diff --git a/lib/gpu/lal_base_atomic.cpp b/lib/gpu/lal_base_atomic.cpp
index 3d47df8a92..da54f1dca3 100644
--- a/lib/gpu/lal_base_atomic.cpp
+++ b/lib/gpu/lal_base_atomic.cpp
@@ -64,15 +64,12 @@ int BaseAtomicT::init_atomic(const int nlocal, const int nall,
   } else
     _nbor_data=&(nbor->dev_nbor);
 
-  int success=device->init(*ans,false,false,nlocal,host_nlocal,nall,nbor,
-                           maxspecial,_gpu_host,max_nbors,cell_size,false,
-                           _threads_per_atom);
+  int success=device->init(*ans,false,false,nlocal,nall,maxspecial);
   if (success!=0)
     return success;
 
   success = device->init_nbor(nbor,nlocal,host_nlocal,nall,maxspecial,_gpu_host,
                   max_nbors,cell_size,false,_threads_per_atom);
-
   if (success!=0)
     return success;
 
diff --git a/lib/gpu/lal_base_charge.cpp b/lib/gpu/lal_base_charge.cpp
index 94e7502c55..a3ec710baa 100644
--- a/lib/gpu/lal_base_charge.cpp
+++ b/lib/gpu/lal_base_charge.cpp
@@ -65,15 +65,12 @@ int BaseChargeT::init_atomic(const int nlocal, const int nall,
   } else
     _nbor_data=&(nbor->dev_nbor);
 
-  int success=device->init(*ans,true,false,nlocal,host_nlocal,nall,nbor,
-                           maxspecial,_gpu_host,max_nbors,cell_size,false,
-                           _threads_per_atom);
+  int success=device->init(*ans,true,false,nlocal,nall,maxspecial);
   if (success!=0)
     return success;
 
   success = device->init_nbor(nbor,nlocal,host_nlocal,nall,maxspecial,_gpu_host,
                   max_nbors,cell_size,false,_threads_per_atom);
-
   if (success!=0)
     return success;
 
diff --git a/lib/gpu/lal_base_dipole.cpp b/lib/gpu/lal_base_dipole.cpp
index b2a41f10cf..9fc7e1b235 100644
--- a/lib/gpu/lal_base_dipole.cpp
+++ b/lib/gpu/lal_base_dipole.cpp
@@ -66,15 +66,12 @@ int BaseDipoleT::init_atomic(const int nlocal, const int nall,
   } else
     _nbor_data=&(nbor->dev_nbor);
 
-  int success=device->init(*ans,true,true,nlocal,host_nlocal,nall,nbor,
-                           maxspecial,_gpu_host,max_nbors,cell_size,false,
-                           _threads_per_atom);
+  int success=device->init(*ans,true,true,nlocal,nall,maxspecial);
   if (success!=0)
     return success;
 
   success = device->init_nbor(nbor,nlocal,host_nlocal,nall,maxspecial,_gpu_host,
                   max_nbors,cell_size,false,_threads_per_atom);
-
   if (success!=0)
     return success;
 
diff --git a/lib/gpu/lal_base_dpd.cpp b/lib/gpu/lal_base_dpd.cpp
index 3f71c820c7..eb5c2088a6 100644
--- a/lib/gpu/lal_base_dpd.cpp
+++ b/lib/gpu/lal_base_dpd.cpp
@@ -65,9 +65,7 @@ int BaseDPDT::init_atomic(const int nlocal, const int nall,
   } else
     _nbor_data=&(nbor->dev_nbor);
 
-  int success=device->init(*ans,false,false,nlocal,host_nlocal,nall,nbor,
-                           maxspecial,_gpu_host,max_nbors,cell_size,false,
-                           _threads_per_atom,true);
+  int success=device->init(*ans,false,false,nlocal,nall,maxspecial,true);
   if (success!=0)
     return success;
 
diff --git a/lib/gpu/lal_base_ellipsoid.cpp b/lib/gpu/lal_base_ellipsoid.cpp
index 8918a3140c..eea5344e33 100644
--- a/lib/gpu/lal_base_ellipsoid.cpp
+++ b/lib/gpu/lal_base_ellipsoid.cpp
@@ -71,12 +71,15 @@ int BaseEllipsoidT::init_base(const int nlocal, const int nall,
 
   _threads_per_atom=device->threads_per_atom();
 
-  int success=device->init(*ans,false,true,nlocal,host_nlocal,nall,nbor,
-                           maxspecial,_gpu_host,max_nbors,cell_size,true,
-                           1);
+  int success=device->init(*ans,false,true,nlocal,nall,maxspecial);
   if (success!=0)
     return success;
 
+  success = device->init_nbor(nbor,nlocal,host_nlocal,nall,maxspecial,_gpu_host,
+                  max_nbors,cell_size,true,1);
+  if (success!=0)
+    return success;
+  
   ucl_device=device->gpu;
   atom=&device->atom;
 
diff --git a/lib/gpu/lal_base_three.cpp b/lib/gpu/lal_base_three.cpp
index ba28d697cc..0510b84d92 100644
--- a/lib/gpu/lal_base_three.cpp
+++ b/lib/gpu/lal_base_three.cpp
@@ -78,15 +78,12 @@ int BaseThreeT::init_three(const int nlocal, const int nall,
   if (_threads_per_atom*_threads_per_atom>device->warp_size())
     return -10;
 
-  int success=device->init(*ans,false,false,nlocal,host_nlocal,nall,nbor,
-                           maxspecial,_gpu_host,max_nbors,cell_size,false,
-                           _threads_per_atom);
+  int success=device->init(*ans,false,false,nlocal,nall,maxspecial);
   if (success!=0)
     return success;
 
   success = device->init_nbor(nbor,nlocal,host_nlocal,nall,maxspecial,_gpu_host,
                   max_nbors,cell_size,false,_threads_per_atom);
-
   if (success!=0)
     return success;
 
diff --git a/lib/gpu/lal_device.cpp b/lib/gpu/lal_device.cpp
index 411e19a78a..3b7f393056 100644
--- a/lib/gpu/lal_device.cpp
+++ b/lib/gpu/lal_device.cpp
@@ -246,11 +246,8 @@ int DeviceT::set_ocl_params(char *ocl_vendor) {
 template <class numtyp, class acctyp>
 int DeviceT::init(Answer<numtyp,acctyp> &ans, const bool charge,
                   const bool rot, const int nlocal,
-                  const int host_nlocal, const int nall,
-                  Neighbor *nbor, const int maxspecial,
-                  const int gpu_host, const int max_nbors,
-                  const double cell_size, const bool pre_cut,
-                  const int threads_per_atom, const bool vel) {
+                  const int nall, const int maxspecial,
+                  const bool vel) {
   if (!_device_init)
     return -1;
   if (sizeof(acctyp)==sizeof(double) && gpu->double_precision()==false)
diff --git a/lib/gpu/lal_device.h b/lib/gpu/lal_device.h
index 68d88a3182..4287967210 100644
--- a/lib/gpu/lal_device.h
+++ b/lib/gpu/lal_device.h
@@ -53,8 +53,38 @@ class Device {
                   const int t_per_atom, const double cell_size,
                   char *vendor_string, const int block_pair);
 
-  /// Initialize the device for Atom and Neighbor storage
-  /** \param rot True if quaternions need to be stored
+  /// Initialize the device for Atom storage
+  /** \param charge True if charges need to be stored 
+    * \param rot True if quaternions need to be stored
+    * \param nlocal Total number of local particles to allocate memory for
+    * \param host_nlocal Initial number of host particles to allocate memory for
+    * \param nall Total number of local+ghost particles
+    *
+    * Returns:
+    * -  0 if successfull
+    * - -1 if fix gpu not found
+    * - -3 if there is an out of memory error
+    * - -4 if the GPU library was not compiled for GPU
+    * - -5 Double precision is not supported on card **/
+  int init(Answer<numtyp,acctyp> &a, const bool charge, const bool rot,
+           const int nlocal, const int nall, const int maxspecial,
+           const bool vel=false);
+
+  /// Initialize the device for Atom storage only
+  /** \param nlocal Total number of local particles to allocate memory for
+    * \param nall Total number of local+ghost particles
+    *
+    * Returns:
+    * -  0 if successfull
+    * - -1 if fix gpu not found
+    * - -3 if there is an out of memory error
+    * - -4 if the GPU library was not compiled for GPU
+    * - -5 Double precision is not supported on card **/
+  int init(Answer<numtyp,acctyp> &ans, const int nlocal, const int nall);
+
+  /// Initialize neighbor list storage and build
+  /** \param charge True if charges need to be stored
+    * \param rot True if quaternions need to be stored
     * \param nlocal Total number of local particles to allocate memory for
     * \param host_nlocal Initial number of host particles to allocate memory for
     * \param nall Total number of local+ghost particles
@@ -73,25 +103,6 @@ class Device {
     * - -3 if there is an out of memory error
     * - -4 if the GPU library was not compiled for GPU
     * - -5 Double precision is not supported on card **/
-  int init(Answer<numtyp,acctyp> &a, const bool charge, const bool rot,
-           const int nlocal, const int host_nlocal, const int nall,
-           Neighbor *nbor, const int maxspecial, const int gpu_host,
-           const int max_nbors, const double cell_size, const bool pre_cut,
-           const int threads_per_atom, const bool vel=false);
-
-  /// Initialize the device for Atom storage only
-  /** \param nlocal Total number of local particles to allocate memory for
-    * \param nall Total number of local+ghost particles
-    *
-    * Returns:
-    * -  0 if successfull
-    * - -1 if fix gpu not found
-    * - -3 if there is an out of memory error
-    * - -4 if the GPU library was not compiled for GPU
-    * - -5 Double precision is not supported on card **/
-  int init(Answer<numtyp,acctyp> &ans, const int nlocal, const int nall);
-
-  /// Initialize neighbor list build -- callback function from pair
   int init_nbor(Neighbor *nbor, const int nlocal,
                   const int host_nlocal, const int nall,
                   const int maxspecial, const int gpu_host,