1.0-8174 for Linux x86 released

[WerNA]

Originally Posted by LubosD

BSoD-when-VT-switch-with-OpenGL-app-running:
Great, it does not crash anymore! Thanks!

However it does output some Xids:

NVRM: Xid: 13, 0002 beef3097 00004097 00000208 07080228 00000002
NVRM: Xid: 27,  L1 -> L0
NVRM: Xid: 13, 0002 beef3097 00004097 00000208 07080228 00000002
NVRM: Xid: 27,  L1 -> L0

Exactly the same situation. No lockups with new driver and 2.6.14, just the first run of Cone3 test produces Xid's. But with 2.6.15-rc2 I had serious crash after running Cone3 - the system didn't response even via ssh.

I've made a simple test - Doom3 demo -> timedemo demo1 (1024x768)
(MSI K8N neo4, nForce4 Ultra, AMD3200+@2300MHz, Leadtek 6800GT@400/1000)

with 7676: UQ - 92.0, HQ - 90.6

with 8174: UQ - 94.9 ,HQ - 93.6

a little, but not very much =) So, the main benefit of this driver for me is a system stability.

Thanks, nVidia!

[neuschnee]

Originally Posted by LubosD

I have 100% stability with RenderAccel, Composite and OpenGL at once. Only a few glitches with Kopete messenger but it looks more like a Composite bug.

What card do you have?

0000:01:00.0 VGA compatible controller: nVidia Corporation NV34 [GeForce FX 5200] (rev a1)

[LubosD]

Originally Posted by neuschnee

0000:01:00.0 VGA compatible controller: nVidia Corporation NV34 [GeForce FX 5200] (rev a1)

NV40 here. Seems that RenderAccel/Composite is quite buggy for "worse" GPUs.

[WerNA]

oops... After 2(!) days I've got lockup when working with Maple. This ssh helps me to reboot my box. The funny thing is that there are no Xid errors. Just X eats 99% of CPU. =(((

switched back to 71.74, wich is rock solid (X runs week and more)

[neuschnee]

Originally Posted by LubosD

NV40 here. Seems that RenderAccel/Composite is quite buggy for "worse" GPUs.

Funny how my "worse" GPU has never had a problem with acceleration in Windows.

Crashes w/composite I would understand... because composite sucks.

Anyway.. I documented the bug more here: http://www.nvnews.net/vbulletin/showthread.php?t=61292

[kirill]

ACPI is indeed a big concern for all the laptop users. It is extremely useful to be able to suspend to hard drive, otherwise, you have to shutdown all you programs while you move between work and home. In that sense failure to provide suspend to hard drive (hibernate) functionality completely defeats the point of having an NVIDIA based laptop/tablet.

I hope NDIVIA will fix this.

[pzgren]

Originally Posted by neuschnee

Funny how my "worse" GPU has never had a problem with acceleration in Windows.

Little bugs, little errors or bad BIOS implemation by the producers get quirks or corrections in the windows driver. You will never see or heard from these under win....then the people come to Linux and say: "hey, it worked under win!".
A good exp. was, when it come to 3DMARK benches, some producer fake their drivers to gain more speed...

Marcus

[neuschnee]

Originally Posted by pzgren

Little bugs, little errors or bad BIOS implemation by the producers get quirks or corrections in the windows driver. You will never see or heard from these under win....then the people come to Linux and say: "hey, it worked under win!".

ZzZzz.. oh please... gimme a break. This is simple acceleration for autohinting, not even a 3D problem. This is not the result of a "bad BIOS". I also did not "come to Linux", but actually, "came to Windows".

I invite you to find some actual proof of this rubbish you're claiming. Find a report of a "bad BIOS" issue that the Windows driver fixed and Linux driver didn't. I think you're talking out of your ass, most likely. Oh, I also was not aware that a GPU is the same thing as a BIOS.

The real reason you don't hear whining from Windows users is because NVIDIA is obviously more devoted to their Windows drivers (and I'm not complaining, it makes sense to me and I understand why). You also don't hear whining over things like Composite, because NT's "composite"-like functionality isn't buggy.

[dev5]

Originally Posted by joaorf

I have the same problem.
When have you generated the nvidia-bug-report.log if the system freezes when X starts?

you can make report after reboot, it just conatin your system info. Add xorg.log.0 (previous one) and maybe xorg.conf

[mattia123]

Just for the record, on my desktop running current Debian testing libc6 (2.3.5-8) the installer detects the wrong TLS mechanisms and installs classic tls thus segfaulting every application using it. Kernel is vanilla 2.6.14.3.
So --force-tls=new fixed it.

And BTW, using an "nVidia Corporation NV17 [GeForce4 MX 440] (rev a3)" I need to use NoRenderAccel (with the 7* driver series I needed to completely disable GLX).

[SilverXXX]

Is a switch to EXA accel possible?

[CondorDes]

Originally Posted by netllama

Enhancements to nvidia-settings will be provided in a future driver release. I don't believe that any NVIDIA employees stated that 1.0-8174 would contain such enhancements.

No, and that's the problem. (a) You never tell us what you're working on, and (b) when you finally do get around to issuing a release, it addresses none (or very few) of the issues that are being complained about on this forum.

What does "a future driver release" mean? The next release, or some release in 10 years or so?

Originally Posted by netllama

Fixes for the composite extension's support in the nvidia driver are also planned for a future driver release. Please keep in mind that many compositor issues are not nvidia driver bugs, but rather X bugs, or window manager bugs.

This is the first I've heard of such bugs, and I follow (at least) some of the X developers' weblogs through FreeDesktop. Have you (meaning nVidia) made any effort to push these bugs back upstream? Can you cite mailing list messages or bug reports, please? (Not necessarily ones you've pushed, just ones that have caused problems)

If you haven't made any effort to connect with the X.org developers about the issues you're encountering in Composite, you have no right to complain.

Please don't misunderstand -- I have no intention of attacking you (or anyone else at nVidia, for that matter). It's just that nVidia is being very tight-lipped, and as a user that has been unsatisfied with the quality of your driver since the 6xxx series, I would expect to hear more about your plans for improving it.

Thank you.