xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 04:26:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

Disable prefix cache by default for benchmark (#18639)

Browse Source 
Signed-off-by: cascade812 <cascade812@outlook.com>
This commit is contained in:
cascade 2025-05-27 05:06:34 -07:00 committed by GitHub
parent 06a0338015
commit aaa4ac1c95
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
benchmarks/benchmark_latency.py
View File

@ -189,5 +189,8 @@ if __name__ == "__main__":
)
parser = EngineArgs.add_cli_args(parser)
# V1 enables prefix caching by default which skews the latency
# numbers. We need to disable prefix caching by default.
parser.set_defaults(enable_prefix_caching=False)
args = parser.parse_args()
main(args)

3
vllm/benchmarks/latency.py
View File

@ -80,6 +80,9 @@ def add_cli_args(parser: argparse.ArgumentParser):
)
parser = EngineArgs.add_cli_args(parser)
# V1 enables prefix caching by default which skews the latency
# numbers. We need to disable prefix caching by default.
parser.set_defaults(enable_prefix_caching=True)
def main(args: argparse.Namespace):