pytorch

mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 12:54:11 +08:00

Files

Ke Wen b291dc9684 [2/N][SymmMem] Add MemPool allocator and tests (#161471 )

(Porting most of #161008)

Hooking SymmetricMemory Allocator to MemPool so that user can create symmetric tensors with regular `torch.zeros`, `torch.arange` etc factories. Also so that our ops can have functional variants that create `out` tensors on symmetric memory.

To end users, this PR supports a python UI as follows:
```
allocator = symm_mem.get_mempool_allocator(device)
mempool = torch.cuda.MemPool(allocator)
with torch.cuda.use_mem_pool(mempool):
    tensor = torch.arange(numel, dtype=dtype, device=device)
```

Added tests for both use cases above.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/161471
Approved by: https://github.com/ngimel
ghstack dependencies: #161470

2025-08-28 06:31:29 +00:00

core

Record the XPU and XCCL build settings in the compiled binary (#147161 )

2025-05-20 09:21:39 +00:00

perfkernels

[2/N] Fix cppcoreguidelines-init-variables suppression (#146237 )

2025-06-19 23:26:42 +00:00

serialize

[2/N] Fix cppcoreguidelines-init-variables suppression (#146237 )

2025-06-19 23:26:42 +00:00

utils

[3/N] Use internal linkage in C++ files (#151297 )