tiny refactor pcg split op registration (#16863)
This commit is contained in:
@@ -1,6 +1,17 @@
|
||||
# Adapted from https://github.com/vllm-project/vllm/blob/v0.10.0/vllm/compilation/compilation_config.py
|
||||
|
||||
from typing import List
|
||||
from typing import Callable, List, Optional
|
||||
|
||||
SPLIT_OPS = []
|
||||
|
||||
|
||||
def register_split_op(op_name: Optional[str] = None):
|
||||
def decorator(op_func: Callable):
|
||||
name = op_name or op_func.__name__
|
||||
SPLIT_OPS.append(f"sglang.{name}")
|
||||
return op_func
|
||||
|
||||
return decorator
|
||||
|
||||
|
||||
# TODO(Yuwei): support better compile config support
|
||||
@@ -15,11 +26,8 @@ class CompilationConfig:
|
||||
self.capture_sizes = capture_sizes
|
||||
self.compiler = compiler
|
||||
self.enable_debug_mode = enable_debug_mode
|
||||
self.split_ops = [
|
||||
"sglang.unified_attention_with_output",
|
||||
"sglang.gdn_with_output",
|
||||
"sglang.inplace_all_reduce",
|
||||
]
|
||||
self.split_ops = []
|
||||
self.split_ops.extend(SPLIT_OPS)
|
||||
|
||||
def add_split_op(self, op: str):
|
||||
self.split_ops.append(op)
|
||||
|
||||
Reference in New Issue
Block a user