Simplify dist init and only init if needed.

deepspeed/runtime/engine.py

@@ -125,21 +125,21 @@ def __init__(self,
         if dist_init_required is None:
             dist_init_required = not dist.is_initialized()
 
-        if self._in_aml():
-            self._set_environment_variables_for_nccl_backend(args)
-        else:
-            self._mpi_check(args, dist_init_required)
-
         self.dist_backend = "nccl"
+
         if dist_init_required:
-            if not dist.is_initialized():
-                logger.info("Initializing torch distributed with backend: {}".format(
-                    self.dist_backend))
-                dist.init_process_group(backend=self.dist_backend)
+            # discover using mpi4py if user specifies the flag
+            if hasattr(args, 'deepspeed_mpi') and args.deepspeed_mpi:
+                self._mpi_check(args, dist_init_required)
             else:
-                logger.warning(
-                    "Was given dist_init_required=True but detected that torch"
-                    "distributed was already initialized, cannot initialize twice.")
+                # detect if we are in Azure ML environment
+                if self._in_aml():
+                    self._set_environment_variables_for_nccl_backend(args)
+
+            # DeepSpeed will initialize torch distributed only if the user has not already intialized it.
+            logger.info("Initializing torch distributed with backend: {}".format(
+                self.dist_backend))
+            dist.init_process_group(backend=self.dist_backend)
 
         self._do_args_sanity_check(args)
         self._configure_with_arguments(args, mpu)
@@ -203,7 +203,7 @@ def __init__(self,
         self.unflatten = util_ops.unflatten
 
     def _in_aml(self):
-        # read and environment variable to detect if we are using an Azure ML environment
+        # read AzureML environment variable to detect if we are using an Azure ML environment
         if 'AZUREML_EXPERIMENT_ID' in os.environ:
             return True
         else:
@@ -246,43 +246,42 @@ def _set_environment_variables_for_nccl_backend(self,
                         os.environ['MASTER_PORT']))
 
     def _mpi_check(self, args, dist_init_required):
-        if hasattr(args, 'deepspeed_mpi') and args.deepspeed_mpi:
-            from mpi4py import MPI
-            import subprocess
-            comm = MPI.COMM_WORLD
-            rank = comm.Get_rank()
-            world_size = comm.Get_size()
-
-            master_addr = None
-            if rank == 0:
-                hostname_cmd = ["hostname -I"]
-                result = subprocess.check_output(hostname_cmd, shell=True)
-                master_addr = result.decode('utf-8').split()[0]
-            master_addr = comm.bcast(master_addr, root=0)
-
-            # Determine local rank by assuming hostnames are unique
-            proc_name = MPI.Get_processor_name()
-            all_procs = comm.allgather(proc_name)
-            local_rank = sum([i == proc_name for i in all_procs[:rank]])
-
-            os.environ['RANK'] = str(rank)
-            os.environ['WORLD_SIZE'] = str(world_size)
-            args.local_rank = local_rank
-            os.environ['MASTER_ADDR'] = master_addr
-            os.environ['MASTER_PORT'] = TORCH_DISTRIBUTED_DEFAULT_PORT
-
-            logger.info(
-                "Discovered MPI settings of world_rank={}, local_rank={}, world_size={}, master_addr={}, master_port={}"
-                .format(os.environ['RANK'],
-                        args.local_rank,
-                        os.environ['WORLD_SIZE'],
-                        os.environ['MASTER_ADDR'],
-                        os.environ['MASTER_PORT']))
-
-            if not dist_init_required and dist.is_initialized():
-                assert dist.get_rank() == rank, "MPI rank {} does not match torch rank {}".format(rank, dist.get_rank())
-                assert dist.get_world_size() == world_size, "MPI world size {} does not match torch world size {}".format(
-                    world_size, dist.get_world_size())
+        from mpi4py import MPI
+        import subprocess
+        comm = MPI.COMM_WORLD
+        rank = comm.Get_rank()
+        world_size = comm.Get_size()
+
+        master_addr = None
+        if rank == 0:
+            hostname_cmd = ["hostname -I"]
+            result = subprocess.check_output(hostname_cmd, shell=True)
+            master_addr = result.decode('utf-8').split()[0]
+        master_addr = comm.bcast(master_addr, root=0)
+
+        # Determine local rank by assuming hostnames are unique
+        proc_name = MPI.Get_processor_name()
+        all_procs = comm.allgather(proc_name)
+        local_rank = sum([i == proc_name for i in all_procs[:rank]])
+
+        os.environ['RANK'] = str(rank)
+        os.environ['WORLD_SIZE'] = str(world_size)
+        args.local_rank = local_rank
+        os.environ['MASTER_ADDR'] = master_addr
+        os.environ['MASTER_PORT'] = TORCH_DISTRIBUTED_DEFAULT_PORT
+
+        logger.info(
+            "Discovered MPI settings of world_rank={}, local_rank={}, world_size={}, master_addr={}, master_port={}"
+            .format(os.environ['RANK'],
+                    args.local_rank,
+                    os.environ['WORLD_SIZE'],
+                    os.environ['MASTER_ADDR'],
+                    os.environ['MASTER_PORT']))
+
+        if not dist_init_required and dist.is_initialized():
+            assert dist.get_rank() == rank, "MPI rank {} does not match torch rank {}".format(rank, dist.get_rank())
+            assert dist.get_world_size() == world_size, "MPI world size {} does not match torch world size {}".format(
+                world_size, dist.get_world_size())
 
     def pld_enabled(self):
         return self._config.pld_enabled