[MITgcm-devel] autodiff_store/restore

[Patrick Heimbach]

Hi Martin,

this was written to combine I/O from multiple variable into single array
which seems to improve performance on some platforms.

It's not followed throughout, so you can keep new stores separate
(unless you anticipate them to be part of a "core" set of stores,
in which case it'd be better to add them in autodiff_store/restore).

Your warning is unrelated,
most likely due to missing initialization of those fields.

-p.

On May 9, 2011, at 9:50 AM, Martin Losch wrote:

> Hi Matt, and others,
> 
> can you remind me why we need autodiff_store/restore?
> 
> I want to include a new control variable (like surface forcing) and I am wondering if I need to include it into this autodiff_store/restore scheme. 
> I am getting warnings that my xx_newctrl0 and 1 are overwritten by the top-level routine. Can this be related?
> 
> Martin
> 
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel

---
Patrick Heimbach | heimbach at mit.edu | http://www.mit.edu/~heimbach
MIT | EAPS 54-1518 | 77 Massachusetts Ave | Cambridge MA 02139 USA
FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE patrick.heimbach