Repository navigation
feat(convolution): add accelerated split weight gradients - #1083
Merged
Merged
Conversation
Contributor
Author
|
@laggui here is the cubek change for the accelerated split wgrad burn pr |
laggui
reviewed
Nov 5, 2026
Comment on lines
+480
to
+481
| #[cube(comptime)] | ||
| channels: u32, |
Member
There was a problem hiding this comment.
channels receives padded_channels, which is a problem extent rather than a kernel setting. Marking it as comptime creates separate JIT entries for channel counts that could otherwise a kernel.
This field should be runtime, consistent with other problem sizes.
| /// The output has shape `[out_channels, ...kernel, in_channels]`, with rank `N + 2`. | ||
| /// `SPATIAL_LAST` stores partial columns with spatial axes inside the channel axis. | ||
| /// F16 and BF16 partial sums stay F32 until the final output write. | ||
| pub fn launch_accelerated<const N: usize, const SPATIAL_LAST: bool>( |
Member
There was a problem hiding this comment.
Could this be a named strategy value instead of a boolean const generic, following the existing convolution APIs?
Passing it as a cubecl comptime argument would retain specialized GPU code and let callers select the ordering through a single launch call.
laggui
approved these changes
Nov 6, 2026
laggui
left a comment
laggui
left a comment
Member
There was a problem hiding this comment.
Thanks, LGTM!
Can you update the burn PR?
laggui
changed the title
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Instructions
Changes
Extend the split-weight gradient pipeline to support an accelerated kernel
Burn PR
tracel-ai/burn#6319