Re: Installing NAMD on a 64-bit linux (NVIDIA CUDA)

From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Mon Jan 30 2012 - 12:24:40 CST

On Mon, Jan 30, 2012 at 1:15 PM, Shyno Mathew <sm3334_at_columbia.edu> wrote:
> Hi,
> I have problems while compiling NAMD. Initially I tried installing version
> 2.8: Linux-x86_64-CUDA (NVIDIA CUDA acceleration) as my machine has NVIDIA
> with CUDA.
> Since I got errors, I tried re-installing NAMD from the source code. I
> followed the instructions for 64-bit linux workstations.
> Still I get the following errors :
> 1. During the Build and test the Charm++/Converse library (MPI version):
>   "mpirun -n 4 ./pgm"
>
> /pgm: error while loading shared libraries: libmpi_cxx.so.0: cannot open
> shared object file: No such file or directory
> ./pgm: error while loading shared libraries: libmpi_cxx.so.0: cannot open
> shared object file: No such file or directory
> ./pgm: error while loading shared libraries: libmpi_cxx.so.0: cannot open
> shared object file: No such file or directory
> ./pgm: error while loading shared libraries: libmpi_cxx.so.0: cannot open
> shared object file: No such file or directory
>
> i ignored this and moved on.

that is bad.
the error message indicates that likely the mpirun you are using is
not matching the MPI library that you are using.
the same will happen to other programs including NAMD.
you should resolve that first.

> 2. during compilation
> ./config Linux-x86_64-g++ --charm-arch mpi-linux-x86_64
> initial error: bash: ./config: /bin/csh: bad interpreter: No such file or
> directory
> #changed csh to bash
> then i got a bunch of errors!
>
> ./config: line 2: goto: command not found
> ./config: line 3: syntax:: command not found

you cannot swap csh to bash.
they have a different syntax.

> This script will create a build directory for NAMD.
>
> Usage: config [<build_dir>/]<arch>[.comment] [options]
>
> Options (most defaults are set in arch subdirectory):
>  --charm-base <Charm++ base build directory containing src>
>    (defaults to ./charm-6.3.2 or ./charm)
>  --charm-arch <subdirectory of Charm++ base build directory>
>   (such as mpi-linux-x86_64-mpicxx or multicore-linux64)
>  --charm-opts <options to pass to charmc (quote list)>
>  --with-tcl (default)  --without-tcl
>  --tcl-prefix <directory containing Tcl lib and include>
>   (defaults to ./tcl or ~/tcl)
>  --with-fftw (default)  --without-fftw
>  --fftw-prefix <directory containing FFTW 2.1.3 lib and include>
>    (defaults to ./fftw or ~/fftw)
>  --cxx <C++ compiler command>
>  --cxx-opts <C++ compiler options (quote list)>
>  --cxx-noalias-opts <C++ compiler options (quote list)>
>  --cxx-thread-opts <C++ compiler options (quote list)>
>  --cc <C compiler command>
>  --cc-opts <C compiler options (quote list)>
>  --with-debug  (sets all opts to -g)
>  --arch-suffix <suffix for NAMD platform printed at startup>
>   (the following are experimental features)
>  --without-memopt (default)  --with-memopt
>  --without-cuda (default)  --with-cuda
>   (do NOT use CUDA-enabled Charm++, NAMD does not need it)
>  --cuda-prefix <directory containing CUDA bin, lib, and include>
> ./config: line 38: syntax error near unexpected token ``uname -s`'
> ./config: line 38: `    switch ( `uname -s` )'
>
> I am not sure if I need to modify config. Any help will be really
> appreciated.

you need to follow the instructions
more closely and find a (local) expert
to explain the things to you that don't
work (they have nothing to do with NAMD).

compiling software requires at least a minimal
understanding of the tools that you are using,
and even more so, if you are ignoring or changing things.

..there ain't no escape from the blues...

axel.

> Sincerely,
> Shyno Mathew
> PhD student
> Department of Chemical Engineering
> Columbia University
>

-- 
Dr. Axel Kohlmeyer
akohlmey_at_gmail.com  http://goo.gl/1wk0
College of Science and Technology
Temple University, Philadelphia PA, USA.

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:09 CST