Re: [PLUG] CPU Scaling

[Keith via plug <plug@lists.phillylinux.org>]

On 12/13/21 1:28 AM, Syeed Ali via plug wrote:
> On Mon, 13 Dec 2021 00:33:38 -0500 (EST)
> "Keith C. Perry via plug" <plug@lists.phillylinux.org> wrote:
>
> > Walt, here's the kernel documentation link...
> >
> > https://www.kernel.org/doc/html/v4.14/admin-guide/pm/cpufreq.html
> >
> > As Rich said, this has been around for a long time now. Typically
> > you're not going to set frequencies directly and instead just change
> > the scaling governor to "performance" if you want to max things out.
> I've had experience with the "performance" setting, and I can confirm
> that it turns your computer into a space heater.
I had the same experience previously but it nice that see if you need 
and want to max out your system, it can be done rather easily.
> It did grant a significant performance boost compared to the "ondemand"
> governor setting when I was desperate for extra frames per second when
> using wine many years ago on underpowered hardware.
>
> I would not recommend ever using that setting since it's the
> CPU-equivalent of crypto-mining with your GPU; it's constantly
> redlining and stressing hardware.
I used to HAVE to do this when using my old Turion 64 x2 laptop when I 
was doing studio work.  In particular, when doing any audio work that 
needs to be "sample accurate".  I would have to max out CPU AND go into 
real time kernel scheduling.  The Turion could achieve that but it was 
pretty much the minimum hardware spec I came across that would do what I 
needed in Linux.  There were still times when I hard to reboot to start 
clean to get the hardware to cooperate but it gets there.  I'm pretty 
sure this was right around the time cgroups were becoming pretty solid 
but I switched to an FX-8350 system by then so I didn't really play 
around with RT scheduling to a group (which I think would have been less 
taxing than pressing the entire system into real time scheduling).
> -
>
> As an aside, the OP may be interested in investigating the scheduler
> being used:
>
> https://www.kernel.org/doc/html/v5.5/scheduler/index.html
+1 on this.  Other than needed a real time environment, I haven't found 
much difference but I should probably play around with in again.  
There's some weird things I see that can reliable occur on my systems 
that I'm wondering if changing the system or I/O scheduler (i.e. disk) 
might resolve.  It is pretty slick Linux users can play with these 
things to really get their system (kernel) dialed in.
> ___________________________________________________________________________
> Philadelphia Linux Users Group         --        http://www.phillylinux.org
> Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
> General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug

--
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
Keith C. Perry, MS E.E.
Managing Member, DAO Technologies LLC
(O) +1.215.525.4165 x2033
(M) +1.215.432.5167
www.daotechnologies.com

___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug