Skip to content

[Fix] fix FlashMLA cudagraph config #4591

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
wants to merge 1 commit into from
Closed

[Fix] fix FlashMLA cudagraph config #4591

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
7 changes: 5 additions & 2 deletions python/sglang/srt/layers/attention/flashmla_backend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -92,7 +92,7 @@ def init_forward_metadata(self, forward_batch: ForwardBatch):
if forward_batch.forward_mode.is_decode_or_idle():
if spec_info is None:
max_seqlen_pad = triton.cdiv(
forward_batch.seq_lens.max().item(), PAGE_SIZE
forward_batch.decode_seq_lens_cpu.max().item(), PAGE_SIZE
)
block_kv_indices = torch.full(
(bs, max_seqlen_pad),
Expand Down Expand Up @@ -196,6 +196,7 @@ def init_forward_metadata_capture_cuda_graph(
def init_forward_metadata_replay_cuda_graph(
self,
bs: int,
num_kv_heads: int,
req_pool_indices: torch.Tensor,
seq_lens: torch.Tensor,
seq_lens_sum: int,
Expand All @@ -206,8 +207,10 @@ def init_forward_metadata_replay_cuda_graph(
):

if forward_mode.is_decode_or_idle():
assert seq_lens_cpu is not None
seq_lens = seq_lens[:bs]
max_seqlen_pad = triton.cdiv(seq_lens.max().item(), PAGE_SIZE)
seq_lens_cpu = seq_lens_cpu[:bs]
max_seqlen_pad = triton.cdiv(seq_lens_cpu.max().item(), PAGE_SIZE)
create_flashmla_kv_indices_triton[(bs,)](
self.req_to_token,
req_pool_indices[:bs],
Expand Down
Loading