Skip to content

Perf: allow tf32 datatype for matmul #4499

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

Closed
caic99 wants to merge 1 commit into from
Closed

Perf: allow tf32 datatype for matmul #4499

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
3 changes: 2 additions & 1 deletion deepmd/pt/entrypoints/main.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -103,7 +103,8 @@ def get_trainer(
finetune_links=None,
):
multi_task = "model_dict" in config.get("model", {})

# https://pytorch.org/docs/stable/notes/cuda.html#tf32-on-ampere
torch.backends.cuda.matmul.allow_tf32 = True
# Initialize DDP
local_rank = os.environ.get("LOCAL_RANK")
if local_rank is not None:
Expand Down