Skip to content

[P/D] force with_prefill true after allreduce in kv producer #3768

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
merged 2 commits into from
Oct 29, 2025
Merged

[P/D] force with_prefill true after allreduce in kv producer #3768

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1ceb857
force with_prefill true after allreduce in kv producer
liziyu179 Oct 25, 2025
7de561b
fix prefill
liziyu179 Oct 27, 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
2 changes: 1 addition & 1 deletion vllm_ascend/distributed/mooncake_layerwise_connector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1151,7 +1151,7 @@ def save_kv_layer(self, layer_name: str, kv_layer: Tuple[torch.Tensor,
connector_metadata: MooncakeLayerwiseConnectorMetadata,
**kwargs) -> None:
"""MooncakeLayerwiseConnector does not save explicitly."""
if self.kv_role == 'kv_producer':
if self.kv_role == 'kv_producer' and connector_metadata.request.keys():
if self.pd_head_ratio != 1:
if self.current_layer != 0:
self.completion_event.wait()
Expand Down
8 changes: 4 additions & 4 deletions vllm_ascend/worker/model_runner_v1.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2574,6 +2574,10 @@ def _dummy_run(
tp_size = self.vllm_config.parallel_config.tensor_parallel_size
num_tokens = math.ceil(num_tokens / tp_size) * tp_size

# Force dummy run on prefill stage when this node is deemed as kv producer.
if self.is_kv_producer and not self.is_kv_consumer:
with_prefill = True

# Padding for DP
(num_tokens, num_tokens_across_dp,
with_prefill) = self._sync_metadata_across_dp(num_tokens,
Expand Down Expand Up @@ -2622,10 +2626,6 @@ def _dummy_run(
num_scheduled_tokens = np.array(num_scheduled_tokens_list,
dtype=np.int32)

# Force dummy run on prefill stage when this node is deemed as kv producer.
if self.is_kv_producer and not self.is_kv_consumer:
with_prefill = True

if not self.in_profile_run and self.dynamic_eplb:
self.eplb_updator.forward_before()

Expand Down
Loading