fenced-code blocks are labeled and
the lint warning is resolved.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 5da7ab12-313c-41a6-addf-d66869c1bf48
📒 Files selected for processing (2)
.claude/skills/update-cutlass-dsl/SKILL.mdscripts/setup_test_env.sh
| ``` | ||
| # Before | ||
| nvidia-cutlass-dsl>=4.3.4 | ||
|
|
||
| # After (example) | ||
| nvidia-cutlass-dsl>=4.4.2 | ||
| ``` |
There was a problem hiding this comment.
Add fenced-code languages to satisfy markdownlint (MD040).
Line 53 and Line 137 use unlabeled fenced blocks; add a language (e.g., text) to avoid lint warnings.
🛠️ Suggested doc-only fix
-```
+```text
# Before
nvidia-cutlass-dsl>=4.3.4
# After (example)
nvidia-cutlass-dsl>=4.4.2
```diff
-```
+```text
Push PR to main
|
+-- pr-test.yml (runs immediately)
| \-- reads old Docker tag from ci/docker-tags.yml
| \-- uses old Docker image
| \-- BUT setup_test_env.sh installs CUTLASS_DSL_VERSION override <-- new version used here
|
+-- release-ci-docker.yml (triggered by requirements.txt change)
\-- builds new Docker image with new version baked in
\-- auto-creates PR to update ci/docker-tags.yml
\-- after that PR merges, future CI uses new image natively
</details>
Also applies to: 137-149
<details>
<summary>🧰 Tools</summary>
<details>
<summary>🪛 markdownlint-cli2 (0.21.0)</summary>
[warning] 53-53: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
</details>
</details>
<details>
<summary>🤖 Prompt for AI Agents</summary>
Verify each finding against the current code and only fix it if needed.
In @.claude/skills/update-cutlass-dsl/SKILL.md around lines 53 - 59, The
unlabeled fenced code blocks in SKILL.md (the example blocks around the
"Before/After" snippet and the PR flow diagram near lines 53 and 137) trigger
markdownlint MD040; update each triple-backtick fence to include a language
(e.g., change totext) for both the small Before/After snippet and the
larger PR flow block so the fenced-code blocks are labeled and the lint warning
is resolved.
</details>
<!-- fingerprinting:phantom:poseidon:hawk -->
<!-- This is an auto-generated comment by CodeRabbit -->
- setup_test_env.sh: only add [cu13] extra for CUDA 13, plain install for CUDA 12 - Update skill docs to match NVIDIA's official installation guide Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
limin2021
force-pushed
the
update_dsl_4.4
branch
from
f3edb43 to
bc3b93c
Compare
|
[FAILED] Pipeline #46590486: 13/20 passed |
|
/bot run |
|
[FAILED] Pipeline #46615561: 13/20 passed |
|
Seems there are till GB200/300 failures due to cutedsl errors in CI. |
nv-yunzheq
requested review from
IwakuraRein,
dhiraj113 and
jiahanc
as code owners
|
/bot run |
|
[FAILED] Pipeline #46648743: 13/20 passed |
|
The remaining failed kernel is:
on cu129 for both b200 and b300. |
7 participants
Update requirements.txt to require >=4.4.2 and add CI override via setup_python.env so tests use the new version immediately without waiting for Docker image rebuild.
📌 Description
Summary
requirements.txtto requirenvidia-cutlass-dsl>=4.4.2(was>=4.3.4)CUTLASS_DSL_VERSIONoverride support inscripts/setup_test_env.shCUTLASS_DSL_VERSION=4.4.2inci/setup_python.envso CI tests use the new version immediately without waiting for Docker imagerebuild
Test plan
CUTLASS_DSL_VERSIONfromci/setup_python.env🔍 Related Issues
🚀 Pull Request Checklist
Thank you for contributing to FlashInfer! Before we review your pull request, please make sure the following items are complete.
✅ Pre-commit Checks
pre-commitby runningpip install pre-commit(or used your preferred method).pre-commit install.pre-commit run --all-filesand fixed any reported issues.🧪 Tests
unittest, etc.).Reviewer Notes
Summary by CodeRabbit