Switch on/off nvidia card Asus UL30Vt, UL50V, UL80V laptops

[ekarasik]

The acpi_call module works on 10.04 as well. Actually, on 10.04 it worked even with nouveau running, but it may cause issues. So, basically, the steps for 10.04 should be the same. If you run into any problem, feel free to ask for help

[Crysler]

Yes, It works. You can follow these steps: (I'm just study from this thread)

1st step (from @ekarasik):

echo options nouveau modeset=0 | sudo tee -a /etc/modprobe.d/nouveau-kms.conf 

sudo update-initramfs -u

2sd step:


look here for more detail instructions

After this step, you'll have a "acpi_call.ko" and "test_off.sh" (You must fix nouveau as in step 1, If you didn't this module does not work).

3rd step:

sudo cp acpi_call.ko /lib/modules/`uname -r`/kernel/

sudo gedit /etc/modules

Add this line after "lp" line: (as in my case)

acpi_call

------------------------------------------------------------------------
and go to terminal type in:

sudo depmod
sudo modprobe acpi_call

(Sorry, I've just checked and I miss this part, that is the reason the module can't load every time machine start.)
------------------------------------------------------------------------

This step help us auto load "acpi_call" module every time we boot our box.

4th step:


Make sure "test_off.sh" will call every time we boot. You can go to:

System > Preferences > Startup Applications > Startup Program > Add

Name: whatever you want.
Command: Browse to "test_off.sh"
> Add > Close.

Done. Work like a charm.

MNCvn

Hello! I'm a complete noob in Ubuntu, and I want to shut down my discrete Nvidia graphics on my Asus UL80VT. I tried using this method, and I'm stuck at this part:

  sudo cp acpi_call.ko /lib/modules/`uname -r`/kernel/

It says

 radu@radu-UL80VT:~$ sudo cp acpi_call.ko /lib/modules/2.6.35-22-generic/kernel/
cp: cannot stat `acpi_call.ko': No such file or directory

Any help ?

[mirko_r]

sorry, my error, looks like I lacked consistency with MNCvn steps. I corrected that now.

Replace it to

#!/bin/sh

case "$1" in
        resume|thaw)
        /lib/modules/`uname -r`/kernel/acpi_call/test_off.sh
        ;;
esac

(I just placed the module on a diffrent place on my PC, because with MNCvn you will have to copy the module to every new kernel you install, and I tend to test development kernels, so this is quite frasturating for me)

I hope this one will work

Hi,

thanks for the response.

I replaced the code in the 99_nvidia file but it doesn't work either. The module is loaded, but the ./testff.sh isn't executed when I return from suspend mode, I think.
When I execute test_off.sh in /lib/modules/`uname -r`/kernel/acpi_call/ manually, everything works fine.

Do you have any further suggestions what went wrong?

@Crysler: is it possible that you're in a wrong directory when you try to copy the file, i.e. you downloaded it to "Downloads" and try to copy it from /home/user/?

[cobra11]

Can someone pleasse fix it that it will work when come from suspend?
This code doesnt work..

#!/bin/sh

case "$1" in
        resume|thaw)
        /lib/modules/`uname -r`/kernel/acpi_call/test_off.sh
        ;;
esac 

[cobra11]

I fixed it it works from hibernation or from suspend. Acpi_call script needs to be copied to /lib/modules/`unaame -r`/kernel/acpi_call/ than you need to create script in /etc/pm/sleep.d/ which is called 99-nvidia-off.sh and add code:

#!/bin/sh 

case "$1" in 
        resume|thaw) 
        /lib/modules/`uname -r`/kernel/acpi_call/test_off.sh 
        ;; 
esac

[christianco]

EDIT: It's the BIOS - I had an early version #204 flashed. Upgrade to 211 and it works great.

-----------------------------

there is a lot in here about shutting the nvidia off, but my trouble is that I can't enable it with nvidia drivers in 10.10/32 bit (ul30vt).

When I install nvidia hardware drivers, then enter "compatibility" mode in the bios, and then boot, X server just hangs. startx from tt1 errors out: device not found. the same is true if I run dpkg-reconfigure xserver, or nvidia-xconfig to generate xorg.conf.

now here's the weird part: lspci | grep VGA reports BOTH cards active. Definitely I am using compatibility mode, so this is a big mystery to me that I haven't seen on any of the forums.

So when i delete my xorg.conf file and starx, viola, i get a full desktop. I assume that the intel card is being used by xserver, but I'm not sure how to tell exactly. Compiz can't be enabled, but Additional Drivers shows the nvidia as active. very confusing to me.

Anyway I have been hacking away at this problem for a long time now and have no clue how to fix it. All the wikis and forums just say that compatibility mode should just work with nvidia drivers enabled. Anyone else have the experience I am having or know how to fix?

thx

[Holymoly]

can someone elaborate on step 2? the part on acpicall module, im just not sure how im suppose to install it, i guess im suppose to make apci_call.ko and test_off.sh appear in /proc/acpi/call, but i cant make a folder there, sooooo...

im pretty new to linux, just installed ubuntu 10.10x64, im assuming that i should follow MNCvn and ekarasik's steps right? sry its just the steps all posted all over the thread idk which one works lol

[TheMalinowski]

i have also some problems with step 2
i'm getting an error while processing "make"

root@mali-netbook:~/nvg# make acpi_call
cc     acpi_call.c   -o acpi_call
acpi_call.c:1: fatal error: linux/module.h: No such file or directory
compilation terminated.
make: *** [acpi_call] Fehler 1

(10.10 64bit on ASUS UL80VT)

[yanf.jiang]

I followed the MNcvn's protocol (page 10), it really worked. But I want to share some unseccessfull experience with vga_switcheroo:
http://asusm51ta-with-linux.blogspot.com/
https://help.ubuntu.com/community/HybridGraphics

(1) you might find there's no vgaswitcheroo folder in /sys/kernel/debug. To solve this, first, do not use nvidia driver, if you have installed it, just uninstall it. second: do not prevent nouveau from running. I mean in this acpi_call method, if you are using 10.10, you have to disable nouveau, see(page 9 and 10), but disabling it will cause the absence of vgaswitcheroo folder, although lspci |grep VGA still shows two cards.

(2)
echo ON > /sys/kernel/debug/vgaswitcheroo/switch
echo OFF > /sys/kernel/debug/vgaswitcheroo/switch
these two commands do change the the status shown in the switch file, however, the OFF doesn't really power it off if you look at the power usage.

echo DIGD > /sys/kernel/debug/vgaswitcheroo/switch
echo DDIS > /sys/kernel/debug/vgaswitcheroo/switch

as default, the computer will use the intel card. However, echo DDIS never works even if you logout and login.
if you echo DIGD and logout, the system will freeze. you have to turn it off and restart it.


I am pretty new with linux. that's all I have tested.
I hope someone can help me out with this. It is really nice if we can freely switch between them.

[didymos88]

Can someone pleasse fix it that it will work when come from suspend?
This code doesnt work..

#!/bin/sh

case "$1" in
        resume|thaw)
        /lib/modules/`uname -r`/kernel/acpi_call/test_off.sh
        ;;
esac 

Code is fine, you must be sure of three things:
- there must be test_off.sh file in that directory
- test_off.sh must have executable privileges
- /usr/lib/pm-utils/sleep.d/99_disable_nvidia must have executable privileges also!!!

If everything will be fine, after invoking:

cat /proc/acpi/battery/BAT0/state

you will see present rate around 8000 mW instead of 12000 mW

[edit] everything works on Ubuntu 10.10 64bit