API名称 |
是否支持 |
限制与说明 |
---|---|---|
torch.distributed.destroy_process_group |
|
|
torch.distributed.is_available |
是 |
|
torch.distributed.init_process_group |
是 |
|
torch.distributed.is_initialized |
是 |
|
torch.distributed.is_mpi_available |
是 |
|
torch.distributed.is_nccl_available |
是 |
|
torch.distributed.is_torchelastic_launched |
否 |
|
torch.distributed.Backend |
是 |
|
torch.distributed.get_backend |
是 |
|
torch.distributed.get_rank |
是 |
|
torch.distributed.get_world_size |
是 |
|
torch.distributed.Store |
否 |
|
torch.distributed.TCPStore |
否 |
|
torch.distributed.HashStore |
否 |
|
torch.distributed.FileStore |
否 |
|
torch.distributed.PrefixStore |
否 |
|
torch.distributed.Store.set |
否 |
|
torch.distributed.Store.get |
否 |
|
torch.distributed.Store.add |
否 |
|
torch.distributed.Store.compare_set |
否 |
|
torch.distributed.Store.wait |
否 |
|
torch.distributed.Store.num_keys |
否 |
|
torch.distributed.Store.delete_key |
否 |
|
torch.distributed.Store.set_timeout |
否 |
|
torch.distributed.new_group |
是 |
|
torch.distributed.send |
是 |
支持bf16,fp16,fp32,fp64,uint8,int8,int16,int32,int64,bool |
torch.distributed.recv |
是 |
支持bf16,fp16,fp32,fp64,uint8,int8,int16,int32,int64,bool |
torch.distributed.isend |
是 |
支持bf16,fp16,fp32,fp64,uint8,int8,int16,int32,int64,bool |
torch.distributed.irecv |
是 |
支持bf16,fp16,fp32,fp64,uint8,int8,int16,int32,int64,bool |
torch.distributed.broadcast |
是 |
支持bf16,fp16,fp32,fp64,uint8,int8,int16,int32,int64,bool |
torch.distributed.broadcast_object_list |
是 |
|
torch.distributed.all_reduce |
是 |
支持fp16, fp32, int32, int64, bool |
torch.distributed.reduce |
是 |
支持bf16,fp16,fp32,uint8,int8,int32,int64,bool |
torch.distributed.all_gather |
是 |
支持bf16,fp16,fp32,int8,int32,bool |
torch.distributed.all_gather_object |
是 |
|
torch.distributed.gather |
否 |
|
torch.distributed.gather_object |
否 |
|
torch.distributed.scatter |
否 |
|
torch.distributed.scatter_object_list |
否 |
|
torch.distributed.reduce_scatter |
是 |
支持bf16,fp16,fp32,int8,int32 |
torch.distributed.all_to_all |
是 |
支持fp32 |
torch.distributed.barrier |
是 |
|
torch.distributed.monitored_barrier |
否 |
|
torch.distributed.ReduceOp |
是 |
支持bf16,fp16,fp32,uint8,int8,int32,int64,bool |
torch.distributed.reduce_op |
是 |
支持bf16,fp16,fp32,uint8,int8,int32,int64,bool |