[Mitgcm-support] Re: RE:

Wed Jul 9 15:56:20 EDT 2003

Hi Honggao,

 I think the messages can be fixed by making the options that are used
in the file timers.F for the PWR3 platform active (#define TARGET_PWR3).
These tell the timer routines to call an AIX routine that should be
there on your system too. 

Chris 
On Mon, 2003-01-06 at 11:22, Honggao Liu wrote:
> Hi, Chris:
>     Where are you hiding?
>     Thank you very much for your response!
>     I have resolved the problem that Entity wordlength has undefined type,
> however, I go error of Undefined symbol: .etime and  .cloc in the last step
> of compiling. The error message is as follow:
> 
> ld: 0711-317 ERROR: Undefined symbol: .etime
> ld: 0711-317 ERROR: Undefined symbol: .cloc
> ld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more
> information.
> make: *** [mitgcmuv] Error 8
> 
> I attached the Makefile and make.log file for your review. Let me know what
> is wrong and how to fix it.
> 
> By the way, I alos failed to compile MITgcm on our IBM Power3 machine using
> the SP3 setting.
> 
> Thanks,
> Honggao Liu, Ph.D.
> Applications Consultant, High Performance Computing
> Office of Computing Services
> Louisiana State University
> Phone: +1 225 578 0235
> Email: honggao at lsu.edu
> 
> 
> 
> From:Honggao Liu/honggao/LSU at gulf.mit.edu on 01/02/2003 07:56 PM EST
> 
> 
> Sent by: Chris Hill <cnh at gulf.mit.edu>
> 
> 
> 
> 
> 
> To:  Honggao Liu <honggao at lsu.edu>
> cc:  support at mitgcm.org
> 
> Subject:  Re: RE:
> 
> 
> 
> Hi Honggao,
> 
>  I'm sure it can work. Not quite sure what options are needed though.
>  Here are some clues.
> 
>  WORDLENGTH should be substituted by a numeric value of either 1 or 4 by
> the CPP preprocesor stage.  The CPP setting is almost guaranteed to be
> -DWORDLENGTH=4. i.e. what is shown for the SP3 options -D settings. That
> is what the SP3 uses and this is not something that normally ever
> changes between OS revisions. The fact that you are getting the message
> from the compiler suggests the CPP stage is not happening. Here is what
> you should do
> 
>  1. check the Makefile to see that the .F source files are indeed being
> piped through CPP to create .f files that are then compiled.
> 
>  2. delete all files called *.f ( little f)
> 
>  3. make again
> 
>  4. send mail with your Makefile and the log of your make session
> attached if you following through the above doesn't show you what is
> wrong.
> 
> Chris
> On Thu, 2003-01-02 at 09:17, Honggao Liu wrote:
> > Hi, Chris:
> >
> > Happy New Year!
> >
> > I know there is settings for for AIX + Power3 in the patch8 genmake.
> > However, our system has different machine architectures such as IBM
> Power2,
> > Power3, PowerPC.We have only installed compilers on Power2 machines for
> > users use.
> >
> > I failed to compile the source code because Entity wordlength has
> undefined
> > type in source code "mdsio_reclen.f".
> >
> > "mdsio_reclen.f", 1516-036 (S) Entity wordlength has undefined type.
> > ** mds_reclen   === End of Compilation 1 ===
> > 1501-511  Compilation failed for file mdsio_reclen.f.
> >
> > Can MITgcm Release 1 Patch 8 work on a IBM Power2 machine running AIX
> 4.3?
> > How can I modify the genmake file to solve the undefined type problem on
> > AIX system?
> >
> > Thanks,
> > Honggao
> >
> > From:Honggao Liu/honggao/LSU at plume.mit.edu on 12/23/2002 12:32 PM EST
> >
> >
> > Sent by: "Chris Hill" <cnh at plume.mit.edu>
> >
> > Please respond to <support at mitgcm.org>
> >
> >
> >
> > To: "'Honggao Liu'" <honggao at lsu.edu>, <support at mitgcm.org>
> > cc: <casper at lsu.edu>
> >
> > Subject: RE:
> >
> >
> >
> > Honggao,
> >
> >  I believe the settings for -platform=SP3 should work for AIX + Power3.
> >  They are in the patch8 genmake. Let us know how you get on.
> >
> > Chris
> >
> >
> > -----Original Message-----
> > From: Honggao Liu [mailto:honggao at lsu.edu]
> > Sent: Monday, December 23, 2002 12:18 PM
> > To: support at mitgcm.org
> > Cc: casper at lsu.edu
> > Subject:
> >
> >
> > Dear Sir/Madam:
> >
> > Our environment (CaSPer) at LSU consists of IBM RS/6000 AIX 4.3 systems
> > for interactive and batch work, as well as a cluster of 16
> > dual-processor IBM
> > RS/6000(SP) Power3 nodes for parallel work.
> >
> > We are considering to install MITgcm Release 1 Patch 8 on our CaSPer for
> > a client. Could you tell me if MITgcm Release 1 can be installed on our
> > IBM AIX 4.3 systems? Also, do you have a parallel version of the
> > software for IBM AIX 4.3 is available?
> >
> > I checked the script genmake located in the tools directory, there is no
> > settings for IBM AIX. I edited the genmake to specify the platform as
> > AIX case -platform: set platform=AIX breaksw
> >
> > and then added machine setup corresponding to a IBM AIX machine into
> > Platform specific options as follow: case AIX*:
> >     echo "Configuring for " $platform
> >     set LN         = ( '/bin/ln -s' )
> >     set CPP        = ( '/lib/cpp -P' )
> >     set DEFINES    = ( ${DEFINES} '-DWORDLENGTH=4' )
> >     set FC         = ( 'xlf' )
> >     set FFLAGS     = ( '' )
> >     if ($?IEEE) set FFLAGS     = ( $FFLAGS '-ffloat-store' )
> >     set FOPTIM     = ( '-O3' )
> >     set LINK       = ( 'xlf' )
> >     breaksw
> >
> > However, I failed to compile the source code.
> >
> > "mdsio_reclen.f", 1516-036 (S) Entity wordlength has undefined type.
> > ** mds_reclen   === End of Compilation 1 ===
> > 1501-511  Compilation failed for file mdsio_reclen.f.
> >
> > Could you tell me how to edit the genmake to make it work on our IBM AIX
> > system?
> >
> > I am looking forward to hearing from you soon.
> >
> > Honggao Liu, Ph.D.
> > Applications Consultant, High Performance Computing
> > Office of Computing Services
> > Louisiana State University
> > Phone: +1 225 578 0235
> > Email: honggao at lsu.edu
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
>  - Makefile
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 