[Pw_forum] Compiling problem on itanium2 bull cluster
Federico Iori
iori at theory.polytechnique.fr
Mon Oct 6 16:30:28 CEST 2008
oookey!
thanks so much for the advice.
Btw, as Lorenzo pointed out i didn't specified the compiler. I'm gonna
do now:
FC intel 9.1.045
CC intel 9.1.049
but i have also to add that:
espresso_4.0 doesnt give any problems, nor with make pw nor with make pp
while the espresso_4.0.1 and the next one espresso_4.0.2 give the same
problem i described before.
well...let me try now with the option -no-ipo suggested before.
i will tell you. and also i will communicate this error to the CCRT guys.
thanks to everybody!
Axel Kohlmeyer wrote:
> On Mon, 6 Oct 2008, Federico Iori wrote:
>
> FI> Dear all.
>
> dear federico!
>
> FI> I was compiling my Pwscf on this machine (PLATINE the common name)
>
> [...]
>
> FI> *fortcom: Severe: **Internal compiler error: internal abort** Please report
> FI> this error along with the circumstances in which it occurred in a Software
> FI> Problem Report. Note: File and line given may not be explicit cause of this
> FI> error.
> FI>
> FI> ifort: error: problem during multi-file optimization compilation (code 3)
>
> [...]
>
> FI> doing the ./configure everything seems to work perfectly.
> FI> Then the make pw step works perfectly.
> FI> But in the end i cant compile the pp or pwall stuff.
> FI>
> FI> What do you suggest me to do? Should I bother you or directly the
> FI> technicians in the CCRT centre??
>
> as stated here many times before: an internal compiler error
> is a bug in the _compiler_, not the application. given the
> hardware configuration that you describe, it is very likely that
> you have a broken intel compiler and you need to update it to
> the latest patchlevel. this is a quite common problem.
>
> i also strongly advise to run tests with the compiled executables,
> as there are very high chances that those are miscompiled. these
> days, a successful compilation does not mean, that you have a
> correctly working executables. the itanium architecture is quite
> notorious there, since the compiler has to do more work compared
> to other architectures.
>
> cheers,
> axel.
>
>
>
> FI> Thanks so much in advance!
> FI> federico iori
> FI>
> FI>
> FI>
>
>
--
Dr. Dr. Federico Iori, PhD!
Laboratoire des Solides Irradiés,
Ecole Polytechnique - CEA/DSM, CNRS
91128 Palaiseau, France
www.etsf.polytechnique.fr
phone: +33 (0) 1 69 33 44 85
emails: federico.iori @ polytechnique.edu
iori.federico @ unimore.it
http://www.dsmodena.it/home.html
http://www.comune.mo.it
More information about the users
mailing list