diff --git a/src/Make.py b/src/Make.py
index f39e0bbaf7..3692aaee3a 100755
--- a/src/Make.py
+++ b/src/Make.py
@@ -90,18 +90,19 @@ def switch2str(switches,switch_order):
 # check if compiler works with ccflags on dummy one-line tmpauto.cpp file
 # return 1 if successful, else 0
 # warn = 1 = print warning if not successful, warn = 0 = no warning
+# NOTE: unrecognized -override-limits can leave verride-limits file
 
 def compile_check(compiler,ccflags,warn):
   open("tmpauto.cpp",'w').write("int main(int, char **) {}")
   str = "%s %s -c tmpauto.cpp" % (compiler,ccflags)
   txt = commands.getoutput(str)
+  flag = 1
   if txt or not os.path.isfile("tmpauto.o"):
     flag = 0
     if warn:
       print str
       if txt: print txt
       else: print "compile produced no output"
-  else: flag = 1
   os.remove("tmpauto.cpp")
   if os.path.isfile("tmpauto.o"): os.remove("tmpauto.o")
   return flag
@@ -114,13 +115,13 @@ def link_check(linker,linkflags,warn):
   open("tmpauto.cpp",'w').write("int main(int, char **) {}")
   str = "%s %s -o tmpauto tmpauto.cpp" % (linker,linkflags)
   txt = commands.getoutput(str)
+  flag = 1
   if txt or not os.path.isfile("tmpauto"):
     flag = 0
     if warn:
       print str
       if txt: print txt
       else: print "link produced no output"
-  else: flag = 1
   os.remove("tmpauto.cpp")
   if os.path.isfile("tmpauto"): os.remove("tmpauto")
   return flag
diff --git a/src/USER-CUDA/Install.sh b/src/USER-CUDA/Install.sh
index b963c26216..f6501877a3 100755
--- a/src/USER-CUDA/Install.sh
+++ b/src/USER-CUDA/Install.sh
@@ -47,7 +47,6 @@ action compute_temp_cuda.h
 action compute_temp_partial_cuda.cpp
 action compute_temp_partial_cuda.h
 action cuda.cpp
-action cuda.h
 action cuda_data.h
 action cuda_modify_flags.h
 action cuda_neigh_list.cpp
@@ -157,6 +156,7 @@ action pppm_cuda.cpp pppm.cpp
 action pppm_cuda.h pppm.cpp
 action pppm_old.cpp pppm.cpp
 action pppm_old.h pppm.cpp
+action user_cuda.h
 action verlet_cuda.cpp
 action verlet_cuda.h
 
diff --git a/src/USER-CUDA/atom_vec_atomic_cuda.cpp b/src/USER-CUDA/atom_vec_atomic_cuda.cpp
index a2233fab6e..1ec8e2964a 100644
--- a/src/USER-CUDA/atom_vec_atomic_cuda.cpp
+++ b/src/USER-CUDA/atom_vec_atomic_cuda.cpp
@@ -46,7 +46,7 @@
 #include "fix.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "comm.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/atom_vec_charge_cuda.cpp b/src/USER-CUDA/atom_vec_charge_cuda.cpp
index 016e7fcca6..cecdc70ed1 100644
--- a/src/USER-CUDA/atom_vec_charge_cuda.cpp
+++ b/src/USER-CUDA/atom_vec_charge_cuda.cpp
@@ -46,7 +46,7 @@
 #include "fix.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "comm.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/comm_cuda.cpp b/src/USER-CUDA/comm_cuda.cpp
index 9df27080dc..d8438e9a4c 100644
--- a/src/USER-CUDA/comm_cuda.cpp
+++ b/src/USER-CUDA/comm_cuda.cpp
@@ -35,7 +35,7 @@
 #include "fix.h"
 #include "group.h"
 #include "compute.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "error.h"
 #include "memory.h"
 #include "comm_cuda_cu.h"
diff --git a/src/USER-CUDA/compute_pressure_cuda.cpp b/src/USER-CUDA/compute_pressure_cuda.cpp
index 339374308e..e87264d722 100644
--- a/src/USER-CUDA/compute_pressure_cuda.cpp
+++ b/src/USER-CUDA/compute_pressure_cuda.cpp
@@ -51,7 +51,7 @@
 #include "improper.h"
 #include "kspace.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/compute_temp_cuda.cpp b/src/USER-CUDA/compute_temp_cuda.cpp
index c327e6d004..4b2a203a47 100644
--- a/src/USER-CUDA/compute_temp_cuda.cpp
+++ b/src/USER-CUDA/compute_temp_cuda.cpp
@@ -48,7 +48,7 @@
 #include "fix.h"
 #include "group.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/compute_temp_partial_cuda.cpp b/src/USER-CUDA/compute_temp_partial_cuda.cpp
index 4d5a2fa9df..eb14fc0ccb 100644
--- a/src/USER-CUDA/compute_temp_partial_cuda.cpp
+++ b/src/USER-CUDA/compute_temp_partial_cuda.cpp
@@ -49,7 +49,7 @@
 #include "group.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/cuda.cpp b/src/USER-CUDA/cuda.cpp
index 4afb212853..044ebc37e5 100644
--- a/src/USER-CUDA/cuda.cpp
+++ b/src/USER-CUDA/cuda.cpp
@@ -24,7 +24,7 @@
 #include <cstdlib>
 #include <cstdio>
 #include <cstring>
-#include "cuda.h"
+#include "user_cuda.h"
 #include "atom.h"
 #include "domain.h"
 #include "force.h"
diff --git a/src/USER-CUDA/cuda.h b/src/USER-CUDA/cuda.h
deleted file mode 100644
index 39269d2050..0000000000
--- a/src/USER-CUDA/cuda.h
+++ /dev/null
@@ -1,159 +0,0 @@
-/* ----------------------------------------------------------------------
-   LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator
-
-   Original Version:
-   http://lammps.sandia.gov, Sandia National Laboratories
-   Steve Plimpton, sjplimp@sandia.gov
-
-   See the README file in the top-level LAMMPS directory.
-
-   -----------------------------------------------------------------------
-
-   USER-CUDA Package and associated modifications:
-   https://sourceforge.net/projects/lammpscuda/
-
-   Christian Trott, christian.trott@tu-ilmenau.de
-   Lars Winterfeld, lars.winterfeld@tu-ilmenau.de
-   Theoretical Physics II, University of Technology Ilmenau, Germany
-
-   See the README file in the USER-CUDA directory.
-
-   This software is distributed under the GNU General Public License.
-------------------------------------------------------------------------- */
-
-#ifndef CUDA_H
-#define CUDA_H
-
-#include "pointers.h"
-#include "cuda_shared.h"
-#include "cuda_data.h"
-#include "cuda_precision.h"
-#include <map>
-
-#ifdef _DEBUG
-#define MYDBG(a) a
-#else
-#define MYDBG(a)
-#endif
-
-namespace LAMMPS_NS
-{
-class Cuda : protected Pointers
-{
-  public:
-    Cuda(class LAMMPS*);
-    ~Cuda();
-    //static void setDevice(class LAMMPS*);
-    void allocate();
-
-    void accelerator(int, char**);
-    void activate();
-
-    void setSharedDataZero();
-    void setSystemParams();
-
-    void setDomainParams();
-
-    void checkResize();
-    void evsetup_eatom_vatom(int eflag_atom, int vflag_atom);
-    void uploadAll();
-    void downloadAll();
-    void upload(int datamask);
-    void download(int datamask);
-    void downloadX();
-
-    class CudaNeighList* registerNeighborList(class NeighList* neigh_list);
-    void uploadAllNeighborLists();
-    void downloadAllNeighborLists();
-    void set_neighinit(int dist_check, double triggerneighsq) {
-      shared_data.atom.dist_check = dist_check;
-      shared_data.atom.triggerneighsq = triggerneighsq;
-    }
-    bool decide_by_integrator() {
-      return neighbor_decide_by_integrator  && cu_xhold && finished_setup;
-    }
-    void update_xhold(int &maxhold, double* xhold);
-
-    void setTimingsZero();
-    void print_timings();
-
-    void cu_x_download() {
-      cu_x->download();
-    }
-    bool device_set;
-    bool dotiming;
-    bool dotestatom;
-    int testatom;
-
-    double uploadtime, downloadtime;
-    bool finished_setup, begin_setup;
-    bool oncpu;
-    bool finished_run;
-
-    int self_comm;
-
-    int cuda_exists;
-
-    double extent[6];
-    int* debugdata;
-    // data shared between host code and device code
-    // (number of atoms, device pointers for up- & download)
-    cuda_shared_data shared_data;
-
-    cCudaData<double  , F_FLOAT , x >* cu_q;
-    cCudaData<double  , F_FLOAT , yx>* cu_f;
-    cCudaData<double  , V_FLOAT , x >* cu_mass;
-    cCudaData<double  , V_FLOAT , x >* cu_rmass;
-    cCudaData<double  , V_FLOAT , yx>* cu_v;
-    cCudaData<double  , X_FLOAT , yx>* cu_x;
-    cCudaData<double  , X_FLOAT , yx>* cu_xhold;
-    cCudaData<int     , int     , x >* cu_mask;
-    cCudaData<int     , int     , x >* cu_tag;
-    cCudaData<int     , int     , x >* cu_type;
-    cCudaData<int     , int     , x >* cu_image;
-    cCudaData<double  , ENERGY_FLOAT, x >* cu_eatom;
-    cCudaData<double  , ENERGY_FLOAT, yx>* cu_vatom;
-    cCudaData<double  , ENERGY_FLOAT, x >* cu_virial;
-    cCudaData<double  , ENERGY_FLOAT, x >* cu_eng_vdwl;
-    cCudaData<double  , ENERGY_FLOAT, x >* cu_eng_coul;
-    cCudaData<double  , double  , x >* cu_extent;
-    int* binned_id;
-    cCudaData<int           , int            , xx >* cu_binned_id;
-    int* binned_idnew;
-    cCudaData<int           , int            , xx >* cu_binned_idnew;
-    cCudaData<int           , int            , x >* cu_debugdata;
-    cCudaData<double  , X_FLOAT , x>* cu_radius;
-    cCudaData<double  , F_FLOAT , x>* cu_density;
-    cCudaData<double  , V_FLOAT , yx>* cu_omega;
-    cCudaData<double  , F_FLOAT , yx>* cu_torque;
-    cCudaData<int           , int            , yx >* cu_special;
-    cCudaData<int           , int            , yx >* cu_nspecial;
-    cCudaData<int     , int     , x >* cu_molecule;
-
-
-    cCudaData<X_FLOAT  , X_FLOAT , x>* cu_x_type;
-    X_FLOAT* x_type;
-
-    cCudaData<V_FLOAT  , V_FLOAT , x>* cu_v_radius;
-    V_FLOAT* v_radius;
-
-    cCudaData<V_FLOAT  , V_FLOAT , x>* cu_omega_rmass;
-    V_FLOAT* omega_rmass;
-
-    cCudaData<int     , int     , x >* cu_map_array;
-    int neighbor_decide_by_integrator;
-
-    bool pinned;
-
-    void* copy_buffer;
-    int copy_buffersize;
-
-  private:
-    int pppn;                  // number of GPUs/node
-    int *devicelist;           // IDs of GPUs
-
-    std::map<class NeighList*, class CudaNeighList*> neigh_lists;
-};
-}
-
-#endif // CUDA_H
diff --git a/src/USER-CUDA/cuda_neigh_list.cpp b/src/USER-CUDA/cuda_neigh_list.cpp
index c34fb22ed4..6e05cee274 100644
--- a/src/USER-CUDA/cuda_neigh_list.cpp
+++ b/src/USER-CUDA/cuda_neigh_list.cpp
@@ -27,7 +27,7 @@
 #include <vector>
 #include <map>
 #include <algorithm>
-#include "cuda.h"
+#include "user_cuda.h"
 #include "atom.h"
 #include "error.h"
 
diff --git a/src/USER-CUDA/domain_cuda.cpp b/src/USER-CUDA/domain_cuda.cpp
index d250b5e9d3..43b1dc24a3 100644
--- a/src/USER-CUDA/domain_cuda.cpp
+++ b/src/USER-CUDA/domain_cuda.cpp
@@ -34,7 +34,7 @@
 #include "memory.h"
 #include "error.h"
 
-#include "cuda.h"
+#include "user_cuda.h"
 #include "domain_cu.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/fix_addforce_cuda.cpp b/src/USER-CUDA/fix_addforce_cuda.cpp
index 9503e1d828..d064f0d7b2 100644
--- a/src/USER-CUDA/fix_addforce_cuda.cpp
+++ b/src/USER-CUDA/fix_addforce_cuda.cpp
@@ -32,7 +32,7 @@
 #include "error.h"
 #include "force.h"
 #include "domain.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "memory.h"
 #include "cuda_modify_flags.h"
 
diff --git a/src/USER-CUDA/fix_aveforce_cuda.cpp b/src/USER-CUDA/fix_aveforce_cuda.cpp
index b6b9cb705a..326037404f 100644
--- a/src/USER-CUDA/fix_aveforce_cuda.cpp
+++ b/src/USER-CUDA/fix_aveforce_cuda.cpp
@@ -31,7 +31,7 @@
 #include "update.h"
 #include "respa.h"
 #include "domain.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 #include "variable.h"
 #include "input.h"
diff --git a/src/USER-CUDA/fix_enforce2d_cuda.cpp b/src/USER-CUDA/fix_enforce2d_cuda.cpp
index 8e482c1478..d10edf1cce 100644
--- a/src/USER-CUDA/fix_enforce2d_cuda.cpp
+++ b/src/USER-CUDA/fix_enforce2d_cuda.cpp
@@ -42,7 +42,7 @@
 #include "domain.h"
 #include "respa.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/fix_gravity_cuda.cpp b/src/USER-CUDA/fix_gravity_cuda.cpp
index a7234512d3..34107ed593 100644
--- a/src/USER-CUDA/fix_gravity_cuda.cpp
+++ b/src/USER-CUDA/fix_gravity_cuda.cpp
@@ -30,7 +30,7 @@
 #include "update.h"
 #include "domain.h"
 #include "respa.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 #include "math_const.h"
 #include "error.h"
diff --git a/src/USER-CUDA/fix_nh_cuda.cpp b/src/USER-CUDA/fix_nh_cuda.cpp
index c0beac2182..4ecf317310 100644
--- a/src/USER-CUDA/fix_nh_cuda.cpp
+++ b/src/USER-CUDA/fix_nh_cuda.cpp
@@ -32,7 +32,7 @@
 #include "memory.h"
 #include "error.h"
 #include "math_extra.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "fix_nh_cuda_cu.h"
 #include "cuda_modify_flags.h"
 
diff --git a/src/USER-CUDA/fix_nve_cuda.cpp b/src/USER-CUDA/fix_nve_cuda.cpp
index 992bb5d5d9..fbe85b176f 100644
--- a/src/USER-CUDA/fix_nve_cuda.cpp
+++ b/src/USER-CUDA/fix_nve_cuda.cpp
@@ -43,7 +43,7 @@
 #include "update.h"
 #include "respa.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/fix_set_force_cuda.cpp b/src/USER-CUDA/fix_set_force_cuda.cpp
index afaf5c7e88..e15fb44c03 100644
--- a/src/USER-CUDA/fix_set_force_cuda.cpp
+++ b/src/USER-CUDA/fix_set_force_cuda.cpp
@@ -29,7 +29,7 @@
 #include "respa.h"
 #include "error.h"
 #include "force.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "memory.h"
 #include "cuda_modify_flags.h"
 
diff --git a/src/USER-CUDA/fix_shake_cuda.cpp b/src/USER-CUDA/fix_shake_cuda.cpp
index a8747d7b13..ca2ff81f6a 100644
--- a/src/USER-CUDA/fix_shake_cuda.cpp
+++ b/src/USER-CUDA/fix_shake_cuda.cpp
@@ -32,7 +32,7 @@
 #include "fix_respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 #include "math_const.h"
 
diff --git a/src/USER-CUDA/fix_temp_berendsen_cuda.cpp b/src/USER-CUDA/fix_temp_berendsen_cuda.cpp
index 1e29de8db2..ee08aa3462 100644
--- a/src/USER-CUDA/fix_temp_berendsen_cuda.cpp
+++ b/src/USER-CUDA/fix_temp_berendsen_cuda.cpp
@@ -47,7 +47,7 @@
 #include "modify.h"
 #include "compute.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/fix_temp_rescale_cuda.cpp b/src/USER-CUDA/fix_temp_rescale_cuda.cpp
index 36c75d446b..a0ebb47d12 100644
--- a/src/USER-CUDA/fix_temp_rescale_cuda.cpp
+++ b/src/USER-CUDA/fix_temp_rescale_cuda.cpp
@@ -36,7 +36,7 @@
 #include "modify.h"
 #include "compute.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/fix_temp_rescale_limit_cuda.cpp b/src/USER-CUDA/fix_temp_rescale_limit_cuda.cpp
index 3675ca0160..eb8cf8d948 100644
--- a/src/USER-CUDA/fix_temp_rescale_limit_cuda.cpp
+++ b/src/USER-CUDA/fix_temp_rescale_limit_cuda.cpp
@@ -36,7 +36,7 @@
 #include "modify.h"
 #include "compute.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "cuda_modify_flags.h"
 
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/fix_viscous_cuda.cpp b/src/USER-CUDA/fix_viscous_cuda.cpp
index a95b5453c5..42f27ec987 100644
--- a/src/USER-CUDA/fix_viscous_cuda.cpp
+++ b/src/USER-CUDA/fix_viscous_cuda.cpp
@@ -31,7 +31,7 @@
 #include "respa.h"
 #include "error.h"
 #include "cuda_modify_flags.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 using namespace FixConst;
diff --git a/src/USER-CUDA/modify_cuda.cpp b/src/USER-CUDA/modify_cuda.cpp
index 7ac2687330..82d6d92036 100644
--- a/src/USER-CUDA/modify_cuda.cpp
+++ b/src/USER-CUDA/modify_cuda.cpp
@@ -23,7 +23,7 @@
 #include "group.h"
 #include "update.h"
 #include "domain.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "memory.h"
 #include "error.h"
 
diff --git a/src/USER-CUDA/neigh_full_cuda.cpp b/src/USER-CUDA/neigh_full_cuda.cpp
index ba70497ef1..4f38b4009d 100644
--- a/src/USER-CUDA/neigh_full_cuda.cpp
+++ b/src/USER-CUDA/neigh_full_cuda.cpp
@@ -28,7 +28,7 @@
 #include "group.h"
 #include "error.h"
 #include "cuda_neigh_list.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "neighbor_cu.h"
 #include <cmath>
 using namespace LAMMPS_NS;
diff --git a/src/USER-CUDA/neighbor_cuda.cpp b/src/USER-CUDA/neighbor_cuda.cpp
index 90d015eeba..eb46a44b3c 100644
--- a/src/USER-CUDA/neighbor_cuda.cpp
+++ b/src/USER-CUDA/neighbor_cuda.cpp
@@ -13,7 +13,7 @@
 
 #include "math.h"
 #include "neighbor_cuda.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include "atom.h"
 #include "atom_vec.h"
 #include "domain.h"
diff --git a/src/USER-CUDA/pair_buck_coul_cut_cuda.cpp b/src/USER-CUDA/pair_buck_coul_cut_cuda.cpp
index 00698ed81d..67d6ea8488 100644
--- a/src/USER-CUDA/pair_buck_coul_cut_cuda.cpp
+++ b/src/USER-CUDA/pair_buck_coul_cut_cuda.cpp
@@ -42,7 +42,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_buck_cuda.cpp b/src/USER-CUDA/pair_buck_cuda.cpp
index 6443f92d2f..d341e57259 100644
--- a/src/USER-CUDA/pair_buck_cuda.cpp
+++ b/src/USER-CUDA/pair_buck_cuda.cpp
@@ -42,7 +42,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj96_cut_cuda.cpp b/src/USER-CUDA/pair_lj96_cut_cuda.cpp
index 58d866528a..af362a3b08 100644
--- a/src/USER-CUDA/pair_lj96_cut_cuda.cpp
+++ b/src/USER-CUDA/pair_lj96_cut_cuda.cpp
@@ -57,7 +57,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_cut_coul_cut_cuda.cpp b/src/USER-CUDA/pair_lj_cut_coul_cut_cuda.cpp
index a702344f08..1abed27cb2 100644
--- a/src/USER-CUDA/pair_lj_cut_coul_cut_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_cut_coul_cut_cuda.cpp
@@ -42,7 +42,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_cut_coul_debye_cuda.cpp b/src/USER-CUDA/pair_lj_cut_coul_debye_cuda.cpp
index 0ed229fff4..130b450fa8 100644
--- a/src/USER-CUDA/pair_lj_cut_coul_debye_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_cut_coul_debye_cuda.cpp
@@ -42,7 +42,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_cut_cuda.cpp b/src/USER-CUDA/pair_lj_cut_cuda.cpp
index 7f35db0918..68502c84a3 100644
--- a/src/USER-CUDA/pair_lj_cut_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_cut_cuda.cpp
@@ -57,7 +57,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_cut_experimental_cuda.cpp b/src/USER-CUDA/pair_lj_cut_experimental_cuda.cpp
index ffc9533b45..952eeefe14 100644
--- a/src/USER-CUDA/pair_lj_cut_experimental_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_cut_experimental_cuda.cpp
@@ -57,7 +57,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_expand_cuda.cpp b/src/USER-CUDA/pair_lj_expand_cuda.cpp
index 4420fb7c58..df878f0039 100644
--- a/src/USER-CUDA/pair_lj_expand_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_expand_cuda.cpp
@@ -57,7 +57,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_gromacs_coul_gromacs_cuda.cpp b/src/USER-CUDA/pair_lj_gromacs_coul_gromacs_cuda.cpp
index 13b53fb0b4..890455c0a5 100644
--- a/src/USER-CUDA/pair_lj_gromacs_coul_gromacs_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_gromacs_coul_gromacs_cuda.cpp
@@ -42,7 +42,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_gromacs_cuda.cpp b/src/USER-CUDA/pair_lj_gromacs_cuda.cpp
index f50fccf290..5c12e0bdbd 100644
--- a/src/USER-CUDA/pair_lj_gromacs_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_gromacs_cuda.cpp
@@ -42,7 +42,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_lj_smooth_cuda.cpp b/src/USER-CUDA/pair_lj_smooth_cuda.cpp
index ef4750b389..2709b47095 100644
--- a/src/USER-CUDA/pair_lj_smooth_cuda.cpp
+++ b/src/USER-CUDA/pair_lj_smooth_cuda.cpp
@@ -42,7 +42,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/pair_morse_cuda.cpp b/src/USER-CUDA/pair_morse_cuda.cpp
index ec5fabd5e8..b5d1bb11de 100644
--- a/src/USER-CUDA/pair_morse_cuda.cpp
+++ b/src/USER-CUDA/pair_morse_cuda.cpp
@@ -57,7 +57,7 @@
 #include "respa.h"
 #include "memory.h"
 #include "error.h"
-#include "cuda.h"
+#include "user_cuda.h"
 
 using namespace LAMMPS_NS;
 
diff --git a/src/USER-CUDA/verlet_cuda.cpp b/src/USER-CUDA/verlet_cuda.cpp
index d5ba68b7ec..d9fbf710b9 100644
--- a/src/USER-CUDA/verlet_cuda.cpp
+++ b/src/USER-CUDA/verlet_cuda.cpp
@@ -49,7 +49,7 @@
 #include "cuda_wrapper_cu.h"
 #include "thermo.h"
 #include "cuda_pair_cu.h"
-#include "cuda.h"
+#include "user_cuda.h"
 #include <ctime>
 #include <cmath>
 #ifdef _OPENMP
diff --git a/src/accelerator_cuda.h b/src/accelerator_cuda.h
index 19639a6932..a4d9f6bdfd 100644
--- a/src/accelerator_cuda.h
+++ b/src/accelerator_cuda.h
@@ -19,7 +19,7 @@
 
 #ifdef LMP_USER_CUDA
 
-#include "cuda.h"
+#include "user_cuda.h"
 #include "comm_cuda.h"
 #include "domain_cuda.h"
 #include "neighbor_cuda.h"