tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: A simple cpufreq(9)

On Sat, Sep 24, 2011 at 07:53:47PM +0200, Joerg Sonnenberger wrote:
> It's not relevant what the exact clock rate is. It's an approximation.
> Just like the TSC frequency won't be measured the same on every boot.

It would be relevant if the interval would be guaranteed to be uniform.

> Ignoring intermediate values can be literally a lot of unwanted noise.
> On my old laptop, I couldn't play all medium quality H.264 streams at
> smallest CPU frequency. It worked with some of the intermediate levels
> and those create enough heat less, that it makes a difference in terms
> of fan activity. My point is that not every load is switches between
> idle and 100%.

Naturally. But given the lack of proper information, you end up doing crazy
guessing game with the steppings in order to see whether the frequency is at
a sustainable level w.r.t. the load. This is what the Linux governors are all
about. And Linux people are rewriting the ondemand governor since it really
doesn't work that well, even after all these years.

What a boolean gives you is: simplicity and a bias towards performance
(which I think should be the priority on NetBSD generally). This is traded
for minor power consumption increase and possibly heat. Should be fine for
servers and most laptop users.

And as far as x86 is concerned, the power savings from CPU are really coming
from C-states today. So one can debate whether the 4000 LOC complexity of
Linux's cpufreq subsystem is really worth the trouble.

- Jukka.

Home | Main Index | Thread Index | Old Index