Skip to content

[compile] Turn standalone_compile back on #27460

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 24, 2025
Merged

[compile] Turn standalone_compile back on #27460

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions vllm/envs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,7 +131,7 @@
VLLM_DP_RANK: int = 0
VLLM_DP_RANK_LOCAL: int = -1
VLLM_DP_SIZE: int = 1
VLLM_USE_STANDALONE_COMPILE: bool = False
VLLM_USE_STANDALONE_COMPILE: bool = True
VLLM_DP_MASTER_IP: str = ""
VLLM_DP_MASTER_PORT: int = 0
VLLM_MOE_DP_CHUNK_SIZE: int = 256
Expand Down Expand Up @@ -494,10 +494,10 @@ def get_vllm_port() -> int | None:
os.environ.get("VLLM_FLASH_ATTN_VERSION", None)
),
# Feature flag to enable/disable Inductor standalone compile.
# In torch <= 2.7 we ignore this flag; in torch >= 2.8 this is
# disabled by default.
# In torch <= 2.7 we ignore this flag; in torch >= 2.9 this is
# enabled by default.
"VLLM_USE_STANDALONE_COMPILE": lambda: os.environ.get(
"VLLM_USE_STANDALONE_COMPILE", "0"
"VLLM_USE_STANDALONE_COMPILE", "1"
)
== "1",
# Debug pattern matching inside custom passes.
Expand Down