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

[ROCm/rccl-tests commit: bb0f15d407]

projects/rccl-tests/src/common.cu

@@ -310,7 +310,7 @@ testResult_t CheckData(struct threadArgs* args, ncclDataType_t type, ncclRedOp_t
 
   *wrongElts = 0;
   for (int i=0; i < args->nGpus; i++) *wrongElts += wrongPerGpu[i];
-  hipFree(wrongPerGpu);
+  hipHostFree(wrongPerGpu);
 
   if (args->reportErrors && *wrongElts) args->errors[0]++;
   return testSuccess;
@@ -1169,6 +1169,8 @@ testResult_t run() {
     errors[t] = bw_count[t] = 0;
   }
 
+  fflush(stdout);
+
   const char* timeStr = report_cputime ? "cputime" : "time";
   PRINT("#\n");
   PRINT("# %10s  %12s  %8s  %6s  %6s           out-of-place                       in-place          \n", "", "", "", "", "");