Skip to content

fix: npu free host cache #78

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
weixiao-huang merged 3 commits into from
Jan 19, 2026
Merged

fix: npu free host cache #78

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4b6d203
fix: npu free host cache
Jan 15, 2026
e767a80
fix pre-commit
Jan 15, 2026
ca65a22
add comments
Jan 19, 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: 7 additions & 1 deletion checkpoint_engine/ps.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -407,7 +407,13 @@ def _unpin(t: torch.Tensor):
del self._memory_pool[checkpoint_name]
# see https://github.com/pytorch/pytorch/blob/31d5c675394705f8a6bc767f80ae14bf4f01246b/torch/csrc/cuda/Module.cpp#L2018
# this works by using torch>=2.5.0
torch._C._host_emptyCache()
if self.device_manager.device_type == "cuda":
torch._C._host_emptyCache()
else:
# torch._C._host_emptyCache() is not supported on NPU, so we call gc.collect() to empty host cache.
import gc

gc.collect()

def gather_metas(self, checkpoint_name: str):
"""
Expand Down