[Reland][Blackwell] Decide TMEMCopy compatibility for block scales early#10571
Open
masahi wants to merge 9 commits into
Open
[Reland][Blackwell] Decide TMEMCopy compatibility for block scales early#10571masahi wants to merge 9 commits into
masahi wants to merge 9 commits into
Conversation
… into scale-tmem-copy-rewrite-revisit-ws-fix
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
#10530 caused perf regression for WS kernels, since WS passes were doing an exact SMEM-encoding equality check which was broken after tensordesc and local_alloc encodings become nominally different (
#nvmma_sharedvs#shared_linear) even though they describe the equivalent layout. We ended up placinglocal_allocafter desc_load into the default partition, which generates unncessarylocal_load/local_storeand synchronizations. This is fixed by the commit c366e57 using LinearLayout equivalence instead.Confirmed no perf regression on a WS version of
python/tutorials/10-block-scaled-matmul.pyandpython/triton_kernels/bench/bench_mlp.py.