fixing the ar_with_bias test issue when running rccl-tests (#1912)

* fixing the AR_With_Bias issue when running rccl-tests

[ROCm/rccl commit: e738c03e39]
Este commit está contenido en:
mberenjk authored and GitHub committed 2025-10-13 13:58:21 -07:00
1 padre d51ed2fdfd
commit 433251272b
6 files changed
+27 -16

No files matched your search

+10 -1
Ver fichero
@@ -500,7 +500,14 @@ ncclResult_t ncclPrepareTasks(struct ncclComm* comm, bool* algoNeedConnect, bool
}
NCCLCHECK(getAlgoInfo(comm, &agg, collNetSupport, nvlsSupport, nTasksPerChannel, simInfo));
agg.devFuncId = ncclDevFuncId(agg.func, agg.opDev.op, agg.datatype, agg.algorithm, agg.protocol, agg.pipeline);
if(agg.func==ncclFuncAllReduce && agg.acc != nullptr)
{
agg.devFuncId = ncclDevFuncId(ncclFuncAllReduceWithBias, agg.opDev.op, agg.datatype, agg.algorithm, agg.protocol, agg.pipeline);
}
else
{
agg.devFuncId = ncclDevFuncId(agg.func, agg.opDev.op, agg.datatype, agg.algorithm, agg.protocol, agg.pipeline);
}
if (agg.devFuncId < 0) {
WARN("%s: unsupported collective. Please ensure the collective has been enabled in build.", __func__);
return ncclInvalidUsage;
@@ -523,6 +530,7 @@ ncclResult_t ncclPrepareTasks(struct ncclComm* comm, bool* algoNeedConnect, bool
struct ncclTaskColl* next = aggBeg->next;
aggBeg->algorithm = agg.algorithm;
aggBeg->protocol = agg.protocol;
aggBeg->acc = agg.acc;
aggBeg->pipeline = agg.pipeline;
if (aggBeg->protocol == NCCL_PROTO_LL) aggBeg->trafficBytes *= 4;
aggBeg->nMaxChannels = agg.nMaxChannels;
@@ -574,6 +582,7 @@ ncclResult_t ncclPrepareTasks(struct ncclComm* comm, bool* algoNeedConnect, bool
struct ncclDevWorkColl devWork = {};
devWork.sendbuff = (void*)task->sendbuff;
devWork.recvbuff = (void*)task->recvbuff;
devWork.acc = (void*)task->acc;
devWork.sendbuffOffset = task->sendbuffOffset;
devWork.recvbuffOffset = task->recvbuffOffset;
devWork.sendbuffRmtAddrs = task->sendbuffRmtAddrs;