[Fix][Relax] Lower non-contiguous WebGPU cumsum - #20133
Open
akaashrp wants to merge 1 commit into
Open
Conversation
akaashrp
changed the title
Contributor
Author
|
@tvm-bot rerun |
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR adds a direct WebGPU lowering for inclusive
cumsumalong a non-innermost axis. Previously, these operations used TOPI’s generic transpose-based GPU scan. That lowering materializes a full-size transposed tensor and expands the scan into multiple WebGPU dispatches. This is unnecessarily resource-heavy for WebGPU, particularly when the scan extent is symbolic.The new lowering reshapes any known-rank input into
[outer, scan, inner]without transposing its data. Each GPU thread handles one(outer, inner)pair and scans the middle dimension serially. The result is then reshaped back to its original shape. The existing parallel kernel is still in use for innermost-axis scans. The new kernel is a correctness fallback that exposes parallelism acrossouter * innerbut does not parallelize work along the scan dimension. Exclusive scans continue to use the existing lowering.