I am not really sure what goes wrong in your case, but find it worth
mentioning that I recently built Amber with the same setup as you did
(Ubuntu 12.04, CUDA sdk-driver-box 5.0.35) but using the Intel compiler
12.1.3 instead of GCC. This worked fine.
Cheers,
Jan-Philip
On 03/20/2013 06:56 PM, Weinzierl, Robert O J wrote:
> Dear all,
>
> I am currently trying to compile the GPU-version of pmemd.cuda on a Ubuntu 12.4 system with installed Nvidia driver (version 310.40) and CUDA Toolbox (version 5.0.35).
>
> The early stages of the installation
>
> cd $AMBERHOME
> make clean
> ./configure -cuda gnu
> make install
>
> appear to work well, but the compilation stops at a later stage with the following error messages:
>
> make[4]: Leaving directory `/home/rojw/amber12/src/pmemd/src/cuda'
> gfortran -O3 -mtune=native -DCUDA -Duse_SPFP -o pmemd.cuda gbl_constants.o gbl_datatypes.o state_info.o file_io_dat.o mdin_ctrl_dat.o mdin_ewald_dat.o mdin_debugf_dat.o prmtop_dat.o inpcrd_dat.o dynamics_dat.o img.o nbips.o parallel_dat.o parallel.o gb_parallel.o pme_direct.o pme_recip_dat.o pme_slab_recip.o pme_blk_recip.o pme_slab_fft.o pme_blk_fft.o pme_fft_dat.o fft1d.o bspline.o pme_force.o pbc.o nb_pairlist.o nb_exclusions.o cit.o dynamics.o bonds.o angles.o dihedrals.o extra_pnts_nb14.o runmd.o loadbal.o shake.o prfs.o mol_list.o runmin.o constraints.o axis_optimize.o gb_ene.o veclib.o gb_force.o timers.o pmemd_lib.o runfiles.o file_io.o bintraj.o binrestart.o pmemd_clib.o pmemd.o random.o degcnt.o erfcfun.o nmr_calls.o nmr_lib.o get_cmdline.o master_setup.o pme_alltasks_setup.o pme_setup.o ene_frc_splines.o gb_alltasks_setup.o nextprmtop_section.o angles_ub.o dihedrals_imp.o cmap.o charmm.o charmm_gold.o findmask.o remd.o multipmemd.o remd_exchg.o amd.o gbsa.o \
> ./cuda/cuda.a -L/usr/local/cuda-5.0//lib64 -L/usr/local/cuda-5.0//lib -lcurand -lcufft -lcudart -L/home/rojw/amber12/lib -L/home/rojw/amber12/lib -lnetcdf
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_startup_: error: undefined reference to 'operator new(unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_startup_: error: undefined reference to 'operator delete(void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_shutdown_: error: undefined reference to 'operator delete(void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_init_: error: undefined reference to 'operator new[](unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_init_: error: undefined reference to 'operator new[](unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_init_: error: undefined reference to 'operator delete[](void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_init_: error: undefined reference to 'operator delete[](void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_amd_setup_: error: undefined reference to 'operator new[](unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_amd_setup_: error: undefined reference to 'operator new(unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_amd_setup_: error: undefined reference to 'operator delete(void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_constraint_molecule_list_setup_: error: undefined reference to 'operator new[](unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_constraint_molecule_list_setup_: error: undefined reference to 'operator new(unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_constraint_molecule_list_setup_: error: undefined reference to 'operator new(unsigned long)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_constraint_molecule_list_setup_: error: undefined reference to 'operator delete[](void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_constraint_molecule_list_setup_: error: undefined reference to 'operator delete[](void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function gpu_constraint_molecule_list_setup_: error: undefined reference to 'operator delete(void*)'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function _GLOBAL__sub_I_gpu_startup_: error: undefined reference to 'std::ios_base::Init::Init()'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function _GLOBAL__sub_I_gpu_startup_: error: undefined reference to 'std::ios_base::Init::~Init()'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function typeinfo for GpuBuffer<double>: error: undefined reference to 'vtable for __cxxabiv1::__class_type_info'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function typeinfo for GpuBuffer<float2>: error: undefined reference to 'vtable for __cxxabiv1::__class_type_info'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function typeinfo for GpuBuffer<float>: error: undefined reference to 'vtable for __cxxabiv1::__class_type_info'
> ./cuda/cuda.a(gpu.o):gpu.cpp:function typeinfo for GpuBuffer<unsigned int>: error: undefined reference to 'vtable for __cxxabiv1::__class_type_info'
> ./cuda/cuda.a(gpu.o)(.eh_frame+0x16293): error: undefined reference to '__gxx_personality_v0'
> collect2: ld returned 1 exit status
> make[3]: *** [pmemd.cuda] Error 1
> make[3]: Leaving directory `/home/rojw/amber12/src/pmemd/src'
> make[2]: *** [cuda] Error 2
> make[2]: Leaving directory `/home/rojw/amber12/src/pmemd'
> make[1]: *** [cuda] Error 2
> make[1]: Leaving directory `/home/rojw/amber12/src'
> make: *** [install] Error 2
>
>
> I would be very grateful for any advice about what might be going wrong on at that stage. The serial versions of AMBER12 and AmberTools installed without problems, so these errors are specifically associated with compiling the GPU-based version.
>
> Robert
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Mar 20 2013 - 12:00:02 PDT
