[CP2K:2202] Re: ERRORL2 in qs_loc_methods:centers_spreads_berry processor

[Jörg Saßmannshausen]

Dear Axel,

the reason why I orignally did not try your suggestion (which, I think, you 
have posted before) was I wanted to make the most or the i7 chip I have (or 
the E5520 on the actual computational node).

I actully tried is just 4 mins. ago:

ifort -c -D__INTEL -D__FFTSG -D__FFTW3 -D__LIBINT -I/opt/intel/fce/11.1/046/include/ -I/opt/fftw3/3.2.1/include -O2 -unroll -march=pentium3 -mtune=core2 -pc64 -fpp -heap-arrays 
64 -free -D__COMPILE_ARCH="\"Linux-x86-64-intel\"" -D__COMPILE_DATE="\"Tue 
Aug  4 20:19:45 BST 
2009\"" -D__COMPILE_HOST="\"linuxtut02\"" -D__COMPILE_LASTCVS="\"D\"" /home/sassy/build/cp2k/rt/cp2k/makefiles/../src/semi_empirical_parameters.F
: catastrophic error: **Internal compiler error: segmentation violation signal 
raised** Please report this error along with the circumstances in which it 
occurred in a Software Problem Report.  Note: File and line given may not be 
explicit cause of this error.

compilation aborted 
for /home/sassy/build/cp2k/rt/cp2k/makefiles/../src/semi_empirical_int_utils.F 
(code 3)
make[1]: *** [semi_empirical_int_utils.o] Error 3
make[1]: *** Waiting for unfinished jobs....

I had to add the -free in order to make it work. So, once I reach an ICE I 
don't board it (unless it is in Germany :D ) but report the bug to Intel. 
Which I actually would love to do but the last time I tried it their website 
did not work properly.

The OS is Debian Lenny with gcc-4.3.2 installed. It was actually the seriell 
build and not the parallel build.

Right now I am not too concerned about it, I put that down as one of the Intel 
compiler problems. So I don't really want to wast too much time on that to be 
honest.

All the best from Glasgow!

Jörg


On Montag 03 August 2009 Axel wrote:
> On Aug 3, 5:06 pm, Jörg Saßmannshausen
>
> <jorg.sassm... at strath.ac.uk> wrote:
> > Dear all,
> >
> > let me jump into this discussion (I am german, I am suppose to jump the
> > queue anyway :D )
> >
> > I had some issues with the Intel 11.0.84 and 11.1.046 compiler which
> > basically gave ICE. So right now I am still using the 10.1.021 as I know
> > it is building ok the testjobs are ok (for me).
>
> joerg,
> please try the following flags with intel 11.
> they don't give an ICE for me (n.b. i tell it to
> not use any new fancy features of newer cpus and
> that seems to do the trick).
>
> DFLAGS   = -D__INTEL -D__FFTSG -D__FFTW3 -D__LIBINT \
>            -D__parallel -D__BLACS -D__SCALAPACK
> CPPFLAGS = -traditional -C $(DFLAGS) -P
> FCFLAGS  = $(DFLAGS) -O2 -unroll -march=pentium3 -mtune=core2 -pc64 -
> fpp -heap-arrays 64
> FCFLAGS2 = $(DFLAGS) -O0 -unroll -march=pentium3 -mtune=core2 -pc64 -
> fpp -heap-arrays 64
>
> [...]
>
> # and this one needed me to turn off optimization at some point in
> time,
> # didn't have the time to check whether it is still needed.
>
> graphcon.o: graphcon.F
>         $(FC) -c $(FCFLAGS2) $<
>
>
> greetings from a sea of packaging chips and surrounded by bubblewrap,
>    axel.
>
> > A know that Intel has got different subversions and whereas one might
> > work, another might break the code.
> > I am using the compiler flags which are supplied with the code for a
> > Linux x86-64 intel sopt/popt build.
> >
> > Just my 2 raindrops from Scotland!
> >
> > Jörg
> >
> > On Montag 03 August 2009 Teodoro Laino wrote:
> > > the best test for checking the optimization is to compile everything
> > > with -O0 (the differences between -O3 and -O2 are not so many..).
> > > If it does work with -O0 than you may  trigger which file is causing
> > > the problem and switch off optimization only for that particular file.
> > >
> > > Teo
> > >
> > > berosek wrote:
> > > > Dear Teo,
> > > > thank you very much for your help.
> > > > The -O3 optimalisation isn't the problem because of the same problem
> > > > is with -O2.
> > > > I'll try out other versions of intel compiler.
> > > > Ones more thank you.
> > > > Regards,
> > > > Berosek
> >
> > --
> > *************************************************************
> > Jörg Saßmannshausen
> > Research Fellow
> > University of Strathclyde
> > Department of Pure and Applied Chemistry
> > 295 Cathedral St.
> > Glasgow
> > G1 1XL
> >
> > email: jorg.sassm... at strath.ac.uk
> > web:http://sassy.formativ.net
> >
> > Please avoid sending me Word or PowerPoint attachments.
> > Seehttp://www.gnu.org/philosophy/no-word-attachments.html
>
> 


-- 
*************************************************************
Jörg Saßmannshausen
Research Fellow
University of Strathclyde
Department of Pure and Applied Chemistry
295 Cathedral St.
Glasgow
G1 1XL

email: jorg.sassm... at strath.ac.uk
web: http://sassy.formativ.net

Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html