Re: [RFC PATCH v1 0/3] Scaled statistics using APERF/MPERF in x86

[Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>]

* Arjan van de Ven <arjan@infradead.org> [2008-05-27 07:19:00]:

> \> > 
> > > that's a case where it really makes sense; it's the case where the
> > > thing that controls the cpu P-state actually learns about how much
> > > work was done to reevaluate what the cpu frequency should be going
> > > forward. Eg it's a case of comparing actual frequency (APERF/MPERF)
> > > to see what's useful to set next.
> > > IDA makes this all needed due to the dynamic nature of the concept
> > > of "frequency".
> > 
> > Scaled statistics relative to maximum CPU capacity is just a method of
> > exposing the actual CPU utilisation of applications independent of CPU
> > frequency changes.
> > 
> > Reason behind the metric is same as the above fact that you have
> > mentioned.  The CPU frequency governors cannot make decisions only
> > based on idle time ratio.  It needs to know current utilisation (used
> > cycles) relative to maximum capacity so that the frequency can be
> > changed to next higher level.
> > 
> > Higher level management software that wants to control CPU capacity
> > externally will need similar information.
> >
> I entirely understand that desire.

Good :)

> But you're not giving it that information!
> The patch is giving it a really poor approximation, an approximation
> that will get worse and worse in upcoming cpu generations.

I agree that power capping and acceleration makes the metric
approximate.  But was are trying to be as accurate and meaningful as
APERF/MPERF ratio is in the processor hardware.

Can I state the problem like this:  The metric is as accurate and
meaningful as APERF/MEPRF ratio, but the interpretation of the metric
is subject to the knowledge of power constraint or acceleration
currently in effect.

--Vaidy