[Mitgcm-support] Re: internal_wave

mitgcm-support at dev.mitgcm.org mitgcm-support at dev.mitgcm.org
Wed Jul 9 15:56:09 EDT 2003


Great - we will delete the misleading README!

Chris 
Theo Gerkema wrote:
> 
> Dear Chris,
> 
> Many thanks for your help! The code runs now perfectly and reproduces
> the results from the original output.txt.
> Once we have a modified genmake file I'll send it to you.
> 
> Best regards, Theo.
> 
> cnh wrote:
> 
> > Hi Theo,
> >
> >  I think the README is out of date. The steps you should follow are
> > outlined
> > here
> 
> > http://mitgcm.org/sealion/online_documents/node90.html (reproduced
> > below)
> >
> >  For this procedure to work on a NEC platform, the genmake script needs
> > an entry
> > for that platform (if you're happy to send it to us we will add it to
> > the
> > list already in there). The manual section
> >
> > http://mitgcm.org/sealion/online_documents/node92.html
> >
> > describes how to add a new platform setting to genmake. The example
> > in the genmake file for case SP3: shows many of the different options.
> > These are normally enough.
> >
> >  I did a quick check of the relase1_p9/verification/internal_wave
> > compile and run on a local Linux machine. It came out with the
> > expected KE values etc... so I think its just the README
> > problem.
> >
> > Chris
> >
> > >
> > >
> > >
> > >              3.5 Building the code
> > >
> > >              To compile the code, we use the make program. This uses a file (Makefile) that allows us to
> > >              pre-process source files, specify compiler and optimization options and also figures out any file
> > >              dependencies. We supply a script (genmake), described in section 3.5.2, that automatically
> > >              creates the Makefile for you. You then need to build the dependencies and compile the code.
> > >
> > >              As an example, let's assume that you want to build and run experiment verification/exp2. The are
> > >              multiple ways and places to actually do this but here let's build the code in
> > >              verification/exp2/input:
> > >
> > >              % cd verification/exp2/input
> > >
> > >              First, build the Makefile:
> > >
> > >              % ../../../tools/genmake -mods=../code
> > >
> > >              The command line option tells genmake to override model source code with any files in the
> > >              directory ./code/.
> > >
> > >              If there is no .genmakerc in the input directory, you have to use the following options when
> > >              invoking genmake:
> > >
> > >              % ../../../tools/genmake  -mods=../code
> > >
> > >              Next, create the dependencies:
> > >
> > >              % make depend
> > >
> > >              This modifies Makefile by attaching a [long] list of files on which other files depend. The purpose
> > >              of this is to reduce re-compilation if and when you start to modify the code. make depend also
> > >              created links from the model source to this directory.
> > >
> > >              Now compile the code:
> > >
> > >              % make
> > >
> > >              The make command creates an executable called mitgcmuv.
> > >
> > >              Now you are ready to run the model. General instructions for doing so are given in section 3.6.
> > >              Here, we can run the model with:
> > >
> > >              ./mitgcmuv > output.txt
> > >
> > >              where we are re-directing the stream of text output to the file output.txt.
> >
> > Theo Gerkema wrote:
> > >
> > > Dear Chris,
> > >
> > > Many thanks for your reply and for asking Sonya to help us;
> > > her help will be much appreciated!
> > >
> > > I'm sorry to bother you with technical questions concerning
> > > the code,  but I'm rather stuck at the moment and would be most
> > > grateful if you or one of your collegues could give me a suggestion on
> > > how to proceed.
> > >
> > > I've been trying to run the internal_wave example by following the
> > > instructions in verification/internal_wave/README, but
> > > encountered some problems.  One of which was that I could not find
> > > the file /verification/internal_wave/code/set_obcs.F  which was to
> > > be copied to /model/src/, so I've taken the file named most
> > > similarly instead (/code/obcs_calc.F), without of course quite
> > > knowing what the consequences are.
> > > Furthermore, when trying to make an executable (in bin), it appeared
> > > that the file verification/internal_wave/code/CPP_EEOPTIONS.h
> > > was needed, which I've then copied to eesupp/inc/.
> > > The executable could then be made, and a run was done (20 time steps).
> > > However, when I compared the outcome with the test results
> > > (/verification/internal_wave/results/output.txt), there appeared to be
> > > gigantic differences, for example, after 20 steps
> > > ke_volint = 1.4100262662647E+02 in the test example
> > > whereas my run gives
> > > ke_volint = 2.4735222463419E-17
> > >
> > > The differences may be due to the above two modifications with
> > > respect to the README file, but I don't see how I could verify this.
> > > Another potential source of problems may be that we have had to
> > > change a number of things in the Makefile by hand, in order to run it at
> > >
> > > the supercomputer (NEC SX-5, Idris centre). This involves a change
> > > in the path of the CPP =  statement, an addition for the C-compiler
> > > (CC=sxcc),
> > > changes in "f77" (becoming: sxmpif90) -- all of which I think are
> > > harmless -- and a removal of all flags after FFLAGS=
> > > and FOPTIM=
> > > (Whether these last two things can be done with impunity I'm less sure
> > > of,
> > > but we were advised to do so; in fact, we were unable make an executable
> > > when
> > > they were included).
> > >
> > > Could you (or one of your collegues) give me some advice on how to
> > > make the code run for this example? Many thanks.
> > >
> > > Best regards, Theo.



More information about the MITgcm-support mailing list