Installing Lammps 15-Nov version in Ubuntu 12.04

Dear All:

I have been trying to install lammps 15-Nov version in Ubuntu 12.04, but it has always failed at this point. Can anyone help me with the solution.

Here is what ccmake gives

BIN2C /usr/local/cuda-7.0/bin/bin2c
BUILD_DOC OFF
BUILD_EXE ON
BUILD_LIB OFF
BUILD_MPI ON
BUILD_OMP ON
CMAKE_AR /usr/bin/ar
CMAKE_BUILD_TYPE Release
CMAKE_COLOR_MAKEFILE ON
CMAKE_CXX_COMPILER /usr/bin/c++
CMAKE_CXX_FLAGS
CMAKE_CXX_FLAGS_DEBUG -g
CMAKE_CXX_FLAGS_MINSIZEREL -Os -DNDEBUG
CMAKE_CXX_FLAGS_RELEASE -O3 -DNDEBUG
CMAKE_CXX_FLAGS_RELWITHDEBINFO -O2 -g -DNDEBUG
CMAKE_C_COMPILER /usr/bin/cc
CMAKE_C_FLAGS
CMAKE_C_FLAGS_DEBUG -g
CMAKE_C_FLAGS_MINSIZEREL -Os -DNDEBUG
CMAKE_C_FLAGS_RELEASE -O3 -DNDEBUG
CMAKE_C_FLAGS_RELWITHDEBINFO -O2 -g -DNDEBUG
CMAKE_EXE_LINKER_FLAGS
CMAKE_EXE_LINKER_FLAGS_DEBUG
CMAKE_EXE_LINKER_FLAGS_MINSIZE
CMAKE_EXE_LINKER_FLAGS_RELEASE
CMAKE_EXE_LINKER_FLAGS_RELWITH
CMAKE_EXPORT_COMPILE_COMMANDS OFF
CMAKE_INSTALL_BINDIR bin
CMAKE_INSTALL_DATADIR
CMAKE_INSTALL_DATAROOTDIR share
CMAKE_INSTALL_DOCDIR
CMAKE_INSTALL_INCLUDEDIR include
CMAKE_INSTALL_INFODIR
CMAKE_INSTALL_LIBDIR lib
CMAKE_INSTALL_LIBEXECDIR libexec
CMAKE_INSTALL_LOCALEDIR
CMAKE_INSTALL_LOCALSTATEDIR var
CMAKE_INSTALL_MANDIR
CMAKE_INSTALL_OLDINCLUDEDIR /usr/include
CMAKE_INSTALL_PREFIX /usr/local
CMAKE_INSTALL_SBINDIR sbin
CMAKE_INSTALL_SHAREDSTATEDIR com
CMAKE_INSTALL_SYSCONFDIR etc
CMAKE_LINKER /usr/bin/ld
CMAKE_MAKE_PROGRAM /usr/bin/make
CMAKE_MODULE_LINKER_FLAGS
CMAKE_MODULE_LINKER_FLAGS_DEBU
CMAKE_MODULE_LINKER_FLAGS_MINS
CMAKE_MODULE_LINKER_FLAGS_RELE
CMAKE_MODULE_LINKER_FLAGS_RELW
CMAKE_NM /usr/bin/nm
CMAKE_OBJCOPY /usr/bin/objcopy
CMAKE_OBJDUMP /usr/bin/objdump

CMAKE_RANLIB /usr/bin/ranlib CMAKE_SHARED_LINKER_FLAGS
CMAKE_SHARED_LINKER_FLAGS_DEBU
CMAKE_SHARED_LINKER_FLAGS_MINS
CMAKE_SHARED_LINKER_FLAGS_RELE
CMAKE_SHARED_LINKER_FLAGS_RELW
CMAKE_SKIP_INSTALL_RPATH OFF
CMAKE_SKIP_RPATH OFF
CMAKE_STATIC_LINKER_FLAGS
CMAKE_STATIC_LINKER_FLAGS_DEBU
CMAKE_STATIC_LINKER_FLAGS_MINS
CMAKE_STATIC_LINKER_FLAGS_RELE
CMAKE_STATIC_LINKER_FLAGS_RELW
CMAKE_STRIP /usr/bin/strip
CMAKE_VERBOSE_MAKEFILE OFF
CUDA_64_BIT_DEVICE_CODE ON
CUDA_ATTACH_VS_BUILD_RULE_TO_C ON
CUDA_BUILD_CUBIN OFF
CUDA_BUILD_EMULATION OFF
CUDA_CUDART_LIBRARY /usr/local/cuda-7.0/lib64/libcudart.so
CUDA_CUDA_LIBRARY /usr/lib/x86_64-linux-gnu/libcuda.so
CUDA_GENERATED_OUTPUT_DIR
CUDA_HOST_COMPILATION_CPP ON
CUDA_HOST_COMPILER /usr/bin/cc
CUDA_NVCC_EXECUTABLE /usr/local/cuda-7.0/bin/nvcc
CUDA_NVCC_FLAGS
CUDA_NVCC_FLAGS_DEBUG
CUDA_NVCC_FLAGS_MINSIZEREL
CUDA_NVCC_FLAGS_RELEASE
CUDA_NVCC_FLAGS_RELWITHDEBINFO
CUDA_PROPAGATE_HOST_FLAGS ON
CUDA_SDK_ROOT_DIR /home/chakraborty/NVIDIA_CUDA-7.0_Samples
CUDA_SEPARABLE_COMPILATION OFF
CUDA_TARGET_CPU_ARCH
CUDA_TOOLKIT_INCLUDE /usr/local/cuda-7.0/include
CUDA_TOOLKIT_ROOT_DIR /usr/local/cuda-7.0
CUDA_TOOLKIT_TARGET_DIR /usr/local/cuda-7.0
CUDA_USE_STATIC_CUDA_RUNTIME ON
CUDA_VERBOSE_BUILD OFF
CUDA_VERSION 7.0
CUDA_cublas_LIBRARY /usr/local/cuda-7.0/lib64/libcublas.so
CUDA_cudart_static_LIBRARY /usr/local/cuda-7.0/lib64/libcudart_static.a
CUDA_cufft_LIBRARY /usr/local/cuda-7.0/lib64/libcufft.so
CUDA_cupti_LIBRARY /usr/local/cuda-7.0/extras/CUPTI/lib64/libcupti.so
CUDA_curand_LIBRARY /usr/local/cuda-7.0/lib64/libcurand.so
CUDA_cusolver_LIBRARY /usr/local/cuda-7.0/lib64/libcusolver.so
CUDA_cusparse_LIBRARY /usr/local/cuda-7.0/lib64/libcusparse.so
CUDA_nppc_LIBRARY /usr/local/cuda-7.0/lib64/libnppc.so
CUDA_nppi_LIBRARY /usr/local/cuda-7.0/lib64/libnppi.so
CUDA_npps_LIBRARY /usr/local/cuda-7.0/lib64/libnpps.so
CUDA_rt_LIBRARY /usr/lib/x86_64-linux-gnu/librt.so
CUDPP_OPT ON
DOWNLOAD_VORO ON
ENABLE_COVERAGE OFF
ENABLE_SANITIZE_ADDRESS OFF ENABLE_SANITIZE_THREAD OFF
ENABLE_SANITIZE_UNDEFINED OFF
ENABLE_TESTING OFF
FFMPEG_EXECUTABLE /usr/bin/ffmpeg
FFT FFTW3
FFTW3_INCLUDE_DIR /usr/include
FFTW3_LIBRARY /usr/lib/x86_64-linux-gnu/libfftw3.so
FFT_PACK array
FFT_SINGLE OFF
GPU_API cuda
GPU_ARCH sm_30
GPU_PREC mixed
GZIP_EXECUTABLE /bin/gzip
HDF5_CXX_COMPILER_EXECUTABLE /usr/bin/h5c++
HDF5_C_COMPILER_EXECUTABLE /usr/bin/h5cc
HDF5_C_INCLUDE_DIR /usr/include
HDF5_DIFF_EXECUTABLE /usr/bin/h5diff
HDF5_DIR HDF5_DIR-NOTFOUND
HDF5_Fortran_COMPILER_EXECUTAB /usr/bin/h5fc
HDF5_IS_PARALLEL OFF
HDF5_dl_LIBRARY_DEBUG HDF5_dl_LIBRARY_DEBUG-NOTFOUND
HDF5_dl_LIBRARY_RELEASE /usr/lib/x86_64-linux-gnu/libdl.so
HDF5_hdf5_LIBRARY_DEBUG HDF5_hdf5_LIBRARY_DEBUG-NOTFOUND
HDF5_hdf5_LIBRARY_RELEASE /usr/lib/x86_64-linux-gnu/libhdf5.so
HDF5_m_LIBRARY_DEBUG HDF5_m_LIBRARY_DEBUG-NOTFOUND
HDF5_m_LIBRARY_RELEASE /usr/lib/x86_64-linux-gnu/libm.so
HDF5_pthread_LIBRARY_DEBUG HDF5_pthread_LIBRARY_DEBUG-NOTFOUND
HDF5_pthread_LIBRARY_RELEASE /usr/lib/x86_64-linux-gnu/libpthread.so
HDF5_z_LIBRARY_DEBUG HDF5_z_LIBRARY_DEBUG-NOTFOUND
HDF5_z_LIBRARY_RELEASE /usr/lib/x86_64-linux-gnu/libz.so
JPEG_INCLUDE_DIR /usr/include
JPEG_LIBRARY /usr/lib/x86_64-linux-gnu/libjpeg.so
LAMMPS_EXCEPTIONS OFF
LAMMPS_LONGLONG_TO_LONG OFF
LAMMPS_MACHINE
LAMMPS_MEMALIGN 64
LAMMPS_SIZES smallbig
MATH_LIBRARIES m
MPIEXEC /usr/bin/mpiexec
MPIEXEC_MAX_NUMPROCS 2
MPIEXEC_NUMPROC_FLAG -np
MPIEXEC_POSTFLAGS
MPIEXEC_PREFLAGS
MPI_CXX_COMPILER /usr/bin/mpicxx
MPI_CXX_COMPILE_FLAGS
MPI_CXX_INCLUDE_PATH /usr/lib/openmpi/include;/usr/lib/openmpi/include/openmpi
MPI_CXX_LIBRARIES /usr/lib/libmpi_cxx.so;/usr/lib/libmpi.so;/usr/lib/x86_64-linux-gnu/libdl.so;/usr/lib/x86_64-linux-gnu/libhwloc.so
MPI_CXX_LINK_FLAGS
MPI_C_COMPILER /usr/bin/mpicc
MPI_C_COMPILE_FLAGS
MPI_C_INCLUDE_PATH /usr/lib/openmpi/include;/usr/lib/openmpi/include/openmpi
MPI_C_LIBRARIES /usr/lib/libmpi.so;/usr/lib/x86_64-linux-gnu/libdl.so;/usr/lib/x86_64-linux-gnu/libhwloc.so
MPI_C_LINK_FLAGS
MPI_EXTRA_LIBRARY /usr/lib/libmpi.so;/usr/lib/x86_64-linux-gnu/libdl.so;/usr/lib/x86_64-linux-gnu/libhwloc.so

MPI_LIBRARY /usr/lib/libmpi_cxx.so
OCL_TUNE kepler
OpenCL_INCLUDE_DIR /usr/local/cuda-7.0/include
OpenCL_LIBRARY /usr/lib/x86_64-linux-gnu/libOpenCL.so
OpenMP_CXX_FLAGS -fopenmp
OpenMP_C_FLAGS -fopenmp
PKG_ASPHERE ON
PKG_BODY ON
PKG_CLASS2 ON
PKG_COLLOID ON
PKG_COMPRESS ON
PKG_CONFIG_EXECUTABLE /usr/bin/pkg-config
PKG_CORESHELL ON
PKG_DIPOLE ON
PKG_GPU ON
PKG_GRANULAR ON
PKG_KIM OFF
PKG_KOKKOS OFF
PKG_KSPACE ON
PKG_LATTE OFF
PKG_MANYBODY ON
PKG_MC OFF
PKG_MEAM OFF
PKG_MESSAGE OFF
PKG_MISC OFF
PKG_MOLECULE ON
PKG_MPIIO OFF
PKG_MSCG OFF
PKG_OPT OFF
PKG_PERI OFF
PKG_POEMS OFF
PKG_PYTHON OFF
PKG_QEQ OFF
PKG_REAX OFF
PKG_REPLICA ON
PKG_RIGID ON
PKG_SHOCK ON
PKG_SNAP ON PKG_USER-MANIFOLD OFF
PKG_USER-MEAMC OFF
PKG_USER-MESO OFF
PKG_USER-MGPT OFF
PKG_USER-MISC OFF
PKG_USER-MOFFF OFF
PKG_USER-MOLFILE ON
PKG_USER-NETCDF OFF
PKG_USER-OMP OFF
PKG_USER-PHONON OFF
PKG_USER-PLUMED OFF
PKG_USER-PTM OFF
PKG_USER-QMMM OFF
PKG_USER-QTB OFF
PKG_USER-QUIP OFF
PKG_USER-REAXC OFF
PKG_USER-SCAFACOS OFF
PKG_USER-SDPD ON
PKG_USER-SMD OFF
PKG_USER-SMTBQ OFF
PKG_USER-SPH ON
PKG_USER-TALLY OFF
PKG_USER-UEF OFF
PKG_USER-VTK OFF
PKG_VORONOI ON
PNG_LIBRARY_DEBUG PNG_LIBRARY_DEBUG-NOTFOUND
PNG_LIBRARY_RELEASE PNG_LIBRARY_RELEASE-NOTFOUND
PNG_PNG_INCLUDE_DIR PNG_PNG_INCLUDE_DIR-NOTFOUND
VORO_INCLUDE_DIR VORO_INCLUDE_DIR-NOTFOUND
VORO_LIBRARY VORO_LIBRARY-NOTFOUND
WITH_FFMPEG ON
WITH_GZIP ON
WITH_JPEG ON
WITH_PNG OFF
ZLIB_INCLUDE_DIR /usr/include
ZLIB_LIBRARY_DEBUG ZLIB_LIBRARY_DEBUG-NOTFOUND
ZLIB_LIBRARY_RELEASE /usr/lib/x86_64-linux-gnu/libz.so

This is the out put of the error. I have tried the stable as well as the unstable version. I have tried cuda-7 as well as cuda 8. The cards are K20M and K40. cmake version is 3.4 and gcc 4.8

9%] Building NVCC (Device) object CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_scan_app.cu.o
nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified
CMake Error at cuda_compile_generated_radixsort_app.cu.o.cmake:207 (message):
Error generating
/home/chakraborty/lammps-15Nov18/build/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o

make[2]: *** [CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o] Error 1
make[2]: *** Waiting for unfinished jobs…
nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified
CMake Error at cuda_compile_generated_scan_app.cu.o.cmake:207 (message):
Error generating
/home/chakraborty/lammps-15Nov18/build/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini/./cuda_compile_generated_scan_app.cu.o

make[2]: *** [CMakeFiles/cuda_compile.dir/__/lib/gpu/cudpp_mini/cuda_compile_generated_scan_app.cu.o] Error 1
make[1]: *** [CMakeFiles/gpu.dir/all] Error 2
make: *** [all] Error 2

Thanks a lot in advance.

Best

Dipanjan

Dear All:

I have been trying to install lammps 15-Nov version in Ubuntu 12.04, but it has always failed at this point. Can anyone help me with the solution.

i can only give you some generic recommendations, that might make it easier to locate the problem. for that purpose, please simplify your setup and try a compilation with only the GPU package enabled.

cd /path/to/lammps/
mkdir build-gpu
cd build-gpu
cmake …/cmake -DPKG_GPU=1 -DGPU_API=cuda
make

if make still fails, run the following command and report the part showing the failing command back here (the error message you quoted is not much help unless we know the exact command line that make issued).

make VERBOSE=1

axel.

Dear Axel,

Thank you very much for the reply. The error still persists and with the verbose command gave the following output:

/usr/local/bin/cmake -H/home/chakraborty/lammps-22Aug18/cmake -B/home/chakraborty/lammps-22Aug18/build-cuda --check-build-system CMakeFiles/Makefile.cmake 0
/usr/local/bin/cmake -E cmake_progress_start /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/progress.marks
make -f CMakeFiles/Makefile2 all
make[1]: Entering directory /home/chakraborty/lammps-22Aug18/build-cuda' make -f CMakeFiles/gpu.dir/build.make CMakeFiles/gpu.dir/depend make[2]: Entering directory /home/chakraborty/lammps-22Aug18/build-cuda’
[ 0%] Building NVCC (Device) object CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o
cd /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini && /usr/local/bin/cmake -E make_directory /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/.
cd /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini && /usr/local/bin/cmake -D verbose:BOOL=1 -D build_configuration:STRING=Release -D generated_ -D generated_cubin_ -P /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o.cmake – Removing /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o /usr/local/bin/cmake -E remove /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o – Generating dependency file: /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o.NVCC-depend /usr/local/cuda-7.0/bin/nvcc -M -D__CUDACC__ /home/chakraborty/lammps-22Aug18/lib/gpu/cudpp_mini/radixsort_app.cu -o /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o.NVCC-depend -m64 -DLAMMPS_SMALLBIG -DLAMMPS_MEMALIGN=64 -DLAMMPS_JPEG -DLAMMPS_GZIP -DLAMMPS_FFMPEG -Xcompiler ,"-fopenmp","-O3","-DNDEBUG" <BOOL::-Xcompiler=-fPIC> -DUNIX -O3 -Xptxas -v --use_fast_math -DUCL_CUDADR -arch=sm_30 -D_SINGLE_DOUBLE -DNVCC -I/home/chakraborty/lammps-22Aug18/lib/gpu -I/home/chakraborty/lammps-22Aug18/build-cuda/lib/gpu -I/home/chakraborty/lammps-22Aug18/lib/gpu/cudpp_mini -I/usr/local/cuda-7.0/include -I/usr/lib/openmpi/include -I/usr/lib/openmpi/include/openmpi -I/usr/include nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified CMake Error at cuda_compile_generated_radixsort_app.cu.o.cmake:207 (message): Error generating /home/chakraborty/lammps-22Aug18/build-cuda/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o make[2]: *** [CMakeFiles/cuda_compile.dir/__/lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o] Error 1 make[2]: Leaving directory /home/chakraborty/lammps-22Aug18/build-cuda' make[1]: *** [CMakeFiles/gpu.dir/all] Error 2 make[1]: Leaving directory /home/chakraborty/lammps-22Aug18/build-cuda’ make: *** [all] Error 2

Hope this helps in figuring out the issue.

Once again thanks a lot for your support.

Best

Dipanjan

Dear Axel,

Thank you very much for the reply. The error still persists and with the verbose command gave the following output:

/usr/local/bin/cmake -H/home/chakraborty/lammps-22Aug18/cmake -B/home/chakraborty/lammps-22Aug18/build-cuda --check-build-system CMakeFiles/Makefile.cmake 0

this suggests that you are trying to compile LAMMPS version 22 August 2018 and NOT 15 November 2018 as you are claiming. there is a BIG difference there. the 22 August 2018 version has several known flaws in the cmake build environment (no surprise, it was just added). in the 15 November 2018 version all known cmake related issues have been addressed. i would also recommend to use the latest available CUDA version for your OS, we have limited options to test on old Linux/CUDA combinations (and perhaps also consider updating your out-of-support Linux OS).

axel.

Dear Axel,
Terribly sorry I forgot about the version of lammps. I have tried the same with 15-Nov version. It gives the same error. Here is the output of the make VERBOSE=1 command.

chakraborty@…8142…:~/lammps-15Nov18/build-gpu$ make VERBOSE=1
/usr/local/bin/cmake -H/home/chakraborty/lammps-15Nov18/cmake -B/home/chakraborty/lammps-15Nov18/build-gpu --check-build-system CMakeFiles/Makefile.cmake 0
Re-run cmake file: Makefile older than: CMakeFiles/cuda_compile_cubin.dir/gpu/cuda_compile_cubin_generated_lal_pppm_d.cu.cubin.depend
– Running check for auto-generated files from make-based build system
– Generating style_angle.h…
– Generating style_atom.h…
– Generating style_body.h…
– Generating style_bond.h…
– Generating style_command.h…
– Generating style_compute.h…
– Generating style_dihedral.h…
– Generating style_dump.h…
– Generating style_fix.h…
– Generating style_improper.h…
– Generating style_integrate.h…
– Generating style_kspace.h…
– Generating style_minimize.h…
– Generating style_nbin.h…
– Generating style_npair.h…
– Generating style_nstencil.h…
– Generating style_ntopo.h…
– Generating style_pair.h…
– Generating style_reader.h…
– Generating style_region.h…
– Generating lmpinstalledpkgs.h…
– Building package: GPU
– The following packages have been found:

  • MPI
  • OpenMP
  • JPEG
  • Threads
  • CUDA

– <<< Build configuration >>>
Build type Release
Install path /usr/local
Compilers and Flags:
C++ Compiler /usr/bin/c++
Type GNU
Version 4.8.4
C++ Flags -fopenmp -O3 -DNDEBUG
Defines LAMMPS_SMALLBIG;LAMMPS_MEMALIGN=64;LAMMPS_JPEG;LAMMPS_GZIP;LAMMPS_FFMPEG

– Static libraries
– Link libraries: /usr/lib/libmpi_cxx.so;/usr/lib/libmpi.so;/usr/lib/x86_64-linux-gnu/libdl.so;/usr/lib/x86_64-linux-gnu/libhwloc.so;/usr/lib/x86_64-linux-gnu/libjpeg.so;m;gpu
– Using mpi with headers in /usr/lib/openmpi/include;/usr/lib/openmpi/include/openmpi and /usr/lib/libmpi_cxx.so;/usr/lib/libmpi.so;/usr/lib/x86_64-linux-gnu/libdl.so;/usr/lib/x86_64-linux-gnu/libhwloc.so
– GPU Api: CUDA
– GPU Arch: sm_30
– GPU Precision: MIXED
– Configuring done
– Generating done
– Build files have been written to: /home/chakraborty/lammps-15Nov18/build-gpu
/usr/local/bin/cmake -E cmake_progress_start /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/progress.marks
make -f CMakeFiles/Makefile2 all
make[1]: Entering directory /home/chakraborty/lammps-15Nov18/build-gpu' make -f CMakeFiles/gpu.dir/build.make CMakeFiles/gpu.dir/depend make[2]: Entering directory /home/chakraborty/lammps-15Nov18/build-gpu’
[ 0%] Building NVCC (Device) object CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o
cd /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini && /usr/local/bin/cmake -E make_directory /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/.
cd /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini && /usr/local/bin/cmake -D verbose:BOOL=1 -D build_configuration:STRING=Release -D generated_file:STRING=/home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o -D generated_cubin_file:STRING=/home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o.cubin.txt -P /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o.cmake
– Removing /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o
/usr/local/bin/cmake -E remove /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o
– Generating dependency file: /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o.NVCC-depend
/usr/local/cuda-7.0/bin/nvcc -M -D__CUDACC__ /home/chakraborty/lammps-15Nov18/lib/gpu/cudpp_mini/radixsort_app.cu -o /home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir//lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o.NVCC-depend -m64 -DLAMMPS_SMALLBIG -DLAMMPS_MEMALIGN=64 -DLAMMPS_JPEG -DLAMMPS_GZIP -DLAMMPS_FFMPEG -Xcompiler ,"-fopenmp","-O3","-DNDEBUG" <BOOL::-Xcompiler=-fPIC> -DUNIX -O3 -Xptxas -v --use_fast_math -DUCL_CUDADR -arch=sm_30 -D_SINGLE_DOUBLE -DNVCC -I/home/chakraborty/lammps-15Nov18/lib/gpu -I/home/chakraborty/lammps-15Nov18/build-gpu/lib/gpu -I/home/chakraborty/lammps-15Nov18/lib/gpu/cudpp_mini -I/usr/local/cuda-7.0/include -I/home/chakraborty/lammps-15Nov18/src -I/usr/lib/openmpi/include -I/usr/lib/openmpi/include/openmpi -I/usr/include
nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified
CMake Error at cuda_compile_generated_radixsort_app.cu.o.cmake:207 (message):
Error generating
/home/chakraborty/lammps-15Nov18/build-gpu/CMakeFiles/cuda_compile.dir/
/lib/gpu/cudpp_mini/./cuda_compile_generated_radixsort_app.cu.o

make[2]: *** [CMakeFiles/cuda_compile.dir/__/lib/gpu/cudpp_mini/cuda_compile_generated_radixsort_app.cu.o] Error 1
make[2]: Leaving directory /home/chakraborty/lammps-15Nov18/build-gpu' make[1]: *** [CMakeFiles/gpu.dir/all] Error 2 make[1]: Leaving directory /home/chakraborty/lammps-15Nov18/build-gpu’
make: *** [all] Error 2

There are some softwares which I require that is tied to the old os. Unfortunately I do not have the budget to make two separate systems — one with the older softwares and the other with the new OS. Any help in this matter would be greatly appreciated.
Thank you in advance.
Best
Dipanjan

Dipanjan Chakraborty
Assistant Professor
Department of Physics
Indian Institute of Science Education and Research (IISER) Mohali.
Manauli-140306
Punjab

please try the following experiment:

edit the file cmake/CMakeLists.txt

on line 1474 change:

<BOOL:${BUILD_SHARED_LIBS}:-Xcompiler=-fPIC>

into

-Xcompiler=-fPIC

and then try compiling LAMMPS again with: make VERBOSE=1
and report the output, if it fails again.

thanks,
axel.

… and please also send me the output of:

grep SHARED CMakeCache.txt

thanks,
axel.

please download this snapshot https://github.com/akohlmey/lammps/archive/cmake-gpu-enhancements.tar.gz
and try building LAMMPS from it. this includes a workaround for your CMake issue and in addition allows to build one LAMMPS executable that supports multiple GPU architectures (perhaps even at the same time, but i would recommend against it).

your compilation failure seems to be caused by a bug in your version of CMake, which creates an incorrect makefile. we can work around it with acceptable effort.

as noted before, you have a very old (and unsupported!) OS and generally and old software stack, so please consider upgrading (at least cmake ;-))

axel.

FYI, we have just a few minutes ago posted a new patch release of LAMMPS on github (availability on lammps.sandia.gov may take until tomorrow) with improved GPU package support in CMake (and more) and improved compatibility with different CMake versions.

axel.