Dev gpu fix

[fjansson]

Fixes after merging the Ruisdael and RRTMGP branches.

[lsoucasse]

Hi @fjansson, just checkout your branch, waiting for snellius to test it...

[lsoucasse]

Hi @fjansson, Currently the botany case crashes with this commit because of tracers being not read properly (though I modified the nameoption file and scalar.inp file with Caspar). Weirdly this reading problem only appear on the GPU and not on the CPU. I will try to add your moment calculation change into my old radiation branch to see if it fixes the issue of having Nan values in stat fields.

[fjansson]

@lsoucasse I got Botany working with this branch:
add this to namoptions. Capitaliztion of the scalar names is important:

&NAMTRACERS                                                                                                                 
ltracers = .true.                                                                                                           
tracernames = 'nr', 'qr'                                                                                                    
/ 

edit scalar.inp.001, change the second line to

#height               nr          qr        

here there should be three words separated by white space, and capitalization might matter again. (There's a tolower call somewhere in the initialization but not everywhere).

[lsoucasse]

@lsoucasse I got Botany working with this branch: add this to namoptions. Capitaliztion of the scalar names is important:

&NAMTRACERS                                                                                                                 
ltracers = .true.                                                                                                           
tracernames = 'nr', 'qr'                                                                                                    
/ 

edit scalar.inp.001, change the second line to

#height               nr          qr        

here there should be three words separated by white space, and capitalization might matter again. (There's a tolower call somewhere in the initialization but not everywhere).

Hi @fjansson, I can read the tracers in the botany case, but still have the issue with my other column test cases while following the same procedure, formatting issue I suppose.

So I ran a couple of iterations of the botany case on the gpu interactive node that worked fine! But then running on a regular node gives me again a netcdf error when writing statistics, I am curious to know if you experience it or not.