Skip to content

[TRTLLM-9431][perf] Enable multistream for Linear Attention in Qwen3-… #9696

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
nv-guomingz merged 1 commit into from
Dec 8, 2025
Merged

[TRTLLM-9431][perf] Enable multistream for Linear Attention in Qwen3-… #9696

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
33 changes: 25 additions & 8 deletions tensorrt_llm/_torch/models/modeling_qwen3_next.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -647,11 +647,10 @@ def fused_gdn_gating(

class Qwen3NextGatedDeltaNet(nn.Module):

def __init__(
self,
model_config: ModelConfig[Qwen3NextConfig],
layer_idx: Optional[int] = None,
):
def __init__(self,
model_config: ModelConfig[Qwen3NextConfig],
aux_stream: torch.cuda.Stream,
layer_idx: Optional[int] = None):
super().__init__()
config = model_config.pretrained_config
self.model_config = model_config
Expand Down Expand Up @@ -778,6 +777,12 @@ def __init__(
force_dynamic_quantization=model_config.force_dynamic_quantization,
use_cute_dsl_blockscaling_mm=False)

self.event_dict = {
key: torch.cuda.Event()
for key in [EventType.Main, EventType.Attention]
}
self.aux_stream = aux_stream

def fix_query_key_value_ordering(self, mixed_qkvz, mixed_ba):
"""
Derives `query`, `key` and `value` tensors from `mixed_qkvzba`.
Expand Down Expand Up @@ -1032,8 +1037,19 @@ def forward(
ssm_states[state_indices_p] = 0
# conv_states[state_indices_p] = 0 # not necessary

projected_states_qkvz = self.in_proj_qkvz(hidden_states)
projected_states_ba = self.in_proj_ba(hidden_states)
def _compute_projected_states_qkvz():
return self.in_proj_qkvz(hidden_states)

def _compute_projected_states_ba():
return self.in_proj_ba(hidden_states)

projected_states_qkvz, projected_states_ba = maybe_execute_in_parallel(
_compute_projected_states_qkvz,
_compute_projected_states_ba,
self.event_dict[EventType.Main],
self.event_dict[EventType.Attention],
self.aux_stream,
)

# Use fused kernel when possible to avoid elementwise ops
if self.num_v_heads // self.num_k_heads in [1, 2,
Expand Down Expand Up @@ -1098,7 +1114,8 @@ def __init__(
super().__init__()
self.model_config = model_config
config = model_config.pretrained_config
self.linear_attn = Qwen3NextGatedDeltaNet(model_config, layer_idx)
self.linear_attn = Qwen3NextGatedDeltaNet(model_config, aux_stream,
layer_idx)

self.mapping = model_config.mapping
self.enable_attention_dp = self.mapping.enable_attention_dp
Expand Down