NCCL 2.4.6-1

Added detection of IBM/Power NVLink bridge device.
    Add NUMA support to PCI distance calculations.
    Added NCCL_IGNORE_CPU_AFFINITY env var.
    Fix memory leaks; GithubIssue#180
    Compiler warning fix; GithubIssue#178
    Replace non-standard variable length arrays. GithubIssue#171
    Fix Tree+Shared Memory crash. GithubPR#185
    Fix LL cleanup hang during long running DL jobs.
    Fix NCCL_RINGS environment variable handling.
    Added extra checks to catch repeat calls to ncclCommDestroy() GithubIssue#191
    Improve bootstrap socket connection reliability at scale.
    Fix hostname hashing issue. GithubIssue#187
    Code cleanup to rename all non device files from *.cu to *.cc


[ROCm/rccl commit: f40ce73e89]

projects/rccl/src/include/enqueue.h

@@ -1,5 +1,5 @@
 /*************************************************************************
- * Copyright (c) 2015-2018, NVIDIA CORPORATION. All rights reserved.
+ * Copyright (c) 2015-2019, NVIDIA CORPORATION. All rights reserved.
  *
  * See LICENSE.txt for license information
  ************************************************************************/
@@ -10,6 +10,12 @@
 #include "core.h"
 #include "group.h"
 
+// Channels / LL tuning
+#define NCCL_LL_CHANNEL_THRESHOLD 8 // Per thread size before we start increasing nrings
+#define NCCL_THREAD_THRESHOLD 64  // Per thread size before we switch to non-LL
+#define NCCL_THREAD_THRESHOLD_PREVOLTA 32 // Per thread size before we switch to non-LL for pre-Volta archs
+#define NCCL_LL_MIN_NTHREADS 64
+
 ncclResult_t ncclEnqueueCheck(struct ncclInfo* info);
 ncclResult_t ncclCpuBarrierIn(ncclComm_t comm, int* isLast);
 ncclResult_t ncclCpuBarrierLast(ncclComm_t comm);