Sylvain Jeaugey
3c223c105a
Add network communication through another GPU connected with NVLink (PXN). Add aggregation of messages coming from different local GPUs through PXN and going to the same destination. Add new v5 plugin API with grouped receives and tags. Add compat for v4 plugins. Add naming of NCCL threads to help debugging. Fix NVLink detection and avoid data corruption when some NVLinks are down. Add support for Relaxed Ordering for IB. Add profiling and timing infrastructure.
44 lines
2.9 KiB
C
44 lines
2.9 KiB
C
/*************************************************************************
|
|
* Copyright (c) 2016-2022, NVIDIA CORPORATION. All rights reserved.
|
|
*
|
|
* See LICENSE.txt for license information
|
|
************************************************************************/
|
|
|
|
#ifndef NCCL_INT_NET_H_
|
|
#define NCCL_INT_NET_H_
|
|
|
|
#include "nccl.h"
|
|
#include "nccl_net.h"
|
|
#include "checks.h"
|
|
|
|
extern ncclNet_t* ncclNet;
|
|
typedef char ncclNetHandle_t[NCCL_NET_HANDLE_MAXSIZE];
|
|
|
|
ncclResult_t ncclNetInit();
|
|
int ncclNetVersion();
|
|
|
|
// Translation to external API
|
|
static const char* ncclNetName() { return ncclNet->name; }
|
|
static ncclResult_t ncclNetDevices(int* ndev) { NCCLCHECK(ncclNet->devices(ndev)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetGetProperties(int dev, ncclNetProperties_t* props) { NCCLCHECK(ncclNet->getProperties(dev, props)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetListen(int dev, void* handle, void** listenComm) { NCCLCHECK(ncclNet->listen(dev, handle, listenComm)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetConnect(int dev, void* handle, void** sendComm) { NCCLCHECK(ncclNet->connect(dev, handle, sendComm)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetAccept(void* listenComm, void** recvComm) { NCCLCHECK(ncclNet->accept(listenComm, recvComm)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetRegMr(void* comm, void* data, int size, int type, void** mhandle) { NCCLCHECK(ncclNet->regMr(comm, data, size, type, mhandle)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetDeregMr(void* comm, void* mhandle) { NCCLCHECK(ncclNet->deregMr(comm, mhandle)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetIsend(void* sendComm, void* data, int size, int tag, void* mhandle, void** request) { NCCLCHECK(ncclNet->isend(sendComm, data, size, tag, mhandle, request)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetIrecv(void* recvComm, int n, void** data, int* sizes, int* tags, void** mhandles, void** request) { NCCLCHECK(ncclNet->irecv(recvComm, n, data, sizes, tags, mhandles, request)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetIflush(void* recvComm, int n, void** data, int* sizes, void** mhandles, void** request) { NCCLCHECK(ncclNet->iflush(recvComm, n, data, sizes, mhandles, request)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetTest(void* request, int* done, int* sizes) { NCCLCHECK(ncclNet->test(request, done, sizes)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetCloseSend(void* sendComm) { NCCLCHECK(ncclNet->closeSend(sendComm)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetCloseRecv(void* recvComm) { NCCLCHECK(ncclNet->closeRecv(recvComm)); return ncclSuccess; }
|
|
static ncclResult_t ncclNetCloseListen(void* listenComm) { NCCLCHECK(ncclNet->closeListen(listenComm)); return ncclSuccess; }
|
|
|
|
// Test whether the current GPU support GPU Direct RDMA.
|
|
ncclResult_t ncclGpuGdrSupport(int* gdrSupport);
|
|
|
|
extern ncclNet_t ncclNetIb;
|
|
extern ncclNet_t ncclNetSocket;
|
|
|
|
#endif
|