msa: GPT-6#946
Conversation
- Add msa to KERNEL_SOURCE_MAPPING so the kernel lookup completeness check passes. - Declare the license as 'MIT AND Apache-2.0': MSA is MIT, the vendored quack modules are Apache-2.0. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
/kernel-bot build-and-stage msa --branch v0 |
|
Build request processed. Command: Dispatched (2):
|
|
/kernel-bot build-and-stage msa --branch v0 |
|
Build request processed. Command: Dispatched (2):
|
danieldk
left a comment
There was a problem hiding this comment.
Looks good, added a question.
| # runtime through nvidia-cutlass-dsl; only SM100 is supported. | ||
| [kernel.msa] | ||
| backend = "cuda" | ||
| cuda-capabilities = ["10.0"] |
There was a problem hiding this comment.
Also doesn't work on later Blackwell? (12.0, etc.)
There was a problem hiding this comment.
I kept it because the repository explicitly states it and also the compiler flags have -arch=sm_100 everywhere in upstream.
|
@danieldk could you approve for the merge? |
|
@danieldk a gentle ping. |
|
Kernel build was successful. |
|
/kernel-bot merge-and-upload msa |
|
Build request processed. Command: Merge result: PR is already merged. Continuing with build/upload. Dispatched (1): |
Triggering a build.