feat(mma): add fp16@fp16->fp32 mma and unit tests #101
Conversation
This file contains hidden or 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
Add mma with half-precision (FP16) inputs and fp32 accumulators for `mma.sync.aligned` instructions including: - mma_AB_base - mma_ABt_base - mma_AtB_base - mma_AtBt_base Add unit tests to ensure the correctness: - test_mma_AB_half - test_mma_ABt_half - test_mma_AtB_half - test_mma_AtBt_half
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Outdated suggestions cannot be applied.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.
feat(mma): add half-precision MMAs for automotive devices and training
Add FP16 variants of matrix multiply operations benefiting non-Hopper
devices including NVIDIA Orin (sm_87) and Ada Lovelace (sm_89) automotive
edge devices. In addition, backward passes during training can benefit from
higher precisions. These variants provide higher precision compared to BF16
when needed.
Key changes:
mma.sync.aligned.m16n8k16.row.col.f32.f16.f16.f32interfacesmma_interfaces for register tilesrt_base<half,...>(AB, ABt, AtB, AtBt)
Tested on NVIDIA A100, Ada, and H100 platforms.