Revert "[distributed] remove pynccl's redundant stream (vllm-project#11744)"

This reverts commit 635b897.

Author: gshtras

tests/distributed/test_pynccl.py

@@ -137,8 +137,9 @@ def worker_fn_with_cudagraph():
         # run something in the default stream to initialize torch engine
         a = torch.ones((4, 4), device=f'cuda:{pynccl_comm.rank}')
         torch.cuda.synchronize()
-        with torch.cuda.graph(graph), \
-            pynccl_comm.change_state(enable=True):
+        with torch.cuda.graph(
+                graph, stream=pynccl_comm.stream), pynccl_comm.change_state(
+                    enable=True):
             a_out = pynccl_comm.all_reduce(a)
         torch.cuda.synchronize()
         graph.replay()

vllm/distributed/device_communicators/pynccl.py

@@ -51,6 +51,7 @@ def __init__(
         if self.world_size == 1:
             self.available = False
             self.disabled = True
+            self.stream = None
             return
         try:
             self.nccl = NCCLLibrary(library_path)
@@ -59,6 +60,7 @@ def __init__(
             # e.g. in a non-GPU environment
             self.available = False
             self.disabled = True
+            self.stream = None
             return
 
         self.available = True
@@ -96,12 +98,12 @@ def __init__(
         with torch.cuda.device(device):
             self.comm: ncclComm_t = self.nccl.ncclCommInitRank(
                 self.world_size, self.unique_id, self.rank)
+            self.stream = torch.cuda.Stream()
 
-            stream = torch.cuda.current_stream()
             # A small all_reduce for warmup.
             data = torch.zeros(1, device=device)
             self.all_reduce(data)
-            stream.synchronize()
+            self.stream.synchronize()
             del data
 
     def all_reduce(self,
@@ -120,7 +122,7 @@ def all_reduce(self,
         out_tensor = torch.empty_like(in_tensor)
 
         if stream is None:
-            stream = torch.cuda.current_stream()
+            stream = self.stream
         self.nccl.ncclAllReduce(buffer_type(in_tensor.data_ptr()),
                                 buffer_type(out_tensor.data_ptr()),
                                 in_tensor.numel(),
@@ -142,7 +144,7 @@ def all_gather(self,
             f"this nccl communicator is created to work on {self.device}, "
             f"but the input tensor is on {input_tensor.device}")
         if stream is None:
-            stream = torch.cuda.current_stream()
+            stream = self.stream
         self.nccl.ncclAllGather(
             buffer_type(input_tensor.data_ptr()),
             buffer_type(output_tensor.data_ptr()), input_tensor.numel(),
@@ -163,7 +165,7 @@ def reduce_scatter(self,
             f"this nccl communicator is created to work on {self.device}, "
             f"but the input tensor is on {input_tensor.device}")
         if stream is None:
-            stream = torch.cuda.current_stream()
+            stream = self.stream
         self.nccl.ncclReduceScatter(
             buffer_type(input_tensor.data_ptr()),
             buffer_type(output_tensor.data_ptr()), output_tensor.numel(),
@@ -178,7 +180,7 @@ def send(self, tensor: torch.Tensor, dst: int, stream=None):
             f"this nccl communicator is created to work on {self.device}, "
             f"but the input tensor is on {tensor.device}")
         if stream is None:
-            stream = torch.cuda.current_stream()
+            stream = self.stream
         self.nccl.ncclSend(buffer_type(tensor.data_ptr()), tensor.numel(),
                            ncclDataTypeEnum.from_torch(tensor.dtype), dst,
                            self.comm, cudaStream_t(stream.cuda_stream))
@@ -190,7 +192,7 @@ def recv(self, tensor: torch.Tensor, src: int, stream=None):
             f"this nccl communicator is created to work on {self.device}, "
             f"but the input tensor is on {tensor.device}")
         if stream is None:
-            stream = torch.cuda.current_stream()
+            stream = self.stream
         self.nccl.ncclRecv(buffer_type(tensor.data_ptr()), tensor.numel(),
                            ncclDataTypeEnum.from_torch(tensor.dtype), src,
                            self.comm, cudaStream_t(stream.cuda_stream))
@@ -202,7 +204,7 @@ def broadcast(self, tensor: torch.Tensor, src: int, stream=None):
             f"this nccl communicator is created to work on {self.device}, "
             f"but the input tensor is on {tensor.device}")
         if stream is None:
-            stream = torch.cuda.current_stream()
+            stream = self.stream
         if src == self.rank:
             sendbuff = buffer_type(tensor.data_ptr())
             # NCCL requires the sender also to have a receive buffer
@@ -215,17 +217,25 @@ def broadcast(self, tensor: torch.Tensor, src: int, stream=None):
                                 self.comm, cudaStream_t(stream.cuda_stream))
 
     @contextmanager
-    def change_state(self, enable: Optional[bool] = None):
+    def change_state(self,
+                     enable: Optional[bool] = None,
+                     stream: Optional[torch.cuda.Stream] = None):
         """
         A context manager to change the state of the communicator.
         """
         if enable is None:
             # guess a default value when not specified
             enable = self.available
 
+        if stream is None:
+            stream = self.stream
+
         old_disable = self.disabled
+        old_stream = self.stream
 
+        self.stream = stream
         self.disabled = not enable
         yield
 
         self.disabled = old_disable
+        self.stream = old_stream

vllm/distributed/parallel_state.py

@@ -310,7 +310,8 @@ def graph_capture(
             if not pynccl_comm:
                 maybe_pynccl_context = nullcontext()
             else:
-                maybe_pynccl_context = pynccl_comm.change_state()
+                maybe_pynccl_context = pynccl_comm.change_state(
+                    stream=torch.cuda.current_stream())
             with maybe_pynccl_context:
                 yield graph_capture_context