12th gen performance core vs efficiency core windows utilization

[Leaffy]

I'm just a normal end used. Forgive if my question sounds wrong, but hope it makes sense to some. If anything, teach me, if possible.

On i7 12th gen, windows itself, where is it running? 

Windows core services, does it run on P cores or E cores?

 

Which is the main core that windows runs on(all the cores services, right from boot up) ?

Does windows boot up and run on performance cores, and when programmes lauched, it shall choose to use perf core or efficiency cores or,

is it vice versa; windows runs on E core and uses P cores, let say for gaming?

 

I'm sorry for bad english + sorry for repeating same question. I fear its hard make sense.

[igormp]

Windows itself is composed of many services, and a process doesn't need to stay stuck onto a single core, it can move around cores, so, answering your questions:

5 minutes ago, Leaffy said:

So, windows itself, where is it running?

As said above, in any core it feels like to, there's a part of the OS called the scheduler which is responsible for that.

6 minutes ago, Leaffy said:

Like that, windows core services, does it run on P cores or E cores?

Likely on the E cores when you have other demanding stuff running, and P cores when it needs to do something more intensive and you're most idling.

[Leaffy]

11 minutes ago, igormp said:

Windows itself is composed of many services, and a process doesn't need to stay stuck onto a single core, it can move around cores, so, answering your questions:

As said above, in any core it feels like to, there's a part of the OS called the scheduler which is responsible for that.

Likely on the E cores when you have other demanding stuff running, and P cores when it needs to do something more intensive and you're most idling.

Yeap, dynamic. Scheduler's the job. Noted.

So, core services itself,(stuff related to drivers, kernel) will be moved around into any core it feels? Stability issue? 

OR, lets look this way,                          

Initially, as windows boots up(fresh install), won't there be a default value; core(P or E), right?  cache allocation for E and P?

                                                                                             ;Cant be random, unlike core clock, as long not 0hz.

 

 

You said likely E cores for this(thought the same, windows might not require lots resource, but i've no education on computers).

 

 

Even on current gens, i was told not all cores were created equally. Think physically, windows gonna use one of the core a lot more.

Let say there is 2 core, A(left) and B(right).

When we press that start button, and it starts to process electrical signal, it can't be one times its on A and the next time on B. Or, am wrong?

I heard the first will be made to withstand higher single core boost.

 

Finally, thanks if you have the patience. I'm just curious and trying to understand.

 

 

 

 

 

[igormp]

4 minutes ago, Leaffy said:

So, core services itself,(stuff related to drivers, kernel) will be moved around into any core it feels?

Yes, pretty much.

4 minutes ago, Leaffy said:

Stability issue? 

Not really, you can always store and load back the register's contents between context switches.

5 minutes ago, Leaffy said:

Initially, as windows boots up(fresh install), won't there be a default value; core(P or E), right?

Right off the boot, yes, there's a core that will be initialized first, then you'll jump into 32-bit mode, initialize other cores, jump again to 64-bit, yadda yadda, by the point you get to the login screen all of the stuff is already loaded and that's irrelevant.

6 minutes ago, Leaffy said:

Think physically, windows gonna use one of the core a lot more.

Only if there's any heuristic to tell windows to do so. Intel's patch to Windows' scheduler gives it hints as to when it should load stuff onto the E or P cores.

7 minutes ago, Leaffy said:

When we press that start button, and it starts to process electrical signal, it can't be one times its on A and the next time on B. Or, am wrong?

Oh boy, you're into a really deep rabbit hole here. 

 

Go into the 3rd volume here, chapter 8, section 4: https://cdrdv2.intel.com/v1/dl/getContent/671200

 

It goes into detail explaining  exactly what you're asking:

[Leaffy]

18 hours ago, igormp said:

Yes, pretty much.

Not really, you can always store and load back the register's contents between context switches.

Right off the boot, yes, there's a core that will be initialized first, then you'll jump into 32-bit mode, initialize other cores, jump again to 64-bit, yadda yadda, by the point you get to the login screen all of the stuff is already loaded and that's irrelevant.

Only if there's any heuristic to tell windows to do so. Intel's patch to Windows' scheduler gives it hints as to when it should load stuff onto the E or P cores.

Oh boy, you're into a really deep rabbit hole here. 

 

Go into the 3rd volume here, chapter 8, section 4: https://cdrdv2.intel.com/v1/dl/getContent/671200

 

It goes into detail explaining  exactly what you're asking:

Thank you very much.