Daniel B (fejenagy.delete@this.gmail.com) on September 13, 2021 5:20 am wrote:
> And the whole scheduling complexity that comes with it all is just awkward, is it not?
> Either the scheduler needs to learn the application behaviour
> or the application should come tagged, but then the ISV is expected to understand
> hardware and system energy efficiency, which they don't.

In a lot of multithreaded applications, the threads which do less can/would be executed on the smaller cores, but usually because it is a single application the threads which do more (and are executing on a fast core) are most of the time waiting for the results of the smaller threads.
Which usually follow a saying of mine, an application executes as fast as the slowest cores available on the system.
Isn't it time to optimise power saving in idle, and re-instate the race to sleep pattern?

Obviously smaller cores can be dedicated to real time tasks (segregated by the scheduler), or some independent applications not part of the user experience. If you have multiple processors on a system (radio management of mobile phones), it might be interesting to unify them.