Merge pull request #120 from laekov/ddp-bcast-fix

Fix Broadcast rank bug in DGDP

fmoe/distributed.py

@@ -4,7 +4,7 @@ Supportive modules to conduct distributed training
 import torch
 import torch.nn as nn
 from torch._utils import _flatten_dense_tensors, _unflatten_dense_tensors
-from .utils import get_torch_default_comm
+from .utils import get_torch_default_comm, get_rank_0_in_comm
 
 
 class DistributedGroupedDataParallel(nn.Module):
@@ -97,7 +97,8 @@ class DistributedGroupedDataParallel(nn.Module):
             comm = self.comms[dp_comm]
             datas = [p.data for p in group]
             coalesced = _flatten_dense_tensors(datas)
-            torch.distributed.broadcast(coalesced, 0, group=comm)
+            torch.distributed.broadcast(coalesced,
+                    get_rank_0_in_comm(comm), group=comm)
             torch.cuda.synchronize()
             synced = _unflatten_dense_tensors(coalesced, datas)
             for d, s in zip(datas, synced):

fmoe/utils.py

 r"""
 Utils to play with PyTorch.
 """
+import torch
 import torch.distributed as dist
 
 
@@ -28,3 +29,13 @@ def get_torch_default_comm():
     except Exception as _:
         pass
     raise RuntimeError("Unsupported PyTorch version")
+
+
+def get_rank_0_in_comm(comm):
+    world_size = dist.get_world_size(comm)
+    x = torch.tensor([dist.get_rank()], dtype=torch.int64, device='cuda')
+    ys = [torch.empty_like(x) for _ in range(world_size)]
+    dist.all_gather(ys, x, group=comm)
+    root_rank = ys[0].item()
+    return root_rank
+

tests/test_numerical.py

@@ -71,7 +71,7 @@ class MyMoE(FMoE):
             d_model=d_model,
             gate=NaiveGate,
             world_size=world_size,
-            mp_group=mp_group,
+            slice_group=mp_group,
             top_k=top_k,
         )
         self.experts = _Expert(num_expert, d_model, d_hidden, activation)
@@ -344,6 +344,7 @@ def _test_fmoe_local_ddp(rank, world_size, mp_group, dp_group, world_group):
     model = MyModule().cuda()
     model_ddp = LocalDDP(deepcopy(model),
             mp_group=mp_group, dp_group=dp_group, world_group=world_group)
+    model = deepcopy(model_ddp.module)
     model.set_comm()
     model_ddp.module.set_comm()