[gmx-users] Installation - can I skip auto-vectorization?

[Nathan Moore]

Thanks for the clarification on the install problem.  I'm now dealing with
another issue.  The code dies half-way through make on the
auto-vectorization step.  I'm cross-compiling (on one PowerPC machine, for
a different PowerPC architecture).  The output is,

...
>>> This is the GROMACS code generator for MD & MC inner loops
>>> It will generate single precision C code in file innerc.c
>>> Using gromacs invsqrt code
>>> Inlining gromacs invsqrt code
>>> Using normal solvent optimized loops
>>> Prefetching forces in loops: none
>>> Nonthreaded inner loops
>>> Vectorizing invsqrt in loops:none
Error creating file innerc.c
make[3]: *** [innerc.c] Error 255
make[3]: Leaving directory
`/bglhome/usr2/ntmoore/GROMACS/gromacs-3.2/src/gmxlib'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/bglhome/usr2/ntmoore/GROMACS/gromacs-3.2/src'
make[1]: *** [all] Error 2
make[1]: Leaving directory `/bglhome/usr2/ntmoore/GROMACS/gromacs-3.2/src'
make: *** [all-recursive] Error 1

Is there a special config flag that I need to turn on to disable this
auto-vectorization?  Can I skip this section by inputting specifics of the
chip (ie CFLAGS="-qarch=440d -qtune=440 -O4",  strangely, including those
flags in the following way causes configure to fail:

./configure \
        --enable-mpi \
        F77='xlf' \
        CC='cc' \
        MPICC='cc' \
        CFLAGS="-I/sys/include" \
        CPPFLAGS="-qarch=440d -qtune=440 -O4
-I/home/usr/ntmoore/lib/fftw-mpi-single/include" \
        LDFLAGS="-L/sys/lib -L/home/usr/ntmoore/lib/fftw-mpi-single/lib" \
        LIBS=" -lmpich.rts -lmsglayer.rts -lrts.rts -ldevices.rts -lfftw
-lfftw_mpi -lrfftw -lrfftw_mpi" \
        --build=ppc-linux-gnu \
        --target=ppc-bgl-blrtsgnu \
        --host=powerpc

)?

Thanks again - I do appreciate the help!

Nathan Moore
University of Minnesota Physics



> On Tue, 2005-05-31 at 16:58 -0500, Nathan Moore wrote:
>> Greetings,
>>
>> I'm working to install GROMACS on a parallel machine.  I have MPIch
>> available, as well as the FFTW library.  Several questions have arisen:
>>
>> (1) What is the difference between the value of "LIBS" and "LDFLAGS" in
>> the configure script?  Is, for example, the following ./configure
>> argument
>> acceptable:
>> LDFLAGS="-I/home/usr/ntmoore/lib/fftw-mpi-single/include -L/sys/lib
>> -lmpich.rts -lfftw_mpi"
>
> CPPFLAGS="-I/home/usr/ntmoore/lib/fftw-mpi-single/include"
> LDFLAGS="-L/sys/lib -L/home/usr/ntmoore/lib/fftw-mpi-single/lib"
> LIBS="-lmpich.rts"
> CC=mpicc
>
>>
>> (2) I noticed after compiling fftw that several variations of the
>> library
>> were created.  Which of the available set, {libfftw libfftw_mpi librfftw
>> librfftw_mpi}, should I include?
> configure will find the right one if you give it the right directory
> (see above)
>>
>> (3) Is it possible to replace fftw with a vendor-supplied Lapack fft
>> package?  (essl/cxml/sgimath/sunperf etc)?
> Not yet.
>>
>> (4) any example config scripts from an MPI machine would be greatly
>> appreciated!
>>
>> Best Regards,
>>
>> Nathan Moore
>> University of Minnesota Physics
>>
>>
>> _______________________________________________
>> gmx-users mailing list
>> gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-request at gromacs.org.
> --
> David.
> ________________________________________________________________________
> David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
> Dept. of Cell and Molecular Biology, Uppsala University.
> Husargatan 3, Box 596,          75124 Uppsala, Sweden
> phone:  46 18 471 4205          fax: 46 18 511 755
> spoel at xray.bmc.uu.se    spoel at gromacs.org   http://xray.bmc.uu.se/~spoel
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
>