Re: namd 2.5 compilation error

From: Dong Xu (d1xu_at_ucsd.edu)
Date: Mon Sep 29 2008 - 14:38:12 CDT

On Mon, Sep 29, 2008 at 11:12 AM, Axel Kohlmeyer
<akohlmey_at_cmm.chem.upenn.edu> wrote:
> On Mon, 29 Sep 2008, Dong Xu wrote:
>
> DX> On Sun, Sep 28, 2008 at 11:27 AM, Axel Kohlmeyer
> DX> <akohlmey_at_cmm.chem.upenn.edu> wrote:
> DX> > On Sat, 27 Sep 2008, Dong Xu wrote:
> DX> >
> DX> > DX> Hi,
> DX> > DX>
> DX> > DX> I was trying to compile namd 2.6 with myrinet mpich-gm and pgi
> DX> > DX> compiler. I managed to get charmm++ compile ok and passed megatest.
> DX> >
> DX> > trying to compile NAMD with PGI compilers is a "very bad idea(TM)".
> DX> > the PGI c++ compilers are very problematic. you'd be _much_ better
> DX> > off using GCC.
> DX>
> DX> Hi Alex, the problem is that the system installed mpicc mpiCC mpirun
> DX> etc. are all compiled with pgi compilers. Would it work if I compile
>
> most likely you cannot use gcc directly with that. but nothing
> would stop you to compile your own MPI (using OpenMPI by preference).

I think I will try it. The hurdle is the cluster uses myrinet and thus
mpich-gm library. I have icc/ifort installed in my home dir. Not sure
how difficult to compile mpich-gm with the intel compilers. Any
recommendations?

Dong

>
> DX> namd using gcc? The gcc version is 2.95, I'm not sure if it's too
> DX> outdated?
>
> gcc 2.95 is some unofficial redhat version and generally discouraged
> to be used for anything, not to mention _very_ old.
> when trying to compile software on such old installations, one always
> has to consider whether it is actually worth the effort. usually, the
> hardware is slow as well...
>
> DX> > DX>
> DX> > DX> When I compile namd using Linux-i686-pgCC.arch, I got the following
> DX> > DX> error. Looks like there are some issues with tcl library libtcl8.3.a.
> DX> >
> DX> > it looks like the tcl library has been compiled with a gcc that
> DX> > was newer than the gcc that the pgi compiler runtime was compiled with.
> DX> > you need to recompiled the tcl library with your pgi or better yet
> DX> > compile NAMD with gcc/g++.
> DX>
> DX> Yes, I also realized that I need to recompile tcol if non-gcc
> DX> compilers are used.
>
> no, it is not about non-gcc compilers. it is about the version of
> libc and particularly the header files that go with it that have
> to match. PGI's c/c++ compilers used to be very outdated.
>
> cheers,
> axel.
>
> DX>
> DX> Thanks!
> DX>
> DX> -DX
> DX>
> DX> > DX>
> DX> > DX> Any help will be highly appreciated.
> DX> > DX>
> DX> > DX> -DX
> DX> > DX>
> DX> > DX> charmc> ----------- Module dependency search ------------
> DX> > DX> charmc> ----------- Module dependency search complete ------------
> DX> > DX> Adding registration for module NeighborLB
> DX> > DX> Adding registration for module commlib
> DX> > DX> charmc: Executing mpiCC -D__CHARMC__=1 -c moduleinit14691.C -o moduleinit14691.o
> DX> > DX> charmc: Executing mpiCC -L/home/dong/namd_lib/lib -L/home/dong/tcl/lib
> DX> > DX> -L/home/dong/namd_lib/lib -L/home/dong/fftm
> DX> > DX> C++ prelinker: executing: /usr/pgi/linux86/bin/pgCC
> DX> > DX> -I/home/dong/NAMD_2.6_Source/charm-5.9/mpi-linux/include -DCMC
> DX> > DX> "/usr/include/bits/in.h", line 69: warning: declaration of a member with the
> DX> > DX> same name as its class
> DX> > DX> char ip_opts[40]; /* Actually variable in size. */
> DX> > DX> ^
> DX> > DX>
> DX> > DX> C++ prelinker: executing: /usr/pgi/linux86/bin/pgCC
> DX> > DX> -I/home/dong/NAMD_2.6_Source/charm-5.9/mpi-linux/include -DCMC
> DX> > DX> "/usr/include/bits/in.h", line 69: warning: declaration of a member with the
> DX> > DX> same name as its class
> DX> > DX> char ip_opts[40]; /* Actually variable in size. */
> DX> > DX> ^
> DX> > DX>
> DX> > DX> /home/dong/namd_lib/lib/libtcl8.3.a(tclUnixPipe.o): In function
> DX> > DX> `TclpCreateTempFile':
> DX> > DX> tclUnixPipe.o(.text+0xe2): the use of `tmpnam' is dangerous, better
> DX> > DX> use `mkstemp'
> DX> > DX> /home/dong/namd_lib/lib/libtcl8.3.a(tclBinary.o): In function `.L174':
> DX> > DX> tclBinary.o(.text+0xa4f): undefined reference to `__ctype_b_loc'
> DX> > DX> tclBinary.o(.text+0xad2): undefined reference to `__ctype_b_loc'
> DX> > DX> /home/dong/namd_lib/lib/libtcl8.3.a(tclBinary.o): In function `GetFormatSpec':
> DX> > DX> tclBinary.o(.text+0x1377): undefined reference to `__ctype_b_loc'
> DX> > DX> /home/dong/namd_lib/lib/libtcl8.3.a(tclCmdAH.o): In function `Tcl_CaseObjCmd':
> DX> > DX> tclCmdAH.o(.text+0x16d): undefined reference to `__ctype_b_loc'
> DX> > DX> /home/dong/namd_lib/lib/libtcl8.3.a(tclCmdAH.o): In function `Tcl_FormatObjCmd':
> DX> > DX> tclCmdAH.o(.text+0x233c): undefined reference to `__ctype_b_loc'
> DX> > DX> /home/dong/namd_lib/lib/libtcl8.3.a(tclCmdAH.o)(.text+0x2427): more
> DX> > DX> undefined references to `__ctype_b_loc' follow
> DX> > DX> Fatal Error by charmc in directory /home/dong/NAMD_2.6_Source/Linux-i686-pgCC
> DX> > DX> Command mpiCC -L/home/dong/namd_lib/lib -L/home/dong/tcl/lib
> DX> > DX> -L/home/dong/namd_lib/lib -L/home/dong/fftw/lib -2
> DX> > DX> charmc exiting...
> DX> > DX> rm -f moduleinit14691.C moduleinit14691.o
> DX> > DX> make: *** [namd2] Error 1
> DX> > DX>
> DX> >
> DX> > --
> DX> > =======================================================================
> DX> > Axel Kohlmeyer akohlmey_at_cmm.chem.upenn.edu http://www.cmm.upenn.edu
> DX> > Center for Molecular Modeling -- University of Pennsylvania
> DX> > Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
> DX> > tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425
> DX> > =======================================================================
> DX> > If you make something idiot-proof, the universe creates a better idiot.
> DX> >
> DX>
>
> --
> =======================================================================
> Axel Kohlmeyer akohlmey_at_cmm.chem.upenn.edu http://www.cmm.upenn.edu
> Center for Molecular Modeling -- University of Pennsylvania
> Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
> tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425
> =======================================================================
> If you make something idiot-proof, the universe creates a better idiot.
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:48:22 CST