Hello,
If you do not have hardware with a high enough hardware revision, you'll be
able to compile pmemd.cuda but not run it. Furthermore, SPSP is not precise
enough to yield production runs (it falls apart after only a few steps I've
been told).
I think your link problems may be arising because your g++/gcc is too new to
be used with the cuda code (gcc 4.4 does not work AFAIK). On my linux box
that has a GTS 250 (which can NOT run pmemd.cuda), I can install pmemd.cuda,
but it quits when I try running saying no DP cards exist. However, the
build will only work if I use my gcc-4.2, not my gcc-4.4 (and the associated
gfortran).
All the best,
Jason
On Wed, May 5, 2010 at 10:02 AM, Alan <alanwilter.gmail.com> wrote:
> Hi Jason,
>
> Well, I tried pmemd.cuda on macbookpro. First I know nvida 9600GT is not
> supposed to be supported and this GPU is hardware revision 1.0 (so no
> double
> precision). Besides, nVidia does a hell for Mac by not providing all
> drivers
> in 64 bits.
>
> Knowing all that, even though, I followed your tips and definitely I have
> to
> use configure.advanced. I created a config_data/mac_32_cuda_SPSP.gfortran,
> did
>
> ./configure.advanced mac_32_cuda gfortran nopar
>
> make -f Makefile.advanced install
>
> and compilation failed only the last stage:
>
> [snip]
> a - cudpp_scan_c.o
> gfortran -m32 -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 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 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 \
> -L/usr/local/cuda/lib -L/usr/local/cuda/lib -lcufft -lcudart
> ./cuda/cuda.a -Wl
> Undefined symbols:
> "std::_Rb_tree_increment(std::_Rb_tree_node_base const*)", referenced
> from:
> std::_Rb_tree_const_iterator<std::pair<void* const, int>
> >::operator++() in cuda.a(radixsort_c.o)
> "std::ios_base::Init::Init()", referenced from:
> __static_initialization_and_destruction_0(int, int) in cuda.a(gpu.o)
> "std::ios_base::Init::~Init()", referenced from:
> __ZNSt8ios_base4InitD1Ev$non_lazy_ptr in cuda.a(gpu.o)
> (maybe you meant: __ZNSt8ios_base4InitD1Ev$non_lazy_ptr)
> "operator delete[](void*)", referenced from:
> [snip]
>
> This is not a Cuda issue, it's something related to stdc++ (I guess) but I
> don't know c/c++ much to solve this.
>
> Cheers,
>
> Alan
>
> On Tue, May 4, 2010 at 17:12, Jason Swails <jason.swails.gmail.com> wrote:
>
> > I have tried compiling pmemd.cuda on my macbook. It is difficult. You
> > essentially have to use ./configure.advanced in the $AMBERHOME/src/pmemd
> in
> > my experience. Also make sure you've built gnu 4.3 from macports, since
> I
> > don't think cuda works with 4.4 and later...
> >
> > Then, to use the nvcc for mac (which only compiles in 32-bit), you have
> to
> > add -m32 to all of the flags (cflags, cppflags, fflags, etc.).
> >
> > That's enough to get started, I suppose.
> >
>
>
>
> --
> Alan Wilter S. da Silva, D.Sc. - CCPN Research Associate
> Department of Biochemistry, University of Cambridge.
> 80 Tennis Court Road, Cambridge CB2 1GA, UK.
> >>http://www.bio.cam.ac.uk/~awd28 <http://www.bio.cam.ac.uk/%7Eawd28><<
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed May 05 2010 - 07:30:08 PDT
