[Cleanup] Refactor profiling env vars into a CLI config (#29912)

Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
Signed-off-by: Benjamin Chislett <chislett.ben@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Benjamin Chislett
2025-12-09 13:29:33 -05:00
committed by GitHub
co-authored by gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> Harry Mellor
parent d471b2aff0
commit e858bfe051
22 changed files with 433 additions and 252 deletions
+17
View File
@@ -20,6 +20,7 @@ from vllm.beam_search import (
from vllm.config import (
CompilationConfig,
PoolerConfig,
ProfilerConfig,
StructuredOutputsConfig,
is_init_field,
)
@@ -211,6 +212,7 @@ class LLM:
structured_outputs_config: dict[str, Any]
| StructuredOutputsConfig
| None = None,
profiler_config: dict[str, Any] | ProfilerConfig | None = None,
kv_cache_memory_bytes: int | None = None,
compilation_config: int | dict[str, Any] | CompilationConfig | None = None,
logits_processors: list[str | type[LogitsProcessor]] | None = None,
@@ -282,6 +284,20 @@ class LLM:
else:
structured_outputs_instance = StructuredOutputsConfig()
if profiler_config is not None:
if isinstance(profiler_config, dict):
profiler_config_instance = ProfilerConfig(
**{
k: v
for k, v in profiler_config.items()
if is_init_field(ProfilerConfig, k)
}
)
else:
profiler_config_instance = profiler_config
else:
profiler_config_instance = ProfilerConfig()
# warn about single-process data parallel usage.
_dp_size = int(kwargs.get("data_parallel_size", 1))
_distributed_executor_backend = kwargs.get("distributed_executor_backend")
@@ -324,6 +340,7 @@ class LLM:
mm_processor_kwargs=mm_processor_kwargs,
pooler_config=pooler_config,
structured_outputs_config=structured_outputs_instance,
profiler_config=profiler_config_instance,
compilation_config=compilation_config_instance,
logits_processors=logits_processors,
**kwargs,
+7 -10
View File
@@ -5,7 +5,7 @@
from fastapi import APIRouter, FastAPI, Request
from fastapi.responses import Response
import vllm.envs as envs
from vllm.config import ProfilerConfig
from vllm.engine.protocol import EngineClient
from vllm.logger import init_logger
@@ -35,15 +35,12 @@ async def stop_profile(raw_request: Request):
def attach_router(app: FastAPI):
if envs.VLLM_TORCH_PROFILER_DIR:
profiler_config = getattr(app.state.args, "profiler_config", None)
assert profiler_config is None or isinstance(profiler_config, ProfilerConfig)
if profiler_config is not None and profiler_config.profiler is not None:
logger.warning_once(
"Torch Profiler is enabled in the API server. This should ONLY be "
"used for local development!"
"Profiler with mode '%s' is enabled in the "
"API server. This should ONLY be used for local development!",
profiler_config.profiler,
)
elif envs.VLLM_TORCH_CUDA_PROFILE:
logger.warning_once(
"CUDA Profiler is enabled in the API server. This should ONLY be "
"used for local development!"
)
if envs.VLLM_TORCH_PROFILER_DIR or envs.VLLM_TORCH_CUDA_PROFILE:
app.include_router(router)