[utils] support detection of number of processes on current node (#723)

colossalai/context/parallel_context.py

@@ -2,6 +2,8 @@
 # -*- encoding: utf-8 -*-
 
 import random
+import socket
+from collections import Counter
 from typing import Union
 
 import numpy as np
@@ -45,6 +47,7 @@ class ParallelContext(metaclass=SingletonMeta):
         self.data_parallel_size = 1
         self.pipeline_parallel_size = 1
         self.tensor_parallel_size = 1
+        self.num_processes_on_current_node = -1
         self.virtual_pipeline_parallel_size = None
         self.virtual_pipeline_parallel_rank = None
 
@@ -81,6 +84,13 @@ class ParallelContext(metaclass=SingletonMeta):
         else:
             raise TypeError("Invalid type for config, only dictionary or string is supported")
 
+    def detect_num_processes_on_current_node(self):
+        hostname = socket.gethostname()
+        hostname_list = [None for _ in range(self.get_world_size(ParallelMode.GLOBAL))]
+        dist.all_gather_object(hostname_list, hostname, group=self.get_group(ParallelMode.GLOBAL))
+        counter = Counter(hostname_list)
+        self.num_processes_on_current_node = counter[hostname]
+
     @staticmethod
     def _check_parallel_mode(parallel_mode: ParallelMode):
         assert isinstance(parallel_mode, ParallelMode)

colossalai/initialize.py

@@ -102,6 +102,9 @@ def launch(config: Union[str, Path, Config, Dict],
         # if local rank is not given, calculate automatically
         gpc.set_device(local_rank)
 
+    # set the number of processes running on the same node
+    gpc.detect_num_processes_on_current_node()
+
     gpc.set_seed(seed)
 
     if verbose:
@@ -398,15 +401,17 @@ def initialize(model: nn.Module,
         else:
             scatter_gather = False
         if use_interleaved:
-            schedule = InterleavedPipelineSchedule(gpc.config.NUM_MICRO_BATCHES, 
+            schedule = InterleavedPipelineSchedule(gpc.config.NUM_MICRO_BATCHES,
-                                                   gpc.config.model.num_chunks, tensor_shape=tensor_shape, scatter_gather_tensors=scatter_gather)
+                                                   gpc.config.model.num_chunks,
+                                                   tensor_shape=tensor_shape,
+                                                   scatter_gather_tensors=scatter_gather)
         else:
             schedule = PipelineSchedule(gpc.config.NUM_MICRO_BATCHES,
-                                        tensor_shape=tensor_shape, scatter_gather_tensors=scatter_gather)
+                                        tensor_shape=tensor_shape,
+                                        scatter_gather_tensors=scatter_gather)
     else:
         schedule = NonPipelineSchedule()
 
-
     if gradient_handler_cfg is None:
         gradient_handlers = None
         if verbose and not isinstance(model, DDP):