AMBER Archive (2008)

Subject: Re: AMBER: Amber10 on Leopard with Intel compilers, "Undefined symbols" error

From: Yu Chen (chen_at_hhmi.umbc.edu)
Date: Wed Apr 23 2008 - 13:50:54 CDT

>> I made some progress on Intel compilers with version 10(seems can't use
>> 64bit compiler with AmberTools). But got stuck with some "Undefined
>> symbols, "_ts_module_mp_boptts_", referenced from:
>> _ts_module_mp_boptts_$non_lazy_ptr in libdivcon.a" ... problems
>
> The workaround is to give to "-nosanderidc" flag to configure_amber. The
> divcon people may be able to start using macosx, but until that happens, or
> someone else debugs this, you will need to disable the divcon part. This is
> unlikely to be a limitation for most calculations.

Thanks, Dr. Case. Will try that.

>
>> then I edit the config_amber.h file replace the gcc with icc and g++ with
>> icpc (config_amber.h is attached too)
>
> I don't think there is any need to do this, but maybe it doesn't hurt. There
> are no performance-critical parts of Amber that are in C, so there is really
> no reason that I know of to use icc rather than gcc here.

Yes, I understand this. The reason is that later on, we are going to
compile the parallel version, and our MPI was compiled with icc/ifort,
just try to get everything compiled with same compilers. Since sometimes
the mixed compilers could actually lead to more problems from what I have
learned.

>
>> Undefined symbols:
>> "_ts_module_mp_boptts_", referenced from:
>> _ts_module_mp_boptts_$non_lazy_ptr in
>
> The hopeful thing is that these routines are not really needed for QM/MM
> anyway. One of the things on the divcon "to-do" list is to not compile
> things into sander that are not needed (because it unnecessarily exposes
> one to compiler problems). I'll cc this email to qtp to see if someone there
> has time to look at it.
>

Thank you.

> ...thanks for the report, and for being the Mac guinea pig!....dac

Thank you for great support! One thing I forgot to mention, I didn't apply
Dr. Zhang Wei's gleap patch for this.

Regards,

>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

===========================================
Yu Chen
Howard Hughes Medical Institute
Chemistry Building, Rm 182
University of Maryland at Baltimore County
1000 Hilltop Circle
Baltimore, MD 21250

phone: (410)455-1728 (primary)
         (410)455-6347 (secondary)
fax: (410)455-1174
email: chen_at_hhmi.umbc.edu
===========================================
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu