Some Cosmology WUs are wierd and will crash. I can see a few of those in my queues, both legacy and docker types on different machines that are all running at stock speeds. If you have one that badly misbehaving, then abort it.
I know you've said you haven't got time to play around but if you are using 24 cores on a Cosmo WU you are no doubt already doing virtual box stuff. VB is a bit of a pain to set up. Lots of recommendations are not to use the embedded boinc version, but the later one direct from the oracle site
https://www.virtualbox.org/
Once you are set then create an app_config.xml file in the Cosmology project folder of Boinc Data
Code:
<app_config>
<project_max_concurrent>2</project_max_concurrent>
<app_version>
<app_name>camb_boinc2docker</app_name>
<plan_class>vbox64_mt</plan_class>
<avg_ncpus>4</avg_ncpus>
</app_version>
<app_version>
<app_name>lsplitsims</app_name>
<plan_class>vbox64_mt</plan_class>
<avg_ncpus>4</avg_ncpus>
</app_version>
</app_config>
Set your max concurrent, and avg ncpus to suit your PC and anything else you still want working. For me 4 cpu per WU is the minimum, 2 WUs at a time on a 12 thread CPU is a nice balance. Virtual Box will take over if you let it, and Cosmology can also use a fair bit of RAM. Create the app_config and restart boinc. You may need to abort any few Cosmo WUs in part progress as virtual boinc gets confused by the change in cpus available.