[MITgcm-devel] netcdf on Sun

Martin Losch mlosch at awi-bremerhaven.de
Fri Dec 10 03:49:28 EST 2004


Hi,
I am still having problems on the $%^&* SunFile 15k (just thinking 
about how much cash AWI has put down for this toaster makes me sick):
I can now compile everything with gmake, gawk und this fix in 
mnc_readparms (#undef HAVE_MNCCDIR, what is this mnccdir anyway and why 
do I  not have it?), but when I run this
mprun -np $number_of_processors ./mitgcmuv
(number_of_processors=20) all standard out is directed to the terminal 
(I mean of all 20 processors that I use, all with prefix (PID.TID 
0000.0001)) and I get only model output from one processor/tile: 
T.0000000000.001.001.??ta etc.

What am I doing wrong? I checked with my sysadmins, and they claim that 
I am doing everything right on the OS side.

Martin

PS. Ed, as usual it is probably not possible for me to get an account 
for you on this machine, in case your were about to ask.
PPS. this the optfile I am using:

> #!/bin/bash
> #
> #  $Header: /u/gcmpack/MITgcm/tools/build_options/sunfire+mpi,v 1.3 
> 2003/11/13 22:03:39 edhill Exp $
> #
>
> S64='$(TOOLSDIR)/set64bitConst.sh'
> DEFINES='-DWORDLENGTH=4'
>
> AWK='gawk'
> MAKE='gmake'
> MAKEDEPEND=makedepend
>
> FC='mpf77'
> LINK='mpf77'
> CPP='/usr/ccs/lib/cpp -P'
>
> FFLAGS='-stackvar -e -u -xtypemap=real:64,double:64,integer:32'
> FOPTIM='-dalign -O4 -xarch=native'
> CFLAGS='-dalign -xO4 -xarch=native'
> NOOPTFLAGS='-dalign -O0 -xarch=native'
>
> INCLUDES='-I/opt/SUNWhpc/include -I/usr/local/include'
> LIBS='-L/opt/SUNWhpc/lib -lmpi -lthread -lsocket -lnsl 
> -L/usr/local/lib'


On Dec 9, 2004, at 5:04 PM, Martin Losch wrote:

> Ed,
>
> the optfile sunfile+mpi that is in the repository seems to work unless 
> you want netcdf support, in which case you can point it there. But I 
> can check in another one for my platform that has the path included 
> (or modify the existing one).
>
> But two problems remain on my particular platform:
> model>uname -a
> SunOS model 5.9 Generic_112233-11 sun4u sparc SUNW,Sun-Fire-15000
>
> 1. AWK=awk in genmake2 does not work. I had to change that line in 
> genmake2 to
> AWK=gawk. That should probably be possible on the command line 
> (similar to -make=gmake)
> 2. After successful compilation with mnc enabled, the link step 
> failed, because it couldn't find
> routine mnccdir in mnc_readparms.F . I set HAVE_MNCCDIR in 
> mnc_readparms.F to undefined and now I have an executable. (I am 
> waiting in the queue, probably because I requested 30 CPUs).
>
> Martin
>
> On Dec 9, 2004, at 3:44 PM, Ed Hill wrote:
>
>> On Thu, 2004-12-09 at 15:14 +0100, Martin Losch wrote:
>>> Sorry about this, I take everthing back about the netcdf problem, 
>>> that
>>> was fixed by using gmake.
>>> But the awk problem remains. I had to change AWK=awk into AWK=gawk in
>>> the tools/genmake2 to make it work.
>>
>>
>> Hi Martin,
>>
>> Yes, those good Gnu tools should always be preferred!  ;-)
>>
>> So I assume you have it working now?  If not, please let me know what
>> errors are reported and I'll do whatever I can to help fix it.
>>
>> And *please* check in any custom optfiles you make to
>> tools/build_options.  There is even a tool at
>> tools/suggest_optfile_names to help people figure out optfile names 
>> for
>> their system.
>>
>> thanks,
>> Ed
>>
>> -- 
>> Edward H. Hill III, PhD
>> office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
>>              Cambridge, MA 02139-4307
>> emails:  eh3 at mit.edu                ed at eh3.com
>> URLs:    http://web.mit.edu/eh3/    http://eh3.com/
>> phone:   617-253-0098
>> fax:     617-253-4464
>>
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://dev.mitgcm.org/mailman/listinfo/mitgcm-devel
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-devel




More information about the MITgcm-devel mailing list