xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-11 01:25:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

Revert "[Bugfix] default set cuda_graph_sizes to max_num_seqs for v1 engine" (#20128)

Browse Source
This commit is contained in:
Michael Goin 2025-06-26 23:32:22 +09:00 committed by GitHub
parent 27c065df50
commit 1f5d178e9c
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
vllm/config.py
View File

@ -2042,12 +2042,11 @@ class SchedulerConfig:
NOTE: This will be replaced by speculative config in the future; it is NOTE: This will be replaced by speculative config in the future; it is
present to enable correctness tests until then.""" present to enable correctness tests until then."""
cuda_graph_sizes: list[int] = field(default_factory=list) cuda_graph_sizes: list[int] = field(default_factory=lambda: [512])
"""Cuda graph capture sizes """Cuda graph capture sizes, default is 512.
1. if none provided, then default set to [max_num_seqs] 1. if one value is provided, then the capture list would follow the
2. if one value is provided, then the capture list would follow the
pattern: [1, 2, 4] + [i for i in range(8, cuda_graph_sizes + 1, 8)] pattern: [1, 2, 4] + [i for i in range(8, cuda_graph_sizes + 1, 8)]
3. more than one value (e.g. 1 2 128) is provided, then the capture list 2. more than one value (e.g. 1 2 128) is provided, then the capture list
will follow the provided list.""" will follow the provided list."""
delay_factor: float = 0.0 delay_factor: float = 0.0
@ -2212,10 +2211,6 @@ class SchedulerConfig:
self.max_num_partial_prefills, self.max_long_partial_prefills, self.max_num_partial_prefills, self.max_long_partial_prefills,
self.long_prefill_token_threshold) self.long_prefill_token_threshold)
# If cuda_graph_sizes is not specified, default set to [max_num_seqs].
if not self.cuda_graph_sizes:
self.cuda_graph_sizes = [self.max_num_seqs]
@model_validator(mode='after') @model_validator(mode='after')
def _verify_args(self) -> Self: def _verify_args(self) -> Self:
if (self.max_num_batched_tokens < self.max_model_len if (self.max_num_batched_tokens < self.max_model_len