GPU-Initiated Communication: Dissecting Down to the Bone

Javid Baydamirli, Ismayil Ismayilov, Kaan Oktay +1

GPU-initiated communication lets GPU threads post RDMA operations directly to the NIC. It underpins NVSHMEM, NCCL GIN, and DeepEP, which serve the fine-grained, latency-critical co…