mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 12:21:27 +01:00
a259fbf72c
Follows #134655 Pull Request resolved: https://github.com/pytorch/pytorch/pull/136634 Approved by: https://github.com/Skylion007
17 lines
546 B
C++
17 lines
546 B
C++
#include <torch/csrc/lazy/core/config.h>
|
|
|
|
// TODO(whc) unclear if this is useful, has only been tested as true
|
|
// NOLINTNEXTLINE(misc-use-internal-linkage)
|
|
C10_DEFINE_bool(
|
|
torch_lazy_ts_tensor_update_sync,
|
|
true,
|
|
"Use synchronous copy inside _copy_from op");
|
|
|
|
// TODO(whc) we need to hook up these flags in a more useful way
|
|
// possibly also keep LTC_TS_CUDA env working?
|
|
// NOLINTNEXTLINE(misc-use-internal-linkage)
|
|
C10_DEFINE_bool(
|
|
torch_lazy_ts_cuda,
|
|
false,
|
|
"Use cuda device for torchscript backend (instead of CPU)");
|