It’s actually not hardcoded, but we we’re using either min(95, sched_get_priority_max(SCHED_RR)) for SCHED_RR (when using real time threads) or sched_get_priority_max(SCHED_RR) for SCHED_FIFO when using real time threads failed.
This also hasn’t changed over the updates.
Seems more a side effect of [Regression]: Linux audio thread priorities not set correctly on startup, only when manually updating "Multi Core" value
So there’s definitely something fishy. I’ll check and fix that. Thanks for the report and figuring out all the details!