Talk:CPU frequency scaling

From ArchWiki
Revision as of 00:29, 26 January 2011 by Thestinger (talk | contribs)
Jump to navigation Jump to search

Hi, I would like to have some feedback regarding my addition to this page: Cpufrequtils#Cpufrequtils and Laptop Mode Tools.

Thank you.


Laptop mode

I think it's good. What we need, at least in my opinion, is to make this page a bit more pedagogic. As for now I see users becoming confused as to what module(s) they need and how to set it up. A user in the forum misunderstood this wiki page and struggled because he/she did add practically all possible modules to rc.conf.

p4-clockmod + ondemand governor issues

Hello, I'd like to point out that there's an issue with the p4-clockmod driver combined with the ondemand governor, as seen here:

In short, the advice given in this wiki article isn't actually working(anymore) and for a newcomer(like me) there isn't really a working substitute(as in re-compiling the kernel and modules).


I found that acpi events were not processed until I installed acpid and added that to the daemon list (I removed laptop-mode).

Move to cpufrequtils

I'm not sure why this page that specifically covers the cpufrequtils package was moved from cpufrequtils to this more generic title. The article title should match the package/utility described within, in my opinion. -- pointone 18:41, 25 January 2011 (EST)

The page contains incorrect information right now - it confuses the kernel's cpufreq with the userspace cpufrequtils package. All of the stuff on the page about drivers/modules, governors, /sys, etc. is independent of cpufrequtils. cpufrequtils provides a daemon script to set the governors and 3 userspace utilities, which offer similar functionality to the /sys filesystem and powertop. The two methods on the page for using laptop-mode-tools and acpid to set the governor are completely independent of cpufrequtils, the page just incorrectly refers to cpufreq as the userspace package. I moved the page here after someone resurrected the old version that got merged into this article, because they were right that cpufrequtils != cpufreq, but there was also no reason to have 2 pages describing the same stuff. thestinger 19:29, 25 January 2011 (EST)