Skip to content

Use stream-aware scratch buffer in CUDA DeformConv #31642

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Akshay Sonawane (apsonawane) merged 5 commits into from
Aug 7, 2026
Merged

Use stream-aware scratch buffer in CUDA DeformConv #31642

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
0316ba0
Use stream-aware scratch buffer in CUDA DeformConv
apsonawane Aug 4, 2026
99790e7
Merge branch 'main' into msrc/deformconv-colbuffer-stream-fix
apsonawane Aug 5, 2026
9f234ae
Fix CUDA plugin scratch buffer stream call in deform conv
apsonawane Aug 5, 2026
9df2ff3
Fix CUDA DeformConv build break
apsonawane Aug 5, 2026
7fffc2f
Fix PR31642 CUDA plugin stream accessor build break
apsonawane Aug 6, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 1 addition & 3 deletions onnxruntime/core/providers/cuda/nn/deform_conv.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -164,9 +164,7 @@ Status DeformConv<T>::ComputeInternal(OpKernelContext* context) const {
const int64_t col_stride = static_cast<int64_t>(n_parallel_imgs) * output_image_size;
const int64_t col_buffer_size = (C * kernel_size) * col_stride;

AllocatorPtr alloc;
ORT_RETURN_IF_ERROR(context->GetTempSpaceAllocator(&alloc));
auto col_buffer = IAllocator::MakeUniquePtr<T>(alloc, SafeInt<size_t>(col_buffer_size));
auto col_buffer = GetScratchBuffer<T>(SafeInt<size_t>(col_buffer_size), GetComputeStream(context));

const T* Xdata = X->Data<T>();
const T* Wdata = W->Data<T>();
Expand Down
Loading