[Mitgcm-support] Re: checkpoint40

[mitgcm-support at dev.mitgcm.org]

Not so fast you guys! c40 is broken.

The model tests out fine under testscript but when I use MPI
it bombs badly.

e.g. convert exp4 to 2x2 multi-processors and the buoyancyRelation
variable gets over-written. Adding array bounds checking causes it
to stop in exch_rs_sendput.f which has been called with strange
values of iMin,iMax,...

Not good. I wouldn't celebrate yet.

A.