On Sat, Oct 18, 2014 at 12:27 AM, Andreas Goetz <agoetz.sdsc.edu> wrote:
> >> -nofftw3 is there because fftw3 requires a minimum of GNU 4.3 and Intel
> >> 9. It used to be that the minimum compiler version we supported was GNU
> >> 4.1.2. I’d be happy to get rid of this restriction.
> >
> > I think we should assume compilers able to handle fftw3. I also note
> that
> > pmemd uses -DPUBFFT even though fftw3 has been configured and is
> available,
> > say with gnu compilers chosen. Does anyone know why this is?
>
> To add to this discussion, the current master fails to configure fftw3 if
> Intel MPI is used. I am not familiar with fftw3 so have not spent time
> trying to fix this. The -nofftw3 flag, though, came in handy.
>
> $> ./configure -intelmpi intel
> Checking for updates...
> Checking for available patches online. This may take a few seconds...
>
> Available AmberTools 15 patches:
>
> No patches available
>
> Available Amber 15 patches:
>
> No patches available
> Searching for python2... Found python2.7: /usr/local/bin/python2.7
> Detected icc version 14
> Detected ifort version 14
> Using Intel MKL libraries in
> /server-home3/netbin/intel2013/composer_xe_2013_sp1.0.080/mkl/lib/intel64
> MKL Version 10 or greater assumed.
>
> Testing the icc compiler:
> icc -fpic -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -o testp testp.c
> OK
>
> Testing the icpc compiler:
> icpc -fpic -o testp testp.cpp
> OK
>
> Testing the ifort compiler:
> ifort -fpic -O0 -o testp testp.f
> OK
>
> Testing mixed C/Fortran compilation:
> icc -fpic -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -c -o testp.c.o
> testp.c
> ifort -fpic -O0 -c -o testp.f.o testp.f
> icc -fpic -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -o testp
> testp.c.o testp.f.o -lifport -lifcore -lsvml
> OK
>
> Testing pointer size:
> icc -fpic -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -o
> test_pointer_size test_pointer_size.c
> Detected 64 bit operating system.
>
> Testing flex: OK
>
> Checking NetCDF...
> Using bundled NetCDF library.
> Using existing NetCDF in
> '/server-home1/agoetz/sources/AMBER/amber.intel'
>
> Checking for zlib: OK
>
> Checking for libbz2: OK
>
> Configuring MKL FFTW for PMEMD...
>
> MKL_HOME set to
> /server-home3/netbin/intel2013/composer_xe_2013_sp1.0.080/mkl
>
> Configuring fftw-3.3 (may be time-consuming)...
>
> Error: FFTW configure returned 1
> FFTW configure failed! Check the fftw3_config.log file
> in the /server-home1/agoetz/sources/AMBER/amber.intel/AmberTools/src
> directory.
> Configure failed due to the errors above!
>
> $> tail AmberTools/src/fftw3_config.log
> checking whether stripping libraries is possible... yes
> checking if libtool supports shared libraries... yes
> checking whether to build shared libraries... yes
> checking whether to build static libraries... yes
> checking for ocamlbuild... no
> checking for mpicc... mpicc
>
Uh oh... this isn't good. Intel MPI should be pulling mpiicc and mpiifort
as the compilers. I didn't see any option to specify this (maybe with the
CC and FC vars?) to the fftw-3.3 configure script, so I'm not sure if this
will work.
Perri already tagged this as bug 286 on bugzilla; I'll add you to the CC
list there:
http://bugzilla.ambermd.org/show_bug.cgi?id=286
I don't know of anyone except you guys at SDSC that actually has access to
the Intel MPI, so testing changes will most likely fall to you...
All the best,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Sat Oct 18 2014 - 10:30:02 PDT
