[MITgcm-devel] vectorization of exf_bulk_largeyeager04.F

[Jean-Michel Campin]

Hi Martin,

There are many platform/compiler where we only get 11 digits for this
experiment, so I would not worry too much (and most of the other numbers,
except cg2d, are 16), and you can check your changes.
And It would be good to generate a new output.txt on faulks with g77, 
since the switch to gfortran might take some time.
Cheers,
Jean-Michel

On Sun, May 16, 2010 at 11:24:50AM +0200, Martin Losch wrote:
> Hi Jean-Michel,
> 
> I have (finally) changed exf_bulk_largeyeager04 to give better performance on a vector computer (as has been done for exf_bulkformulae long ago). The only test that uses this code is the forward test of global_ocean.cs32x15. Three of the 4 sub-test pass (16), but the forth only gives 11digits of agreement (on faulks). Guess, which one it is ... yes, input.icedyn, see below. I cannot reproduce this difference on my apple laptop with gfortran (however, I do deviate from the reference to begin with, expected with this platform).
> 
> I don't see any fundamental difference between input.icedyn and the other experiments, except for the dynamic sea ice, but why should the dynamics be affected by bulk_largeyeager04? The wind stress is converted to wind speed independently.
> 
> What should I do. Should I still go ahead and check this in (+change the output.txt.icedyn), or in the light of your planned change in reference compiler from g77 to gfortran, should I wait ...?
> 
> 
> Martin
> 
> 
> 
> default 13  ----T-----  ----S-----  ----U-----  ----V-----  --PTR 01--  --PTR 02--  --PTR 03--  --PTR 04--  --PTR 05--
> G D M    c        m  s        m  s        m  s        m  s        m  s        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  .  m  m  e  .  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  i  a  a  d  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  .  n  x  n  .  n  x  n  .  n  x  n  .  n  x  n  .  n  x  n  .
> 
> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 14 16 16 16 16 16  .  .  .  .  .  .  .  . pass  global_ocean.cs32x15
> Y Y Y Y>11<16 16 16 16 16 16 16 16 16 16 14 16 16 16 14 16  .  .  .  .  .  .  .  . FAIL  global_ocean.cs32x15.icedyn
> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 14 16 16 16 16 16  .  .  .  .  .  .  .  . pass  global_ocean.cs32x15.thsice
> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 16 16 16 16 14 16  .  .  .  .  .  .  .  . pass  global_ocean.cs32x15.viscA4
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel