# Introduce more Experiments to the icon CD/CI Pipeline

###### tags: `shaping-archive`

Developers: -

Support: -

Appetite: -

## Background

So far, only a single experiment is tested in the icon-exclaim CD/CI pipeline; `exp.mch_ch_r04b09_dsl`. While there is nothing wrong with that experiment, it features a small domain only, and omits code paths that are relevant to exclaim (e.g. global mode) or both exclaim and MCH operations (high CFL numbers). Hence, more experiments need to be introduced.

## Goals

* Introduce a reasonable set of experiments to the `icon-exclaim`CD/CI pipeline
* At least for one experiment, possibly one with a rather large domain, set something up s.t. the `LOG.exp_name.o` files are archived for the _substitution_ runs, in order to have a log over time w.r.t to the wallclock times

## Non-Goals

* Do not try to set up automatic plotting of runtimes or other performance plots in jenkins

## Known Tasks

* A large number of experiments come "for free" once spack has been introduced on balfrin, see task **TODO**
* It could be the case that one or more experiment does not probtest. If this should be the case, prioritize fixing those experiments.
* However, it is not clear if all of these experiments are required (to cover all relevant code paths and configurations for exclaim + MCH operations), or if more experiments are required
* It is also not clear if the set of experiments should be the same for the daily and PR jenkins plan
* `!$acc wait` statements and `timer_start()` calls need to be moved to the correct place in the diffusion, velocity advection and solve nonhydro modules for the second goal