mirror of
https://git.datalinker.icu/vllm-project/vllm.git
synced 2026-01-14 11:54:29 +08:00
96b9aa5aa0
Signed-off-by: morrison-turnansky <mturnans@redhat.com> Signed-off-by: Morrison Turnansky <mturnans@redhat.com> Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
60 lines
2.0 KiB
Python
60 lines
2.0 KiB
Python
# SPDX-License-Identifier: Apache-2.0
|
|
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
|
|
|
import time
|
|
|
|
from vllm.config import CompilationConfig, CompilationMode, VllmConfig
|
|
from vllm.logger import init_logger
|
|
|
|
logger = init_logger(__name__)
|
|
|
|
context_manager = None
|
|
torch_compile_start_time: float = 0.0
|
|
|
|
|
|
def start_monitoring_torch_compile(vllm_config: VllmConfig):
|
|
global torch_compile_start_time
|
|
torch_compile_start_time = time.time()
|
|
|
|
compilation_config: CompilationConfig = vllm_config.compilation_config
|
|
path = vllm_config.compile_debug_dump_path()
|
|
if compilation_config.mode == CompilationMode.VLLM_COMPILE and path:
|
|
import depyf
|
|
|
|
path.mkdir(parents=True, exist_ok=True)
|
|
global context_manager
|
|
context_manager = depyf.prepare_debug(path.as_posix())
|
|
context_manager.__enter__()
|
|
|
|
|
|
def end_monitoring_torch_compile(vllm_config: VllmConfig):
|
|
compilation_config: CompilationConfig = vllm_config.compilation_config
|
|
if compilation_config.mode == CompilationMode.VLLM_COMPILE:
|
|
logger.info(
|
|
"torch.compile takes %.2f s in total", compilation_config.compilation_time
|
|
)
|
|
global context_manager
|
|
if context_manager is not None:
|
|
context_manager.__exit__(None, None, None)
|
|
context_manager = None
|
|
|
|
|
|
cudagraph_capturing_enabled: bool = True
|
|
|
|
|
|
def validate_cudagraph_capturing_enabled():
|
|
# used to monitor whether a cudagraph capturing is legal at runtime.
|
|
# should be called before any cudagraph capturing.
|
|
# if an illegal cudagraph capturing happens, raise an error.
|
|
global cudagraph_capturing_enabled
|
|
if not cudagraph_capturing_enabled:
|
|
raise RuntimeError(
|
|
"CUDA graph capturing detected at an inappropriate "
|
|
"time. This operation is currently disabled."
|
|
)
|
|
|
|
|
|
def set_cudagraph_capturing_enabled(enabled: bool):
|
|
global cudagraph_capturing_enabled
|
|
cudagraph_capturing_enabled = enabled
|