mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-25 13:12:45 +08:00
383 lines
15 KiB
Python
383 lines
15 KiB
Python
import atexit
|
|
import faulthandler
|
|
import multiprocessing
|
|
import platform
|
|
import signal
|
|
import traceback
|
|
from abc import ABC, abstractmethod
|
|
from pathlib import Path
|
|
from queue import Queue
|
|
from typing import TYPE_CHECKING, Generator, List, Optional, Union
|
|
|
|
import numpy as np
|
|
import torch
|
|
|
|
from tensorrt_llm.logger import logger, set_level
|
|
|
|
from .._utils import mpi_world_size
|
|
from ..bindings import executor as tllm
|
|
from ..builder import Engine
|
|
from ..disaggregated_params import DisaggregatedParams
|
|
from ..llmapi.llm_utils import KvCacheRetentionConfig
|
|
from ..llmapi.mpi_session import (MpiSession, external_mpi_comm_available,
|
|
need_spawn_mpi_workers)
|
|
from ..llmapi.utils import (AsyncQueue, enable_llm_debug,
|
|
enable_worker_single_process_for_tp1, print_colored,
|
|
print_colored_debug)
|
|
from ..sampling_params import BatchedLogitsProcessor, SamplingParams
|
|
from .postproc_worker import PostprocParams, PostprocWorkerConfig
|
|
from .request import GenerationRequest, LoRARequest, PromptAdapterRequest
|
|
from .result import GenerationResult, IterationStatsResult
|
|
from .utils import ProcessPoolExecutorSession, RequestError, has_event_loop
|
|
|
|
if TYPE_CHECKING:
|
|
from .proxy import ExecutorBindingsProxy
|
|
from .worker import ExecutorBindingsWorker
|
|
|
|
__all__ = [
|
|
"GenerationExecutor",
|
|
"CppExecutorError",
|
|
]
|
|
|
|
if enable_llm_debug():
|
|
# Mainly enable more detailed logging from cpp runtime.
|
|
set_level("info")
|
|
|
|
|
|
class CppExecutorError(RuntimeError):
|
|
|
|
def __init__(self, message: Optional[str] = None):
|
|
self.message = message
|
|
self.stack_trace = traceback.format_exc()
|
|
super().__init__(message)
|
|
|
|
def __str__(self):
|
|
return f"{self.message}\nStack trace:\n{self.stack_trace}"
|
|
|
|
|
|
class GenerationExecutor(ABC):
|
|
|
|
def __init__(self,
|
|
num_postprocess_workers: int = 0,
|
|
postprocess_tokenizer_dir: Optional[str] = None,
|
|
is_llm_executor: Optional[bool] = None):
|
|
self.postproc_config = PostprocWorkerConfig(
|
|
num_postprocess_workers=num_postprocess_workers,
|
|
postprocess_tokenizer_dir=postprocess_tokenizer_dir)
|
|
|
|
self._stats = None
|
|
self.stats_queue = None
|
|
|
|
atexit.register(self.shutdown)
|
|
|
|
# This is used to capture the exceptions from the threads.
|
|
self._error_queue = Queue()
|
|
|
|
# A flag to avoid calling shutdown() recursively. This happens when the background threads raise errors.
|
|
self.doing_shutdown = False
|
|
|
|
self._last_client_id: int = 1
|
|
|
|
self._iter_stats_result = None
|
|
|
|
# whether it's the executor instance of LLM API
|
|
self._is_llm_executor = is_llm_executor
|
|
|
|
@abstractmethod
|
|
def submit(self, request: GenerationRequest) -> GenerationResult:
|
|
pass
|
|
|
|
@abstractmethod
|
|
def abort_request(self, request_id: int) -> None:
|
|
pass
|
|
|
|
def generate_async(
|
|
self,
|
|
prompt_token_ids: List[int],
|
|
sampling_params: SamplingParams,
|
|
query_token_ids: Optional[Union[torch.Tensor, np.ndarray,
|
|
list]] = None,
|
|
lora_request: Optional[LoRARequest] = None,
|
|
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
|
|
streaming: bool = False,
|
|
prompt_tuning_config: Optional[list] = None,
|
|
kv_cache_retention_config: Optional[KvCacheRetentionConfig] = None,
|
|
disaggregated_params: Optional[DisaggregatedParams] = None,
|
|
postproc_params: Optional[PostprocParams] = None
|
|
) -> GenerationResult:
|
|
"""Generate output for the given prompt token ids in the asynchronous mode.
|
|
Asynchronous generation accepts single prompt only.
|
|
"""
|
|
assert isinstance(prompt_token_ids[0], int)
|
|
assert isinstance(sampling_params, SamplingParams)
|
|
|
|
if self._is_llm_executor:
|
|
if self._iter_stats_result is None:
|
|
# singleton to store cpp runtime stats
|
|
self._iter_stats_result = IterationStatsResult()
|
|
else:
|
|
# expect more engine stats whenever new prompts are submitted
|
|
self._iter_stats_result.mark_undone()
|
|
|
|
if postproc_params:
|
|
postproc_params.postproc_args.num_prompt_tokens = len(
|
|
prompt_token_ids)
|
|
result = self.submit(
|
|
GenerationRequest(
|
|
prompt_token_ids,
|
|
sampling_params=sampling_params,
|
|
postproc_params=postproc_params,
|
|
query_token_ids=query_token_ids,
|
|
lora_request=lora_request,
|
|
prompt_adapter_request=prompt_adapter_request,
|
|
streaming=streaming,
|
|
prompt_tuning_config=prompt_tuning_config,
|
|
kv_cache_retention_config=kv_cache_retention_config,
|
|
disaggregated_params=disaggregated_params))
|
|
return result
|
|
|
|
def generate(
|
|
self,
|
|
prompt_token_ids: Union[List[int], List[List[int]]],
|
|
sampling_params: Union[SamplingParams, List[SamplingParams]],
|
|
query_token_ids: Optional[Union[torch.Tensor, np.ndarray, list]] = None,
|
|
lora_request: Optional[Union[LoRARequest, List[LoRARequest]]] = None,
|
|
prompt_adapter_request: Optional[Union[
|
|
PromptAdapterRequest, List[PromptAdapterRequest]]] = None,
|
|
disaggregated_params: Optional[DisaggregatedParams] = None,
|
|
) -> Union[GenerationResult, List[GenerationResult]]:
|
|
"""Generate output for the given prompt token ids in the synchronous mode.
|
|
Synchronous generation accepts either single prompt or batched prompts.
|
|
"""
|
|
unbatched = isinstance(prompt_token_ids[0], int)
|
|
|
|
if unbatched:
|
|
prompt_token_ids = [prompt_token_ids]
|
|
if query_token_ids:
|
|
query_token_ids = [query_token_ids]
|
|
|
|
futures = []
|
|
for i, p in enumerate(prompt_token_ids):
|
|
if isinstance(sampling_params, list):
|
|
sp = sampling_params[i]
|
|
else:
|
|
sp = sampling_params
|
|
if isinstance(lora_request, list):
|
|
lora_req = lora_request[i]
|
|
else:
|
|
lora_req = lora_request
|
|
if isinstance(prompt_adapter_request, list):
|
|
pa_req = prompt_adapter_request[i]
|
|
else:
|
|
pa_req = prompt_adapter_request
|
|
future = self.generate_async(
|
|
p,
|
|
sampling_params=sp,
|
|
query_token_ids=query_token_ids,
|
|
lora_request=lora_req,
|
|
prompt_adapter_request=pa_req,
|
|
streaming=False,
|
|
disaggregated_params=disaggregated_params)
|
|
futures.append(future)
|
|
|
|
for future in futures:
|
|
future.result()
|
|
|
|
if unbatched:
|
|
futures = futures[0]
|
|
|
|
return futures
|
|
|
|
def _get_next_client_id(self):
|
|
# (self._last_client_id + 1) % UINT64_MAX
|
|
self._last_client_id = (self._last_client_id + 1) & ((1 << 64) - 1)
|
|
return self._last_client_id
|
|
|
|
def _handle_background_error(self, error: Optional[Exception | str] = None):
|
|
""" Process the errors from the threads or processes.
|
|
NOTE: This should be called in the main thread.
|
|
"""
|
|
if error is not None:
|
|
# For details please refer to the comment of `GenerationResult.error`
|
|
if isinstance(error, Exception):
|
|
# Serious error from background thread or process
|
|
if enable_llm_debug():
|
|
print_colored(
|
|
f"Got background error: {repr(error)}, will shutdown the LLM instance\n",
|
|
"red")
|
|
self.shutdown()
|
|
raise error
|
|
elif isinstance(error, str):
|
|
if enable_llm_debug():
|
|
print_colored(f"Got per-request error: {repr(error)}\n",
|
|
"red")
|
|
print_colored(str(traceback.extract_stack()) + "\n", "red")
|
|
# A per-request error, can be captured and ignored
|
|
raise RequestError(error)
|
|
|
|
# Here we raise the first error in the queue. This method will be called repeatedly and user can choose to catch
|
|
# more than one error.
|
|
if not self._error_queue.empty():
|
|
e = self._error_queue.get()
|
|
self._error_queue.task_done()
|
|
self.shutdown()
|
|
# We can catch some exceptions here.
|
|
raise e
|
|
|
|
@abstractmethod
|
|
def shutdown(self):
|
|
pass
|
|
|
|
@property
|
|
def enable_postprocess_parallel(self) -> bool:
|
|
return self.postproc_config.enabled
|
|
|
|
def create_stats_queue(self):
|
|
# Stats queue is created during first submission to ensure event loop exists if it is needed.
|
|
if not self._stats:
|
|
if has_event_loop():
|
|
self._stats = AsyncQueue()
|
|
self.stats_queue = self._stats.sync_q
|
|
self.stats_aqueue = self._stats
|
|
else:
|
|
self._stats = Queue()
|
|
self.stats_queue = self._stats
|
|
self.stats_aqueue = None
|
|
|
|
def get_stats(self, timeout: float) -> List[dict]:
|
|
"""
|
|
Get iteration statistics from the runtime.
|
|
Args:
|
|
timeout (float): Max wait time in seconds when retrieving stats from queue.
|
|
Returns:
|
|
List[dict]: A list of runtime stats as dict.
|
|
"""
|
|
assert self._iter_stats_result is not None, "IterationStatsResult is not properly instantiated."
|
|
|
|
self._iter_stats_result.set_timeout(timeout)
|
|
return self._iter_stats_result.get_results()
|
|
|
|
def aget_stats(self, timeout: float) -> IterationStatsResult:
|
|
"""
|
|
Get iteration statistics from the runtime.
|
|
Returns:
|
|
IterationStatsResult: An async iterable object containing runtime stats.
|
|
"""
|
|
assert self._iter_stats_result is not None, "IterationStatsResult is not properly instantiated."
|
|
|
|
self._iter_stats_result.set_timeout(timeout)
|
|
return self._iter_stats_result
|
|
|
|
@staticmethod
|
|
def create(
|
|
engine: Union[Path, Engine],
|
|
executor_config: Optional[tllm.ExecutorConfig] = None,
|
|
batched_logits_processor: Optional[BatchedLogitsProcessor] = None,
|
|
model_world_size: int = 1,
|
|
world_size: int = 0,
|
|
mpi_session: Optional[MpiSession] = None,
|
|
reuse_mpi_comm: bool = False,
|
|
return_logits: bool = False,
|
|
postproc_worker_config: Optional[PostprocWorkerConfig] = None,
|
|
is_llm_executor: Optional[bool] = None,
|
|
) -> Union["ExecutorBindingsProxy", "ExecutorBindingsWorker"]:
|
|
# local imports to avoid cyclic importing
|
|
from .proxy import ExecutorBindingsProxy
|
|
from .worker import ExecutorBindingsWorker
|
|
|
|
if world_size == 0:
|
|
world_size = mpi_world_size()
|
|
|
|
if world_size > 1 and world_size < model_world_size:
|
|
raise RuntimeError(
|
|
"Cannot instantiate Generator for engine built "
|
|
f"for {model_world_size} ranks, while currently running "
|
|
f"on {world_size} ranks.")
|
|
|
|
postproc_worker_config = postproc_worker_config or PostprocWorkerConfig(
|
|
)
|
|
|
|
if postproc_worker_config.enabled:
|
|
print_colored_debug(
|
|
f"Using {postproc_worker_config.num_postprocess_workers} postprocess parallel processes.\n",
|
|
"green")
|
|
|
|
worker_kwargs = {
|
|
"engine": engine,
|
|
"executor_config": executor_config,
|
|
"batched_logits_processor": batched_logits_processor,
|
|
}
|
|
|
|
# The case where the Python main process is launched by mpirun
|
|
mpirun_launch = external_mpi_comm_available(model_world_size)
|
|
# The case where the Python main process utilizes mpi4py to spawn MPI workers
|
|
spawn_workers = need_spawn_mpi_workers(model_world_size)
|
|
if spawn_workers or (mpirun_launch and reuse_mpi_comm):
|
|
if reuse_mpi_comm:
|
|
assert mpi_session is not None, "reuse_mpi_comm requires an external MPI session"
|
|
return ExecutorBindingsProxy(
|
|
worker_kwargs,
|
|
model_world_size=model_world_size,
|
|
mpi_session=mpi_session,
|
|
postproc_worker_config=postproc_worker_config,
|
|
is_llm_executor=is_llm_executor)
|
|
|
|
# WAR: For the performance of gathering logits, we use single process worker
|
|
# for TP1 to avoid the large overhead of IPC.
|
|
# WAR: Developers can enable this manually, this will be easier for TP1
|
|
# debugging. We will introduce a better solution in the future.
|
|
if return_logits or enable_worker_single_process_for_tp1():
|
|
logger.warning(
|
|
"Using single process worker for TP1, this may hurt streaming generation performance."
|
|
)
|
|
return ExecutorBindingsWorker(**worker_kwargs,
|
|
is_llm_executor=is_llm_executor)
|
|
|
|
# For single-gpu case:
|
|
# Partition the workload to multiple process for streaming performance.
|
|
# While this requires uses to protect their entrypoint to
|
|
# `if __name__ == "__main__":`.
|
|
if not platform.system() == 'Windows':
|
|
return ExecutorBindingsProxy(
|
|
worker_kwargs,
|
|
model_world_size=model_world_size,
|
|
mpi_session=None, # use mpi4py
|
|
postproc_worker_config=postproc_worker_config,
|
|
is_llm_executor=is_llm_executor)
|
|
else:
|
|
ctx = multiprocessing.get_context("spawn")
|
|
# The ProcessPoolExecutorSession is used to support Windows, as mpi4py cannot.
|
|
mpi_session = ProcessPoolExecutorSession(n_workers=1,
|
|
mp_context=ctx)
|
|
return ExecutorBindingsProxy(
|
|
worker_kwargs,
|
|
model_world_size=model_world_size,
|
|
mpi_session=mpi_session,
|
|
postproc_worker_config=postproc_worker_config,
|
|
is_llm_executor=is_llm_executor)
|
|
|
|
def wait_first_completed(
|
|
self, futures: List[GenerationResult]
|
|
) -> Generator[GenerationResult, None, None]:
|
|
wait_set = set(futures)
|
|
|
|
# clear already-finished requests
|
|
for f in futures:
|
|
if f._done:
|
|
wait_set.pop(f)
|
|
yield f
|
|
|
|
# wait remaining active requests
|
|
while len(wait_set) > 0:
|
|
fut = wait_set.pop()
|
|
if fut.request_id not in self._results:
|
|
yield fut
|
|
else:
|
|
wait_set.add(fut)
|
|
|
|
|
|
if enable_llm_debug():
|
|
print_colored("LLM debug mode enabled.\n", "yellow")
|
|
# This will dump all the alive threads when the process is interrupted by SIGINT.
|
|
faulthandler.register(signal.SIGINT, all_threads=True)
|