xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-22 07:11:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

[perf] fix perf regression from #12253 (#12380)

Browse Source 
Signed-off-by: youkaichao <youkaichao@gmail.com>
This commit is contained in:
youkaichao 2025-01-24 11:34:27 +08:00 committed by GitHub
parent 0e74d797ce
commit 6dd94dbe94
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/worker/model_runner.py
View File

@ -455,7 +455,6 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
self.enable_prompt_adapter = (self.runner.prompt_adapter_config self.enable_prompt_adapter = (self.runner.prompt_adapter_config
is not None) is not None)
self.multi_modal_input_mapper = self.runner.multi_modal_input_mapper self.multi_modal_input_mapper = self.runner.multi_modal_input_mapper
self.decode_only = True
# Attention metadata inputs. # Attention metadata inputs.
if self.attn_backend is not None: if self.attn_backend is not None:
@ -477,6 +476,10 @@ class ModelInputForGPUBuilder(ModelRunnerInputBuilderBase[ModelInputForGPU]):
finished_requests_ids: Optional[List[str]] = None) -> None: finished_requests_ids: Optional[List[str]] = None) -> None:
self.finished_requests_ids = finished_requests_ids self.finished_requests_ids = finished_requests_ids
# if the current batch is decode-only.
# will be set to False if there is any non-decode request.
self.decode_only = True
# Intermediate data (data in CPU before going to GPU) for # Intermediate data (data in CPU before going to GPU) for
# the current sequence group. # the current sequence group.
self.inter_data_list: List[ self.inter_data_list: List[