diff --git a/src/exchangeAMRDonors.C b/src/exchangeAMRDonors.C
index eba12c0..d863f8c 100644
--- a/src/exchangeAMRDonors.C
+++ b/src/exchangeAMRDonors.C
@@ -342,7 +342,7 @@ void tioga::checkComm(void)
       sndPack[i].intData[1]=1;
     }
   //
-  pc_cart->sendRecvPackets(sndPack,rcvPack);
+  pc_cart->sendRecvPackets2(sndPack,rcvPack);
   pc_cart->clearPackets(sndPack,rcvPack);
   TIOGA_FREE(sndMap);
   TIOGA_FREE(rcvMap);
diff --git a/src/exchangeBoxes.C b/src/exchangeBoxes.C
index f3ad56b..2df1aa9 100644
--- a/src/exchangeBoxes.C
+++ b/src/exchangeBoxes.C
@@ -240,7 +240,7 @@ void tioga::exchangeBoxes(void)
     //           << std::setw(3) << key_recv << " "
     //           << std::setw(3) << key_send << std::endl;
   }
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
 
   for (int k=0; k<nrecv; k++) {
     int m=0;
diff --git a/src/exchangeDonors.C b/src/exchangeDonors.C
index 4e50f66..2bdfbf6 100644
--- a/src/exchangeDonors.C
+++ b/src/exchangeDonors.C
@@ -35,7 +35,6 @@ void tioga::exchangeDonors(void)
   // and receiving
   //
   pc->getMap(&nsend,&nrecv,&sndMap,&rcvMap);
-  if (nsend == 0) return;
   //
   // create packets to send and receive
   // and initialize them to zero
@@ -74,7 +73,8 @@ void tioga::exchangeDonors(void)
   //
   // communicate donors (comm1)
   //
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
+  //if (nsend == 0) return;
   // Initialize linked lists and populate donor data from rcvPack
   for (int ib=0;ib<nblocks;ib++) {
     auto& mb = mblocks[ib];
@@ -156,7 +156,7 @@ void tioga::exchangeDonors(void)
   // comm 2
   // notify donors that they are accepted (for now)
   //
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
 
   std::vector<int> ninterp(nblocks,0);
   for(int k=0;k<nrecv;k++)
@@ -228,7 +228,7 @@ void tioga::exchangeDonors(void)
   //
   // communciate cancellation data comm 3
   //
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
 
   for (int k=0; k<nrecv; k++) {
     int m = 0;
@@ -279,7 +279,7 @@ void tioga::exchangeDonors(void)
   // comm 4
   // final receptor data to set iblanks
   //
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
   //
   for(int ib=0;ib<nblocks;ib++)
     mblocks[ib]->clearIblanks();
diff --git a/src/exchangeSearchData.C b/src/exchangeSearchData.C
index cc661b5..58a8e62 100644
--- a/src/exchangeSearchData.C
+++ b/src/exchangeSearchData.C
@@ -113,7 +113,7 @@ void tioga::exchangeSearchData(int at_points)
         sndPack[k].realData[m++] = real_data[ii][j];
     }
   }
-  pc->sendRecvPackets(sndPack, rcvPack);
+  pc->sendRecvPackets2(sndPack, rcvPack);
 
   // Reset MeshBlock data structures
   for (int ib=0;ib<nblocks;ib++) {
diff --git a/src/parallelComm.C b/src/parallelComm.C
index b00f246..c2c93cb 100644
--- a/src/parallelComm.C
+++ b/src/parallelComm.C
@@ -212,6 +212,139 @@ void parallelComm::sendRecvPacketsAll(PACKET *sndPack, PACKET *rcvPack)
 //   TIOGA_FREE(status);
 // }
 
+void parallelComm::sendRecvPackets2(PACKET *sndPack,PACKET *rcvPack)
+{
+  int i;
+  int *sint,*sreal,*rint,*rreal;
+  //
+  sint=(int *)malloc(sizeof(int)*numprocs);
+  sreal=(int *) malloc(sizeof(int)*numprocs);
+  rint=(int *)malloc(sizeof(int)*numprocs);
+  rreal=(int *) malloc(sizeof(int)*numprocs);
+  // remove when using stl vectors and just init the vectors to 0
+  for(i=0;i<numprocs;i++){
+    sint[i]=sreal[i]=0;
+    rint[i]=rreal[i]=0;
+  }
+  for(i=0;i<nsend;i++){
+    sint[sndMap[i]]=sndPack[i].nints;			
+    sreal[sndMap[i]]=sndPack[i].nreals;
+  }
+  //
+  MPI_Alltoall(sint,1,MPI_INT,rint,1,MPI_INT,scomm);
+  MPI_Alltoall(sreal,1,MPI_INT,rreal,1,MPI_INT,scomm);
+  //
+  for(i=0;i<nrecv;i++) {
+    rcvPack[i].nints=rint[rcvMap[i]];
+    rcvPack[i].nreals=rreal[rcvMap[i]];
+  }
+
+  int all_snd_nints = std::accumulate(sint, sint + numprocs, 0);
+  int all_rcv_nints = std::accumulate(rint, rint + numprocs, 0);
+  int *all_snd_intData, *all_rcv_intData;
+  all_snd_intData=(int *) malloc(sizeof(int)*all_snd_nints);
+  all_rcv_intData=(int *) malloc(sizeof(int)*all_rcv_nints);
+  for (int i = 0; i < all_snd_nints; i++) {
+    all_snd_intData[i] = 0;
+  }
+  for (int i = 0; i < all_rcv_nints; i++) {
+    all_rcv_intData[i] = 0;
+  }
+  std::vector<int> snd_int_displs(numprocs+1, 0);
+  std::vector<int> rcv_int_displs(numprocs+1, 0);
+  for (int i=1; i <= numprocs; i++) {
+    snd_int_displs[i] = snd_int_displs[i-1] + sint[i-1];
+    rcv_int_displs[i] = rcv_int_displs[i-1] + rint[i-1];
+  }
+  for (int i=0; i < nsend; i++) {
+    int displ = snd_int_displs[sndMap[i]];
+    for(int j=0; j < sint[sndMap[i]]; j++){
+      all_snd_intData[displ+j] = sndPack[i].intData[j];
+    }
+  }
+  MPI_Request int_request;
+  MPI_Ialltoallv(all_snd_intData,
+                sint,
+                snd_int_displs.data(),
+                MPI_INT,
+                all_rcv_intData,
+                rint,
+                rcv_int_displs.data(),
+                MPI_INT,
+                scomm,
+                &int_request);
+
+  int all_snd_nreals = std::accumulate(sreal, sreal + numprocs, 0);
+  int all_rcv_nreals = std::accumulate(rreal, rreal + numprocs, 0);
+  REAL *all_snd_realData, *all_rcv_realData;
+  all_snd_realData=(REAL *) malloc(sizeof(REAL)*all_snd_nreals);
+  all_rcv_realData=(REAL *) malloc(sizeof(REAL)*all_rcv_nreals);
+  for (int i = 0; i < all_snd_nreals; i++) {
+    all_snd_realData[i] = 0;
+  }
+  for (int i = 0; i < all_rcv_nreals; i++) {
+    all_rcv_realData[i] = 0;
+  }
+  std::vector<int> snd_real_displs(numprocs+1, 0);
+  std::vector<int> rcv_real_displs(numprocs+1, 0);
+  for (int i=1; i <= numprocs; i++) {
+    snd_real_displs[i] = snd_real_displs[i-1] + sreal[i-1];
+    rcv_real_displs[i] = rcv_real_displs[i-1] + rreal[i-1];
+  }
+  for (int i=0; i < nsend; i++) {
+    int displ = snd_real_displs[sndMap[i]];
+    for(int j=0; j < sreal[sndMap[i]]; j++){
+      all_snd_realData[displ+j] = sndPack[i].realData[j];
+    }
+  }
+  MPI_Request real_request;
+  MPI_Ialltoallv(all_snd_realData,
+                sreal,
+                snd_real_displs.data(),
+                MPI_DOUBLE,
+                all_rcv_realData,
+                rreal,
+                rcv_real_displs.data(),
+                MPI_DOUBLE,
+                scomm,
+                &real_request);
+
+  // FIXME: here and above I think I should move this a bit lower
+  MPI_Wait(&int_request, MPI_STATUS_IGNORE);
+  for(i=0;i<nrecv;i++){
+    if (rcvPack[i].nints > 0) {
+      rcvPack[i].intData=(int *) malloc(sizeof(int)*rcvPack[i].nints);
+    }
+    if (rcvPack[i].nreals > 0) {
+      rcvPack[i].realData=(REAL *) malloc(sizeof(REAL)*rcvPack[i].nreals);
+    }
+  }
+
+  // FIXME: here and above I think I should move this a bit lower
+  MPI_Wait(&real_request, MPI_STATUS_IGNORE);
+  for (int i=0; i < nrecv; i++) {
+    int displ = rcv_int_displs[rcvMap[i]];
+    for(int j=0; j < rint[rcvMap[i]]; j++){
+      rcvPack[i].intData[j] = all_rcv_intData[displ+j];
+    }
+  }
+  for (int i=0; i < nrecv; i++) {
+    int displ = rcv_real_displs[rcvMap[i]];
+    for(int j=0; j < rreal[rcvMap[i]]; j++){
+      rcvPack[i].realData[j] = all_rcv_realData[displ+j];
+    }
+  }
+
+  TIOGA_FREE(all_snd_intData);
+  TIOGA_FREE(all_rcv_intData);
+  TIOGA_FREE(all_snd_realData);
+  TIOGA_FREE(all_rcv_realData);
+  TIOGA_FREE(sint);
+  TIOGA_FREE(sreal);
+  TIOGA_FREE(rint);
+  TIOGA_FREE(rreal);
+}
+
 void parallelComm::sendRecvPackets(PACKET *sndPack,PACKET *rcvPack)
 {
   int i;
diff --git a/src/parallelComm.h b/src/parallelComm.h
index c941a1e..6796e7a 100644
--- a/src/parallelComm.h
+++ b/src/parallelComm.h
@@ -58,6 +58,8 @@ class parallelComm {
 
   void sendRecvPacketsAll(PACKET *sndPack, PACKET *rcvPack);
 
+  void sendRecvPackets2(PACKET *sndPack, PACKET *rcvPack);
+
   void sendRecvPackets(PACKET *sndPack, PACKET *rcvPack);
 
   void sendRecvPacketsCheck(PACKET *sndPack, PACKET *rcvPack);
diff --git a/src/tioga.C b/src/tioga.C
index 3405419..02e4984 100644
--- a/src/tioga.C
+++ b/src/tioga.C
@@ -492,7 +492,8 @@ void tioga::performConnectivity(void)
   this->myTimer("tioga::exchangeDonors",1);
   //this->reduce_fringes();
   //outputStatistics();
-  MPI_Allreduce(&ihigh,&ihighGlobal,1,MPI_INT,MPI_MAX,scomm);
+  //MPI_Allreduce(&ihigh,&ihighGlobal,1,MPI_INT,MPI_MAX,scomm);
+  ihighGlobal = 0;
   //if (ihighGlobal) {
   this->myTimer("tioga::getCellIblanks",0);
   for (int ib=0;ib<nblocks;ib++) {
@@ -743,7 +744,7 @@ void tioga::dataUpdate(int nvar,int interptype, int at_points)
   fp=NULL;
   //
   pc->getMap(&nsend,&nrecv,&sndMap,&rcvMap);
-  if (nsend==0) return;
+  //if (nsend==0) return;
   sndPack=(PACKET *)malloc(sizeof(PACKET)*nsend);
   rcvPack=(PACKET *)malloc(sizeof(PACKET)*nrecv);
   //
@@ -807,7 +808,7 @@ void tioga::dataUpdate(int nvar,int interptype, int at_points)
   //
   // communicate the data across
   //
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
   //
   // decode the packets and update the data
   //
@@ -968,7 +969,7 @@ void tioga::getReceptorInfo(std::vector<int>& receptors)
 
   if (nsend == 0) {
       receptors.clear();
-      return;
+      //return;
   }
 
   //
@@ -1037,7 +1038,7 @@ void tioga::getReceptorInfo(std::vector<int>& receptors)
     }
   }
 
-  pc->sendRecvPackets(sndPack, rcvPack);
+  pc->sendRecvPackets2(sndPack, rcvPack);
 
   int rsize=0;
   for (int k=0; k<nrecv; k++) {
@@ -1225,7 +1226,7 @@ void tioga::reduce_fringes(void)
   //
   // communciate cancellation data comm 3
   //
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
   //
   for (int k=0; k<nrecv; k++) {
     int m = 0;
@@ -1275,7 +1276,7 @@ void tioga::reduce_fringes(void)
   // comm 4
   // final receptor data to set iblanks
   //
-  pc->sendRecvPackets(sndPack,rcvPack);
+  pc->sendRecvPackets2(sndPack,rcvPack);
   //
   for(int ib=0;ib<nblocks;ib++)
     mblocks[ib]->clearIblanks();