Skip to content

Use cudaMemcpyAsync rather than kernel when possible #1088

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
merged 1 commit into from
Nov 12, 2025
Merged

Use cudaMemcpyAsync rather than kernel when possible #1088

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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: 2 additions & 2 deletions include/matx/core/tensor.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -741,7 +741,7 @@ class tensor_t : public detail::tensor_impl_t<T,RANK,Desc> {

int dev;
cudaGetDevice(&dev);
#if CUDA_VERSION <= 12000
#if CUDART_VERSION <= 12000
cudaMemPrefetchAsync(this->Data(), this->desc_.TotalSize() * sizeof(T), dev, stream);
#else
cudaMemLocation loc;
Expand All @@ -765,7 +765,7 @@ class tensor_t : public detail::tensor_impl_t<T,RANK,Desc> {
{
MATX_NVTX_START("", matx::MATX_NVTX_LOG_API)

#if CUDA_VERSION <= 12000
#if CUDART_VERSION <= 12000
cudaMemPrefetchAsync(this->Data(), this->desc_.TotalSize() * sizeof(T), cudaCpuDeviceId,
stream);
#else
Expand Down
22 changes: 22 additions & 0 deletions include/matx/operators/base_operator.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -197,6 +197,28 @@ namespace matx

tp->TransformExec(tp->Shape(), ex);
}
else if constexpr (is_tensor_view_v<typename T::tensor_type> && is_tensor_view_v<typename T::op_type> && is_cuda_executor_v<Ex>) {
// If we are doing a tensor to tensor assignment we should prefer cudaMemcpyAsync instead of a kernel
if (detail::check_aliased_memory(tp->get_lhs(), tp->get_rhs(), true)) {
MATX_THROW(matxInvalidParameter, "Possible aliased memory detected: LHS and RHS memory ranges overlap");
}

if (tp->get_lhs().IsContiguous() && tp->get_rhs().IsContiguous() && tp->get_lhs().Rank() == tp->get_rhs().Rank()) {
MATX_ASSERT_STR(tp->get_lhs().Bytes() >= tp->get_rhs().Bytes(), matxInvalidSize, "LHS tensor is smaller than RHS tensor in assignment");
MATX_LOG_TRACE("Copying {} bytes from {} to {} using cudaMemcpyAsync",
tp->get_lhs().Bytes(), reinterpret_cast<void*>(tp->get_rhs().Data()), reinterpret_cast<void*>(tp->get_lhs().Data()));
cudaMemcpyAsync(reinterpret_cast<void*>(tp->get_lhs().Data()),
reinterpret_cast<void*>(tp->get_rhs().Data()),
tp->get_rhs().Bytes(),
cudaMemcpyDefault,
ex.getStream());
}
else {
MATX_LOG_TRACE("Copying {} bytes from {} to {} using kernel",
tp->get_lhs().Bytes(), reinterpret_cast<void*>(tp->get_rhs().Data()), reinterpret_cast<void*>(tp->get_lhs().Data()));
ex.Exec(*tp);
}
}
else {
if (detail::check_aliased_memory(tp->get_lhs(), tp->get_rhs(), true)) {
MATX_THROW(matxInvalidParameter, "Possible aliased memory detected: LHS and RHS memory ranges overlap");
Expand Down