[MITgcm-support] Optimizing the 2D pressure solver

[Dimitris Menemenlis]

> but my guess is that the MITgcm gurus have already worked this routine and
> that further optimization of it won't provide speedups of more than 1%

Actually, some eight years ago, Ralf Giering had experimented heavily with
"cg2dChkResFreq" type of flags and in some architectures, for example, the old 
HP Exemplar, they provided significant speed improvement.  Ralf had also written 
a direct solver, which was both faster and more accurate than cg2d.  But these 
changes never got sufficiently polished to make it to the main trunk of the 
MITgcm code.

D.

-- 
Dimitris Menemenlis <menemenlis at sbcglobal.net>
5056 Oakwood Ave, La Canada, CA 91011-2450
tel/fax: 818-790-6735;   cell: 818-625-6498