Add new NVLS algorithm for allreduce using NVLink SHARP (intra-node only).
Add new config options: cgaClusterSize, minCTAs, maxCTAs, netName.
Enable LL128 when we use PXN to close rings.
NVTX3 includes update.
Fix crash when one CollNet (SHARP) rail fails to initialize.


[ROCm/rccl commit: 5d3ab08b69]
This commit is contained in:
Sylvain Jeaugey
2023-02-27 02:48:21 -08:00
parent e6e8f2555c
commit 8dcf8e8720
72 changed files with 4541 additions and 2391 deletions
+1
View File
@@ -25,6 +25,7 @@ ncclResult_t bootstrapSend(void* commState, int peer, int tag, void* data, int s
ncclResult_t bootstrapRecv(void* commState, int peer, int tag, void* data, int size);
ncclResult_t bootstrapBarrier(void* commState, int *ranks, int rank, int nranks, int tag);
ncclResult_t bootstrapIntraNodeAllGather(void* commState, int *ranks, int rank, int nranks, void* allData, int size);
ncclResult_t bootstrapIntraNodeBroadcast(void* commState, int *ranks, int rank, int nranks, int root, void* bcastData, int size);
ncclResult_t bootstrapClose(void* commState);
ncclResult_t bootstrapAbort(void* commState);
#endif