[SM121] Enable native block-scaled dot_scaled for DGX Spark (GB10)#10010
Merged
Conversation
masahi
reviewed
Apr 13, 2026
SM121 (GB10 DGX Spark) supports the same mma.sync block-scaled instructions as SM120 (RTX 5090) but was excluded from the native lowering path by exact compute capability checks. Without this fix, dot_scaled on SM121 falls through to DecomposeScaledBlocked which upcasts to bf16 — ~10 TFLOPS vs ~270 TFLOPS with native mma.sync block-scaled FP4. Tested on GB10 with both MXFP4 (scale_vec::2X, ue8m0) and NVFP4 (scale_vec::4X, ue4m3).
ianbarber
force-pushed
the
sm121-dot-scaled
branch
from
8caf39d to
f40d381
Compare
plognjen
pushed a commit
to plognjen/triton
that referenced
this pull request
Apr 14, 2026
…riton-lang#10010) SM121 (GB10 DGX Spark) supports the same mma.sync block-scaled instructions as SM120 (RTX 5090) but was excluded from the native lowering path by exact compute capability checks. Without this fix, dot_scaled on SM121 falls through to DecomposeScaledBlocked which upcasts to bf16 — ~10 TFLOPS vs ~270 TFLOPS with native mma.sync block-scaled FP4. Tested on GB10 with both MXFP4 (scale_vec::2X, ue8m0) and NVFP4 (scale_vec::4X, ue4m3). # New contributor declaration - [ x] I am not making a trivial change, such as fixing a typo in a comment. - [ x] I have written a PR description following these [rules](https://cbea.ms/git-commit/#why-not-how). - [ x] I have run `pre-commit run --from-ref origin/main --to-ref HEAD`. - Select one of the following. - [ ] I have added tests. - `/test` for `lit` tests - `/unittest` for C++ tests - `/python/test` for end-to-end tests - [ x ] This PR does not need a test because current test paths cover the flow, though there are no GB10s in CI to verify AFAIK it does work for me. - Select one of the following. - [x ] I have not added any `lit` tests.
raymondtay
pushed a commit
to raymondtay/triton
that referenced
this pull request
Apr 18, 2026
…riton-lang#10010) SM121 (GB10 DGX Spark) supports the same mma.sync block-scaled instructions as SM120 (RTX 5090) but was excluded from the native lowering path by exact compute capability checks. Without this fix, dot_scaled on SM121 falls through to DecomposeScaledBlocked which upcasts to bf16 — ~10 TFLOPS vs ~270 TFLOPS with native mma.sync block-scaled FP4. Tested on GB10 with both MXFP4 (scale_vec::2X, ue8m0) and NVFP4 (scale_vec::4X, ue4m3). # New contributor declaration - [ x] I am not making a trivial change, such as fixing a typo in a comment. - [ x] I have written a PR description following these [rules](https://cbea.ms/git-commit/#why-not-how). - [ x] I have run `pre-commit run --from-ref origin/main --to-ref HEAD`. - Select one of the following. - [ ] I have added tests. - `/test` for `lit` tests - `/unittest` for C++ tests - `/python/test` for end-to-end tests - [ x ] This PR does not need a test because current test paths cover the flow, though there are no GB10s in CI to verify AFAIK it does work for me. - Select one of the following. - [x ] I have not added any `lit` tests.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
SM121 (GB10 DGX Spark) supports the same mma.sync block-scaled instructions as SM120 (RTX 5090) but was excluded from the native lowering path by exact compute capability checks.
Without this fix, dot_scaled on SM121 falls through to DecomposeScaledBlocked which upcasts to bf16 — ~10 TFLOPS vs ~270 TFLOPS with native mma.sync block-scaled FP4.
Tested on GB10 with both MXFP4 (scale_vec::2X, ue8m0) and NVFP4 (scale_vec::4X, ue4m3).
New contributor declaration
[ x] I am not making a trivial change, such as fixing a typo in a comment.
[ x] I have written a PR description following these
rules.
[ x] I have run
pre-commit run --from-ref origin/main --to-ref HEAD.Select one of the following.
/testforlittests/unittestfor C++ tests/python/testfor end-to-end testsSelect one of the following.
littests.