[MITgcm-support] compiling error in the latest sources of MITgcm on irix64!
Patrick Heimbach
heimbach at MIT.EDU
Tue Jul 26 13:32:36 EDT 2005
Probably the best bet might be the optfile
irix64_ip35_f77+mpi_mnc_gfdl
i.e. in the optfile you're using try adding the line
FC_NAMEMANGLE='#define FC_NAMEMANGLE(X) X ## _'
Might be worth a try.
-p.
On Tue, 2005-07-26 at 13:20, Ed Hill wrote:
> On Tue, 2005-07-26 at 12:50 -0400, Van Thinh Nguyen wrote:
> >
> > I tried to update the old sources of MITgcm (using cvs update command),
> > after that I re-compile my old model (this is compiled and ran (mpi)
> > without any error by old sources), then I've got following error:
> > ---
> > ld64: ERROR 33 : Unresolved text symbol "timenow_" -- 1st referenced by timers.o.
>
> > Anyone may know about this error, please give me an advice.
>
> Hi Van Thinh,
>
> The problem is that your Fortran compiler doesn't support the timenow()
> function and there are at least two ways that you can fix it:
>
> 1) quick and easy fix : use the genmake2 "-ignoretime" option to
> turn off the timing code entirely -- MITgcm will run just
> fine (you will get correct results) but the information about
> how much time is spent in various subroutines will be useless
>
> 2) difficult but more satisfying fix : look at all the optfiles
> in MITgcm/tools/build_options and figure out which values of CC
> and FC_NAMEMANGLE will allow you to compile and call C code
> from your Fortran subroutines.
>
> Ed
--
--------------------------------------------------------
Patrick Heimbach Massachusetts Institute of Technology
FON: +1/617/253-5259 EAPS, Room 54-1518
FAX: +1/617/253-4464 77 Massachusetts Avenue
mailto:heimbach at mit.edu Cambridge MA 02139
http://www.mit.edu/~heimbach/ USA
More information about the MITgcm-support
mailing list