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

2.10.3-1

Browse Source 
Add support for bfloat16.
Add ncclAvg reduction operation.
Improve performance for aggregated operations.
Improve performance for tree.
Improve network error reporting.
Add NCCL_NET parameter to force a specific network.
Add NCCL_IB_QPS_PER_CONNECTION parameter to split IB traffic onto multiple queue pairs.
Fix topology detection error in WSL2.
Fix proxy memory elements affinity (improve alltoall performance).
Fix graph search on cubemesh topologies.
Fix hang in cubemesh during NVB connections.


[ROCm/rccl commit: 7e51592129]
This commit is contained in:
Ke Wen
2021-07-08 14:12:04 -07:00
parent 3dd3a1ca66
commit 58a9f19dc3
52 changed files with 3496 additions and 2469 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/rccl/src/graph/xml.h
+13 -1
View File
@@ -38,7 +38,7 @@ struct ncclXml {
/* File functions */
#define NCCL_TOPO_XML_VERSION 1
ncclResult_t ncclTopoGetXmlFromFile(const char* xmlTopoFile, struct ncclXml* xml);
ncclResult_t ncclTopoGetXmlFromFile(const char* xmlTopoFile, struct ncclXml* xml, int warn);
ncclResult_t ncclTopoDumpXmlToFile(const char* xmlTopoFile, struct ncclXml* xml);
#define NCCL_GRAPH_XML_VERSION 1
ncclResult_t ncclTopoGetXmlGraphFromFile(const char* xmlGraphFile, struct ncclXml* xml);
@@ -137,6 +137,18 @@ static ncclResult_t xmlSetAttr(struct ncclXmlNode* node, const char* attrName, c
return ncclSuccess;
}
static ncclResult_t xmlSetAttrIfUnset(struct ncclXmlNode* node, const char* attrName, const char* value) {
int index;
NCCLCHECK(xmlGetAttrIndex(node, attrName, &index));
if (index != -1) return ncclSuccess;
index = node->nAttrs++;
strncpy(node->attrs[index].key, attrName, MAX_STR_LEN);
node->attrs[index].key[MAX_STR_LEN] = '\0';
strncpy(node->attrs[index].value, value, MAX_STR_LEN);
node->attrs[index].value[MAX_STR_LEN] = '\0';
return ncclSuccess;
}
static ncclResult_t xmlSetAttrInt(struct ncclXmlNode* node, const char* attrName, const int value) {
int index;
NCCLCHECK(xmlGetAttrIndex(node, attrName, &index));