Skip to content
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

[fix] fix bug of argmax.py when the dim-th dimension is less than BLOCK_M #476

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

[fix] fix bug of argmax.py when the dim-th dimension is less than BLOCK_M #476

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
11 changes: 10 additions & 1 deletion src/flag_gems/ops/argmax.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,6 +124,15 @@ def argmax(inp, dim=None, keepdim=False, *, dtype=None):
M = math.prod(shape[:dim])
K = inp.numel() // M // N

# The special case of `M=1`
if M == 1:
shape_list = list(shape)
shape_list[dim] = 1
out_index = torch.zeros(shape_list, dtype=torch.int64, device=inp.device)
if not keepdim:
out_index = torch.squeeze(out_index, dim)
return out_index

inp = inp.contiguous()

shape_list = list(shape)
Expand All @@ -133,7 +142,7 @@ def argmax(inp, dim=None, keepdim=False, *, dtype=None):
out_index = torch.squeeze(out_index, dim)

grid = lambda meta: (
triton.cdiv(M, meta["BLOCK_M"]),
triton.cdiv(M, min(meta["BLOCK_M"], M)),
K,
)
with torch_device_fn.device(inp.device):
Expand Down
Loading