Skip to content

bugfix: skip empty safetensors file in inplace_pin_memory #79

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
blahgeek merged 2 commits into from
Jan 20, 2026
Merged

bugfix: skip empty safetensors file in inplace_pin_memory #79

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
0eef081
misc: translate cuda error code to string when pin and unpin
HubertZhang Jan 19, 2026
b4bed4b
bugfix: skip empty safetensors file when inplace pin memory
HubertZhang 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
10 changes: 9 additions & 1 deletion checkpoint_engine/pin_memory.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -209,7 +209,9 @@ def _pin(t: torch.Tensor):
torch.cuda.set_device(device_index)
cudart = torch.cuda.cudart()
r = cudart.cudaHostRegister(t.data_ptr(), t.numel() * t.element_size(), 0)
assert r == 0, f"pin memory error, error code: {r}"
if r != 0:
error_msg = cudart.cudaGetErrorString(r)
raise RuntimeError(f"pin memory error, error code: {r}, error message: {error_msg}")

# TODO: should only support /dev/shm? but we found files in disk also work?
size = os.stat(file_path).st_size
Expand Down Expand Up @@ -254,6 +256,12 @@ def _pin(t: torch.Tensor):
# Remove the file after successfully loading. This will avoid doubling the memory usage.
# We assume files in /dev/shm/ are temporary files. So it's safe to remove them after loading.
os.remove(file_path)
if not metas:
# TODO: should we still return this buffer?
Copy link
Collaborator

@blahgeek blahgeek Jan 20, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah skipping this buffer seems to be a better idea, to prevent future bugs. (not a hard requirement

assert buffer.nbytes == 0, f"buffer nbytes {buffer.nbytes} should be 0"
logger.warning(f"[rank{rank}] no metas found in {file_path}, skip pin memory")
return MemoryBuffer(buffer=buffer, size=buffer.nbytes, metas=[], manually_pinned=False)

_pin(buffer)
logger.info(
f"[rank{rank}] inplace pin memory for file {file_path} finished, size {buffer.nbytes / 1024 / 1024:.2f}MiB"
Expand Down
6 changes: 5 additions & 1 deletion checkpoint_engine/ps.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -391,7 +391,11 @@ def _unpin(t: torch.Tensor):
)
cudart = torch.cuda.cudart()
r = cudart.cudaHostUnregister(t.data_ptr())
assert r == 0, f"unpin memory error, error code: {r}"
if r != 0:
error_msg = cudart.cudaGetErrorString(r)
raise RuntimeError(
f"unpin memory error, error code: {r}, error message: {error_msg}"
)

# if the checkpoint is pinned by cudaHostRegister manually, we need to unpin it manually
try:
Expand Down