diff --git a/configure.ac b/configure.ac
index 835f3f9..de4949c 100644
--- a/configure.ac
+++ b/configure.ac
@@ -256,7 +256,7 @@ AS_IF([test "$HAVE_OPENACC_OFFLOAD" = "yes"], [
           CFLAGS="$CFLAGS -fopenacc"
           ;;
     *nvc*)
-          CFLAGS="$CFLAGS -acc"
+          CFLAGS="$CFLAGS -acc=gpu"
           ;;
   esac
 
@@ -266,7 +266,7 @@ AS_IF([test "$HAVE_OPENACC_OFFLOAD" = "yes"], [
           FCFLAGS="$FCFLAGS -fopenacc"
           ;;
     *nvfortran*)
-          FCFLAGS="$FCFLAGS -acc"
+          FCFLAGS="$FCFLAGS -acc=gpu"
           ;;
   esac
 
@@ -282,7 +282,7 @@ AS_IF([test "$HAVE_CUBLAS_OFFLOAD" = "yes"], [
           CFLAGS="$CFLAGS -fopenacc"
           ;;
     *nvc*)
-          CFLAGS="$CFLAGS -acc"
+          CFLAGS="$CFLAGS -acc=gpu"
           ;;
   esac
 
@@ -292,7 +292,7 @@ AS_IF([test "$HAVE_CUBLAS_OFFLOAD" = "yes"], [
           FCFLAGS="$FCFLAGS -fopenacc"
           ;;
     *nvfortran*)
-          FCFLAGS="$FCFLAGS -acc"
+          FCFLAGS="$FCFLAGS -acc=gpu"
           ;;
   esac
 ])
diff --git a/org/qmckl_jastrow.org b/org/qmckl_jastrow.org
index 7b5bcee..35003f5 100644
--- a/org/qmckl_jastrow.org
+++ b/org/qmckl_jastrow.org
@@ -5706,7 +5706,7 @@ qmckl_exit_code qmckl_compute_tmp_c_acc_offload (
   const int64_t size_e = walk_num*(cord_num+1)*elec_num*elec_num;
   const int64_t size_n = walk_num*(cord_num+1)*nucl_num*elec_num; 
 
-  #pragma acc parallel create(tmp_c[0:size_tmp_c]) copyout(tmp_c [0:size_tmp_c]) copyin(een_rescaled_e[0:size_e], een_rescaled_n[0:size_n])
+  #pragma acc parallel copyout(tmp_c [0:size_tmp_c]) copyin(een_rescaled_e[0:size_e], een_rescaled_n[0:size_n])
   {
       #pragma acc loop independent gang worker vector
       for (int64_t i=0 ; i<size_tmp_c ; ++i)
@@ -6178,7 +6178,7 @@ qmckl_exit_code qmckl_compute_dtmp_c_acc_offload (
   const int64_t size_n = walk_num*(cord_num+1)*nucl_num*elec_num; 
   const int64_t size_e = walk_num*(cord_num+1)*elec_num*4*elec_num;
 
-  #pragma acc parallel create(dtmp_c[0:size_dtmp_c]) copyout(dtmp_c [0:size_dtmp_c]) copyin(een_rescaled_e_deriv_e[0:size_e], een_rescaled_n[0:size_n])
+  #pragma acc parallel copyout(dtmp_c [0:size_dtmp_c]) copyin(een_rescaled_e_deriv_e[0:size_e], een_rescaled_n[0:size_n])
   {
       #pragma acc loop independent gang worker vector
       for (int64_t i=0 ; i<size_dtmp_c ; ++i)