Skip to content

Optimize prepare_lens by removing device transfer #289

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
iforgetmyname merged 2 commits into from
Dec 29, 2025
Merged

Optimize prepare_lens by removing device transfer #289

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f73d7f1
Optimize prepare_lens by removing device transfer
shengzhaotian Dec 29, 2025
2cf3a42
Refactor prepare_lens to simplify tensor conversion
shengzhaotian Dec 29, 2025
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 python/sgl_kernel_npu/sgl_kernel_npu/fla/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -135,10 +135,8 @@ def wrapper(*args: Any, **kwargs: Any) -> Any:

@tensor_cache
def prepare_lens(cu_seqlens: torch.LongTensor) -> torch.LongTensor:
device = cu_seqlens.device
cu_seqlens = cu_seqlens.to(torch.int64).cpu().clone()
tmp = cu_seqlens[1:] - cu_seqlens[:-1]
return tmp.to(device)
cu_seqlens_i64 = cu_seqlens.to(torch.int64)
return cu_seqlens_i64[1:] - cu_seqlens_i64[:-1]


@tensor_cache
Expand Down