AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'ncclRepo/master' into develop

Browse Source
This commit is contained in:
Aristotelis
2022-06-02 15:27:24 +00:00
parent eef812bed7 7aa1c46fd5
commit e0864e7093
12 changed files with 97 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/transport/net_ib.cc
+1 -1
View File
@@ -224,8 +224,8 @@ ncclResult_t ncclIbInit(ncclDebugLogger_t logFunction) {
ncclIbDevs[ncclNIbDevs].mrCache.slots = NULL;
pthread_create(&ncclIbAsyncThread, NULL, ncclIbAsyncThreadMain, context);
pthread_detach(ncclIbAsyncThread); // will not be pthread_join()'d
ncclSetThreadName(ncclIbAsyncThread, "NCCL IbAsync %2d", ncclNIbDevs);
pthread_detach(ncclIbAsyncThread); // will not be pthread_join()'d
ncclNIbDevs++;
nPorts++;
// [RCCL]
src/transport/p2p.cc
+3 -2
View File
@@ -140,6 +140,7 @@ ncclResult_t p2pCanConnect(int* ret, struct ncclTopoSystem* topo, struct ncclTop
// Setting this to non zero causes P2P to use Reads rather than Writes
NCCL_PARAM(P2pReadEnable, "P2P_READ_ENABLE", -2);
NCCL_PARAM(P2pDirectDisable, "P2P_DIRECT_DISABLE", 0);
static ncclResult_t p2pGetInfo(struct ncclTopoSystem* topo, struct ncclPeerInfo* info1, struct ncclPeerInfo* info2, int* read, int* intermediateRank) {
int p2p;
@@ -209,7 +210,7 @@ ncclResult_t p2pSendSetup(struct ncclComm* comm, struct ncclTopoGraph* graph, st
if (intermediateRank == -1) {
info->rank = myInfo->rank;
if (myInfo->pidHash == peerInfo->pidHash) {
send->conn.direct |= info->read ? NCCL_DIRECT_READ : NCCL_DIRECT_WRITE;
if (ncclParamP2pDirectDisable() == 0) send->conn.direct |= info->read ? NCCL_DIRECT_READ : NCCL_DIRECT_WRITE;
INFO(NCCL_INIT|NCCL_P2P, "Channel %02d : %d[%lx] -> %d[%lx] via P2P/direct pointer%s",
channelId, myInfo->rank, myInfo->busId, peerInfo->rank, peerInfo->busId, useReadStr);
} else {
@@ -254,7 +255,7 @@ ncclResult_t p2pRecvSetup(struct ncclComm* comm, struct ncclTopoGraph* graph, st
if (intermediateRank == -1) {
info->rank = myInfo->rank;
if (myInfo->pidHash == peerInfo->pidHash) {
recv->conn.direct |= info->read ? NCCL_DIRECT_READ : NCCL_DIRECT_WRITE;
if (ncclParamP2pDirectDisable() == 0) recv->conn.direct |= info->read ? NCCL_DIRECT_READ : NCCL_DIRECT_WRITE;
} else {
recv->conn.direct |= info->read ? NCCL_IPC_READ : NCCL_IPC_WRITE;
}