[MITgcm-devel] Latest changes in gmredi_calc_psi_bvp.F

[Jean-Michel Campin]

Hi,

It looks like my changes are causing some diffences with
ifort + optimisation (fast) on beagle:
before (March 14):
> Y Y Y Y>16<16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . pass  front_relax.bvp
and last night:
< Y Y Y Y> 2< 8  4  9  8  7  6 11  8  3  3  3  3  2  2  2  2  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  . FAIL  front_relax.bvp
Without optimisation, the results are identical.

This might be related to the "compiler internal error" problem
reported earlier (Beaufort experiment on mac os thread).

Will take a look at this routine.

Cheers,
Jean-Michel

On Thu, Mar 15, 2012 at 11:18:02AM -0400, Jean-Michel Campin wrote:
> Update of /u/gcmpack/MITgcm/pkg/gmredi
> In directory forge:/tmp/cvs-serv21229
> 
> Modified Files:
> 	gmredi_calc_psi_bvp.F 
> Log Message:
> fix for vectorized & differentiable solve_tridiagonal.F (to become the default)
> 
> 
> _______________________________________________
> MITgcm-cvs mailing list
> MITgcm-cvs at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-cvs