diff --git a/src/include/mscclpp/mscclpp_nccl.h b/src/include/mscclpp/mscclpp_nccl.h
index fbc1b87e04..c4dc7dfa0c 100644
--- a/src/include/mscclpp/mscclpp_nccl.h
+++ b/src/include/mscclpp/mscclpp_nccl.h
@@ -49,4 +49,6 @@ namespace std {
 
 bool operator ==(const ncclUniqueId& a, const ncclUniqueId& b);
 
+bool mscclppCommCompatible(ncclComm_t comm);
+
 #endif
diff --git a/src/init.cc b/src/init.cc
index 0c3297f0d3..c5beab45dd 100644
--- a/src/init.cc
+++ b/src/init.cc
@@ -189,25 +189,6 @@ ncclResult_t ncclGetUniqueId_impl(ncclUniqueId* out) {
   NCCLCHECK(PtrCheck(out, "GetUniqueId", "out"));
   ncclResult_t res = bootstrapGetUniqueId((struct ncclBootstrapHandle*)out);
   TRACE_CALL("ncclGetUniqueId(0x%llx)", (unsigned long long)hashUniqueId(*out));
-  if (rcclParamMscclppEnabled()) {
-#ifdef ENABLE_MSCCLPP
-    NCCLCHECK(res);
-    int dev;
-    CUDACHECK(cudaGetDevice(&dev));
-    hipDeviceProp_t devProp;
-    CUDACHECK(hipGetDeviceProperties(&devProp, dev));
-    if (IsArchMatch(devProp.gcnArchName, "gfx94")) {
-      auto& mscclppUniqueId = mscclpp_uniqueIdMap[*out];
-      res = mscclpp_ncclGetUniqueId(&mscclppUniqueId);
-      TRACE_CALL("mscclpp_ncclGetUniqueId(0x%llx)", (unsigned long long)hashUniqueId(mscclppUniqueId));
-      mscclpp_uniqueIdReverseMap[mscclppUniqueId].insert(*out);
-    } else {
-      WARN("MSCCL++: Cannot enable MSCCL++ on %s architecture", devProp.gcnArchName);
-    }
-#else
-    WARN("MSCCL++: Feature not enabled. ENABLE_MSCCLPP must be defined at compile-time to enable this feature.");
-#endif
-  }
   return res;
 }
 
@@ -1982,27 +1963,43 @@ static ncclResult_t ncclCommInitRankFunc(struct ncclAsyncJob* job_) {
       auto& mscclppUniqueId = mscclpp_uniqueIdMap[origUniqueId];
       mscclpp_uniqueIdMap[job->commId] = mscclppUniqueId;
       mscclpp_uniqueIdReverseMap[mscclppUniqueId].insert(job->commId);
+      ncclCommToUniqueIdMap[comm] = job->commId;
     }
   }
   else
 #endif
   if (rcclParamMscclppEnabled()) {
 #ifdef ENABLE_MSCCLPP
-    hipDeviceProp_t devProp;
-    CUDACHECK(hipGetDeviceProperties(&devProp, cudaDev));
-    comm->mscclppCompatible = IsArchMatch(devProp.gcnArchName, "gfx94");
-    if (comm->mscclppCompatible) {
-      auto& mscclppUniqueId = mscclpp_uniqueIdMap[job->commId];
-      NCCLCHECKGOTO(bootstrapIntraNodeBroadcast(comm->bootstrap, comm->localRankToRank, comm->localRank, comm->localRanks, 0, &mscclppUniqueId, sizeof(mscclppUniqueId)), res, fail);
-      unsigned long long mscclppUniqueIdHash; (void)mscclppUniqueIdHash;
-      TRACE_CALL("bootstrapIntraNodeBroadcast(rank=%d, nranks=%d, root=%d, bcastData=hash:0x%llx)", comm->localRank, comm->localRanks, 0, (mscclppUniqueIdHash = (unsigned long long)hashUniqueId(mscclppUniqueId)));
-      comm->mscclpp_threshold = rcclParamMscclppThreshold();
-      INFO(NCCL_INIT, "MSCCL++: Enabled! Msg size threshold=%zu", comm->mscclpp_threshold);
-      NCCLCHECKGOTO(mscclpp_ncclCommInitRank(&(comm->mscclpp_comm), job->nranks, mscclppUniqueId, job->myrank), res, fail);
-      TRACE_CALL("mscclpp_ncclCommInitRank (*comm=%p, nranks=%d, commId=hash:0x%llx, myrank=%d)", comm->mscclpp_comm, job->nranks, mscclppUniqueIdHash, job->myrank);
-      mscclpp_commToUniqueIdMap[comm->mscclpp_comm] = mscclppUniqueId;
+    if (mscclEnabled() && (comm->topo->mscclEnabled || mscclForceEnabled()) && mscclppCommCompatible(comm)) {
+      hipDeviceProp_t devProp;
+      CUDACHECK(hipGetDeviceProperties(&devProp, cudaDev));
+      comm->mscclppCompatible = IsArchMatch(devProp.gcnArchName, "gfx94");
+      if (comm->mscclppCompatible) {
+        bool mapContainsId = (mscclpp_uniqueIdMap.count(job->commId) > 0);
+        auto& mscclppUniqueId = mscclpp_uniqueIdMap[job->commId];
+        if (comm->localRank == 0 && !mapContainsId) {
+          NCCLCHECKGOTO(mscclpp_ncclGetUniqueId(&mscclppUniqueId), res, fail);
+          TRACE_CALL("mscclpp_ncclGetUniqueId(0x%llx)", (unsigned long long)hashUniqueId(mscclppUniqueId));
+        }
+
+        NCCLCHECKGOTO(bootstrapIntraNodeBroadcast(comm->bootstrap, comm->localRankToRank, comm->localRank, comm->localRanks, 0, &mscclppUniqueId, sizeof(mscclppUniqueId)), res, fail);
+        unsigned long long mscclppUniqueIdHash; (void)mscclppUniqueIdHash;
+        TRACE_CALL("bootstrapIntraNodeBroadcast(rank=%d, nranks=%d, root=%d, bcastData=hash:0x%llx)", comm->localRank, comm->localRanks, 0, (mscclppUniqueIdHash = (unsigned long long)hashUniqueId(mscclppUniqueId)));
+        mscclpp_uniqueIdReverseMap[mscclppUniqueId].insert(job->commId);
+
+        comm->mscclpp_threshold = rcclParamMscclppThreshold();
+        INFO(NCCL_INIT, "MSCCL++: Enabled! Msg size threshold=%zu", comm->mscclpp_threshold);
+
+        NCCLCHECKGOTO(mscclpp_ncclCommInitRank(&(comm->mscclpp_comm), job->nranks, mscclppUniqueId, job->myrank), res, fail);
+        TRACE_CALL("mscclpp_ncclCommInitRank (*comm=%p, nranks=%d, commId=hash:0x%llx, myrank=%d)", comm->mscclpp_comm, job->nranks, mscclppUniqueIdHash, job->myrank);
+        mscclpp_commToUniqueIdMap[comm->mscclpp_comm] = mscclppUniqueId;
+        ncclCommToUniqueIdMap[comm] = job->commId;
+      } else {
+        WARN("MSCCL++: Cannot enable MSCCL++ on %s architecture", devProp.gcnArchName);
+      }
     } else {
-      WARN("MSCCL++: Cannot enable MSCCL++ on %s architecture", devProp.gcnArchName);
+      comm->mscclppCompatible = false;
+      WARN("MSCCL++: Cannot enable MSCCL++; environment is not MSCCL compatible");
     }
 #else
     WARN("MSCCL++: Feature not enabled. ENABLE_MSCCLPP must be defined at compile-time to enable this feature.");
diff --git a/src/misc/msccl/msccl_lifecycle.cc b/src/misc/msccl/msccl_lifecycle.cc
index 2b179fccaf..d186e0c47f 100644
--- a/src/misc/msccl/msccl_lifecycle.cc
+++ b/src/misc/msccl/msccl_lifecycle.cc
@@ -84,6 +84,12 @@ static bool mscclCommCompatible(ncclComm_t comm) {
   return true;
 }
 
+#ifdef ENABLE_MSCCLPP
+bool mscclppCommCompatible(ncclComm_t comm) {
+  return mscclCommCompatible(comm);
+}
+#endif
+
 const char *mscclFuncNames[] = {
             "mscclFuncReduce",
             "mscclFuncBroadcast",