[Refactor]Abstract Platform Interface for Distributed Backend and Add xccl Support for Intel XPU (#19410)
Signed-off-by: dbyoung18 <yang5.yang@intel.com> Signed-off-by: Kunshang Ji <kunshang.ji@intel.com> Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
This commit is contained in:
@@ -11,6 +11,7 @@ from vllm.config import VllmConfig
|
||||
from vllm.distributed.parallel_state import get_pp_group, get_tp_group
|
||||
from vllm.logger import init_logger
|
||||
from vllm.model_executor.utils import set_random_seed
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.v1.core.sched.output import SchedulerOutput
|
||||
from vllm.v1.outputs import ModelRunnerOutput
|
||||
@@ -58,7 +59,8 @@ class CPUWorker(Worker):
|
||||
# Initialize the distributed environment.
|
||||
init_worker_distributed_environment(self.vllm_config, self.rank,
|
||||
self.distributed_init_method,
|
||||
self.local_rank, "gloo")
|
||||
self.local_rank,
|
||||
current_platform.dist_backend)
|
||||
# Set random seed.
|
||||
set_random_seed(self.model_config.seed)
|
||||
|
||||
|
||||
@@ -157,7 +157,8 @@ class Worker(WorkerBase):
|
||||
# Initialize the distributed environment.
|
||||
init_worker_distributed_environment(self.vllm_config, self.rank,
|
||||
self.distributed_init_method,
|
||||
self.local_rank)
|
||||
self.local_rank,
|
||||
current_platform.dist_backend)
|
||||
# Set random seed.
|
||||
set_random_seed(self.model_config.seed)
|
||||
|
||||
|
||||
@@ -18,6 +18,7 @@ from vllm.distributed import (ensure_model_parallel_initialized,
|
||||
from vllm.logger import init_logger
|
||||
from vllm.lora.request import LoRARequest
|
||||
from vllm.model_executor import set_random_seed
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.utils import STR_DTYPE_TO_TORCH_DTYPE, cdiv
|
||||
from vllm.v1.attention.backends.pallas import TPU_HEAD_SIZE_ALIGNMENT
|
||||
from vllm.v1.core.sched.output import SchedulerOutput
|
||||
@@ -300,7 +301,7 @@ class TPUWorker:
|
||||
rank=rank,
|
||||
local_rank=local_rank,
|
||||
distributed_init_method=distributed_init_method,
|
||||
backend="gloo",
|
||||
backend=current_platform.dist_backend,
|
||||
)
|
||||
ensure_model_parallel_initialized(
|
||||
parallel_config.tensor_parallel_size,
|
||||
|
||||
Reference in New Issue
Block a user