Did The NVIDIA 295.40 Linux Driver Fall Off A Cliff?

[Gusar]

maybe with kernel 3.4 support.

Kernel 3.4 support is easy, all it requires is a small patch: http://www.nvnews.net/vbulletin/show...77&postcount=4. Out of the box support would be cool of course, but needing to apply a patch is not that big of a deal.

[Kano]

I know that the patch is small, but i did not get why 295.40 still required this hack.

[thefirstm]

8600m GT here. I have no performance or stability issues. Seems about the same as the 295.33 driver.

[birdie]

All the machines I maintain work fine on 295.40, 295.33 so this is beginning to look a bit like more AMD clownery.

At first I wanted to say f*ck you, but then I decided to give you several links:

http://www.nvnews.net/vbulletin/showthread.php?t=173214

http://www.nvnews.net/vbulletin/showthread.php?t=171282

http://www.nvnews.net/vbulletin/showthread.php?t=177310

Now, will you please shut the f*ck up (in light that NVIDIA confirmed it's a regression in their drivers)?

[AlbertP]

The 610M/635M cards mentioned in the article are not Kepler. They are rebranded Fermi chips introduced back in December along with the 630M which has been supported since 295.33.

I haven't tried 295.40 yet on my hardware.

[Gusar]

Nvidia responds - it's an issue hitting specific cards: http://www.nvnews.net/vbulletin/show...10#post2546510. Interesting that the Geforce 6 series is supposedly among them, by my Go 6600 is unaffected. Maybe cos I'm forcing it into the lowest performance level?

Anyway... gordboy, AMD clownery huh? You're gonna seriously have to improve your trolling skills.

[emptythevoid]

Upgraded to Ubuntu 12.04, and it is updating me to nvidia 295.40. This causes the computer so not shut down properly. Not all the time, but frequently. It just freezes on a black screen, forcing me to hold the button to power it off (https://bugs.launchpad.net/ubuntu/+s...rs/+bug/940564). Also, Flash videos on Youtube do not play back properly (https://bugs.launchpad.net/ubuntu/+s...ee/+bug/968647) and it seems to be related to the nvidia drivers. None of this happened when I was on 280 driver.

[Gusar]

295.49 is out, which fixes all this stuff.

Well, except flash, that's for Adobe to fix (yeah, as if that'll ever happen). Several workarounds described in this thread: http://www.nvnews.net/vbulletin/showthread.php?t=177380. Probably the easiest is disabling acceleration as a whole, but then, you're without acceleration (duh ). I went for the vdpau_trace hack. Using hardware decoding is too crashy, even when disabling overlay (yeah, I know I said something different in that thread ). The hack at least gives you hardware presentation, while fixing the blue people.

[md1032]

295.49 is out, which fixes all this stuff.

Well, except flash, that's for Adobe to fix (yeah, as if that'll ever happen). Several workarounds described in this thread: http://www.nvnews.net/vbulletin/showthread.php?t=177380. Probably the easiest is disabling acceleration as a whole, but then, you're without acceleration (duh ). I went for the vdpau_trace hack. Using hardware decoding is too crashy, even when disabling overlay (yeah, I know I said something different in that thread ). The hack at least gives you hardware presentation, while fixing the blue people.

There's also this: http://lists.freedesktop.org/archive...ay/000022.html

I haven't tried it.

[Gusar]

There's also this: http://lists.freedesktop.org/archive...ay/000022.html

I haven't tried it.

Hey, cool! It does the same thing as the vdpau_trace hack, but in a better way that's a lot less hackish. It doesn't apply to libvdpau-0.4.1 though. So no test right now. Tomorrow I'll whip up a libvdpau-git PKGBUILD to test it.