Skip to content

[Performance] qwen3-next improve causal conv1d in prefill phase #10595

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
zhyncs merged 4 commits into from
Sep 18, 2025
Merged

[Performance] qwen3-next improve causal conv1d in prefill phase #10595

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
fdbebd6
improve causal conv1d in prefill phase
jychen21 Sep 18, 2025
dc0ef6e
Merge branch 'main' into prefill_causal_conv1d
liz-badada Sep 18, 2025
26089fb
code style
jychen21 Sep 18, 2025
37ca485
Merge branch 'main' into prefill_causal_conv1d
liz-badada Sep 18, 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
5 changes: 4 additions & 1 deletion python/sglang/srt/layers/attention/hybrid_linear_attn_backend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,9 @@
from sglang.srt.layers.attention.fla.fused_sigmoid_gating_recurrent import (
fused_sigmoid_gating_delta_rule_update,
)
from sglang.srt.layers.attention.mamba.causal_conv1d import (
causal_conv1d_fn as causal_conv1d_fn_sgl,
)
from sglang.srt.layers.attention.mamba.causal_conv1d_triton import (
causal_conv1d_fn,
causal_conv1d_update,
Expand Down Expand Up @@ -334,7 +337,7 @@ def forward_extend(
mixed_qkv_processed.transpose(1, 2).contiguous().view(seq_len, -1)
)
else:
mixed_qkv = causal_conv1d_fn(
mixed_qkv = causal_conv1d_fn_sgl(
mixed_qkv.transpose(0, 1),
conv_weights,
bias,
Expand Down
Loading