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

[2/N][torch.compile] make compilation cfg part of vllm cfg (#10383)

Browse Source 
Signed-off-by: youkaichao <youkaichao@gmail.com>
This commit is contained in:
youkaichao
2024-11-16 18:02:14 -08:00
committed by GitHub
parent 661a34fd4f
commit 4fd9375028
27 changed files with 359 additions and 283 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
vllm/envs.py
View File

@@ -69,7 +69,6 @@ if TYPE_CHECKING:
VLLM_SKIP_P2P_CHECK: bool = False
VLLM_TORCH_COMPILE_LEVEL: int = 0
VLLM_TORCH_COMPILE_CONFIG: Optional[str] = None
VLLM_CUSTOM_OPS: List[str] = []
VLLM_DISABLED_KERNELS: List[str] = []
VLLM_USE_V1: bool = False
VLLM_ENABLE_V1_MULTIPROCESSING: bool = False
@@ -217,18 +216,6 @@ environment_variables: Dict[str, Callable[[], Any]] = {
"VLLM_TORCH_COMPILE_CONFIG":
lambda: os.environ.get("VLLM_TORCH_COMPILE_CONFIG", None),
# Fine-grained control over which custom ops to enable/disable.
# Use 'all' to enable all, 'none' to disable all.
# Also specify a list of custom op names to enable (prefixed with a '+'),
# or disable (prefixed with a '-').
# Examples:
# - 'all,-op1' to enable all except op1
# - 'none,+op1,+op2' to enable only op1 and op2
# By default, all custom ops are enabled when running without Inductor
# and disabled when running with Inductor (compile_level >= Inductor).
"VLLM_CUSTOM_OPS":
lambda: os.environ.get("VLLM_CUSTOM_OPS", "").replace(" ", "").split(","),
# local rank of the process in the distributed setting, used to determine
# the GPU device id
"LOCAL_RANK":