In Megatron-LM repo https://github.com/NVIDIA/Megatron-LM/blob/v3.0.2/megatron/mpu/initialize.py#L62, there are three positions will create pg through torch.distributed.new_group.
If I set os.environ["NCCL_SHARP_DISABLE"] = "1" after data parallel, the expect result is data parallel pg will allocate sharp resources, the model parallel pg and the tensor parallel pg will not allocate sharp resources.
But from repo https://github.com/Mellanox/nccl-rdma-sharp-plugins/blob/master/src/sharp_plugin.c#L252 and my experiment, debug log reports "SHARP: Set to disable on this communicator" and all pg can not allocate sharp resources, this is not in line with expectations.
Could you check this problem ?
In Megatron-LM repo https://github.com/NVIDIA/Megatron-LM/blob/v3.0.2/megatron/mpu/initialize.py#L62, there are three positions will create pg through torch.distributed.new_group.
If I set os.environ["NCCL_SHARP_DISABLE"] = "1" after data parallel, the expect result is data parallel pg will allocate sharp resources, the model parallel pg and the tensor parallel pg will not allocate sharp resources.
But from repo https://github.com/Mellanox/nccl-rdma-sharp-plugins/blob/master/src/sharp_plugin.c#L252 and my experiment, debug log reports "SHARP: Set to disable on this communicator" and all pg can not allocate sharp resources, this is not in line with expectations.
Could you check this problem ?