Commits

Commits on Apr 25, 2025

Refactor to use torch.accelerator.device_index instead of torch.cuda.device for generic device context manager (#148880)

guangyey
authored and
pytorchmergebot
committed
Add torch.accelerator.device_index as accelerator's device switch context (#148864)

guangyey
authored and
pytorchmergebot
committed

Commits on Apr 22, 2025

Deprecate host allocator legacy APIs (#151437)

guangyey
authored and
pytorchmergebot
committed

Commits on Apr 18, 2025

Add HostAllocator as the unified parent class (#151431)

guangyey
authored and
pytorchmergebot
committed

Commits on Apr 17, 2025

Refine host caching allocator (#151403)

guangyey
authored and
pytorchmergebot
committed

Commits on Apr 13, 2025

Generalize poison fork logic for each device backend (#144664)

guangyey
authored and
pytorchmergebot
committed

Commits on Apr 10, 2025

Delegate torch.accelerator.device_count to torch.xxx.device_count for multi-process usage (#149924)

guangyey
authored and
pytorchmergebot
committed
Document poison fork note for accelerator APIs (#147507)

guangyey
authored and
pytorchmergebot
committed
Generalize poison fork logic for each device backend (#144664)

guangyey
authored and
pytorchmergebot
committed

Commits on Apr 7, 2025

Generalize poison fork logic for each device backend (#144664)

guangyey
authored and
pytorchmergebot
committed

Commits on Mar 24, 2025

Add default XPU toolkit path to CMake (#149270)

guangyey
authored and
pytorchmergebot
committed
Reuse format_size utils (#149383)

guangyey
authored and
pytorchmergebot
committed

Commits on Mar 17, 2025

Refine XPU oneDNN context manager API (#147349)

guangyey
authored and
pytorchmergebot
committed

Commits on Feb 13, 2025

Friendly handle mem_get_info's runtime error message (#146899)

guangyey
authored and
pytorchmergebot
committed

Commits on Jan 31, 2025

[Break XPU] Fix Inductor cuda bias UT (#145934)

guangyey
authored and
pytorchmergebot
committed

Commits on Jan 29, 2025

Filter out iGPU if dGPU is found on XPU (#144378)

guangyey
authored and
pytorchmergebot
committed

Commits on Jan 23, 2025

Generalize pin memory logic for accelerator when non blocking copy happened (#143783)

guangyey
authored and
pytorchmergebot
committed
Introduce a new API isAcceleratorExcluded (#144959)

guangyey
authored and
pytorchmergebot
committed

Commits on Jan 17, 2025

Use torch with statement in torch distributed module (#144951)

guangyey
authored and
pytorchmergebot
committed

Commits on Jan 14, 2025

[RELAND] Generalize at::manual_seed for all accelerators (#144370)

guangyey
authored and
pytorchmergebot
committed

Commits on Jan 10, 2025

Commits on Dec 31, 2024

Refine CUDA Stream priority (#143849)

guangyey
authored and
pytorchmergebot
committed
Add get_stream_from_external API for CUDA backend (#143799)

guangyey
authored and
pytorchmergebot
committed
Add get_stream_from_external API for XPU backend (#141123)

guangyey
authored and
pytorchmergebot
committed
Add low priority XPU Stream (#141119)

guangyey
authored and
pytorchmergebot
committed
Refine XPU external Stream (#142347)

guangyey
authored and
pytorchmergebot
committed

Commits on Dec 23, 2024

Fix torch.accelerator api abort when passing invaild device (#143550)

guangyey
authored and
pytorchmergebot
committed

Commits on Dec 16, 2024

[RELAND] Add device-agnostic runtime Device/Stream C++ API (#138677)

guangyey
authored and
pytorchmergebot
committed
[RELAND] Add UTs for accelerator device-agnostic runtime APIs (#133572)

guangyey
authored and
pytorchmergebot
committed
[MPS] Support torch.accelerator.synchronize() on mps (#143171)

guangyey
authored and
pytorchmergebot
committed

Commits on Dec 11, 2024

[RELAND] Add device-agnostic runtime Device/Stream C++ API (#138677)

guangyey
authored and
pytorchmergebot
committed
[RELAND] Add UTs for accelerator device-agnostic runtime APIs (#133572)

guangyey
authored and
pytorchmergebot
committed