notebook with lunar?

[Zbigniew Luszpinski]

Friday 07 of March 2008 13:58:40 Jean-Michel Bruenn wrote:
> > k8-temp is kernel module, govenors also are configurable in kernel - I
> > choose ondemand as default one. You can use cpufrequtils lunar module to
> > control govenors by hand.
> >
> > I can send you my kernel config for AMD to take everything the factory
> > gives.
> >
> > My last Intel was Pentium75MHz. Then I used AMD Duron/AthlonXP/Athlon64.
> > All work great. The cool feature of AMD since AthlonXP/64 you can have
> > both 3dnow/sse together and C'n'Q which makes CPU cold (my machine have 1
> > fan).
>
> Yep, please mail me your kernel config. I haven't tried arround with
> cpufrequtils right now, but k8-temp is in kernel and seems it's working,
> dmesg is showing some things ;-)

I will mail config when come back home.

cpufrequtils are easy:

$ cpufreq-info
reads present state:
analyzing CPU 0:
  driver: powernow-k8
  CPUs which need to switch frequency at the same time: 0
  hardware limits: 1000 MHz - 1.80 GHz
  available frequency steps: 1.80 GHz, 1000 MHz
  available cpufreq governors: ondemand, performance
  current policy: frequency should be within 1000 MHz and 1.80 GHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 1000 MHz.

# cpufreq-set -g performance

(which always must be run as root) on my machine changes default ondemand 
govenor to performance. However since I started to use ondemand govenor I do 
not use performance govenor anymore (and will remove any governor except 
ondemad on next kernel update).
The ondemand behaviour can be very good observed by lining big module. At idle 
CPU time check cpufreq-info to see it displays current CPU freq: 1000 MHz. 
Then lin something big and on second console during lin run cpufreq-info to 
see that CPU freq jumped to maximum (1800 MHz on my 3000+ CPU).
After lining the CPU freq will go down to 1000 again waiting for another big 
job.

k8temp module is accessible via lm_sensors-2.6 app or ksensors not kernel logs 
or something else. You only should take care of loading k8temp module before 
running sensor reader app (I have k8temp compiled in kernel so do not care 
about modules).

$ sensors
k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp:
             +32°C

(this is for my Athlon64 3000+ Venice)

> > activate vesafb console in kernel. Then add to kernel boot parameters:
> > vga=773 video=vesafb:mtrr:4,ywrap
> > (vga=773 is 1024x768x256 resolution which is perfect for me; you can pick
> > up other vesa mode). Do _not_ use nvidiafb in kernel because you will not
> > boot to X (open nvidiafb vs proprietary nvidia X driver conflict).
>
> ah, okay, ty.

Here is doc how to calculate mode number:
/usr/src/linux/Documentation/fb/vesafb.txt
On my Geforce 6150 the 256 color (8bpp) looks to be faster than 64k or 16M 
colors modes. To my surprise 4bpp modes do not work.
If you do not watch movies on vesa console without having/using X there is no 
need for using more than 8bpp modes.

I got reports that since Geforce7 there are some card manufacturers who use 
stripped video bios where vesa was cut here and there resulting in black 
screen instead of vesafb console. Picking up card with not stripped video 
bios is pure luck. If your video bios is smaller than 63kb (like 58 or 61 kb 
size) it is stripped. Nobody except manufacturer know what was cut so on 
smaller video bioses the vesafb may work or may not.

> > HDA is integrated in nForce 630. Use alsa hda driver.
> > options snd-hda-intel model=acer
>
> i haven't tried model=acer right now - and i don't have much experience
> with modules where do i set this options? (normally i have everything i
> need statically in my kernel heh)

/etc/modules.d/alsa
But I do not know if this is place recommended by Lunar fathers. :-)
Any Lunar daddy comment on this? :-D

zbiggy