diff --git a/Makefile b/Makefile
index 6c927cf832feaf1f51aeb578f146f4b1d7e73b80..f8ed68f892a75d61fc1863dedd609ab2f8d9ddcc 100755
--- a/Makefile
+++ b/Makefile
@@ -20,8 +20,8 @@ endif
 
 LINKER=$(MPICC)
 
-FFTW_MPI_INC = 
-FFTW_MPI_LIB = 
+FFTW_MPI_INC = -I/opt/cray/pe/fftw/3.3.10.5/x86_rome/include
+FFTW_MPI_LIB = -L/opt/cray/pe/fftw/3.3.10.5/x86_rome/lib
 
 CFLAGS += -I./
 
@@ -63,11 +63,11 @@ OPT += -DPHASE_ON
 # SELECT THE GRIDDING KERNEL: GAUSS, GAUSS_HI_PRECISION, KAISERBESSEL
 
 
-OPT += -DGAUSS_HI_PRECISION
+#OPT += -DGAUSS_HI_PRECISION
 
 #OPT += -DGAUSS
 
-#OPT += -DKAISERBESSEL
+OPT += -DKAISERBESSEL
 
 # ========================================================
 
@@ -92,7 +92,7 @@ OPT += -DGAUSS_HI_PRECISION
 #OPT += -DCUFFTMP
 
 # FULL NVIDIA GPU SUPPORT - Recommended for full NVIDIA GPU code execution
-OPT += -DFULL_NVIDIA
+#OPT += -DFULL_NVIDIA
 ifeq (FULL_NVIDIA,$(findstring FULL_NVIDIA,$(OPT)))
 OPT += -DCUDACC -DNCCL_REDUCE -DCUFFTMP
 endif
@@ -109,7 +109,7 @@ endif
 #OPT += -DRCCL_REDUCE
 
 # FULL AMD GPU SUPPORT - Recommended for full AMD GPU code execution
-#OPT += -DFULL_AMD
+OPT += -DFULL_AMD
 ifeq (FULL_AMD,$(findstring FULL_AMD,$(OPT)))
 OPT += -DHIPCC -DRCCL_REDUCE -D__HIP_PLATFORM_AMD__
 endif
@@ -384,9 +384,9 @@ ifeq (RCCL_REDUCE,$(findstring RCCL_REDUCE,$(OPT)))
 EXEC_EXT := $(EXEC_EXT)_acc-reduce
 LINKER=$(MPIC++)
 FLAGS=$(OPTIMIZE_AMD) $(CFLAGS)
-LIBS=$(AMDLIB) 
+LIBS=$(AMDLIB_3) 
 $(OBJ_RCCL_REDUCE): $(DEPS_RCCL_REDUCE)
-	$(MPIC++) $(FLAGS) $(OPT) -c $^ $(CFLAGS) $(LIBS)
+	$(HIPCC) $(FLAGS) $(OPT) -c $^ $(CFLAGS) $(LIBS)
 OBJ += $(OBJ_RCCL_REDUCE)
 endif
 
diff --git a/gridding_rccl.hip.cpp b/gridding_rccl.hip.cpp
index dba42eeec2580d387322856e21392998c25d7791..e54fa4d6d0a28f83d4c42716d50036466541eedb 100755
--- a/gridding_rccl.hip.cpp
+++ b/gridding_rccl.hip.cpp
@@ -92,7 +92,7 @@ void gridding_data(){
   if (rank == 0) ncclGetUniqueId(&id);
   MPI_Bcast((void *)&id, sizeof(id), MPI_BYTE, 0, MPI_COMM_WORLD);
 
-  hipSetDevice(local_rank);
+  int h = hipSetDevice(local_rank);
 
   int n = hipMalloc(&grid_gpu, 2*param.num_w_planes*xaxis*yaxis * sizeof(double));
   n = hipMalloc(&gridss_gpu, 2*param.num_w_planes*xaxis*yaxis * sizeof(double));
diff --git a/w-stacking.hip.cpp b/w-stacking.hip.cpp
index 37db97cd94eaa78afc9133602cd93f7f287fd326..5c376f3c7d8a6566d84593bbeafe7dd0162edfe9 100755
--- a/w-stacking.hip.cpp
+++ b/w-stacking.hip.cpp
@@ -7,7 +7,7 @@
 #include <stdio.h>
 
 #ifdef __HIPCC__
-#include "allvars_nccl.hip.hpp"
+#include "allvars_rccl.hip.hpp"
 #endif
 
 #include "proto.h"