On Thu, 25 Mar 2004 14:18:45 +0100 Mattia Dongili <[EMAIL PROTECTED]> wrote:
> On Wed, Mar 24, 2004 at 01:16:47PM +0100, paolo wrote: > > > How do you determine that? I mean which measurements have you tried to > state that frequency doesn't change? /proc/cpuinfo is often wrong, try > with /usr/bin/bogomips > > bye > -- > mattia > :wq! > Hodie octavo Kalendas Apriles MMDCCLVII ab urbe condita est > Thanks a lot the mistake was to don't understand that /proc/cpuinfo is broken! I tried with bogomips and when i set manually different speeds i have different values of bogomips. For my processor i have a mean factor of 1,2 between bogomips and the processor's avaible speed that i have set. So it's wrong the 2.0 factor of the bogomips mini-howto. Another consideration is that with the cpufreqd, bogomips fails with the follow message: Calibrating delay loop.. failed. (i believe because in the meantime the speed change due to bogomips) Now my last question is if the cpufreqd take in account /proc/acpi/ac_adapter/AC0/state, because in my laptop it is broken (it's always on-line). If it's true i have to take in account 8 point less in cpufreqd.conf (it's a really good job my congratulations to the developers! :-) Again thanks 1000 BogoMips! Ciao Paolo Ps see you at the debian-italian list
