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

[torch.compile] caching of config fields should be opt-out by default (#26468)

Browse Source 
Signed-off-by: vnadathur <glvikramn@gmail.com>
Signed-off-by: WorldExplored <srreyansh.sethi@gmail.com>
Signed-off-by: Srreyansh Sethi <srreyansh.sethi@gmail.com>
Signed-off-by: Srreyansh Sethi <107075589+WorldExplored@users.noreply.github.com>
Co-authored-by: WorldExplored <srreyansh.sethi@gmail.com>
Co-authored-by: Srreyansh Sethi <107075589+worldexplored@users.noreply.github.com>
Co-authored-by: vnadathur <236933696+vnadathur@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
This commit is contained in:
vnadathur
2025-11-19 06:13:54 -08:00
committed by GitHub
parent 2c8b9182b5
commit 1ffe934c8a
11 changed files with 599 additions and 190 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/compilation/pass_manager.py
View File

@@ -127,7 +127,7 @@ class PostGradPassManager(CustomGraphPass):
affects compilation caching. Its uuid depends on the UUIDs of all
dependent passes and the pass config. See InductorPass for more info.
"""
state = {"pass_config": self.pass_config.uuid(), "passes": []}
state = {"pass_config": self.pass_config.compute_hash(), "passes": []}
for pass_ in self.passes:
state["passes"].append(pass_.uuid())
state["passes"].append(self.fix_functionalization.uuid())