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

Merge remote-tracking branch 'nccl/master' into HEAD

Browse Source 
[ROCm/rccl commit: c4edc257b0]
This commit is contained in:
Wenkai Du
2020-04-27 17:16:54 +00:00
parent 5f57e6b466 c43022b9d8
commit 9813d67cd1
8 changed files with 50 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/rccl/src/graph/search.cc
+5 -2
View File
@@ -944,7 +944,10 @@ ncclResult_t ncclTopoDumpGraphs(struct ncclTopoSystem* system, int ngraphs, stru
return ncclSuccess;
}
ncclResult_t ncclTopoGetNetDev(struct ncclTopoGraph* graph, int dir, int channelId, int* dev) {
*dev = graph->inter[(channelId%graph->nChannels)*2+dir];
ncclResult_t ncclTopoGetNetDev(struct ncclTopoSystem* system, struct ncclTopoGraph* graph, int rank, int channelId, int* dev) {
int channel = channelId%graph->nChannels;
int ngpus = system->nodes[GPU].count;
int index = graph->intra[channel*ngpus] == rank ? 0 : 1;
*dev = graph->inter[channel*2+index];
return ncclSuccess;
}
projects/rccl/src/graph/topo.cc
+3 -4
View File
@@ -313,8 +313,8 @@ ncclResult_t ncclTopoAddGpu(struct ncclXmlNode* xmlGpu, struct ncclTopoSystem* s
return ncclSuccess;
}
struct kvDict kvDictPciClass[] = { { "0x060400", PCI }, { "0x068000", NVS }, { "0x068001", CPU }, { "0x030200", GPU }, { "0x030000", GPU }, { "0x038000", GPU }, { "0x020700", NIC }, { "0x020000", NIC }, { NULL, 0 } };
struct kvDict kvDictPciGen[] = { { "2.5 GT/s", 15 }, { "5 GT/s", 30 }, { "8 GT/s", 60 }, { "16 GT/s", 120 }, { NULL, 0 } }; // x100 Mbps per lane
struct kvDict kvDictPciClass[] = { { "0x060400", PCI }, { "0x068000", NVS }, { "0x068001", CPU }, { "0x030200", GPU }, { "0x030000", GPU }, { "0x038000", GPU }, { "0x020700", NIC }, { "0x020000", NIC }, { NULL, PCI /* Default fallback value */ } };
struct kvDict kvDictPciGen[] = { { "2.5 GT/s", 15 }, { "5 GT/s", 30 }, { "8 GT/s", 60 }, { "16 GT/s", 120 }, { NULL, 60 /* Default fallback */ } }; // x100 Mbps per lane
ncclResult_t ncclTopoAddPci(struct ncclXmlNode* xmlPci, struct ncclTopoSystem* system, struct ncclTopoNode* parent) {
const char* str;
@@ -366,8 +366,6 @@ ncclResult_t ncclTopoAddPci(struct ncclXmlNode* xmlPci, struct ncclTopoSystem* s
// Manage cases where speed was not indicated in /sys
if (width == 0) width = 16;
if (strlen(str) == 0 || strcasecmp(str, "Unknown speed") == 0) str = "8 GT/s";
NCCLCHECK(kvConvertToInt(str, &speed, kvDictPciGen)); // Values in 100Mbps, per lane (we want GB/s in the end)
NCCLCHECK(ncclTopoConnectNodes(node, parent, LINK_PCI, width*speed/80.0));
@@ -589,6 +587,7 @@ ncclResult_t ncclTopoGetSystem(struct ncclComm* comm, struct ncclTopoSystem** sy
NCCLCHECK(int64ToBusId(comm->peerInfo[r].busId, busId));
struct ncclXmlNode* node;
NCCLCHECK(ncclTopoFillGpu(xml, busId, &node));
if (node == NULL) continue;
NCCLCHECK(xmlSetAttrInt(node, "rank", r));
NCCLCHECK(xmlInitAttrInt(node, "gdr", comm->peerInfo[r].gdrSupport));
}
projects/rccl/src/graph/xml.cc
+1 -1
View File
@@ -578,7 +578,7 @@ ncclResult_t ncclTopoGetXmlFromGpu(struct ncclXmlNode* pciNode, nvmlDevice_t nvm
NCCLCHECK(xmlSetAttrInt(gpuNode, "dev", dev));
}
NCCLCHECK(xmlGetAttrInt(gpuNode, "dev", &dev));
if (dev == -1) return ncclSuccess;
if (dev == -1) { *gpuNodeRet = NULL; return ncclSuccess; }
NCCLCHECK(xmlGetAttrIndex(gpuNode, "sm", &index));
if (index == -1) {
projects/rccl/src/graph/xml.h
+3 -2
View File
@@ -218,8 +218,9 @@ static ncclResult_t kvConvertToInt(const char* str, int* value, struct kvDict* d
}
d++;
}
WARN("KV Convert to int : could not find value of '%s' in dictionary", str);
return ncclInternalError;
INFO(NCCL_GRAPH, "KV Convert to int : could not find value of '%s' in dictionary, falling back to %d", str, d->value);
*value = d->value;
return ncclSuccess;
}
static ncclResult_t kvConvertToStr(int value, const char** str, struct kvDict* dict) {
struct kvDict* d = dict;