[Mitgcm-support] different results with different nb of procs

[mitgcm-support at dev.mitgcm.org]

Hi Chris, Alistair, Patrick, Jean-Michel,

I am running my sector model on the pleiades network and I found something 
surprising: I did two identical simulations except that one was run 
interactively on bonanza with 4 procs while the other was run in one of the 
8 procs queue and I obtain...different results! 

I'm wondering whether any of you had this kind of experience before and 
whether that means that there is a bug in my code (I guess it does!)

Thanks for any insight you might have.

Daniel