关于深度学习:NCCL源码解析②Bootstrap网络连接的建立

作者|KIDGINBROOK
更新|潘丽晨

上次介绍到 rank0 的机器生成了 ncclUniqueId，并实现了机器的 bootstrap 网络和通信网络的初始化，这节接着看下所有节点间 bootstrap 的连贯是如何建设的。

rank0 节点执行 ncclGetUniqueId 生成 ncclUniqueId，通过 mpi 将 Id 播送到所有节点，而后所有节点都会执行 ncclCommInitRank，这里其余节点也会进行初始化 bootstrap 网络和通信网络的操作，而后会执行到 ncclCommInitRankSync。

ncclResult_t ncclCommInitRankSync(ncclComm_t* newcomm, int nranks, ncclUniqueId commId, int myrank, int cudaDev) {
  ncclResult_t res;
 
  CUDACHECK(cudaSetDevice(cudaDev));
  NCCLCHECKGOTO(commAlloc(newcomm, nranks, myrank), res, cleanup);
  NCCLCHECKGOTO(initTransportsRank(*newcomm, &commId), res, cleanup);
  NCCLCHECKGOTO(devCommSetup(*newcomm), res, cleanup);
 
  INFO(NCCL_INIT,"comm %p rank %d nranks %d cudaDev %d busId %x - Init COMPLETE", *newcomm, myrank, nranks, (*newcomm)->cudaDev, (*newcomm)->busId);
 
  return ncclSuccess;
cleanup:
  if ((*newcomm) && (*newcomm)->bootstrap) bootstrapAbort((*newcomm)->bootstrap);
  *newcomm = NULL;
  return res;
}

ncclComm_t 是指向 ncclComm 的指针，ncclComm 是一个大杂烩，蕴含了通信用到的所有上下文信息，外面的字段等用到的时候再介绍，而后通过 commAlloc 调配 newcom，并且实现初始化，比方以后是哪个卡，对应的 pcie busid 是什么，而后执行 initTransportsRank。

static ncclResult_t initTransportsRank(struct ncclComm* comm, ncclUniqueId* commId) {
  // We use 3 AllGathers
  // 1. {peerInfo, comm}
  // 2. ConnectTransport[nranks], ConnectValue[nranks]
  // 3. {nThreads, nrings, compCap, prev[MAXCHANNELS], next[MAXCHANNELS] }
 
  int rank = comm->rank;
  int nranks = comm->nRanks;
  uint64_t commHash = getHash(commId->internal, NCCL_UNIQUE_ID_BYTES);
  TRACE(NCCL_INIT, "comm %p, commHash %lx, rank %d nranks %d - BEGIN", comm, commHash, rank, nranks);
  NCCLCHECK(bootstrapInit(commId, rank, nranks, &comm->bootstrap));
 
  // AllGather1 - begin
  struct {
    struct ncclPeerInfo peerInfo;
    struct ncclComm* comm;
  } *allGather1Data;
 
  NCCLCHECK(ncclCalloc(&allGather1Data, nranks));
  allGather1Data[rank].comm = comm;
  struct ncclPeerInfo* myInfo = &allGather1Data[rank].peerInfo;
  NCCLCHECK(fillInfo(comm, myInfo, commHash));
  NCCLCHECK(bootstrapAllGather(comm->bootstrap, allGather1Data, sizeof(*allGather1Data)));
 
  NCCLCHECK(ncclCalloc(&comm->peerInfo, nranks+1)); // Extra rank to represent CollNet root
  for (int i = 0; i < nranks; i++) {memcpy(comm->peerInfo+i, &allGather1Data[i].peerInfo, sizeof(struct ncclPeerInfo));
    if ((i != rank) && (comm->peerInfo[i].hostHash == myInfo->hostHash) && (comm->peerInfo[i].busId == myInfo->busId)) {WARN("Duplicate GPU detected : rank %d and rank %d both on CUDA device %x", rank, i, myInfo->busId);
      return ncclInvalidUsage;
    }
  }

看下 bootstrapInit：

ncclResult_t bootstrapInit(ncclUniqueId * id, int rank, int nranks, void** commState) {ncclNetHandle_t* netHandle = (ncclNetHandle_t*) id;
  bool idFromEnv = getenv("NCCL_COMM_ID") != NULL;
  struct extState* state;
  NCCLCHECK(ncclCalloc(&state, 1));
  state->rank = rank;
  state->nranks = nranks;
  *commState = state;
 
  TRACE(NCCL_INIT, "rank %d nranks %d", rank, nranks);
 
  struct extInfo info = {0};
  info.rank = rank;
  info.nranks = nranks;
  void *tmpSendComm, *tmpRecvComm;
  // Pass the remote address to listen via info
  if (idFromEnv) {memcpy(&info.extHandleListen, netHandle, sizeof(ncclNetHandle_t));
    memcpy(&info.extHandleListenRoot, netHandle, sizeof(ncclNetHandle_t));
  }
  // listen will return the local address via info (specify interface type 'findSubnetIf')
  state->dev = idFromEnv ? findSubnetIf : 0;
  void* extBstrapListenCommRoot;
  NCCLCHECK(bootstrapNetListen(state->dev, &info.extHandleListen, &state->extBstrapListenComm));
  NCCLCHECK(bootstrapNetListen(state->dev, &info.extHandleListenRoot, &extBstrapListenCommRoot));
 
  // stagger connection times to avoid an overload of the root at very high rank counts
  if (nranks > 128) {
    long msec = rank;
    struct timespec tv;
    tv.tv_sec = msec / 1000;
    tv.tv_nsec = 1000000 * (msec % 1000);
    TRACE(NCCL_INIT, "rank %d delaying connection to root by %ld msec", rank, msec);
    (void) nanosleep(&tv, NULL);
  }
 
  // send info on my listening socket to root
  NCCLCHECK(bootstrapNetConnect(state->dev, netHandle, &tmpSendComm));
  NCCLCHECK(bootstrapNetSend(tmpSendComm, &info, sizeof(info)));
  NCCLCHECK(bootstrapNetCloseSend(tmpSendComm));
 
  // get info on my "next" rank in the bootstrap ring from root
}

首先看下 commState，即 ncclComm 的 bootstrap，类型为 extState。

struct extState {
  void* extBstrapListenComm;
  void* extBstrapRingRecvComm;
  void* extBstrapRingSendComm;
  ncclNetHandle_t* peerBstrapHandles;
  struct unexConn* unexpectedConnections;
  int rank;
  int nranks;
  int dev;
};

其中 extBstrapRingSendComm 是以后节点连贯 next 的 socket 连贯，extBstrapRingRecvComm 是以后节点和 prev 节点的 socket 连贯，extBstrapListenComm 是以后节点的监听 socket，peerBstrapHandles 是所有 rank 的 ip port（对应 extBstrapListenComm），dev 默认为 0，示意用第几个 ip 地址。

而后通过 bootstrapNetListen 创立 extHandleListen 和 extHandleListenRoot 两个 bootstrap comm，如前文所述，bootstrap comm 其实就是保留了 fd，这里创立两个 comm 的起因是 extHandleListen 是 rank 之间理论应用的 bootstrap 连贯，extHandleListenRoot 是 rank0 节点和其余所有 rank 进行通信应用的连贯。

static ncclResult_t bootstrapNetListen(int dev, ncclNetHandle_t* netHandle, void** listenComm)

bootstrapNetListen 函数上节有介绍过，会获取到第 dev 个以后机器的 ip，而后 listen 获取监听 fd，将 ip port 写到 nethandle，获取到的 bootstrap comm 写到 listencomm。

而后将 rank，nrank，extHandleListen 和 extHandleListenRoot 写到 extInfo 里。

struct extInfo {
  int rank;
  int nranks;
  ncclNetHandle_t extHandleListenRoot;
  ncclNetHandle_t extHandleListen;
};

netHandle 为 ncclUniqueId，即 rank0 的 ip port，而后通过 bootstrapNetConnect 创立 bootstrap send comm，类比 bootstrapNetListen，bootstrapNetConnect 就是建设到 netHandle 的 socket 连贯，将 socket 写到 sendComm 里，这里 dev 并没有用到。

static ncclResult_t bootstrapNetConnect(int dev, ncclNetHandle_t* netHandle, void** sendComm)

而后通过 bootstrapNetSend 将 extInfo 发送进来，即发给 rank0：

static ncclResult_t bootstrapNetSend(void* sendComm, void* data, int size) {struct bootstrapNetComm* comm = (struct bootstrapNetComm*)sendComm;
  NCCLCHECK(socketSend(comm->fd, &size, sizeof(int)));
  NCCLCHECK(socketSend(comm->fd, data, size));
  return ncclSuccess;
}

其中 socketSend 就是执行 send 接口发送数据。

而后通过 bootstrapNetCloseSend 敞开 fd。

rank0 收到数据后会做什么工作呢，回顾一下，rank0 的节执行 ncclGetUniqueId 生成 ncclUniqueId，其中在执行 bootstrapCreateRoot 的最初会启动一个线程执行 bootstrapRoot。

static void *bootstrapRoot(void* listenComm) {
  struct extInfo info;
  ncclNetHandle_t *rankHandles = NULL;
  ncclNetHandle_t *rankHandlesRoot = NULL; // for initial rank <-> root information exchange
  ncclNetHandle_t zero = {0}; // for sanity checking
  void* tmpComm;
  ncclResult_t res;
  setFilesLimit();
 
  TRACE(NCCL_INIT, "BEGIN");
  /* Receive addresses from all ranks */
  int nranks = 0, c = 0;
  do {NCCLCHECKGOTO(bootstrapNetAccept(listenComm, &tmpComm), res, out);
    NCCLCHECKGOTO(bootstrapNetRecv(tmpComm, &info, sizeof(info)), res, out);
    NCCLCHECKGOTO(bootstrapNetCloseRecv(tmpComm), res, out);
 
    if (c == 0) {
      nranks = info.nranks;
      NCCLCHECKGOTO(ncclCalloc(&rankHandles, nranks), res, out);
      NCCLCHECKGOTO(ncclCalloc(&rankHandlesRoot, nranks), res, out);
    }
 
    if (nranks != info.nranks) {WARN("Bootstrap Root : mismatch in rank count from procs %d : %d", nranks, info.nranks);
      goto out;
    }
 
    if (memcmp(&zero, &rankHandlesRoot[info.rank], sizeof(ncclNetHandle_t)) != 0) {WARN("Bootstrap Root : rank %d of %d ranks has already checked in", info.rank, nranks);
      goto out;
    }
 
    // Save the connection handle for that rank
    memcpy(rankHandlesRoot+info.rank, info.extHandleListenRoot, sizeof(ncclNetHandle_t));
    memcpy(rankHandles+info.rank, info.extHandleListen, sizeof(ncclNetHandle_t));
 
    ++c;
    TRACE(NCCL_INIT, "Received connect from rank %d total %d/%d",  info.rank, c, nranks);
  } while (c < nranks);
  TRACE(NCCL_INIT, "COLLECTED ALL %d HANDLES", nranks);
 
  // Send the connect handle for the next rank in the AllGather ring
  for (int r=0; r<nranks; ++r) {int next = (r+1) % nranks;
    void *tmpSendComm;
    NCCLCHECKGOTO(bootstrapNetConnect(0, rankHandlesRoot+r, &tmpSendComm), res, out);
    NCCLCHECKGOTO(bootstrapNetSend(tmpSendComm, rankHandles+next, sizeof(ncclNetHandle_t)), res, out);
    NCCLCHECKGOTO(bootstrapNetCloseSend(tmpSendComm), res, out);
  }
  TRACE(NCCL_INIT, "SENT OUT ALL %d HANDLES", nranks);
 
out:
  bootstrapNetCloseListen(listenComm);
  if (rankHandles) free(rankHandles);
  if (rankHandlesRoot) free(rankHandlesRoot);
 
  TRACE(NCCL_INIT, "DONE");
  return NULL;
}

listenComm 是上一个博文中 rank0 创立的监听 fd，bootstrapNetAccept 是从 listenComm 中获取一个新连贯，应用新连贯的 fd 创立 recvcomm。

static ncclResult_t bootstrapNetAccept(void* listenComm, void** recvComm)

而后通过 bootstrapNetRecv 读取 tmpComm 的数据，即其余 rank 发送来的 extInfo，而后保留其余 rank 的 extHandleListen 和 extHandleListenRoot，这个时候 rank0 就获取到其余所有 rank 的 ip 和 port 了。

获取完所有 rank 的 info 之后开始建环，将节点 (r+1) % nranks 的 extHandleListen 发送给节点 r，就是说将节点 r 的 next 节点的 nethandle 发送给节点 r。这里能够看出，每个节点创立了两个 listen comm，其中 rank0 应用 extHandleListenRoot 进行通信，其余节点之间通过 extHandleListen 进行通信。

而后再回去接着看 bootstrapInit。

ncclResult_t bootstrapInit(ncclUniqueId * id, int rank, int nranks, void** commState) {
  // get info on my "next" rank in the bootstrap ring from root
  ncclNetHandle_t extHandleNext;
  NCCLCHECK(bootstrapNetAccept(extBstrapListenCommRoot, &tmpRecvComm));
  NCCLCHECK(bootstrapNetRecv(tmpRecvComm, &extHandleNext, sizeof(extHandleNext)));
  NCCLCHECK(bootstrapNetCloseRecv(tmpRecvComm));
  NCCLCHECK(bootstrapNetCloseListen(extBstrapListenCommRoot));
 
  NCCLCHECK(bootstrapNetConnect(state->dev, &extHandleNext, &state->extBstrapRingSendComm));
  // Accept the connect request from the previous rank in the AllGather ring
  NCCLCHECK(bootstrapNetAccept(state->extBstrapListenComm, &state->extBstrapRingRecvComm));
 
  // AllGather all listen handlers
  NCCLCHECK(ncclCalloc(&state->peerBstrapHandles, nranks));
  memcpy(state->peerBstrapHandles+rank, info.extHandleListen, sizeof(ncclNetHandle_t));
  NCCLCHECK(bootstrapAllGather(state, state->peerBstrapHandles, sizeof(ncclNetHandle_t)));
 
  TRACE(NCCL_INIT, "rank %d nranks %d - DONE", rank, nranks);
 
  return ncclSuccess;
}

接着所有 rank 都会在 extHandleListenRoot 上接管新连贯创立 tmpRecvComm，而后接管到以后 rank 的 next 的 ip，port；而后连贯 next 创立 bscomm 到 state->extBstrapRingSendComm，接管 prev 的连贯创立 bscomm 到 state->extBstrapRingRecvComm，到当初 bootstrap 网络连接就齐全建设起来了，如下图：

最初 gather 所有 rank 的 ip port，首先将本人的 nethandle 放到 peerBstrapHandles 的对应地位，如下所示。

而后执行 bootstrapAllGather：

ncclResult_t bootstrapAllGather(void* commState, void* allData, int size) {struct extState* state = (struct extState*)commState;
  char* data = (char*)allData;
  int rank = state->rank;
  int nranks = state->nranks;
 
  TRACE(NCCL_INIT, "rank %d nranks %d size %d", rank, nranks, size);
 
  /* Simple ring based AllGather
   * At each step i receive data from (rank-i-1) from left
   * and send previous step's data from (rank-i) to right
   */
  for (int i=0; i<nranks-1; i++) {size_t rslice = (rank - i - 1 + nranks) % nranks;
    size_t sslice = (rank - i + nranks) % nranks;
 
    // Send slice to the right
    NCCLCHECK(bootstrapNetSend(state->extBstrapRingSendComm, data+sslice*size, size));
    // Recv slice from the left
    NCCLCHECK(bootstrapNetRecv(state->extBstrapRingRecvComm, data+rslice*size, size));
  }
 
  TRACE(NCCL_INIT, "rank %d nranks %d size %d - DONE", rank, nranks, size);
  return ncclSuccess;
}

每一次将本人的 data 发送给对应的 rank，而后接管其余 rank 发送过去的 data，如下图。

第一步：

第二步：

到这里每个 rank 就都有了全局所有 rank 的 ip port。

最初总结一下，本节次要创立了 bootstrap 环形网络连接，并保留到 ncclComm 里。

欢送 Star、试用 OneFlow 最新版本：https://github.com/Oneflow-Inc/oneflow/