4 ms·
>Unless you can turn them off? Then the background tasks need to run on the performance cores slowing down your main workload. > Then why pollute the die with
by Havoc 3y ago
>Unless you can turn them off?
Then the background tasks need to run on the performance cores slowing down your main workload.
> Then why pollute the die with these?
That's partially the point - they're much much smaller than P cores physically. So if you can move some non-critical stuff off the P cores at a very low die space cost then that's a win.
...all of which assumes correct & intelligent scheduling which I gather isn't always the case
- varispeed 3y agoWhat you saying makes sense, but I am thinking that this may be false economy. It of course depends on the nature of background tasks, but likely they would have a chance to finish quicker on a performance core than on efficiency core and if main workload depends on those e.g. something generating an event, you still are probably going to get a performance penalty greater than if everything was running on P cores or if main task needs I/O and slow background process is blocking it for longer.