Are there ANY working nVidia drivers for Quantal?

[sixonetonoffun]

Have you made any progress? There is more then one way to install the most recent driver Ubuntu doesn't always have the one I need. The manual install is not difficult to do. Simply follow the guide posted above choose the option your most comfortable with.

[Spadge]

I see there's some new Nvidia drivers out - gtkperf gives me: Total time: 5.22

I'm calling them "good".

[bamyasi]

Have you made any progress? There is more then one way to install the most recent driver Ubuntu doesn't always have the one I need. The manual install is not difficult to do. Simply follow the guide posted above choose the option your most comfortable with.

Unfortunately, I did not. And yes, I have tried each and every version of the Nvidia binary blob available in the standard repositories as well as in x-updates PPA (I've already wrote about this) but to no avail. The only thing I haven't tried yet is manual install from Nvidia-provided tarball, which I do not want to touch. Maintaining several desktops with maually installed core system drivers isn't exactly what I wanted ending up with.

I guess my options now is to wait for next driver version to be released by x-updates team and in the meantime, I'll try to find another Nvidia card and test it with the current drivers to see if the issue is specific to this particular Gigabyte model we have in all those new destops. I will post any feedback here as soon as I have it.

Thank you all who responded and offered help and advice.

[bamyasi]

I see there's some new Nvidia drivers out - gtkperf gives me: Total time: 5.22

I'm calling them "good".

Which version?

[SteveRiley]

I don't have my nVidia-equipped laptop with me at the moment, but I'll try to reconstruct a few ideas from memory.

For a long time, I was unable to use nVidia's binary blob on my ThinkPad T520 because it had problems with many Optimus-equipped devices even when Optimus was disabled. Eventually that problem disappeared somewhere around the arrival of the 304.xx series. With Optimus firmly disabled, the thing finally installed. However, X kept falling back to VESA, or would sometimes fail to load altogether after I logged into the display manager.

Eventually, I discovered that the nVidia driver can get flummoxed by the presence of RandR configuration files. The Display and Monitor settings applet creates the file ~/.kde/share/config/krandrrc. I logged out of KDE, logged into a TTY, deleted that krandrrc file and any leftover /etc/X11/xorg.conf* files, and rebooted. The nVidia logo flickered briefly, I logged in, and -- woo hoo! -- everything started working.

At this point, I was able to use the nVidia settings utility to properly configure native resolutions of the two external digital monitors attached to my docking station. This procedure wrote a correctly formatted xorg.conf. Furthermore, I was pleased to see that the driver will enter a detection mode if the probed monitors fail to match xorg.conf. I notice this when I boot the laptop when not docked. The laptop's LCD operates at native resolution, and when I look through log files, I can see evidence of the detection mode being triggered.

All in all, I'm now very happy with the way it's working. And even though I'm using a laptop, not a desktop, since I've disabled Optimus, I'd expect that the desktop configuration experience would be similar. Perhaps you can try the same procedure I've described here.

Oh, and I use the Xorg Edgers PPA for drivers, not X-Swat.

[Spadge]

Which version?

It says "313.09-0ubuntu1~xedgers~quantal1"

[bamyasi]

SteveRiley: It looks like "every unhappy family is a different story" here since I do not have neither krandrrc nor xorg.conf files on my system. But they might appear in the future since I am going to continue palying with my X.org & KDE desktop configurations. So I'll make sure to write this down just in case, thanks.

I might be onto something else with my desktop issues, however. Today, I replaced the video card with another one, a completely different Nvidia chipset/model/make (ASUS GeForce GT 520 PCI) and yet absolutely nothing changed, same choppy destop behavior. Which made me think it's not a video card/driver problem per se. Examining dmesg output on this system I quickly discovered the inafmous "irq 16: nobody cared" error, followed by a kernel trap. Which gave me tons of google search hits, tracing back all the way to 2002. Looking at /proc/interrupts on all 4 systems showed that Nvidia cards were not using MSI interrupt handling mode at all and instead attempted to share APIC IRQ 16 with the USB controller, which apparentely failed and crashed both kernel drivers.

I have forced MSI enable for Nvidia binary driver by creating /etc/modprobe.d/nvidia.conf file with the following line in it:

options nvidia_current NVreg_EnableMSI=1

After a reboot, MSI interrupts were successfully enabled for the video card and "irq 16: nobody cared" errors vanished from the logs. I made the modprobe configuration changes remotely via ssh from my home computer, so the visual results will be seen tomorrow, as soon as I get to my lab office. But I am pretty certain that I am on the right track at this point and the whole debacle was due to a BIOS PNP/PCI resources configuration bug. I have also downloaded the latest BIOS update for the motherboard and planning to give it a try tomorrow as well.

BTW, BIOS configuration interface on these boards is a total disaster: all bells and whistles with 3D animated graphics and whatnot. It's good there is at least a Classic View button which disables all this crap but it took me some time to locate it because of the initial shock ;-).

[bamyasi]

It says "313.09-0ubuntu1~xedgers~quantal1"

Whoa, I haven't seen this version on the Nvidia web site yet. Looks like it arrived just yesterday. Maybe I should give it a try. Thanks, Spadge.

[gjringo]

For nvidia-graphics-drivers 313.09-0ubuntu1~xedgers~quantal1
Go here:https://launchpad.net/~xorg-edgers/+archive/ppa
and do this:This PPA can be added to your system manually by copying the lines below and adding them to your system's software sources.
Display sources.list entries for:

deb http://ppa.launchpad.net/xorg-edgers/ppa/ubuntu quantal main
deb-src http://ppa.launchpad.net/xorg-edgers/ppa/ubuntu quantal main

[Arran]

I am working with the «nouveau» drivers to my full satisfaction on 12.04 and 12.10. I have to admit, that I have no gags like turning dice of windows falling to part when closing. So, 3-D is activated but probably not much used.

Why don't you try the nouveau and forget about the nvidia driver?

[bamyasi]

I am working with the «nouveau» drivers to my full satisfaction on 12.04 and 12.10. I have to admit, that I have no gags like turning dice of windows falling to part when closing. So, 3-D is activated but probably not much used.

Why don't you try the nouveau and forget about the nvidia driver?

I did try nouveau initially, as they were installed by default. With the desktop effects enabled (default set) it will crash every 10 seconds on random operations like clicking an icon on the panel. With desktop effects disabled completely, it's more stable but desktop still crashes every 30 minutes or so. Oh, and nouveau drivers are also very slow on these computers. Better than binary nvidia ones but still hardly usable.

[bamyasi]

I might be onto something else with my desktop issues, however. Today, I replaced the video card with another one, a completely different Nvidia chipset/model/make (ASUS GeForce GT 520 PCI) and yet absolutely nothing changed, same choppy desktop behavior. Which made me think it's not a video card/driver problem per se. Examining dmesg output on this system I quickly discovered the infamous "irq 16: nobody cared" error, followed by a kernel trap. Which gave me tons of google search hits, tracing back all the way to 2002. Looking at /proc/interrupts on all 4 systems showed that Nvidia cards were not using MSI interrupt handling mode at all and instead attempted to share APIC IRQ 16 with the USB controller, which apparently failed and crashed both kernel drivers.

I have forced MSI enable for Nvidia binary driver by creating /etc/modprobe.d/nvidia.conf file with the following line in it:

options nvidia_current NVreg_EnableMSI=1

After a reboot, MSI interrupts were successfully enabled for the video card and "irq 16: nobody cared" errors vanished from the logs. I made the modprobe configuration changes remotely via ssh from my home computer, so the visual results will be seen tomorrow, as soon as I get to my lab office. But I am pretty certain that I am on the right track at this point and the whole debacle was due to a BIOS PNP/PCI resources configuration bug. I have also downloaded the latest BIOS update for the motherboard and planning to give it a try tomorrow as well.

Ok, mystery solved. This whole can of worms appear to be caused by a broken USB 2.0 controller implementation on the motherboard and/or BIOS bugs in handling legacy USB 2.0 support. Possibly also a kernel bug or more precisely, kernel inability to work around several bugs in particular USB hardware. As a result, USB 2.0 driver would crash during kernel initialization, which caused (among other nasty things) kernel to disable IRQ processing for it. PNP BIOS would initially configure Nvidia driver to share the same IRQ with USB 2.0; when USB 2.0 driver crashed it would bring down video IRQ processing with it. The kernel would then switch to IRQ polling, causing sever performance degradation. After a) enabling MSI interrupts handling for nvidia driver as described above and b) re-connecting USB keyboard and mouse to the USB 3 ports instead of USB 2.0 ones KDE desktop now works as expected, either with or without 3-D effects enabled, and using nvidia_current drivers from X-Updates PPA (version 304.64).

Hope this information would be useful for others with the similar hardware:

Motherboard: Intel Desktop Board DZ77BH-55K
BIOS version: BHZ7710H.86A.0094.2012.1101.1115 11/01/2012

[Spadge]

Wow.

Good work figuring that one out.

Sounds like a total mare.