[Mitgcm-support] Re: c40 merge

mitgcm-support at dev.mitgcm.org mitgcm-support at dev.mitgcm.org
Wed Jul 9 15:42:28 EDT 2003


 I'll add the atmos. threads stuff this afternoon.

 JMC - can you look at Eq Channel. Lets not get it
deleted by the check-in monster!

 I'm happy to wait until c41 for dynamics to be split.
Think we should make sure everything is stable under c40
first. Lets just do the energetics min/max/variance 
stats stuff. Note - cg2d residual is still the most sensitive test 
for bit reproducibility. Its very useful for fighting
multi-threaded and/or multi-process hysteria.

 Patrick - the top-level directory "adjoint" really
looks like it should be under "verification" or
"tools" to me. Any chance we could get this
fixed before c40 is c40?

Chris
 

Alistair Adcroft wrote:
> 
> All done and tested.
> 
> Not too bad. It all worked easily apart from having to upgrade the
> "patch" in hs94.cs and sort out the write(0) and write(*) stuff.
> 
> I haven't even looked to see how the merge/split of forward_step.F
> worked out. It just worked - I'll figure out why later....
> 
> Tasks before c40 tag:
>  o Chris's thread mods need to be included
>  o verification/aim.5l_XZ_Equatorial_Slice is untested in both pre38 and HEAD
>    (if no-one volunteers to sort it out then I'll delete it)
>  o I would dearly like to have a better metric than cg2d iter,res in the
>    std output before c40. Will work on that today.
>  o split dynamics.F into thermodynamics.F and dynamics.F (or should we wait?)
> 
> Can someone kind persons try this out on SUNs and DECs? Email so we know
> who's doing what?
> 
> A.



More information about the MITgcm-support mailing list