[Pallas] Slice store values to match clamped Pallas BlockSpec ref shape#2398
Open
thcmbs wants to merge 1 commit into
Open
[Pallas] Slice store values to match clamped Pallas BlockSpec ref shape#2398thcmbs wants to merge 1 commit into
thcmbs wants to merge 1 commit into
Conversation
meta-cla
Bot
added
the
CLA Signed
thcmbs
changed the title
thcmbs
force-pushed
the
pallas-store-slice-clamped-ref
branch
from
4f4670e to
a7279ac
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
When block_size > tensor.shape[dim], the launcher clamps the BlockSpec to min(block_size, dim_size) but the kernel computes block_size-shaped values. Slice the value before storing to avoid shape mismatch.
Only applies to grid-tiled dimensions that produce ":" in index_str; pl.ds() dimensions are padded instead of clamped and need no slicing.
Unblocks kl_div example on TPU.
Ideally, we might need to work on the autotuner to not fall into that case too often, but we still need to fix it in case a user wants to force it for some reason.