[MITgcm-devel] IBM Regatta and cubed sphere

[Martin Losch]

Hi,

today I managed to compile and run a verification experiment (exp0) 
with mpi on our IBM Regatta. The experiment passed but only after 
manipulation:
1. no cc-compiler was available, old problem, I made softlink cc --> gcc
2. in genmake2:
Error: can't read 
OPTFILE="../../../tools/build_options/aix_0010117a4c00_f77"
reason uname -a returns: AIX edvir1 1 5 0010117A4C00
so I specified sp4 as the optfile
3. Makefile and compilation is fine with
./testreport -ieee -mpi -of ../tools/build_options/sp4 -t exp0
but no number to compare( "Not enough lines of output when searching 
for ..."):
                 T           S           U           V
G D M    c        m  s        m  s        m  s        m  s
E p a R  g  m  m  e  .  m  m  e  .  m  m  e  .  m  m  e  .
N n k u  2  i  a  a  d  i  a  a  d  i  a  a  d  i  a  a  d
2 d e n  d  n  x  n  .  n  x  n  .  n  x  n  .  n  x  n  .

Y Y Y Y -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- N/O   exp0

I had to run the model with the IBM specific command
poe ./mitgcmuv -procs 2
and then I copied STDOUT.0000 to output.txt in exp0/input and reran 
testreport with:
./testreport -ieee -mpi -q -t exp0
which gives:
                 T           S           U           V
G D M    c        m  s        m  s        m  s        m  s
E p a R  g  m  m  e  .  m  m  e  .  m  m  e  .  m  m  e  .
N n k u  2  i  a  a  d  i  a  a  d  i  a  a  d  i  a  a  d
2 d e n  d  n  x  n  .  n  x  n  .  n  x  n  .  n  x  n  .

Y Y Y Y 13 16 16 16 16 16 16 16 16 16 16 12 13 13 13 13 16 pass  exp0
Start time:  Wed May 12 14:46:29 DFT 2004
End time:    Wed May 12 14:46:31 DFT 2004

So that's good.
I did the same with global_ocean.cs32x15 (also with -mpi) and I have 
the same problems as before but now I get:

STOP MON_SOLUTION: STOPPED DUE TO EXTREME VALUES OF SOLUTION
STOP MON_SOLUTION: STOPPED DUE TO EXTREME VALUES OF SOLUTION

The problem as before (last week with my own version of this 
experiment) is near the tile edges (see my last figure), immediately 
for the first time step 72000, w is wrong at the tile edges.

So far I have not been able to test this experiment with MPI on a 
different platform. But I'll keep trying.

Martin