OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-06 12:20:52 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity

clang-format CUDASymmetricMemory.cu (#151260)

Browse Source 
Ported from #146592

Pull Request resolved: https://github.com/pytorch/pytorch/pull/151260
Approved by: https://github.com/Skylion007
This commit is contained in:
Yifu Wang 2025-04-14 12:51:09 -07:00 committed by PyTorch MergeBot
parent 25803d3a22
commit e178a3aa94
1 changed files with 20 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
torch/csrc/distributed/c10d/CUDASymmetricMemory.cu
View File

@ -88,7 +88,9 @@ class IpcChannel {
}
TORCH_CHECK(
sendmsg(socket_, &msg, 0) > 0, "Failed to send fd: ", c10::utils::str_error(errno));
sendmsg(socket_, &msg, 0) > 0,
"Failed to send fd: ",
c10::utils::str_error(errno));
}
int recv_fd() {
@ -256,8 +258,15 @@ void map_block(
namespace c10d {
namespace symmetric_memory {
AllocationRef::AllocationRef(void* ptr, HandleType handle, size_t block_size, int device_idx)
: ptr(ptr), handle(handle), block_size(block_size), device_idx(device_idx) {}
AllocationRef::AllocationRef(
void* ptr,
HandleType handle,
size_t block_size,
int device_idx)
: ptr(ptr),
handle(handle),
block_size(block_size),
device_idx(device_idx) {}
AllocationRef::~AllocationRef() {
#if !defined(USE_ROCM) && defined(PYTORCH_C10_DRIVER_API_SUPPORTED)
@ -572,12 +581,12 @@ int CUDASymmetricMemory::get_world_size() {
}
Block::Block(
c10::intrusive_ptr<AllocationRef> alloc_ref,
int device_idx,
size_t block_size,
size_t buffer_size,
size_t signal_pad_offset,
const std::optional<std::string>& group_name)
c10::intrusive_ptr<AllocationRef> alloc_ref,
int device_idx,
size_t block_size,
size_t buffer_size,
size_t signal_pad_offset,
const std::optional<std::string>& group_name)
: alloc_ref(std::move(alloc_ref)),
device_idx(device_idx),
block_size(block_size),
@ -618,7 +627,8 @@ void* CUDASymmetricMemoryAllocator::alloc(
AT_CUDA_CHECK(cudaMemset(ptr, 0, block_size));
auto alloc_ref = c10::make_intrusive<AllocationRef>(ptr, handle, block_size, device_idx);
auto alloc_ref =
c10::make_intrusive<AllocationRef>(ptr, handle, block_size, device_idx);
auto block = c10::make_intrusive<Block>(
std::move(alloc_ref),
device_idx,