lkml.org 
[lkml]   [2002]   [Aug]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH][2.5.32] CPU frequency and voltage scaling (0/4)
Helge Hafting wrote:
>
> Dave Jones wrote:
> >
> > On Fri, Aug 30, 2002 at 10:04:20AM +0200, Helge Hafting wrote:
> > > An MHz carries more meaning - it is a measurable frequency.
> >
> > It's equally meaningless (in fact, less meaningful).
> > - By your definition my 900MHz VIA C3 is faster than my 800MHz Athlon.
> > (Clue: It isn't).
> I never said such a thing!
> You are right that MHz is useless for telling which
> processor is the fastest. But this discussion wasn't about
> comparing performance.
>
> It was about:
> Should we tell the kernel to run a cpu at "500MHz", or
> "50% of max" in order to (save power|avoid overheating|whatever).
>
> In this case MHz is useful - because that's what the manufacturer
> specifies. That's what you program into cpu or
> motherboard registers, and MHz is what you can measure with an
> oscilloscope in order to verify correct operation of the driver.
>
> Percentages don't buy you anything if you replace
> the cpu with a different one. The other cpu may of course
> have different MHz ratings for "full speed" and "power save|cold
> running"
> but the percentages may very well be different too.
> Some runs cool at 80%, some at 60%...
>
> finally - "full speed" is ill-defined. Some AMD chips have different
> speed ratings for different operating temperatures.
>
> So, I think MHz is the better choice for setting up
> speed policies for cooling and power saving.

If cooling and power saving is what is wanted, why not talk
in terms of degrees or watts? Or are you really trying to
say something like "it is ok to run this much slower to
conserve power/temp/battery".

I really think we need to push down the power/temp to Mhz
conversion to the lowest level. At the user level we should
only be expressing what we want and/or are willing to give
up for it. From this point of view, if you are talking
frequency you are already out of the box.

-g
>
> > - With trickery like AMD's quantispeed ratings, MHz really is a totally
> > meaningless number when relating to performance of a CPU.
> > - A MHz rating is only meaningful across the same vendor/family of CPUs.
>
> This is all fine for the purpose of comparing cpu's, but this
> isn't about such comparisons. I would never compare an
> intel and an amd chip based on frequency, I'd look at how
> well they perform what I want them to do.
>
> > Getting cpufreq's policy interface into something CPU agnostic therefore
> > precludes MHz ratings AFAICS.
> Why? It is not as if cpufreq is being used to tell who
> has the faster machine...
>
> Helge Hafting
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/

--
George Anzinger george@mvista.com
High-res-timers:
http://sourceforge.net/projects/high-res-timers/
Preemption patch:
http://www.kernel.org/pub/linux/kernel/people/rml
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:22    [from the cache]
©2003-2011 Jasper Spaans