biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Add VLLM_ENABLE_INDUCTOR_MAX_AUTOTUNE & VLLM_ENABLE_INDUCTOR_COORDINA… (#25493)

Browse Source 
Signed-off-by: rouchenzi <ruochenwen@gmail.com>
Signed-off-by: rouchenzi <40842833+rouchenzi@users.noreply.github.com>
This commit is contained in:
rouchenzi
2025-09-23 15:17:49 -07:00
committed by GitHub
parent 969b4da3a6
commit eca7be9077
2 changed files with 18 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/compilation/compiler_interface.py
View File

@@ -551,8 +551,9 @@ def set_inductor_config(config, runtime_shape):
if isinstance(runtime_shape, int):
# for a specific batchsize, tuning triton kernel parameters
# can be beneficial
config["max_autotune"] = True
config["coordinate_descent_tuning"] = True
config["max_autotune"] = envs.VLLM_ENABLE_INDUCTOR_MAX_AUTOTUNE
config["coordinate_descent_tuning"] = (
envs.VLLM_ENABLE_INDUCTOR_COORDINATE_DESCENT_TUNING)
class EagerAdaptor(CompilerInterface):