Skip to content

GetPlaceFromPtr support GPUPinnedPlace #78676

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
zhoutianzi666 merged 4 commits into from
Apr 16, 2026
Merged

GetPlaceFromPtr support GPUPinnedPlace #78676

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
e80bd5f
GetPlaceFromPtr support GPUPinnedPlace
ming1753 Apr 14, 2026
cd63bb8
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
ming1753 Apr 14, 2026
7761dad
add test
ming1753 Apr 15, 2026
b673423
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
ming1753 Apr 16, 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
8 changes: 6 additions & 2 deletions paddle/phi/api/lib/tensor_utils.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,8 +38,12 @@ PADDLE_API phi::Place GetPlaceFromPtr(void* data) {
#ifdef PADDLE_WITH_CUDA
cudaPointerAttributes attr = {};
cudaError_t status = cudaPointerGetAttributes(&attr, data);
if (status == cudaSuccess && attr.type == cudaMemoryTypeDevice) {
return phi::GPUPlace(attr.device);
if (status == cudaSuccess) {
if (attr.type == cudaMemoryTypeDevice) {
return phi::GPUPlace(attr.device);
} else if (attr.type == cudaMemoryTypeHost) {
return phi::GPUPinnedPlace();
}
}
#else
hipPointerAttribute_t attr = {};
Expand Down
9 changes: 9 additions & 0 deletions test/cpp/phi/api/test_from_blob.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -107,6 +107,15 @@ TEST(GetPlaceFromPtr, GPU) {
ASSERT_EQ(gpu1_data_place, phi::GPUPlace(1));
std::cout << "gpu1_data_place: " << gpu1_data_place << std::endl;
}

// Test GPUPinnedPlace (cudaMemoryTypeHost)
auto pinned_alloc_ptr =
paddle::GetAllocator(phi::GPUPinnedPlace())->Allocate(sizeof(cpu_data));
float* pinned_data = static_cast<float*>(pinned_alloc_ptr->ptr());
auto pinned_data_place = GetPlaceFromPtr(pinned_data);
ASSERT_EQ(pinned_data_place, phi::GPUPinnedPlace());
std::cout << "pinned_data_place: " << pinned_data_place << std::endl;
pinned_alloc_ptr.release();
}

TEST(from_blob, GPU) {
Expand Down