matmul_nbits: Use GPU_WARP_SIZE_HOST for host side code #65
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
For ROCm device, the host side code needs to call GPU_WARP_SIZE_HOST to query warpsize of the underlying GPU device. Fixes MatMulNBits tests on Navi. Signed-off-by: Jagadish Krishnamoorthy <[email protected]>
jagadish-amd
requested review from
jeffdaily,
xinyazhang,
TedThemistokleous and
causten
| @@ -288,6 +288,7 @@ bool TryMatMul4Bits( | |||
| if (n % kColsPerThreadBlock != 0 || k % 8 != 0 || m > 1) { | |||
| return false; | |||
| } | |||
| const int kWarpSize = GPU_WARP_SIZE_HOST; | |||
There was a problem hiding this comment.
Put a #if around this via how we handle things with USE_ROCM and USE_MIGRAPHX
#if USE_MIGRAPHX || USE_ROCM
const int kWarpSize = GPU_WARP_SIZE_HOST;
#endif
That way if once we upstream this change it wouldn't break another EP if this is specific to AMD
There was a problem hiding this comment.
There's no need to ifdef protect this. CUDA EP also defines GPU_WARP_SIZE_HOST for this reason.
There was a problem hiding this comment.
Ah then my real curiosity is why redefine it here then if that's the case?
jeffdaily
reviewed
Sep 10, 2024
| @@ -288,6 +288,7 @@ bool TryMatMul4Bits( | |||
| if (n % kColsPerThreadBlock != 0 || k % 8 != 0 || m > 1) { | |||
| return false; | |||
| } | |||
| const int kWarpSize = GPU_WARP_SIZE_HOST; | |||
There was a problem hiding this comment.
There's no need to ifdef protect this. CUDA EP also defines GPU_WARP_SIZE_HOST for this reason.
Signed-off-by: Jagadish Krishnamoorthy <[email protected]>
jeffdaily
approved these changes
Sep 10, 2024
jeffdaily
merged commit d7e1c61
into
ROCm:rocm6.3_internal_testing
5 of 10 checks passed
|
@jagadish-amd please file upstream PR for same change. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
For ROCm device, the host side code needs to call GPU_WARP_SIZE_HOST to query warpsize of the underlying GPU device.
Fixes MatMulNBits tests on Navi.