AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

2.6.4-1

Browse Source 
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.


[ROCm/rccl commit: b221128eca]
This commit is contained in:
Sylvain Jeaugey
2020-01-16 16:02:42 -08:00
parent 6034c27655
commit 40adc74496
48 changed files with 3621 additions and 1398 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/rccl/src/debug.cc
+2 -4
View File
@@ -1,5 +1,5 @@
/*************************************************************************
* Copyright (c) 2016-2019, NVIDIA CORPORATION. All rights reserved.
* Copyright (c) 2016-2020, NVIDIA CORPORATION. All rights reserved.
*
* See LICENSE.txt for license information
************************************************************************/
@@ -107,7 +107,6 @@ void ncclDebugInit() {
if (debugFn[0] != '\0') {
FILE *file = fopen(debugFn, "w");
if (file != NULL) {
INFO(NCCL_ALL,"DEBUG file is '%s'", debugFn);
ncclDebugFile = file;
}
}
@@ -125,7 +124,7 @@ void ncclDebugInit() {
*/
void ncclDebugLog(ncclDebugLogLevel level, unsigned long flags, const char *filefunc, int line, const char *fmt, ...) {
if (ncclDebugLevel == -1) ncclDebugInit();
if (ncclDebugNoWarn == 1 && level == NCCL_LOG_WARN) level = NCCL_LOG_INFO;
if (ncclDebugNoWarn != 0 && level == NCCL_LOG_WARN) { level = NCCL_LOG_INFO; flags = ncclDebugNoWarn; }
char hostname[1024];
getHostName(hostname, 1024, '.');
@@ -135,7 +134,6 @@ void ncclDebugLog(ncclDebugLogLevel level, unsigned long flags, const char *file
char buffer[1024];
size_t len = 0;
pthread_mutex_lock(&ncclDebugLock);
if (ncclDebugNoWarn && ncclDebugLevel == NCCL_LOG_WARN) printf("WARN -> INFO\n");
if (level == NCCL_LOG_WARN && ncclDebugLevel >= NCCL_LOG_WARN)
len = snprintf(buffer, sizeof(buffer),
"\n%s:%d:%d [%d] %s:%d NCCL WARN ", hostname, getpid(), gettid(), cudaDev, filefunc, line);