[Bugfix] Include inductor and functorch configs in compilation cache key (#40627)

Signed-off-by: Richard Zou <zou3519@gmail.com>
This commit is contained in:
Richard Zou
2026-04-23 09:52:59 -04:00
committed by GitHub
parent da1e7311ca
commit 424033f4fc
3 changed files with 40 additions and 2 deletions
+2
View File
@@ -56,6 +56,7 @@ def _cold_start(vllm_runner):
def test_moe_startup(monkeypatch, vllm_runner, fresh_vllm_cache, mega_aot_artifact):
monkeypatch.setenv("VLLM_ENABLE_V1_MULTIPROCESSING", "0")
monkeypatch.setenv("VLLM_USE_MEGA_AOT_ARTIFACT", mega_aot_artifact)
monkeypatch.setenv("VLLM_DEEP_GEMM_WARMUP", "skip")
# Cold start in a forked child (must fork before CUDA init).
# This model has 32 identical transformer layers which produce
@@ -235,6 +236,7 @@ def _cold_start_model(vllm_runner, spec: ModelStartupSpec):
@fork_new_process_for_each_test
def test_model_startup(monkeypatch, vllm_runner, fresh_vllm_cache, spec):
monkeypatch.setenv("VLLM_ENABLE_V1_MULTIPROCESSING", "0")
monkeypatch.setenv("VLLM_DEEP_GEMM_WARMUP", "skip")
# Cold start in a forked child (must fork before CUDA init).
ctx = mp.get_context("fork")