[fp8 blockwise training] improve fp8 blockwise gemm perf; add GEMM bench script #2730
+200
−25
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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/2730
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 215e891 with merge base 1526dfe ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
danielvegamyhre
added
the
topic: not user facing
meta-cla
bot
added
the
CLA Signed
danielvegamyhre
force-pushed
the
perf2
branch
2 times, most recently
from
fc0fdaf to
252da1e
Compare
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Outdated suggestions cannot be applied.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.
Repro for triton compiler upcasting fp8 inputs to f16 and doing mma.sync f16.f16.fp32:
python benchmarks/blockwise_fp8_training/bench_gemms.pyand use NCU see associated PTXblockwise_fp8_gemm_1x128_128x1_kernel(accepts "A" scaled with 1x128 granularity and "B" scaled with 128x128 granularity)