Friday, 27 August 2010

Howto Linux Asus U35JC hybrid graphics

Linux Hybrid Graphics: Using acpi_call module to switch on/off discrete graphics card in Linux
Asus U35JC-RX005X with ubuntu lucid lynx:
echo '\_SB.PCI0.PEG1.GFX0._OFF' > /proc/acpi/call

The battery using rate decrease from 4200mW :

echo '\_SB.PCI0.PEG1.GFX0._ON' > /proc/acpi/call
grep rate /proc/acpi/battery/BAT0/state
present rate: 17808 mW
echo '\_SB.PCI0.PEG1.GFX0._OFF' > /proc/acpi/call
grep rate /proc/acpi/battery/BAT0/state
present rate: 13664 mW

Everything else worked fine under Ubuntu for this computer.

Wednesday, 25 August 2010

Asus U/UL series nvidia hybrid graphics and BIOS version

It's been reported that some of the issues with switching the nvidia card on/off the Asus U/UL series in Linux are problematic due to the BIOS version. In one case, updating from 203 to 210 solved the problem. Happy Linuxing...

Friday, 13 August 2010

HOWTO Sony Vaio Z Series (2010) and Ubuntu Maverick alpha-3

Re: [Sony-vaio-z-series] vpcz12 and maverick alpha-3
On Thursday 12 August 2010 15:49:06 Justin Karneges wrote:
> I tried booting from the latest Ubuntu alpha live CD. The initial menu
> sometimes appears, where you can choose to install or do a memory test,
> etc. However, more often than not it doesn't appear and a timeout will
> cause the system to go on ahead with booting the live desktop.

It seems if I press an F-key (e.g. F1, F2, or F3, etc), I can get the menu to
appear most of the time. There have been a couple of occasions when this
doesn't work but the majority of the time it does work. It might be a timing
thing. I think you have to press a key rather quickly.

> The initial menu has the wrong colors (pink-ish), as well as X once it
> loads. Under X, the color distortion is more severe.

If I press F6 and add xforcevesa and nomodeset to the end of the kernel
parameters string (so that the string ends looking like "-- xforcevesa
nomodeset"), color is fixed in X. I'm not sure if the "i915.modeset=0"
option suggested by Martin has any effect. It doesn't seem to for me.

However, X resolution remains incorrect (1024x768, stretched to 1600x900,
looks pretty terrible!). Also, I still can't say for sure if the physical
graphics switch does *anything*. I get the exact same results whether it is
set to speed or stamina at cold boot. I have a feeling this means that I'm
using the same card every time, no matter what I have selected with the
physical switch. At least, this feels more likely to me than both cards
having the exact same color distortions with the exact same kernel params
fix. I wonder which card it is using?

> Once X loads, the touchpad does not work.

Solved by adding the i8042.nopnp kernel param. I don't know if the touchpad
is getting used in its full capacity but it seems to work well enough, and I
can even slide along the side for a wheel effect.

> I can flip over to the VTs and back to X no problem. The VTs have the wrong
> colors but otherwise seem to work fine.

Once the desktop loads, using the above fixes for X, the VTs have the correct
colors.

> I tried suspending (via keyboard in X, mostly blindly) and the system
> appeared to go into suspend, but on wakeup I was greeted with a blank
> screen.

Maybe this was a graphics issue. With the above fixes for X, suspend and wake
now works. Strange behavior I noticed: when X restores after waking, the
screen is no longer stretched, and instead the 1024x768 desktop is centered
1:1 in the middle of the display. Flipping to a VT and back to X will cause
the stretching to restore.

Now that I can actually use the computer, I'll note that wifi also works.

> So that's the current "out of the box" experience with bleeding edge
> Ubuntu.

And just three added kernel parameters later, a usable Live CD experience!
Not too shabby.

Tuesday, 10 August 2010

Acer Timeline 3810T down to 5.8W with Linux

[ubuntu] Acer Timeline 3810T - Page 90 - Ubuntu Forums
Re: Acer Timeline 3810T
I just found another 1.2W of power saving trick in my 3810T, bringing my minimum power usage down from 7.0W to 5.8W.

Powertop used to show the Intel graphics chip as generating 60 interrupts/second:
51,3% ( 60,0) [i915@pci:0000:00:02.0]
But if I do a Suspend-to-RAM (Fn-F4) and wake it up again the 60Hz interrupt from i915 disappears and my power usage drops 1.2 Watts.

Now, the odd thing is that if I start a new instance of Powertop the 60Hz i915 interrupts returns and power usage goes up by 1.2W again. It seems that Powertop is probing the hardware and triggering the i915 60Hz interrupt.
The i915 interrupts is described on Powertop's homepage http://www.lesswatts.org/projects/powertop/known.php#intelgfx but the documented fix (setting "NoDRI" in xorg.conf) doesn't seem to do anything.

My 3810T is configured as follows:
Ubuntu 10.04 with kernel 2.6.32-24 and backports for wireless and alsa.
Compiz configured to minimum bells and whistles.
Core2 Solo CPU U3500@1.40GHz; Harddisk replaced with Intel SSD.
Fan disabled by disconnecting it from the motherboard. (I'm running a temperature monitor that will hibernate the laptop if it reaches 75°C).