[Platform][Dist] Make torch distributed process group extendable (#18763)

Signed-off-by: Mengqing Cao <cmq0113@163.com>

vllm/distributed/utils.py

@@ -5,7 +5,6 @@
 # https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/core/tensor_parallel/utils.py
 # Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
 import dataclasses
-import datetime
 import os
 import pickle
 import socket
@@ -14,14 +13,14 @@ import time
 import uuid
 from collections import deque
 from collections.abc import Sequence
+from datetime import timedelta
 from typing import Any, Optional
 
 import torch
 from torch.distributed import ProcessGroup, TCPStore
 from torch.distributed.distributed_c10d import (Backend, PrefixStore,
                                                 _get_default_timeout,
-                                                _unregister_process_group,
-                                                is_nccl_available)
+                                                _unregister_process_group)
 from torch.distributed.rendezvous import rendezvous
 
 import vllm.envs as envs
@@ -406,7 +405,7 @@ class StatelessProcessGroup:
             port=port,
             world_size=world_size,
             is_master=launch_server,
-            timeout=datetime.timedelta(seconds=store_timeout),
+            timeout=timedelta(seconds=store_timeout),
             use_libuv=False,  # for now: github.com/pytorch/pytorch/pull/150215
             master_listen_fd=listen_fd,
         )
@@ -419,6 +418,43 @@ class StatelessProcessGroup:
             data_expiration_seconds=data_expiration_seconds)
 
 
+def init_gloo_process_group(backend: Backend, prefix_store: PrefixStore,
+                            group_rank: int, group_size: int,
+                            timeout: timedelta) -> ProcessGroup:
+    """
+    Stateless init ProcessGroup with gloo backend compatible with 
+    different torch versions.
+    """
+    if is_torch_equal_or_newer("2.6"):
+        pg = ProcessGroup(
+            prefix_store,
+            group_rank,
+            group_size,
+        )
+    else:
+        options = ProcessGroup.Options(backend=backend)
+        pg = ProcessGroup(
+            prefix_store,
+            group_rank,
+            group_size,
+            options,
+        )
+    from torch.distributed.distributed_c10d import ProcessGroupGloo
+    backend_class = ProcessGroupGloo(prefix_store,
+                                     group_rank,
+                                     group_size,
+                                     timeout=timeout)
+    backend_type = ProcessGroup.BackendType.GLOO
+    device = torch.device("cpu")
+    if is_torch_equal_or_newer("2.6"):
+        # _set_default_backend is supported in torch >= 2.6
+        pg._set_default_backend(backend_type)
+    backend_class._set_sequence_number_for_group()
+
+    pg._register_backend(device, backend_type, backend_class)
+    return pg
+
+
 def stateless_init_torch_distributed_process_group(
         host: str, port: int, rank: int, world_size: int,
         backend: str) -> ProcessGroup:
@@ -468,40 +504,19 @@ def stateless_init_torch_distributed_process_group(
     # different systems (e.g. RPC) in case the store is multi-tenant.
     prefix_store = PrefixStore(init_method, store)
 
-    pg: ProcessGroup = ProcessGroup(
-        prefix_store,
-        group_rank,
-        group_size,
-    )
-
     if backend == "gloo":
-        from torch.distributed.distributed_c10d import ProcessGroupGloo
-        backend_class = ProcessGroupGloo(prefix_store,
-                                         group_rank,
-                                         group_size,
+        return init_gloo_process_group(backend=backend,
+                                       prefix_store=prefix_store,
+                                       group_rank=group_rank,
+                                       group_size=group_size,
+                                       timeout=timeout)
+    from vllm.platforms import current_platform
+    return current_platform.stateless_init_device_torch_dist_pg(
+        backend=backend,
+        prefix_store=prefix_store,
+        group_rank=group_rank,
+        group_size=group_size,
         timeout=timeout)
-        backend_type = ProcessGroup.BackendType.GLOO
-        device = torch.device("cpu")
-    elif backend == "nccl":
-        assert is_nccl_available()
-        from torch.distributed.distributed_c10d import ProcessGroupNCCL
-
-        backend_options = ProcessGroupNCCL.Options()
-        backend_options._timeout = timeout
-
-        backend_class = ProcessGroupNCCL(prefix_store, group_rank, group_size,
-                                         backend_options)
-        backend_type = ProcessGroup.BackendType.NCCL
-        device = torch.device("cuda")
-    else:
-        raise RuntimeError(f"Unsupported torch distributed backend: {backend}")
-
-    pg._set_default_backend(backend_type)
-    backend_class._set_sequence_number_for_group()
-
-    pg._register_backend(device, backend_type, backend_class)
-
-    return pg
 
 
 def stateless_destroy_torch_distributed_process_group(

vllm/platforms/cuda.py

@@ -4,10 +4,13 @@ pynvml. However, it should not initialize cuda context.
 """
 
 import os
+from datetime import timedelta
 from functools import wraps
 from typing import TYPE_CHECKING, Callable, Optional, TypeVar, Union
 
 import torch
+from torch.distributed import PrefixStore, ProcessGroup
+from torch.distributed.distributed_c10d import is_nccl_available
 from typing_extensions import ParamSpec
 
 # import custom ops, trigger op registration
@@ -316,6 +319,36 @@ class CudaPlatformBase(Platform):
     def get_piecewise_backend_cls(cls) -> str:
         return "vllm.compilation.cuda_piecewise_backend.CUDAPiecewiseBackend"  # noqa
 
+    @classmethod
+    def stateless_init_device_torch_dist_pg(
+        cls,
+        backend: str,
+        prefix_store: PrefixStore,
+        group_rank: int,
+        group_size: int,
+        timeout: timedelta,
+    ) -> ProcessGroup:
+        assert is_nccl_available()
+        pg: ProcessGroup = ProcessGroup(
+            prefix_store,
+            group_rank,
+            group_size,
+        )
+        from torch.distributed.distributed_c10d import ProcessGroupNCCL
+
+        backend_options = ProcessGroupNCCL.Options()
+        backend_options._timeout = timeout
+
+        backend_class = ProcessGroupNCCL(prefix_store, group_rank, group_size,
+                                         backend_options)
+        backend_type = ProcessGroup.BackendType.NCCL
+        device = torch.device("cuda")
+        pg._set_default_backend(backend_type)
+        backend_class._set_sequence_number_for_group()
+
+        pg._register_backend(device, backend_type, backend_class)
+        return pg
+
 
 # NVML utils
 # Note that NVML is not affected by `CUDA_VISIBLE_DEVICES`,

vllm/platforms/interface.py

@@ -3,11 +3,13 @@ import enum
 import os
 import platform
 import random
+from datetime import timedelta
 from platform import uname
 from typing import TYPE_CHECKING, NamedTuple, Optional, Union
 
 import numpy as np
 import torch
+from torch.distributed import PrefixStore, ProcessGroup
 
 from vllm.inputs import ProcessorInputs, PromptType
 from vllm.logger import init_logger
@@ -486,6 +488,20 @@ class Platform:
         """
         return "vllm.compilation.base_piecewise_backend.AbstractPiecewiseBackend"  # noqa
 
+    @classmethod
+    def stateless_init_device_torch_dist_pg(
+        cls,
+        backend: str,
+        prefix_store: PrefixStore,
+        group_rank: int,
+        group_size: int,
+        timeout: timedelta,
+    ) -> ProcessGroup:
+        """
+        Init platform-specific torch distributed process group.
+        """
+        raise RuntimeError(f"Unsupported torch distributed backend: {backend}")
+
 
 class UnspecifiedPlatform(Platform):
     _enum = PlatformEnum.UNSPECIFIED

vllm/platforms/rocm.py

 # SPDX-License-Identifier: Apache-2.0
 
 import os
+from datetime import timedelta
 from functools import cache, lru_cache, wraps
 from typing import TYPE_CHECKING, Optional
 
 import torch
+from torch.distributed import PrefixStore, ProcessGroup
+from torch.distributed.distributed_c10d import is_nccl_available
 
 import vllm.envs as envs
 from vllm.logger import init_logger
@@ -387,3 +390,33 @@ class RocmPlatform(Platform):
     @classmethod
     def get_piecewise_backend_cls(cls) -> str:
         return "vllm.compilation.cuda_piecewise_backend.CUDAPiecewiseBackend"  # noqa
+
+    @classmethod
+    def stateless_init_device_torch_dist_pg(
+        cls,
+        backend: str,
+        prefix_store: PrefixStore,
+        group_rank: int,
+        group_size: int,
+        timeout: timedelta,
+    ) -> ProcessGroup:
+        assert is_nccl_available()
+        pg: ProcessGroup = ProcessGroup(
+            prefix_store,
+            group_rank,
+            group_size,
+        )
+        from torch.distributed.distributed_c10d import ProcessGroupNCCL
+
+        backend_options = ProcessGroupNCCL.Options()
+        backend_options._timeout = timeout
+
+        backend_class = ProcessGroupNCCL(prefix_store, group_rank, group_size,
+                                         backend_options)
+        backend_type = ProcessGroup.BackendType.NCCL
+        device = torch.device("cuda")
+        pg._set_default_backend(backend_type)
+        backend_class._set_sequence_number_for_group()
+
+        pg._register_backend(device, backend_type, backend_class)
+        return pg