[Mitgcm-support] Re: EXCH gUnm1,gVnm1 in aim_5l_cs

[mitgcm-support at dev.mitgcm.org]

Jean-Michel Campin wrote:
> It seems that the exchange of gUnm1,gVnm1 at the begining of
> SOLVE_FOR_PRESSURE is also needed for aim_5l_cs with pgf77 (on Linux).

Can we figure out why rather than adding exchanges. It seems that a loop
range is wrong somewhere and so should be fixed. Too many exachanges is
costly and clumsy.

> May be aim_5l_cs/result/output.txt could help to keep
> this experiment working ?

I agree but it uses pgf77 which I don't have so I rarely get to compile it.

A.