Fix: GaussianProjectionBackward dLossDQuat missing warpSum#435
Merged
swahtz merged 2 commits intoopenvdb:mainfrom Feb 9, 2026
Merged
Fix: GaussianProjectionBackward dLossDQuat missing warpSum#435swahtz merged 2 commits intoopenvdb:mainfrom
GaussianProjectionBackward dLossDQuat missing warpSum#435swahtz merged 2 commits intoopenvdb:mainfrom
Conversation
Signed-off-by: Jonathan Swartz <jonathan@jswartz.info>
swahtz
added
bug
swahtz
requested review from
Copilot
and removed request for
harrism and
sifakis
Contributor
There was a problem hiding this comment.
Pull request overview
This PR fixes a critical bug in the CUDA kernel for Gaussian projection backward pass where gradient accumulation for quaternion parameters was incomplete. The missing warpSum operation caused only a single thread's gradients to be used instead of properly summing gradients across all threads in a warp group.
Changes:
- Added missing
warpSum(dLossDQuat, warp_group_g)call before writing quaternion gradients to ensure proper reduction across threads
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
harrism
changed the title
GaussianProjectxionBackward dLossDQuat missing warpSumGaussianProjectionBackward dLossDQuat missing warpSum
harrism
approved these changes
Feb 3, 2026
Signed-off-by: Jonathan Swartz <jonathan@jswartz.info>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixed
dLossDQuatmissing thewarpSumwhere gradients across all cameras need to be summed before writing to output. The other quantities (point, covar, scale) are correctly summed butdLossDQuatwas missing awarpSum, leading to only the gradients from thread rank 0 (1 camera) to be used for this quantity.