From 4d17e20310a256b42409c1b04166a60316a2f075 Mon Sep 17 00:00:00 2001 From: Lu Fang <30275821+houseroad@users.noreply.github.com> Date: Sat, 26 Apr 2025 09:17:58 -0700 Subject: [PATCH] Disable the torch.compile cache checks when VLLM_DISABLE_COMPILE_CACHE=1 (#16573) Signed-off-by: Lu Fang --- vllm/compilation/compiler_interface.py | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/vllm/compilation/compiler_interface.py b/vllm/compilation/compiler_interface.py index bc9e421a660a5..f91e4cf3e89ea 100644 --- a/vllm/compilation/compiler_interface.py +++ b/vllm/compilation/compiler_interface.py @@ -11,6 +11,7 @@ import torch import torch._inductor.compile_fx import torch.fx as fx +import vllm.envs as envs from vllm.config import VllmConfig from vllm.utils import is_torch_equal_or_newer @@ -317,10 +318,14 @@ class InductorAdaptor(CompilerInterface): inner_compile=hijacked_compile_fx_inner, config_patches=current_config) - assert hash_str is not None, ( - "failed to get the hash of the compiled graph") - assert file_path is not None, ( - "failed to get the file path of the compiled graph") + # We treat VLLM_DISABLE_COMPILE_CACHE as the overall switch for torch + # compilation cache. So turn off the checks if we disable the + # compilation cache. + if not envs.VLLM_DISABLE_COMPILE_CACHE: + assert hash_str is not None, ( + "failed to get the hash of the compiled graph") + assert file_path is not None, ( + "failed to get the file path of the compiled graph") return compiled_graph, (hash_str, file_path) def load(self,