[MITgcm-devel] Re: MITgcm vectorization on NEC SX

[Constantinos Evangelinos]

On Mon 26 Nov 2007 12:19, Jean-Michel Campin wrote:

> I have few comments:
> a) regarding I/O: without singleCpuIO, MDSIO read/write
> 1 line at a time ; with singleCpuIO a 2-D field is read/written
> in 1 instruction. Does this make a difference in our run
> (not so easy to check, may be the easiest would be on 1 processor
> only) ?

Yes it does. In fact the small granularity of the writes in global file and 
per tile mode go a long way in explaining why funneling everything through 
processor zero works so well.

Constantinos
-- 
Dr. Constantinos Evangelinos                    Room 54-1518, EAPS/MIT
Earth, Atmospheric and Planetary Sciences       77 Massachusetts Avenue
Massachusetts Institute of Technology           Cambridge, MA 02139
+1-617-324-3386/+1-617-253-4464 (fax)           USA