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

2.7.3-1

Browse Source 
Add support for A100 GPU and related platforms.
Add support for CUDA 11.
Add support for send/receive operations (beta).


[ROCm/rccl commit: 5949d96f36]
This commit is contained in:
Sylvain Jeaugey
2020-05-12 14:40:18 -07:00
parent c43022b9d8
commit 89072c82e5
54 changed files with 2051 additions and 1254 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/rccl/src/graph/xml.cc
+1 -1
View File
@@ -590,7 +590,7 @@ ncclResult_t ncclTopoGetXmlFromGpu(struct ncclXmlNode* pciNode, nvmlDevice_t nvm
NCCLCHECK(xmlGetSub(pciNode, "nvlink", &nvlNode));
if (nvlNode == NULL) {
// NVML NVLink detection
int maxNvLinks = (sm < 60) ? 0 : (sm < 70) ? 4 : 6;
int maxNvLinks = (sm < 60) ? 0 : (sm < 70) ? 4 : (sm < 80) ? 6 : 12;
if (maxNvLinks > 0 && nvmlDev == NULL) {
WARN("No NVML device handle. Skipping nvlink detection.\n");