2.6.4-1

Add support for network collectives.
Add support for XML topology dump/injection.
Add text values for GDR and P2P Levels, including "NVL".
Add speed detection for PCI, Infiniband and Ethernet cards.
Add CPU detection for ARM and AMD CPUs.
Add support for adaptive routing on Infiniband.
Change NET plugin API to v3 : merge PCI path and GPU pointer
  capability into a single structure and add other properties.

src/include/comm.h

@@ -1,5 +1,5 @@
 /*************************************************************************
- * Copyright (c) 2015-2019, NVIDIA CORPORATION. All rights reserved.
+ * Copyright (c) 2015-2020, NVIDIA CORPORATION. All rights reserved.
  *
  * See LICENSE.txt for license information
  ************************************************************************/
@@ -20,8 +20,6 @@ struct cudaLaunchParams {
 };
 #endif
 
-#define DEFAULT_BUFFER_SIZE_BYTES (1LL << 22) /* 4MiB */
-
 #define CACHE_LINE_SIZE 128
 #define MEM_ALIGN 4096
 #define CUDA_IPC_MIN 2097152UL
@@ -91,14 +89,11 @@ struct ncclComm {
   // Channels for collectives
   int nChannels;
 
-  // Only nvlink is used for inter-GPU communication
-  int nvlink;
-
   // Algorithm/Protocols thresholds
   ssize_t threadThresholds[NCCL_NUM_ALGORITHMS][NCCL_NUM_PROTOCOLS];
   float latencies[NCCL_NUM_FUNCTIONS][NCCL_NUM_ALGORITHMS][NCCL_NUM_PROTOCOLS];
   float bandwidths[NCCL_NUM_FUNCTIONS][NCCL_NUM_ALGORITHMS][NCCL_NUM_PROTOCOLS];
-  int maxThreads[NCCL_NUM_PROTOCOLS];
+  int maxThreads[NCCL_NUM_ALGORITHMS][NCCL_NUM_PROTOCOLS];
 
   // An internal CUDA stream for NCCL kernel CGMD launches
   int groupCudaStream;
@@ -136,6 +131,9 @@ struct ncclComm {
   // Global proxy thread
   pthread_t proxyThread;
   struct ncclProxyState proxyState;
+
+  // Whether this communicator uses collNet
+  int collNetSupport;
 };
 
 #endif