Use stream-aware scratch buffer in CUDA DeformConv - #31642
Merged
Akshay Sonawane (apsonawane) merged 5 commits intoAug 7, 2026
Conversation
Replace untagged temp allocation for col_buffer with GetScratchBuffer(..., compute_stream) to preserve StreamAwareBFCArena cross-stream safety.
The CUDA plugin build uses the adapter OpKernelContext, which exposes GetGPUComputeStream() rather than GetComputeStream(). Update the scratch buffer allocation site to use the plugin-compatible stream accessor so the file builds cleanly in the plugin configuration. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Akshay Sonawane (apsonawane)
enabled auto-merge (squash)
August 5, 2026 17:47
Use OpKernelContext::GetComputeStream() for the scratch buffer allocation in the CUDA DeformConv object build. This build path does not expose GetGPUComputeStream(), so the previous call caused a compilation failure. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Tianlei Wu (tianleiwu)
approved these changes
Aug 7, 2026
auto-merge was automatically disabled
August 7, 2026 16:32
Pull request was closed
Akshay Sonawane (apsonawane)
enabled auto-merge (squash)
August 7, 2026 16:33
Akshay Sonawane (apsonawane)
deleted the
msrc/deformconv-colbuffer-stream-fix
branch
August 7, 2026 18:23
This was referenced Aug 12, 2026
Closed
This was referenced Aug 17, 2026
Open
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 pull request updates the scratch buffer allocation method in the
DeformConvCUDA kernel implementation. The main change is a switch from using a generic allocator interface to a more specialized scratch buffer utility, which may improve performance and code clarity.Memory allocation improvements:
IAllocator::MakeUniquePtr<T>and manual allocator retrieval withGetScratchBuffer<T>for allocating thecol_bufferindeform_conv.cc. This streamlines temporary buffer allocation and ensures the buffer is tied to the compute stream.