NVIDIA 173.14.05 Display Driver

[phoronix]

Phoronix: NVIDIA 173.14.05 Display Driver

NVIDIA's last Linux display driver release was just shy of two months ago when they had unveiled the 173.08 Beta driver with compatibility for the GeForce 9800 series and other new GeForce 9 parts. Prior to the 173.08 release was version 171.06, which was another beta release, and the last official release being NVIDIA 169.12 from February. NVIDIA this morning has released the 173.14.05 driver, which marks the return to their old naming convention, but it doesn't bring many changes compared to the recent beta releases and just a regurgitated release announcement.

http://www.phoronix.com/vr.php?view=12408

[[Knuckles]]

Does anyone know if the 2D performance problems from earlier releases were fixed?

[Forge]

The NVIDIA 173.14.05 does ship with "out of the box" support for the Linux 2.6.25 kernel and preliminary support for X.Org 7.4 / X Server 1.5. For those using a Linux 2.6.26-rc kernel, the NVIDIA 173.14.05 driver needs to be patched manually.

Ok, so now it's current kernel ready but next kernel broken? AGAIN???

Now it's starting to look like the Linux kernel devs are actively working to break it, and that would be Very Uncool. Also, when Nvidia breaks, VMware follows. I don't like having those two broken, they are *very* important to the way things get done here. If I have to choose between an older kernel or broken Nvidia/VMware, it's going to be older kernel 10 times out of 10.

Not Cool.

[apaige]

;33709']Does anyone know if the 2D performance problems from earlier releases were fixed?
Nope, they weren't. Default settings are usable, but that's it. It's basically the same as with the last beta release.

[Kano]

Well the patch can be found in the release thread, not really hard...

http://www.nvnews.net/vbulletin/showthread.php?t=113919

[M1AU]

Looks like it broke my EDID detection or somethings on my Ubuntu 8.04 installation.

[remm]

I hate them and their driver, but on my G80 nouveau isn't there yet (I won't comment on the pathetic joke that is nv).

[Xanikseo]

I still get bad frame rates with this compared to 171.06 according to glxgears on my 8600gts. It's 13,500 vs 10,000 FPS. I wonder which I'll go for...

[murphy666]

New release without 2d performance gain for 8xxx series...

THAT'S ENOUGH!!! I waited since oct2007 to get the real deal... my old MX440 work better that my 8600 card...

Now i'm going to drill a hole in it to make sure no one can use this #$%!%# card again and I'm going to buy ATI card.

[marakaid]

Strange driver:
"X Server Display Configuration" does not work on NVIDIA X Server Settings.
glxgears is about 15000+ FPS, on 169.12 is 21300+ FPS !

[remm]

Same as Murphy, and on the desktop, I only ever bought NVidia following 3DFX's demise (and I didn't even abandon them during the FX series ...) :( I'll be buying ATI for a while (with no second though, as the upcoming GTX series seems way to extreme for me anyway), and the reason why is just sad. Oh well ...

[deanjo]

People.....

GLXGEARS IS NOT A BENCHMARK!!!

It can be used to verify that hardware-accelerated libraries are installed correctly, but that's about it.

You want to benchmark it, then use realworld benchmarks. I recently read about a site that happens to be doing a benchmark suite for linux. If I could only find a link to that site.......

[M1AU]

Strange driver:
"X Server Display Configuration" does not work on NVIDIA X Server Settings.

This is what I meant.

glxgears is about 15000+ FPS, on 169.12 is 21300+ FPS !

I get around 13000 with this version on a 7800 GTX

[Silent Storm]

People.....

GLXGEARS IS NOT A BENCHMARK!!!

It can be used to verify that hardware-accelerated libraries are installed correctly, but that's about it.

You want to benchmark it, then use realworld benchmarks. I recently read about a site that happens to be doing a benchmark suite for linux. If I could only find a link to that site.......

To be honest, as a programmer, I expext that if a code is better than other, it works faster (in this context). So, yes, glxgears may not be a benchmark and just tries to render three meshed gears but, if the code accelerating it better, it'll yield better frame rates...

To be more specific, I understand that glxgears just renders 3 objects and a light and uses most basic properties of openGL but this means that glxgears shows the foundation's speed. So if the code yields more frame rates it means most important part of the code, the foundation, has got better. Latest drivers may have more optimized shader compilers or more efficient power management or some nasty memory optimizations and work 3-5 FPS faster in games or whatever...

This optimizations are useless after some point since the foundation is slower and this is the point. I vote for a completely-better driver. Not a optimized here but broke there code mess.

[ltmon]

New release without 2d performance gain for 8xxx series...

THAT'S ENOUGH!!! I waited since oct2007 to get the real deal... my old MX440 work better that my 8600 card...

Now i'm going to drill a hole in it to make sure no one can use this #$%!%# card again and I'm going to buy ATI card.

Having just upgraded my laptop with and ATI card to a Nvidia card this kind of funny to read.

Before you do anything rash with your hardware have a read of the ATI forum here on Phoronix and notice that others are pulling their hair out over various issues and threatening to burn their ATI and switch to a nvidia.

Be well aware that both binary drivers have their issues, sometimes major showstoppers depending on your hardware. Both open source efforts (RadeonHD, Noveau) are a little way off completion, although hopefully closing in on "good enough" status rapidly.

[marakaid]

This is what I meant.



I get around 13000 with this version on a 7800 GTX

I am using 8800GT with E6750 on a G33 mainboard.

[Danniello]

The NVIDIA 173.14.05 does ship with "out of the box" support for the Linux 2.6.25 kernel and preliminary support for X.Org 7.4 / X Server 1.5.

So it means that nVidia driver will work with Fedora 9?
If yes, I'm waiting for livna.org package and finally I could update Fedora:)

[osmanovic100]

The NEW driver is very fast!
Wine + Cod4 = 28-40FPS! :)

before with 173.08
Wine + Cod4 = 19-30FPS!
NVIDIA Thank you!, And request more FPS

[AHSauge]

So it means that nVidia driver will work with Fedora 9?
If yes, I'm waiting for livna.org package and finally I could update Fedora:)
Yes, it works, though there is some text corruption going on :confused: Strangely enough, it only seem to affect some fonts, but not all. I'm not sure if this is my fault or the driver though.

Here's an example of it
http://www.ascdevel.com/ernie/corrupttext.png

[murphy666]

Having just upgraded my laptop with and ATI card to a Nvidia card this kind of funny to read.

Before you do anything rash with your hardware have a read of the ATI forum here on Phoronix and notice that others are pulling their hair out over various issues and threatening to burn their ATI and switch to a nvidia.

Be well aware that both binary drivers have their issues, sometimes major showstoppers depending on your hardware. Both open source efforts (RadeonHD, Noveau) are a little way off completion, although hopefully closing in on "good enough" status rapidly.

The reason I will switch to ATI is that AMD ARE really working with the community to get open source driver usable in 2d/3d for these card. Sadly "Nouveau" driver is not usable at all for me (I know they are working really hard!) so the reason I will switch to ATI is their open source drivers not the closed one.

[taipan]

;33709']Does anyone know if the 2D performance problems from earlier releases were fixed?
Somehow. It's behind intel and ati but kde4 is now usable.
Try this:
Put this in your xorg.conf:
Section "Screen"
........
Option "PixmapCacheSize" "200000"
........
EndSection

if you have a laptop

put this in /etc/modprobe.conf (or create a file in /etc/modprobe.d/)

options nvidia NVreg_RegistryDwords="PerfLevelSrc=0x2222"
options nvidia NVreg_Mobile=1

reboot (or from console (without X started): rmmod nvidia;modprobe nvidia)


run: nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1

I have NVIDIA 8600M GT. Kde now works pretty good.

If you don't have an laptop I suggest you to buy ATI !!!

[Danniello]

Yes, it works, though there is some text corruption going on :confused: Strangely enough, it only seem to affect some fonts, but not all. I'm not sure if this is my fault or the driver though.


And there is already linva.org package.
Time to update Fedora:)
http://livna-dl.reloumirrors.net/fedora/9/i386/kmod-nvidia-173.14.05-2.lvn9.i686.rpm

[Xanikseo]

It seems that the performance difference between 171.06 and 173.14.05 in Nexuiz are very slight (in contrast to the performance in glxgears)

My PTS benchmark (http://global.phoronix-test-suite.com/index.php?k=profile&u=xanikseo-29917-28257-25347)

[_txf_]

Somehow. It's behind intel and ati but kde4 is now usable.
Try this:
Put this in your xorg.conf:
Section "Screen"
........
Option "PixmapCacheSize" "200000"
........
EndSection

if you have a laptop

put this in /etc/modprobe.conf (or create a file in /etc/modprobe.d/)

options nvidia NVreg_RegistryDwords="PerfLevelSrc=0x2222"
options nvidia NVreg_Mobile=1

reboot (or from console (without X started): rmmod nvidia;modprobe nvidia)


run: nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1

I have NVIDIA 8600M GT. Kde now works pretty good.

If you don't have an laptop I suggest you to buy ATI !!!

does it make konqueror work? with ipp @ 2 konq in kde4 rendered "paint drying" slow and opening large images was an exercise in patience

[apaige]

InitialPixmapPlacement=1 works best for large images on my setup.

[ferreira]

Well, all I can say is "lulz"... Now the people suddenly want to embrace ATI's efforts the help the development of radeon/radeonhd drivers by telling others to buy ATI stuff...
I'd risk a statement that people ALWAYS stick to what works... ideology is irrelevant here - I think I've never seen anyone cursing NVIDIA when their driver was working well, no matter if it was closed or whatever.
I am sure NVIDIA will fix the bugs - and then, will you still call for people to buy ATI or return to using what you have?
Anyway... 'twas just a random rant... feel free to flame me if you see it proper...

[naja]

Hi
Any new fix for Powermizer on Notebooks? (6150NOGO)
I am stuck at Performance-Mode.I would like to get my card cooler.
Thanks

[tuxic]

Yes, it works, though there is some text corruption going on :confused: Strangely enough, it only seem to affect some fonts, but not all. I'm not sure if this is my fault or the driver though.

Here's an example of it
http://www.ascdevel.com/ernie/corrupttext.png

I'm getting the same text corruption with this driver. Any solution yet?

[taipan]

does it make konqueror work? with ipp @ 2 konq in kde4 rendered "paint drying" slow and opening large images was an exercise in patience
Yes, is not like intel or ati but is working !!!

[kirù]

Since I left the 100.14.19 driver I had problems with window icons corruption (minimize, maximize, close) and the related bar.
I also had problems with amsn when I try to change my personal picture (X crashes in that moment). Does this driver resolve them?

My card is a 7300 GT.

[Michael]

For those of you really upset (or very pleased) by the driver, we do accept freelance editorials on Phoronix :)

[Shakey_Jake33]

I'm a bit worried about this now. I'm using a 7900 GTO right now, but will be switching to a laptop w/ 8400m from September for university. When people say the new drivers are really bad for 2D on GeForce 8 cards, are we talking general desktop usage, watching videos etc?

[deanjo]

I'm a bit worried about this now. I'm using a 7900 GTO right now, but will be switching to a laptop w/ 8400m from September for university. When people say the new drivers are really bad for 2D on GeForce 8 cards, are we talking general desktop usage, watching videos etc?


A few people have issues with them, most don't, watching video's shouldn't be a problem. (Heck watching 1080p even on a lowly Gforce 7200 can be done). Speaking from first hand experience based using a MacBook Pro which uses the 8600m, everything is running smooth as silk here on the latest drivers.

[deanjo]

For those of you really upset (or very pleased) by the driver, we do accept freelance editorials on Phoronix :)

Do you except editorials on other linux topics as well (such as a distribution review)? If so, how do you wish for them to be submitted? A guide/rules/expectations post would be a nice thing to have stickied in the forums.

[AHSauge]

I'm getting the same text corruption with this driver. Any solution yet?
Haven't found any yet, and I've currently gone back to downgraded Xorg. As far as I know, 3D didn't work for me which is quite odd considering that glxgears actually worked with somewhat high fps. Also,there's the nice fonts too :p One thing I've found out is that the problem doesn't seem to exist with KDE apps (I'm using KDE4). So for instance Firefox and Opera has the problem with the fonts, but not Konqueror. Didn't try running Opera or Firefox in Gnome though, which I probably should to rule out KDE being the problem. If anyone has any bright ideas here I'll happily try them.

Btw: Running a GeForce 7800GS AGP card if it's relevant.

[tuxic]

That's weird. I'm running Gnome and experiencing font corruption throughout all applications. Menu's and even Firefox. 3D works nicely with this driver. I'll continue to look for a solution. Keep you posted.

[murphy666]

Well, all I can say is "lulz"... Now the people suddenly want to embrace ATI's efforts the help the development of radeon/radeonhd drivers by telling others to buy ATI stuff...
I'd risk a statement that people ALWAYS stick to what works... ideology is irrelevant here - I think I've never seen anyone cursing NVIDIA when their driver was working well, no matter if it was closed or whatever.
I am sure NVIDIA will fix the bugs - and then, will you still call for people to buy ATI or return to using what you have?
Anyway... 'twas just a random rant... feel free to flame me if you see it proper...

That's not only a question about ideology, open source driver will be easier to maintain, releases and fixes will come out faster and I'm sure that more people will work on the driver development. And best of all, distributions will include and enable them automatically because the driver will not be restricted anymore.

[AHSauge]

That's weird. I'm running Gnome and experiencing font corruption throughout all applications. Menu's and even Firefox. 3D works nicely with this driver. I'll continue to look for a solution. Keep you posted.
I just read this (http://www.nvnews.net/vbulletin/showthread.php?t=113949). Antialias really helps, and it just happends to be enabled for KDE which makes sense :) Now if I just could get wine to run again ... (it refuses to run with the new driver :()

[Forge]

I'd like to 'me too' this. I'm currently on an 8800GT because it offered the best price/performance/usability, but I'm hoping to go HD 4870 once they are released. The nice thing about open drivers is the long-term nature. You can still fire up a box with a G400 or ancient 3Dfx card without too much work. Older Nvidia cards aren't very well supported, Nvidia is slow to release updates for them and they often miss out on a lot of the new features in newer drivers.

Anyways, come HD 4870 launch, maybe I'll take Michael up on his offer, and document some of my experiences from the POV of a long time Nvidia user (0.9-6 and up!).

[AHSauge]

I just read this (http://www.nvnews.net/vbulletin/showthread.php?t=113949). Antialias really helps, and it just happends to be enabled for KDE which makes sense :) Now if I just could get wine to run again ... (it refuses to run with the new driver :()Ehm ... and that would be what happends when I forget to install the 32bit libs of the driver :o Well, now everything works great! :D Hopefully they'll fix the problem with the fonts in the next version, but for now antialias seem to work, though I must say, it doesn't look to pritty on small text.

[mmhorda]

To be honest, as a programmer, I expext that if a code is better than other, it works faster (in this context). So, yes, glxgears may not be a benchmark and just tries to render three meshed gears but, if the code accelerating it better, it'll yield better frame rates...

To be more specific, I understand that glxgears just renders 3 objects and a light and uses most basic properties of openGL but this means that glxgears shows the foundation's speed. So if the code yields more frame rates it means most important part of the code, the foundation, has got better. Latest drivers may have more optimized shader compilers or more efficient power management or some nasty memory optimizations and work 3-5 FPS faster in games or whatever...

This optimizations are useless after some point since the foundation is slower and this is the point. I vote for a completely-better driver. Not a optimized here but broke there code mess.

you are weird programmer.
Cause glx_gears is not a benchmark. it doesnt have to show more fps just because your card is better.

[Forge]

To be honest, as a programmer, I expext that if a code is better than other, it works faster (in this context). So, yes, glxgears may not be a benchmark and just tries to render three meshed gears but, if the code accelerating it better, it'll yield better frame rates...

To be more specific, I understand that glxgears just renders 3 objects and a light and uses most basic properties of openGL but this means that glxgears shows the foundation's speed. So if the code yields more frame rates it means most important part of the code, the foundation, has got better. Latest drivers may have more optimized shader compilers or more efficient power management or some nasty memory optimizations and work 3-5 FPS faster in games or whatever...

This optimizations are useless after some point since the foundation is slower and this is the point. I vote for a completely-better driver. Not a optimized here but broke there code mess.


Please do not contribute code to anything I care about. Your ideas on performance metrics are absolutely bass-ackwards.

GLXGears is not a benchmark. It takes one very very very tiny part of the OpenGL spec and runs it as fast as it can. Using this as any sort of performance metric is like comparing word processors based on how fast a held key fills a file. Graphics cards accelerate 3D tasks, and if newer drivers do that better, but have a mildly slower GLXGears score, then that just means that they have their priorities straight.

Seriously: You are misunderstanding what GLXGears was written for. The whole idea is that you run it before setting up Utah GLX, and then you run it again afterwards, to see if 3D acceleration was being enabled on your Matrox G200 or NV Riva 128. It should have been removed years and years ago. It no longer matters in any way.

Oh, bonus: You're almost certainly grossly CPU limited. Congratulations.

[b15hop]

Should make a GPU, pixel shaded, glxgears! glxgears2 maybe?

I agree that I just use glxgears to test if OpenGL is working. Although this can be tested in other ways. Running a game would be the best way to truly test it anyway. To get a better understanding of how much of a difference hardware acceleration would make.

[Redeeman]

anyone experiencing XV corruption? after i got this version i now get:
http://img108.imageshack.us/my.php?image=xvproblemkq5.jpg

[b15hop]

I haven't checked video yet. Seems fine otherwise.

[greg]

I just upgraded my box to Ubuntu 8.04 and now I have strange problems with one app (Stepmania) with the new drivers. Background movies in this game totally slow down everything to just a few frames per second. Before that, in Ubuntu 7.10 with the 100.14 drivers, everything was fine and very smooth.

My card is a GeForce 7050 IGP. Sure, it's not the fastest, but it isn't THAT slow -- something as simple as these background videos has to run fine.