Skip to content

Fix cuda memory allocation issue caused by fused_linear_act.py #1822

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
emailweixu merged 1 commit into from
Nov 12, 2025
Merged

Fix cuda memory allocation issue caused by fused_linear_act.py #1822

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
6 changes: 2 additions & 4 deletions alf/ext/fused_linear_act.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,10 +29,7 @@
class StaticState:
workspace = {}
workspace_size = 1024 * 1024 * 8
bias_g = {
idx: torch.tensor([], dtype=torch.float16).cuda(idx)
for idx in range(torch.cuda.device_count())
}
bias_g = {}

@classmethod
def get(cls, name: str, device: torch.device) -> Any:
Expand All @@ -41,6 +38,7 @@ def get(cls, name: str, device: torch.device) -> Any:
cls.workspace[idx] = torch.empty((cls.workspace_size, ),
dtype=torch.uint8,
device=device).cuda(idx)
cls.bias_g[idx] = torch.tensor([], dtype=torch.float16).cuda(idx)
if name == "bias":
return cls.bias_g[idx]
if name == "workspace":
Expand Down