[compile] aot_compile should respect VLLM_DISABLE_COMPILE_CACHE (#36358)

Signed-off-by: Richard Zou <zou3519@gmail.com>
This commit is contained in:
Richard Zou
2026-03-11 03:12:03 -07:00
committed by GitHub
parent c87fb515ed
commit 09b6f99852
4 changed files with 182 additions and 43 deletions
+3
View File
@@ -349,6 +349,9 @@ def reset_compile_wrapper(model: torch.nn.Module) -> None:
compilation_counter.num_cache_entries_updated = 0
compilation_counter.num_compiled_artifacts_saved = 0
compilation_counter.stock_torch_compile_count = 0
compilation_counter.num_aot_compiles = 0
compilation_counter.num_aot_artifacts_saved = 0
compilation_counter.num_aot_artifacts_loaded = 0
# Clear the AOT compiled function so the model is forced to
# recompile on the next call. Without this, decorators.py