Trim NICs when all GPUs are connected by XGMI (#430)

* Trim NICs when all GPUs are connected by XGMI

* Only enable clique with maximum of 2 hops

[ROCm/rccl commit: 29c729d8b6]
This commit is contained in:
Wenkai Du
2021-10-05 18:27:43 -07:00
zatwierdzone przez GitHub
rodzic 55fa23a613
commit 91eca0d7d2
6 zmienionych plików z 159 dodań i 78 usunięć
+2 -2
Wyświetl plik
@@ -576,7 +576,7 @@ ncclResult_t initTransportsRank_1(struct ncclComm* comm, struct allGather1Data_t
bool allXgmi = true;
{ // [RCCL] Check if clique-based kernels can be enabled and initialize CliqueManager
//CliqueManager::cliqueMode_t cliqueMode = CliqueManager::CLIQUE_DISABLED;
if (comm->localRanks == comm->nRanks)
if (comm->localRanks == comm->nRanks && comm->topo->nodes[GPU].nodes[0].gpu.gcn != 910)
{
// Check that all the GPUs have peer access to one another and are XGMI connected
bool hasPeerAccess = true;
@@ -595,7 +595,7 @@ ncclResult_t initTransportsRank_1(struct ncclComm* comm, struct allGather1Data_t
//}
bool isXGMI;
NCCLCHECK(ncclTopoGetLinkType(comm->topo, i, j, &isXGMI));
NCCLCHECK(ncclTopoGetLinkType(comm->topo, i, j, &isXGMI, 1));
allXgmi &= isXGMI;
}
}