[gmx-users] Benching against different compilers - Help needed

[David van der Spoel]

Jones de Andrade wrote:
> Hi all.
> 
> Well, some time ago I decided to bench gromacs on AMD64 machines
> compiled with different compilers (both single and double precision,
> both internal and external blas and lapack, just FFTw 3 for fourier)
> in order to see which one performs better.
> 
> To time, I have available GCC, PGI and Intel compilers. Pathscale is a
> future option.
> 
> Unfortunatelly, I was able to properly compile it just with GCC (all 4
> flavours, despite the fact that the external libraries of ACML didn't
> seem to improve so much the performance as I expected), but failed on
> both PGI and Intel compilers at different points. So, I'm asking for
> help on those compilation issues.
> 
> For Intel, the compilation ends up with a segmentation fault at the
> following point:
> 
> ***********************************************************************************
> ./mknb   -software_invsqrt
>  >>> Gromacs nonbonded kernel generator (-h for help)
>  >>> Generating single precision functions in C.
>  >>> Using Gromacs software version of 1/sqrt(x).
>  make[5]: *** [kernel-stamp] Falha de segmentação
>  make[5]: Leaving directory
> `/home/johannes/src/gromacs/gromacs-3.3/src/gmxlib/nonbonded/nb_kernel'
>  make[4]: ** [all-recursive] Erro 1
> ***********************************************************************************
> 
> And, for PGI, it dies at the following point (after using the
> --without-x --without-motif-libraries -- without-motif-includes flags
> for configure):
> 
> ****************************************************************************************
> source='gmx_sgangle.c' object='gmx_sgangle.lo' libtool=yes \
>  DEPDIR=.deps depmode=none /bin/sh ../../config/depcomp \
>  /bin/sh ../../libtool --mode=compile --tag=CC
> /usr/local/pgi/linux86-64/6.0/bin/pgcc -DHAVE_CONFIG_H -I. -I.
> -I../../src  -I../../include
> -DGMXLIBDIR=\"/usr/local/bin/chemistry/gromacs/PSI//share/top\"
> -I/usr/local/lib64/fftw/pgi/include  -tp=k8-64 -fastsse -Mipa=fast
> -Mipa=inline -c -o gmx_sgangle.lo gmx_sgangle.c
>   /usr/local/pgi/linux86-64/6.0/bin/pgcc -DHAVE_CONFIG_H -I. -I.
> -I../../src -I../../include
> -DGMXLIBDIR=\"/usr/local/bin/chemistry/gromacs/PSI//share/top\"
> -I/usr/local/lib64/fftw/pgi/include -tp=k8-64 -fastsse -Mipa=fast
> -Mipa=inline -c gmx_sgangle.c -o gmx_sgangle.o
>  PGC-S-0090-Scalar data type required for logical expression
> (gmx_sgangle.c: 313)
>  PGC-S-0090-Scalar data type required for logical expression
> (gmx_sgangle.c: 320)
>  PGC/x86-64 Linux/x86-64 6.0-5: compilation completed with severe errors
> **************************************************************************************
> 
Shitty compiler...

replace in the header of the function
matrix box,
by
rvec *box


> Did anyone run into those problems before when trying to compile
> gromacs with those compilers?
> 
> Any help will be very welcome. :)
> 
> Thanks a lot in advance,
> 
> Jones
> 
> P.S.: I'm using the "mixed" run simulation of the "tutor" directory as
> a compilation test.
> _______________________________________________
> 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.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php


-- 
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://folding.bmc.uu.se
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++