Re: 2.6.12-rc4 lots faster than any kernel before?
OK Matthias: I owe you a Pizza, family-size (at least ... :)
On Sun, May 15, 2005 at 01:41:56PM +0200, Matthias Grimm wrote:
> On Sun, 15 May 2005 13:07:58 +0200
> Wolfgang Pfeiffer <roto@gmx.net> wrote:
>
> Hi,
>
> > > >>Am I the only who got impressed?
>
> After all this cheering I wanted to see it with my own eyes. So I installed
> 2.6.12-rc4 last night. What should I say: My machine behaves as slow as before.
> I can't see any improvement in speed.
>
> On the other hand all my hardware seems to run out of the box including ALSA
> and sleep. Ok, on a G3 Pismo sleep haven't been a problem since ages ;-)
>
> > But as I said: I have problems with ALSA (no sound so far) and with
> > pbbuttonsd: I enabled userspace Power Management in the .config, and
> > I'm not sure yet why the speed on the machine (Titanium IV, 867 MHz)
> > is set to ~665 MHz after booting the machine.
>
> Dynamic CPU frequency scaling should only be done by the kernel itself.
> Pbbuttons changes the CPU speed if the power profile changes only and
> the external script 'cpufreq' must be active for this to work. So please
> check if you have a link in /etc/power/event.d called 'cpufreq' linked
> to /etc/power/scripts.d/cpufreq.
No, IIRC didn't have this link ... :)
And does, what you write, mean that at boot-time pbbuttons can't set
the CPU frequency? Please see the test reports, reported below, for more ..
> Because my PowerBook doesn't support frequency scaling, I never
> tested this script sufficiently. I would apprecieate your feedback if you got
> it work.
It works. :))))
After I did that:
# ln -s /etc/power/scripts.d/cpufreq /etc/power/event.d/cpufreq
# kill -HUP `cat /var/run/pbbuttonsd.pid`
Tests:
I unplugged the power-plug connected to the machine, and it switched
back from 867MHz to 667 MHz:
Before:
:$ cat /proc/cpuinfo | grep -i mhz
clock : 867MHz
After pulling the power-plug:
$ cat /proc/cpuinfo | grep -i mhz
clock : 667MHz
I remove the link again and see whether it still works:
# rm /etc/power/event.d/cpufreq
rm: remove symbolic link `/etc/power/event.d/cpufreq'? y
removed `/etc/power/event.d/cpufreq'
root@ 17:33:52:# kill -HUP `cat /var/run/pbbuttonsd.pid`
$ cat /proc/cpuinfo | grep -i mhz
clock : 867MHz
Pulling the power-plug:
:$ cat /proc/cpuinfo | grep -i mhz
clock : 867MHz
Reinserting the link fixed it:
# ln -s /etc/power/scripts.d/cpufreq /etc/power/event.d/cpufreq
root@ 17:37:19:# kill -HUP `cat /var/run/pbbuttonsd.pid`
Then:
pulling the power-plug:
$ cat /proc/cpuinfo | grep -i mhz
clock : 667MHz
reinserting it:
$ cat /proc/cpuinfo | grep -i mhz
clock : 867MHz
MAXIMUM SPEED AFTER REBOOT?
And I'll reboot the machine in a minute to see, whether pbbuttons will
be able with the new settings to automatically set this machine to
maximum speed (Maximum, because it will be connected to the
power-adapter). Until later ...
[Minutes later:]
It didn't work:
/var/log/syslog:
May 15 17:03:53 debby pbbuttonsd: INFO: Script
'/etc/power/pmcs-pbbuttonsd performance ac ' lauched but killed after
4 seconds
And then, after booting:
cat /proc/cpuinfo | grep -i mhz
clock : 667MHz
What's that? Is this why you wrote: "Pbbuttons changes the CPU speed
if the power profile changes only"? So setting
CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE
is the only way to set the CPU to Maximum speed at boot time? (besides
some boot script with something like
cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq > /sys/devices/system/cpu/cpu0/cpufreq/scaling_setspeed
in it? ... but the latter's not what I want, I think ...)
But pulling the power-plug and reconnecting it still works:
$ cat /proc/cpuinfo | grep -i mhz
clock : 867MHz
> Another interresting fact: Up to now I haven't got high cpu loads (100%)
Either me: I never realised any unusual high CPU loads with
2.6.12-rc4. So far. And 99% of my time on this machine I'm on X. With
FVWM.
> with pbuttonsd and kernel 2.6.12 as reported multiple times on this
> list. Maybe it has something to do with hardware components I don't
> have.
Here:
$ cat /proc/cpuinfo
processor : 0
cpu : 7455, altivec supported
clock : 867MHz
revision : 0.2 (pvr 8001 0302)
bogomips : 865.18
machine : PowerBook3,5
motherboard : PowerBook3,5 MacRISC2 MacRISC Power Macintosh
detected as : 80 (PowerBook Titanium IV)
pmac flags : 0000001b
L2 cache : 256K unified
memory : 768MB
pmac-generation : NewWorld
:# lspci
0000:00:0b.0 Host bridge: Apple Computer Inc. UniNorth 1.5 AGP
0000:00:10.0 VGA compatible controller: ATI Technologies Inc Radeon R250 Lf [Radeon Mobility 9000 M9] (rev 01)
0001:10:0b.0 Host bridge: Apple Computer Inc. UniNorth 1.5 PCI
0001:10:17.0 ff00: Apple Computer Inc. KeyLargo Mac I/O (rev 03)
0001:10:18.0 USB Controller: Apple Computer Inc. KeyLargo USB
0001:10:19.0 USB Controller: Apple Computer Inc. KeyLargo USB
0001:10:1a.0 CardBus bridge: Texas Instruments PCI1410 PC card Cardbus Controller (rev 02)
0002:24:0b.0 Host bridge: Apple Computer Inc. UniNorth 1.5 Internal PCI
0002:24:0e.0 FireWire (IEEE 1394): Lucent Microelectronics FW323
0002:24:0f.0 Ethernet controller: Apple Computer Inc. UniNorth GMAC
(Sun GEM) (rev 01)
# modprobe airport -v
insmod /lib/modules/2.6.12-rc4-selinux1/kernel/drivers/net/wireless/hermes.ko
insmod /lib/modules/2.6.12-rc4-selinux1/kernel/drivers/net/wireless/orinoco.ko
insmod /lib/modules/2.6.12-rc4-selinux1/kernel/drivers/net/wireless/airport.ko
and mounting an outside-disk via firewire also didn't hurt, IINM:
top - 18:05:24 up 1:02, 7 users, load average: 0.01, 0.03, 0.00
Tasks: 81 total, 1 running, 80 sleeping, 0 stopped, 0 zombie
Cpu(s): 1.3% us, 0.7% sy, 0.0% ni, 98.0% id, 0.0% wa, 0.0% hi, 0.0% si
A bit later:
top - 18:16:01 up 1:13, 7 users, load average: 0.05, 0.04, 0.01
Tasks: 88 total, 1 running, 87 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0% us, 0.3% sy, 0.0% ni, 99.7% id, 0.0% wa, 0.0% hi, 0.0% si
HTH
Best Regards
Wolfgang
--
Wolfgang Pfeiffer
http://profiles.yahoo.com/wolfgangpfeiffer
Reply to: