[Mitgcm-support] Re: SunOS5.7
mitgcm-support at dev.mitgcm.org
mitgcm-support at dev.mitgcm.org
Wed Jul 9 15:52:48 EDT 2003
Hi Mark,
We are using the mitgcm on SUN machines here at SOC. The problems are
that the newer Sun compilers use a different option syntax to yours, and
also 'cpp' is not in the PATH by default, and the mitgcmuv tool
xmakedepend looks for it as 'cpp'.
Here's how we've overcome the errors you've encountered so far:
Within './tools/genmake' change the following region:
case SunOS:
set LN = ( '/usr/bin/ln -s' )
set CPP = ( '/usr/ccs/lib/cpp -P' )
set MAKEDEPEND = ( makedepend )
set DEFINES = ( ${DEFINES} '-DTARGET_SUN -DWORDLENGTH=4' )
set FFLAGS = ( '-stackvar -explicitpar -vpara -e -u -noautopar
-r8 -fsimple=0' )
set FOPTIM = ( '-dalign -O3' )
set CFLAGS = ( '-dalign -O3' )
set NOOPTFLAGS = ( '-dalign -O0' )
set NOOPTFILES = ( 'barrier.F different_multiple.F
external_fields_load.F ini_vertical_grid.F ini_spherical_polar_grid.F
ini_cori.F mon_printstats_rl.F mon_printstats_rs.F aim_aim2dyn.F
aim_dyn2aim.F aim_aim2dyn_exchanges.F aim_external_fields_load.F
aim_calc_diags.F aim_external_forcing.F aim_do_atmos_physics.F
aim_write_diags.F aim_do_inphys.F ')
breaksw
The changes I've made are to
MAKEDEPEND (to use SUNs makedepend, /usr/openwin/bin/makedepend on my
machine)
FFLAGS (changed the -xtypemap=... form to the older -r8 form)
Removed the references to -xarch=... in FOPTIM and CFLAGS which set the
processor type to optimise for
You are going to be better off with a cheap linux box than the old sparc
10 you have. A 1.3GHz Athlon will run the model faster than the sparc 10
- we have tried it. We have single processor Blade 2000 machines that
run a 100x64x22 model at about the same rate as a 1.8GHz Pentium with
the intel compiler.
Hope this helps,
John
------------------------------------------
John Stark
Applications Programmer
Southampton Oceanography Centre
Tel. +44 (0)23 8059 6571
e-mail jods at soc.soton.ac.uk
------------------------------------------
Mark Inall wrote:
>
> Hi, I've just downloaded MITgcm in the zipped tar form. My interest is
> in fjordic flows of salinity stratified water over steep topographic
> sills, and the generation of internal jumps, bores etc and also
> propagating internal waves. We are predominantly an observational
> group, however I was motivated by Sonia Legg's talk at EGS/AGU last
> week about internal tide/wave generation on a corrugated slope - and
> discussion thereafter with and old friend Tom Haine convinced me to
> have a go with the MITgcm! So I'm trying it out on the only unix machine
> I have to hand right now, an old Sun sparc 10 running SunOS5.7. Am I
> wasting my time? Should I get myself a linux machine? I've tried to
> compile for a few of the verification expts - the following problems
> are encountered each time. Any help you can give would be hugely
> appreciated. Thanks in anticipation.
>
> Mark.
>
> Trying to compile for exp4 the problems so far on the Sun have been
> 1) make depend .......... load of fortran files then write_state.F
> ./mdep22933a: cpp: not found (repeated hundreds of times) 2) make f77
> illegal options -xtypemap and -xarch ld: fatal: symbol referenceing
> errors
>
> Dr Mark Inall
> Lecturer in Marine Physics
> Scottish Association for Marine Science
> Dunstaffnage Marine Laboratory
> Oban PA34 4AD
>
> email: mei at dml.ac.uk
--
More information about the MITgcm-support
mailing list