Skip to content

fix prefill_params when prefill num_reqs > 1024 #1336

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
hiworldwzj merged 1 commit into from
Jun 8, 2026
Merged

fix prefill_params when prefill num_reqs > 1024 #1336

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
4 changes: 2 additions & 2 deletions lightllm/common/basemodel/triton_kernel/gen_prefill_params.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ def _gen_cumsum_pad0_kernel(

for start_index in range(0, size, BLOCK):
current_offs = start_index + offs
in_data = tl.load(b_q_seq_len + offs, mask=current_offs < size, other=0)
in_data = tl.load(b_q_seq_len + current_offs, mask=current_offs < size, other=0)
in_data = tl.cumsum(in_data) + start_value
start_value = tl.max(in_data, 0)
tl.store(b1_cu_q_seq_len + current_offs + 1, in_data, mask=current_offs < size)
Expand All @@ -30,7 +30,7 @@ def _gen_cumsum_pad0_kernel(
start_value = tl.cast(0, tl.int64)
for start_index in range(0, size, BLOCK):
current_offs = start_index + offs
in_data = tl.load(b_kv_seq_len + offs * b_kv_seq_len_stride_0, mask=current_offs < size, other=0)
in_data = tl.load(b_kv_seq_len + current_offs * b_kv_seq_len_stride_0, mask=current_offs < size, other=0)
in_data = tl.cumsum(in_data) + start_value
start_value = tl.max(in_data, 0)
tl.store(b1_cu_kv_seq_len + current_offs + 1, in_data, mask=current_offs < size)
Expand Down
Loading