2.12.10-1

Fix bug with CollNet
Fix bug with zero-bytes send/recv operations
Fix NCCL_PARAM implementation to avoid taking a lock on every call
Fix bug when setting NCCL_IB_QPS_PER_CONNECTION to more than one.
Improve error reporting for network errors.
This commit is contained in:
Sylvain Jeaugey
2022-03-30 02:25:49 -07:00
bovenliggende 44eb40da0e
commit 353e8ba446
16 gewijzigde bestanden met toevoegingen van 203 en 114 verwijderingen
+1
Bestand weergeven
@@ -15,6 +15,7 @@
#define NCCL_AGG_CHANNEL_SIZE (1LL << 21) /* 2 MiB, ideal per-channel size to fully utilize bandwidth */
size_t ncclKernMaxLocalSize();
ncclResult_t ncclKernSetSharedMemoryCarveout(int carveOut);
ncclResult_t ncclEnqueueCheck(struct ncclInfo* info);
ncclResult_t ncclCpuBarrierIn(struct ncclComm* comm, int* isLast);
ncclResult_t ncclCpuBarrierLast(struct ncclComm* comm);