add amip with diagnostic edmf job

[szy21]

Purpose

Adds an AMIP with edmf job to the longrun. Instead of only adding diagnostic edmf to the current amip_target_topo job, I chose to add the most realistic configuration (time-varying insolation and all-sky radiation). Because with edmf it is more than a factor of 2 slower I set t_end to 40days. This one should run, I can check once buildkite works again.

Also changes dt and dt_cpl in amip_targe_topo to be the same.

Closes #574

To-do

Content

---

• I have read and checked the items on the review checklist.

[szy21]

I will revert the pipeline change after testing.

[szy21]

Ah, we need a new ClimaAtmos release for this to work, as it is using a new atmos longrun config file. I can rebase the PR after we release ClimaAtmos and update ClimaCoupler this week.

[LenkaNovak]

@szy21, I know you're waiting for the update for CA, but FYI the macOS problem should be sorted once you rebase. Also, would it be possible to set up this run (but with a prescribed surface) in the Atmos Buildkite CPU longruns, so we can keep track of performance differences?

[LenkaNovak]

These need to be uncommented otherwise good to go.

[szy21]

@szy21, I know you're waiting for the update for CA, but FYI the macOS problem should be sorted once you rebase. Also, would it be possible to set up this run (but with a prescribed surface) in the Atmos Buildkite CPU longruns, so we can keep track of performance differences?

I would prefer to keep the most important jobs in the GPU longrun, as that is what we are focusing on right now. The corresponding one in atmos is different nevertheless, as I am using idealized insolation to be more consistent with the time-invariant boundary condition. I can add a job in the CPU longrun with the exactly same setup as the coupler one for a few weeks or months for comparison.

[LenkaNovak]

@szy21, I know you're waiting for the update for CA, but FYI the macOS problem should be sorted once you rebase. Also, would it be possible to set up this run (but with a prescribed surface) in the Atmos Buildkite CPU longruns, so we can keep track of performance differences?

I would prefer to keep the most important jobs in the GPU longrun, as that is what we are focusing on right now. The corresponding one in atmos is different nevertheless, as I am using idealized insolation to be more consistent with the time-invariant boundary condition. I can add a job in the CPU longrun with the exactly same setup as the coupler one for a few weeks or months for comparison.

That makes sense. And yeah, that would be awesome, thank you! Or I can also do it, if you have a lot on this week. :)

[szy21]

@LenkaNovak CliMA/ClimaAtmos.jl#2624 added the job if you want to take a look