# tensorplay.distributed.distributed_c10d.batch_isend_irecv

Source: https://www.tensorplay.cn/docs/generated/tensorplay.distributed.distributed_c10d.batch_isend_irecv.html

# tensorplay.distributed.distributed_c10d.batch_isend_irecv

tensorplay.distributed.distributed_c10d.batch_isend_irecv(p2p_op_list: [List](https://docs.python.org/3/library/typing.html#typing.List)[[P2POp](tensorplay.distributed.distributed_c10d.P2POp.html#tensorplay.distributed.distributed_c10d.P2POp)]) → [List](https://docs.python.org/3/library/typing.html#typing.List)[[Work](tensorplay.distributed.distributed_c10d.Work.html#tensorplay.distributed.distributed_c10d.Work)][[source]](../_modules/tensorplay/distributed/distributed_c10d.html#batch_isend_irecv)

Send or receive a batch of tensors asynchronously (torch parity).

All operations are treated as a single NCCL group so ordering of sends
vs receives cannot deadlock. Every rank in group must participate.