On Tuesday 24 October 2006 4:19 am, Basil Chupin wrote:

Fred A. Miller wrote:

...

I've also noticed that the performace of the nVidia commercial driver stinks! I have a PCI Express 16X card with the Geforce FX 5300 chipset. In the past, it's proved to be very quick.......NOT now!! Below are my results with glxgears.......this is a P-4 3.2Ghz. box.

Fred

3776 frames in 5.1 seconds = 744.131 FPS 4469 frames in 5.0 seconds = 893.677 FPS 6473 frames in 5.0 seconds = 1294.524 FPS 6480 frames in 5.0 seconds = 1295.932 FPS 6548 frames in 5.0 seconds = 1309.598 FPS

Out of curiosity, Fred, login in as root and do the glxgears thing.

Someone mentioned a day ago that he got different (higher) result by logging in as root. I tried this and found that the result also went up--but by a very small margin. Nevertheless, to me it indicated that something in the user login possibly affects the fps result.

Yes, I read that. I also double checked to make sure my user ID is listed as "video" in the user setup, and it is. I get the same results you do.....not worth noting the difference. Thanks, Fred -- MickySoft, the ultimate corporate parasite.

On Tuesday 24 October 2006 8:07 pm, Joachim Schrod wrote: [snip]

In full screen mode, 1600x1200 at 24bpp, glxgears gets reduced to 55 FPS! This seems to be awfully slow, or is this normal for that card? glxinfo|grep rendering outputs Yes, so 3D accelaration is supposed to be there. Root and non-root achieve the same speed, no it's not a matter of access to /dev/nvidia*

This is on SUSE 10.0, with the nvidia driver in version 1.0.7676, that I installed via YOU.

Would an update to the new 8776 driver version improve the performance?

I doubt it. I d'l the file directly from nVidia......NO change.

How do I do such an update in the best way? Ideally it would survive kernel updates, as the current installation does. Btw, can anybody point me to a description how that is done? I would like to configure my VMware installation the same way, that it survives kernel updates as well.

It's easy enough to do, but from my experience and of others, it hasn't improved over the last driver. Fred -- MickySoft, the ultimate corporate parasite.

On Wed, 2006-10-25 at 02:07 +0200, Joachim Schrod wrote:

Fred A. Miller wrote:

...

On Tuesday 24 October 2006 4:19 am, Basil Chupin wrote:

...

Fred A. Miller wrote:

...

I've also noticed that the performace of the nVidia commercial driver stinks! I have a PCI Express 16X card with the Geforce FX 5300 chipset. In the past, it's proved to be very quick.......NOT now!! Below are my results with glxgears.......this is a P-4 3.2Ghz. box.

Fred

3776 frames in 5.1 seconds = 744.131 FPS 4469 frames in 5.0 seconds = 893.677 FPS 6473 frames in 5.0 seconds = 1294.524 FPS 6480 frames in 5.0 seconds = 1295.932 FPS 6548 frames in 5.0 seconds = 1309.598 FPS

Yes, I read that. I also double checked to make sure my user ID is listed as "video" in the user setup, and it is. I get the same results you do.....not worth noting the difference.

Here too, with a GeForce FX 5200, and a P4 3 GHz, I get roughly 1000 FPS in the small window.

In full screen mode, 1600x1200 at 24bpp, glxgears gets reduced to 55 FPS! This seems to be awfully slow, or is this normal for that card? glxinfo|grep rendering outputs Yes, so 3D accelaration is supposed to be there. Root and non-root achieve the same speed, no it's not a matter of access to /dev/nvidia*

This is on SUSE 10.0, with the nvidia driver in version 1.0.7676, that I installed via YOU.

Would an update to the new 8776 driver version improve the performance?

How do I do such an update in the best way? Ideally it would survive kernel updates, as the current installation does. Btw, can anybody point me to a description how that is done? I would like to configure my VMware installation the same way, that it survives kernel updates as well.

Best, Joachim

-- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany

Just checked my present system, Suse 10.1, Athlon 64 X2 3800+ with w Gig RAM, Geforce 7300 GT PCI Express, 1.0.8762 Nvidia driver w/ nvidiahack.sh patch (for operation with Xen kernel) and default Suse Xen Kernel 2.6.16.21-.25 I get the following with glxgears 24-bit, Small window as it defaults to 43531 frames in 5.0 seconds = 8706.025 FPS 24-bit, 1680X1050 fullscreen 4684 frames in 5.0 seconds = 936.673 FPS Art

Matthias Hopf a écrit :

On Oct 25, 06 02:07:50 +0200, Joachim Schrod wrote:

...

Fred A. Miller wrote:

...

...

...

I've also noticed that the performace of the nVidia commercial driver stinks! I have a PCI Express 16X card with the Geforce FX 5300 chipset. In the past, it's proved to be very quick.......NOT now!! Below are my results with glxgears.......this is a P-4 3.2Ghz. box.

Yes, I read that. I also double checked to make sure my user ID is listed as "video" in the user setup, and it is. I get the same results you do.....not worth noting the difference.

Here too, with a GeForce FX 5200, and a P4 3 GHz, I get roughly 1000 FPS in the small window.

In full screen mode, 1600x1200 at 24bpp, glxgears gets reduced to 55 FPS! This seems to be awfully slow, or is this normal for that card?

Well - it doesn't look *too* bad. Though if you had higher performance in fullscreen windows before, this is a regression. glxgears is actually a valid indicator for fillrate issues, if used fullscreen. Just the higher fps numbers in windowed mode don't make any sense at all.

I find that very bad, with a GeForce 7800 GTX, I get with glxgears 24bbp: -default resolution of glxgears which is about 300x300 pixels. - Full screen is 1280x1024. SuSE 10.1 64 bits Digital entry : 300x300 -> 16606 fps 1280x1024 -> 3410fps SuSE 10.1 64 bits Analog entry : 300x300 -> 16791 fps 1280x1024 -> 3383 fps SuSE 10.1 32 bits Digital entry : 300x300 -> 15710 fps 1280x1024 -> 3410fps My monitor is an DFP EIZO Flexcan L778 with digital and analog entries. Michel.

Matthias Hopf wrote:

On Oct 25, 06 02:07:50 +0200, Joachim Schrod wrote:

...

Here too, with a GeForce FX 5200, and a P4 3 GHz, I get roughly 1000 FPS in the small window.

In full screen mode, 1600x1200 at 24bpp, glxgears gets reduced to 55 FPS! This seems to be awfully slow, or is this normal for that card?

Well - it doesn't look *too* bad. Though if you had higher performance in fullscreen windows before, this is a regression.

I didn't had higher numbers before, this is the driver that I installed anew when I bought the 5200 last year. I know that this ain't no high-end card, and if that's a typical performance, so be it. (My actual problem is more in movies video and audio are sometimes out-of-sync; and I thought that the performance might have something to do with it.)

Someone please open a bug report and post the number of it here, so everybody who's seeing degrading performance (and also who isn't) can attach his numbers.

Since I cannot say if I really have degrading performance, I am reluctant to be the owner of that ticket; but I'm happy to provide test results if that helps. Best, Joachim -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany

I found this thread very interesting and helpful, since I am looking at video cards to upgrade my system. I am curious though, about the frame rates. Since movies look pretty good at 24 fps, and experts have claimed for many years that we can't see changes very much faster than that, what is the advantage of the extreme frame rates most of the newer cards offer? Just trying to learn, -- ED --

Carlos E. R. wrote:

The Thursday 2006-10-26 at 20:52 -0400, Ed McCanless wrote:

...

what is the advantage of the extreme frame rates most of the newer cards offer?

Probably because if it is capable of a high frame rate, it will waste very little time drawing at a lower rate. And anyhow, modern computers do refresh at a higher rate than 24: 90 FPS is not that uncommon, to minimize flicker.

Thanks Carlos, I'm not sure I would notice the difference above 30 or 40 fps, but I was aware that computers (and cards) are capable of much more than that. I thought that perhaps something else in the animation process was enhanced by the higher rates. -- ED --

Rajko M wrote:

On Thursday 26 October 2006 21:26, Ed McCanless wrote:

...

Carlos E. R. wrote:

...

The Thursday 2006-10-26 at 20:52 -0400, Ed McCanless wrote:

...

what is the advantage of the extreme frame rates most of the newer cards offer?

Probably because if it is capable of a high frame rate, it will waste very little time drawing at a lower rate. And anyhow, modern computers do refresh at a higher rate than 24: 90 FPS is not that uncommon, to minimize flicker.

Thanks Carlos, I'm not sure I would notice the difference above 30 or 40 fps, but I was aware that computers (and cards) are capable of much more than that. I thought that perhaps something else in the animation process was enhanced by the higher rates. -- ED --

Higher refresh rates are just eye friendly, for people that spend a lot in front of monitor. Nothing else.

Thanks Rajko; Then, to get back toward the subject, I should get satisfactory improvement with any of the cards discussed in this thread (I am usin SuSe 10.1 on an AMD 64 box). I have integrated video now, and glxgears runs pretty smooth now, although I don't remember the frame rate, it's been a while. -- ED --

On Oct 26, 06 20:52:10 -0400, Ed McCanless wrote:

I found this thread very interesting and helpful, since I am looking at video cards to upgrade my system. I am curious though, about the frame rates. Since movies look pretty good at 24 fps, and experts have claimed for many years that we can't see changes very much faster than that, what is the advantage of the extreme frame rates most of the newer cards offer?

Just trying to learn,

The following numbers of course vary a bit from person to person: - The eye will see changing images as "moving" begining with 12fps (if it is less, it will see them as several still images) - Motion will *start* looking fluently with approx. 24fps (Film is 24fps, NTSC Movie is 23.976fps, PAL Movie is 25fps) - Even with 50fps judder will be eminent for sensible people (IMAX is filmed at 50fps, I personally still think it is juddering) - Images will start to look steady if shown 50 times per second (PAL is 50Hz, NTSC 60Hz, for film each picture is shown three times, so it is actually 72Hz with 24fps) - Even with 75Hz flickering will be eminent for sensible people (though AFAIK all agree that 100Hz is enough) Strangely enough, this doesn't seem to apply to film (because I haven't heard anybody complaining about cinemas with that respect, which use only 72Hz)... - The eye can detect motion with a frequency up to 200Hz. The brain also anticipates changes in the environment at approx. that rate, and gets distracted if the input from the eye doesn't follow its anticipation. (in a virtual environment the perfect update rate would have to be 200fps, else you easily develop a headache. It goes without saying that the typical update rates in these environments of less than 10fps are way below this margin) Additional: - Old games had game logic and video rendering coupled, so in Quake you could run faster if you had an video rendering rate of AFAIR 73fps. - Old games had a lag of 3-4 frames, from input to final reaction on screen. So if you had the game running at 100fps, your received update time was closer to 25fps which is still slow for good reactions. See also the last point of the top list (anticipated changes). Hope this helps Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ labs www.mshopf.de

On Oct 31, 06 16:14:13 +0100, Carlos E. R. wrote:

...

- Images will start to look steady if shown 50 times per second (PAL is 50Hz, NTSC 60Hz, for film each picture is shown three times, so it is actually 72Hz with 24fps)

When I worked in an old projection cabin, film was projected at 48 Hz, 24 fps; ie, each frame was projected and obscured twice, with a mechanical obturator. Things might have improved with the new xenon lighting, I dunno, but I suppose.

Well, I think I remember that I read that long ago as well, but later on I read that the typical shutter discs rotate three times as fast as the film moves nowadays. I guess they couldn't make the film transport mechanism fast enough back then. Today they are using low air pressure devices to suck the individual frames into place... :-P So I guess it's save to assume this improved a bit :) 48Hz is barely enough to have a steady image... CU Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ labs www.mshopf.de

Carlos E. R. wrote:

...

The Tuesday 2006-10-31 at 17:28 +0100, Matthias Hopf wrote:

...

...

...

Well, I think I remember that I read that long ago as well, but later on I read that the typical shutter discs rotate three times as fast as the film moves nowadays. I guess they couldn't make the film transport mechanism fast enough back then. Today they are using low air pressure devices to suck the individual frames into place... :-P

Low air... whatever? I'm outdated. Sigh.

Yes, I saw the air pressure thing on the Science channel. It is what has eliminated most of the vertical scratches form the theater experience.(I feel outdated all the time, dealing with computer technology)

...

Yes, the 35 mm machines I handled used a shutter disk or cylinder, blocking the light source two times per frame. But in order to do that they have to compensate the dark periods with higher brightness, ie, stronger light source... The machine I used had an 2 KW arc lamp (around 65 Amps), and that's pretty hot: if the film stops it burns within seconds. So they invented xenon arc lamps. They are relatively colder, brighter, and much easier to use (the traditional arc lamp uses carbon composites electrodes that get burnt and have to be continuously adjusted to compensate during the session; a faulty machine or human and the spectators get mad). I suspect that only with xenon lamps can the apparent fps be increased.

(http://en.wikipedia.org/wiki/Xenon_arc_lamp, http://en.wikipedia.org/wiki/Carbon_arc_lamp).

I'm surprised that projectors don't use strobes yet to replace the shutter, as with extreme slow motion cameras.

...

The film movement is a different thing: it moves at 24 fps, not continuously, but intermittently. That's what limits its speed. It moves a frame fast, then it is hold in place for the rest of the 1/24 S left, while it is projected into the screen two or three times.

It is unfortunate that that fps of 24 doesn't match the TV rate of 25 or 30. Computers are different.

And, returning a bit to "topic". I wonder if sometime in the future we will see digital cinema, kind of super-high-definition-dvd. The definition in film is tremendous, more so with the 70mm film; coupled with the brightness and size of the screen is something the digital world can not reach yet, AFAIK.

I think we may see something like this in the future. Perhaps through better optics, or some sort of Laser tech.

...

...

...

...

So I guess it's save to assume this improved a bit :) 48Hz is barely enough to have a steady image...

The slight headache is part of going to the movies experience :-)

I guess I'm lucky there. I only get headaches on rare occasions from the computer screen, never at movies. -- ED -- p.s. I messed up again, Carlos. Sorry!

On Wednesday 01 November 2006 20:40, William Gallafent wrote:

It's already here

For more detail: http://en.wikipedia.org/wiki/Digital_cinema -- Bill Gallafent.

On Nov 01, 06 13:32:05 +0100, Carlos E. R. wrote:

...

Well, I think I remember that I read that long ago as well, but later on I read that the typical shutter discs rotate three times as fast as the film moves nowadays. I guess they couldn't make the film transport mechanism fast enough back then. Today they are using low air pressure devices to suck the individual frames into place... :-P

Low air... whatever? I'm outdated. Sigh.

;-)

blocking the light source two times per frame. But in order to do that they have to compensate the dark periods with higher brightness, ie,

They still do. Though the black period is typically shorter now than the lit period.

And, returning a bit to "topic". I wonder if sometime in the future we will see digital cinema, kind of super-high-definition-dvd. The definition in film is tremendous, more so with the 70mm film; coupled with the brightness and size of the screen is something the digital world can not reach yet, AFAIK.

?!? Since we have 3 movie theaters with digital projectors here in Nuremberg now, I don't like going to the regular projektor cinemas any longer. Well, have to, as not all films are shown in digital quality... Digital projection gets rid of all jiddering, the colors are much more vivid, resolution is really good enough for me, even typically better than with film - so for me the digital world is already there.

...

So I guess it's save to assume this improved a bit :) 48Hz is barely enough to have a steady image...

The slight headache is part of going to the movies experience :-)

ROTFL Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ labs www.mshopf.de --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Wednesday 2006-11-01 at 01:10 -0500, Ed McCanless wrote:

As Carlos said, this helps, especially the part about the games. I noticed long ago that there was quite a difference between what I saw as smooth animation and what I perceived as unnoticeable reaction time. And, since we started discussing frame rates, I have started to notice a slight flicker on the monitor, like that of a fluorescent (thank God for spell checkers) bulb.

It is easier to notice if you look with the edge of the vision field, ie, if you aim your sight outside of the monitor. The perifery of the eye has faster "sensors". - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFFSIJwtTMYHG2NR9URAn3vAJ0W5sRxPW3MB+BOjXSmlTMdq18RngCcCM74 8g7C4WLk+NKbr4NDncythlQ= =KWUg -----END PGP SIGNATURE-----

Carlos E. R. wrote:

It is easier to notice if you look with the edge of the vision field, ie, if you aim your sight outside of the monitor. The perifery of the eye has faster "sensors".

Yep, the periphery vision is more sensitive to light in general, as in searching for dim stars at night (astronomy 101). Look to Randall's answer for the reasons. -- ED --

On Wednesday 01 November 2006 17:51, Carlos E. R. wrote:

...

Yep, the periphery vision is more sensitive to light in general, as in searching for dim stars at night (astronomy 101).

That I didn't know.

Not sure this is really true. The reason for looking a little offset from something you want to see at night is because you can see anything where the optic nerve is on the retina. You need to offset a bit so that your focus isn't at the point of the optic nerve. (Aviation 101) "The non-central, peripheral part of the retina perceives light at low levels of illumination. It can actually perceive light at one thousandth the illumination needed by the fovea. Sometimes pilots complain that they may see an object at night only to have it disappear as they look directly at it. What happens is that they shift from peripheral dark- adapted vision to central day light vision. This part of the eye is not able to detect objects at low intensity. Another location of the retina which cannot see at all is the nearby "blind spot" where the optic nerve enters the retina. Looking at objects off center about 15 degrees will correct that loss of vision in this area."

On Wednesday 01 November 2006 16:11, Bruce Marshall wrote:

On Wednesday 01 November 2006 17:51, Carlos E. R. wrote:

...

...

Yep, the periphery vision is more sensitive to light in general, as in searching for dim stars at night (astronomy 101).

That I didn't know.

Not sure this is really true. The reason for looking a little offset from something you want to see at night is because you can see anything where the optic nerve is on the retina. You need to offset a bit so that your focus isn't at the point of the optic nerve.

The fovea and the connection point of the optic nerve bundle are certainly no co-located. If they were, our blind spot would be at the center of our visual field. RRS

Fred A. Miller wrote:

On Thursday 26 October 2006 11:57 am, Matthias Hopf wrote:

...

It shouldn't. Are you using an XVideo or OpenGL output plugin? Movies have typically almost 24fps, so as long as you get more than 24fps in full screen OpenGL apps, this should be enough. Also both xine and mplayer tell you if your system is too slow for playing.

I have to use 16-bit color depth because the driver is so bad! But then, it's been BAD since 10.1.

Fred, have you tried going back several versions of the driver to see what results you get? Cheers. -- I'm dangerous when I know what I'm doing.

Carlos E. R. wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

The Tuesday 2006-10-31 at 20:49 +1100, Basil Chupin wrote:

...

Fred, have you tried going back several versions of the driver to see what results you get?

Rather, go back in SuSE versions. The latest driver installed in 9.3 works perfect in my machine.

Aha, just as I thought. It's the kernel, the kernel, I tell you! :-) . Cheers. -- I'm dangerous when I know what I'm doing.

On Tuesday October 31 2006 4:49 am, Basil Chupin wrote:

Fred A. Miller wrote:

...

On Thursday 26 October 2006 11:57 am, Matthias Hopf wrote:

...

It shouldn't. Are you using an XVideo or OpenGL output plugin? Movies have typically almost 24fps, so as long as you get more than 24fps in full screen OpenGL apps, this should be enough. Also both xine and mplayer tell you if your system is too slow for playing.

I have to use 16-bit color depth because the driver is so bad! But then, it's been BAD since 10.1.

Fred, have you tried going back several versions of the driver to see what results you get?

Sorry I'm late in getting back to you Basil. No, I haven't. I have hope that SOON SUSE will get nVidia to fix the driver as it SHOULD be. That is, equal in performance to the 'Bloze driver. I am also watching ATI, as the buyout of ATI by AMD is now consumated. The "word" is, is that AMD IS going to place the driver code under GPL. We'll see if they do. 'Hope so! Intel has already made their drivers open source, so this would put ample on nVidia to get their heads out of the sand bucket. Fred -- MickySoft, the ultimate corporate parasite.

Carlos E. R. wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

The Monday 2006-11-06 at 05:47 -0500, Fred A. Miller wrote:

...

...

Fred, have you tried going back several versions of the driver to see what results you get? Sorry I'm late in getting back to you Basil. No, I haven't. I have hope that SOON SUSE will get nVidia to fix the driver as it SHOULD be. That is, equal in performance to the 'Bloze driver. I am also watching ATI, as the buyout of ATI by AMD is now consumated. The "word" is, is that AMD IS going to place the driver code under GPL. We'll see if they do. 'Hope so! Intel has already made their drivers open source, so this would put ample on nVidia to get their heads out of the sand bucket.

I have mentioned here that there are reports opened in bugzilla about this, and that they are waiting for your report. If nothing comes, they will close it without doing anything.

And, for the record, the problem seems to be in the linux camp, not the nvidia's.

Perhaps not, in fact probably not at all. I suspect that it *is* nVidia driver. I'm sorry that I haven't posted my observations earlier but I've been tied up with messing around with 10.2 beta1. I have 3 computers all with nVidia cards in them: 6600 5500 5200 Now, when I try and install Suse 10.1 or 10.2 beta1 or Kubuntu 6.10 or even try and run the Live copy of Kubuntu on the computer with the 6600 nVidia card I get garbage on the screen at the point when I have to enter my name/password to logon. The screen looks as if it is using 4-bit graphics with a 640x400 resolution (I'm guessing here of course). I do not have this problem with the computers using the 5500 and the 5200 cards- none at all. In fact I just fired up the kubuntu 6.10 Live to see what driver was being used in Kubuntu. Sure enough, it is the basic nVidia driver, "nv". The common thread I believe is the nVidia driver. I believe that both distros are using the new xorg but if this was the problem then I would have the problem with all nVidia cards and not just the 6600 one. Submitted with the hope that this may contribute in some small way towards rectifying the problem - and so making my life easier the next time I have to install Suse or Kubuntu :-) . Cheers. PS The question obviously has to be asked: if I have this problem with the 6600 card how then am I able to run Suse 10.1? There is a simple answer but I won't take up time now explaining what I have to do. -- "If the terrier and the bariffs are torn down, this economy will grow." George W. Bush January 2000

Carlos E. R. wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

The Tuesday 2006-11-07 at 01:56 +1100, Basil Chupin wrote:

...

...

And, for the record, the problem seems to be in the linux camp, not the nvidia's. Perhaps not, in fact probably not at all. I suspect that it *is* nVidia driver.

Don't tell me, nor the list: tell HIM, in Bugzilla.

Sorry, who is "HIM"? (Do yo mean "HIM", you know, up THERE? :-) .) You spoke about "HIM" waiting on info and would drop the matter if (?)Fred didn't produce anything. So couldn't you please FWD my msg to "HIM" to make sure that this matter is not dropped? Cheers. -- "If the terrier and the bariffs are torn down, this economy will grow." George W. Bush January 2000

Carlos E. R. wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

The Tuesday 2006-11-07 at 02:44 +1100, Basil Chupin wrote:

...

Sorry, who is "HIM"? (Do yo mean "HIM", you know, up THERE? :-) .)

You spoke about "HIM" waiting on info and would drop the matter if (?)Fred didn't produce anything. So couldn't you please FWD my msg to "HIM" to make sure that this matter is not dropped?

The nvidia chap waiting for reports on the bug in the already mentioned bugzilla report. So far, there are no proofs the problem we see has to be blamed on the driver. There are _no_ reports proving that the 3D is slow etiher.

So, if any of you want the matter solved, please report in Bugzilla. They (I'll say "they" this time) are waiting. If there are no reports, the matter will be dropped.

Bug report just filed with nVidia. Cheers. -- "If the terrier and the bariffs are torn down, this economy will grow." George W. Bush January 2000

Robert Cunningham wrote:

Carlos E. R. wrote:

...

The Monday 2006-11-06 at 05:47 -0500, Fred A. Miller wrote:

...

...

Fred, have you tried going back several versions of the driver to see what results you get? Sorry I'm late in getting back to you Basil. No, I haven't. I have hope that SOON SUSE will get nVidia to fix the driver as it SHOULD be. That is, equal in performance to the 'Bloze driver. I am also watching ATI, as the buyout of ATI by AMD is now consumated. The "word" is, is that AMD IS going to place the driver code under GPL. We'll see if they do. 'Hope so! Intel has already made their drivers open source, so this would put ample on nVidia to get their heads out of the sand bucket. I have mentioned here that there are reports opened in bugzilla about this, and that they are waiting for your report. If nothing comes, they will close it without doing anything.

And, for the record, the problem seems to be in the linux camp, not the nvidia's.

...

Basil

I would also like to report that Nvidia is also trying to work on some of this as well. Following a few reports of this on this forum about a month ago I went to Nvidia and downloaded their Linux Beta Driver 1.0-9625 and have been using it on both my AMD-64 system and my junky work computer a AMD 1700+ with a Geforce MX-400 card running 1280x1024 at 101Hz vertical. Running GLX-gears at it's native window, with lots of other stuff running in the background, I get 226 f/min, which is about 10% faster than the 8762 driver that I was running. It has been running rock solid since I installed it and comes with Nvidia's control panel as well.

The beta 9625 was superceded by the 8776 driver which came out on 19 October. The beta driver, and the 8776, mainly fix a security hole but I may be wrong. For me the 8776, the 9625, and their predecessor all work OK once installed on all the 3 nVidia cards I have even though the glxgears revolve slowly - much slower than in versions of Suse prior to 10.0 I would say. I think the problem of slowness in glxgears started with suse 10.0 so what actually changed in the nVidia driver at that time I don't know. Cheers. -- "If the terrier and the bariffs are torn down, this economy will grow." George W. Bush January 2000

On Mon, 2006-11-06 at 21:20 +0100, Carlos E. R. wrote:

...

For me the 8776, the 9625, and their predecessor all work OK once installed on all the 3 nVidia cards I have even though the glxgears revolve slowly - much slower than in versions of Suse prior to 10.0 I would say. I think the problem of slowness in glxgears started with suse 10.0 so what actually changed in the nVidia driver at that time I don't know.

Nothing. The same driver in 9.3 is very fast.

Right now I think everything is strange. I had great fps and no problems with both 8774 and 8776 in Suse 10.0 and default kernel (System was fully updated). I posted my fps to the list. I now have reinstalled Suse 10.0 and fully updated it, this because I´m dual booting with FreeBSD. I reeinstalled 8774 first and 8776 after that. Prettry quick I reinstalled 8774 again... 8774 works fine but (this time) NOT giving the same performance (maybe half the fps if compared to my last 10.0 instalation...) 8776 seems unstable and the performance is ~ the same as 8774. The slow glxgears animation doesn´t affect the fps so I don´t care about that. For an example 8776 currupts my screen when using init 3 <> init 5. I don´t have any error logs though...And I have to much work to do a couple of days from now to dig deeper in to it. Something´s strange though... -- /Ola! Peo -- Registered Linux User #432116, get counted at http://counter.li.org www.whylinuxisbetter.net -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com

On Monday November 06 2006 6:49 am, Carlos E. R. wrote:

The Monday 2006-11-06 at 05:47 -0500, Fred A. Miller wrote:

...

...

Fred, have you tried going back several versions of the driver to see what results you get?

Sorry I'm late in getting back to you Basil. No, I haven't. I have hope that SOON SUSE will get nVidia to fix the driver as it SHOULD be. That is, equal in performance to the 'Bloze driver. I am also watching ATI, as the buyout of ATI by AMD is now consumated. The "word" is, is that AMD IS going to place the driver code under GPL. We'll see if they do. 'Hope so! Intel has already made their drivers open source, so this would put ample on nVidia to get their heads out of the sand bucket.

I have mentioned here that there are reports opened in bugzilla about this, and that they are waiting for your report. If nothing comes, they will close it without doing anything.

And, for the record, the problem seems to be in the linux camp, not the nvidia's.

At this point, I'm in agreement. My suspect is "X". Fred -- MickySoft, the ultimate corporate parasite.

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Tuesday 2006-10-31 at 02:02 -0500, Fred A. Miller wrote:

I have to use 16-bit color depth because the driver is so bad! But then, it's been BAD since 10.1.

Then, What are you waiting to report this in Bugzilla? Do it NOW. The chap from NVidia says that after reading my report (which is inconclusive as to speed) and the emails here there is no proof of slowenes. Logs, please. Numbers. Do it. Now, please. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFFRywHtTMYHG2NR9URAmOSAJ9aFBOOfD0HeJF+3gO0iKZyLCtgmwCgjs9k xdUUdiRpJpRonNyHYZZDhEU= =xdKe -----END PGP SIGNATURE-----

On Oct 25, 06 22:23:24 +0100, Dave Howorth wrote:

I've now installed nvidia's driver and it seems to be working. No crashes so far at least. Here are my results.

Your results look reasonable. Except if somebody finds that an older driver was faster.

1280 x 1024 24-bit

full-screen glxgears 564 frames in 5.0 seconds = 112.800 FPS

default window size glxgears 6938 frames in 5.0 seconds = 1387.600 FPS

Thanks Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ labs www.mshopf.de

On Tue, 2006-10-24 at 02:52 -0700, Steve Reilly wrote:

Is this the same driver you speak of that is installed when I did tiny-nvidia-installer --update?

907 frames in 5.0 seconds = 181.314 FPS 991 frames in 5.0 seconds = 198.052 FPS 995 frames in 5.0 seconds = 198.878 FPS 1000 frames in 5.0 seconds = 199.983 FPS 981 frames in 5.0 seconds = 196.103 FPS 992 frames in 5.0 seconds = 198.380 FPS

this is what I get on an athlon xp 2.2 ghz, although I have no complaints..... World of Warcraft and unreal run perfectly. You get 1300 FPS?? maybe were not using the same driver?

Athlon XP 2200+, Nvidia GForce 420 MX, nvidia driver 1.0-8774 gives: (not full screen) 5757 frames in 5.0 seconds = 1151.400 FPS 6577 frames in 5.0 seconds = 1315.400 FPS 7433 frames in 5.0 seconds = 1486.600 FPS 7409 frames in 5.0 seconds = 1481.800 FPS 7412 frames in 5.0 seconds = 1482.400 FPS -- /Peo -- Registered Linux User #432116, get counted at http://counter.li.org -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com

On Tue, 2006-10-24 at 20:04 +1000, Basil Chupin wrote:

Peo Nilsson wrote:

...

On Tue, 2006-10-24 at 02:52 -0700, Steve Reilly wrote:

...

Is this the same driver you speak of that is installed when I did tiny-nvidia-installer --update?

907 frames in 5.0 seconds = 181.314 FPS 991 frames in 5.0 seconds = 198.052 FPS 995 frames in 5.0 seconds = 198.878 FPS 1000 frames in 5.0 seconds = 199.983 FPS 981 frames in 5.0 seconds = 196.103 FPS 992 frames in 5.0 seconds = 198.380 FPS

this is what I get on an athlon xp 2.2 ghz, although I have no complaints..... World of Warcraft and unreal run perfectly. You get 1300 FPS?? maybe were not using the same driver?

Athlon XP 2200+, Nvidia GForce 420 MX, nvidia driver 1.0-8774 gives:

(not full screen)

5757 frames in 5.0 seconds = 1151.400 FPS 6577 frames in 5.0 seconds = 1315.400 FPS 7433 frames in 5.0 seconds = 1486.600 FPS 7409 frames in 5.0 seconds = 1481.800 FPS 7412 frames in 5.0 seconds = 1482.400 FPS

I think the subject of this thread is the latest, 8776, driver. What do you get from glxgears if you installed 8776?

Cheers.

-- A quick survey at this office revealed that the preferred method of getting rid of unwanted pubic hair is to use dental floss.

One thing I have noticed on this thread with geaars fps, no one says what their resolution is. The gears fps depends highly on the resolution, and the only meaningful comparison is with the fps at a resolution. Set gears to full screen, then state the fps for the resolution of the screen. You will get a much more meaningful comparison that way. Art

On Tue, 2006-10-24 at 15:19 +0200, Carlos E. R. wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

The Tuesday 2006-10-24 at 20:05 +0800, Art Fore wrote:

...

One thing I have noticed on this thread with geaars fps, no one says what their resolution is. The gears fps depends highly on the resolution, and the only meaningful comparison is with the fps at a resolution. Set gears to full screen, then state the fps for the resolution of the screen. You will get a much more meaningful comparison that way.

Don't forget depth (bpp).

- -- Cheers, Carlos E. R.

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Made with pgp4pine 1.76

iD8DBQFFPhLJtTMYHG2NR9URAtScAJ4wOZQ8UnIvrNmXpyZJEtdNSbZLnQCgkrR6 6/rn8gVuNq1SaOscwpoz5wM= =qrto -----END PGP SIGNATURE-----

True. I alway made mine in 24-bit. Art

Joachim Schrod wrote:

Art Fore wrote:

...

One thing I have noticed on this thread with geaars fps, no one says what their resolution is. The gears fps depends highly on the resolution, and the only meaningful comparison is with the fps at a resolution. Set gears to full screen, then state the fps for the resolution of the screen.

Sorry, but how do I do that? Neither glxgears -h nor man glxgears give any hint how one could set it "to full screen".

When glxgears is running simply click on the 'full screen' button- top, right. You know, the normal set of buttons to minimise the window, exit the window. Cheers. -- A quick survey at this office revealed that the preferred method of getting rid of unwanted pubic hair is to use dental floss.

On Wednesday 25 October 2006 00:08, Fred A. Miller wrote:

On Tuesday 24 October 2006 6:04 am, Basil Chupin wrote:

...

...

5757 frames in 5.0 seconds = 1151.400 FPS 6577 frames in 5.0 seconds = 1315.400 FPS 7433 frames in 5.0 seconds = 1486.600 FPS 7409 frames in 5.0 seconds = 1481.800 FPS 7412 frames in 5.0 seconds = 1482.400 FPS

I think the subject of this thread is the latest, 8776, driver. What do you get from glxgears if you installed 8776?

I'm not far off from the above list, but it's still WAY TOO slow. The FX 5300 chipset is able to do a LOT more than that!!

glxgears is not meant to be used as a benchmark, it's only a quick way of seeing if you have 3d acceleration active at all. You really can't use those FPS numbers as a measurement of how fast your card is if "glxinfo|grep rendering" says "yes", then you have 3d acceleration.

May be to late to add these question but for my first usage of the command and next the first search about it. what happens when you get these: Eva03:/home/ikari # glxgears Xlib: extension "GLX" missing on display ":0.0". glxgears: Error: couldn't get an RGB, Double-buffered visual. and these: Eva03:/home/ikari # glxinfo name of display: :0.0 Xlib: extension "GLX" missing on display ":0.0". Xlib: extension "GLX" missing on display ":0.0". Xlib: extension "GLX" missing on display ":0.0". Error: couldn't find RGB GLX visual visual x bf lv rg d st colorbuffer ax dp st accumbuffer ms cav id dep cl sp sz l ci b ro r g b a bf th cl r g b a ns b eat ---------------------------------------------------------------------- Xlib: extension "GLX" missing on display ":0.0". Xlib: extension "GLX" missing on display ":0.0". 0x21 24 tc 1 0 0 c . . 0 0 0 0 0 0 0 0 0 0 0 0 0 None Xlib: extension "GLX" missing on display ":0.0". Xlib: extension "GLX" missing on display ":0.0". 0x22 24 dc 1 0 0 c . . 0 0 0 0 0 0 0 0 0 0 0 0 0 None Found little information about the command I'm really ignorant, but since you mention it and even test it, can you , good people, give me a fast idea of what it means? at least where I can get information about my problem. Thanks Carlos A. PS: Athlon XP 2600+, 512 MB Ram, 80Gb HD, nVidia Geforce 4 MX 4000, with SuSE 9.3 updated using apt-get and video driver: NVIDIA-Linux-x86-1.0-8774-pkg1.run <snip>

glxgears is not meant to be used as a benchmark, it's only a quick way of seeing if you have 3d acceleration active at all. You really can't use those FPS numbers as a measurement of how fast your card is

if "glxinfo|grep rendering" says "yes", then you have 3d acceleration.

-- <snip>

_________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

o.O!! I delete what is useless and keep what is needed ^-^ <snip>

...

what happens when you get these:

Eva03:/home/ikari # glxgears Xlib: extension "GLX" missing on display ":0.0". glxgears: Error: couldn't get an RGB, Double-buffered visual.

I would think you don't have the propietary nvidia driver properly installed.

Instructions here:

http://www.suse.de/~sndirsch/nvidia-installer-HOWTO.html#3

I have to use this: http://www.suse.de/~sndirsch/nvidia-installer-HOWTO.html#7 'cause I use SuSE 9.3 on an Athlon XP 2600+ I'm really late =( I spent the whole week "working" so I couldn't check the list email. After I sent this letter, I kept on reading emails. The advice to: chmod 666 /dev/nvi* did the trick I'm just wondering how =/ also I'm wondering why I had so many nvidia* like nvidia0, nvidia1, nvidia2 and so on. Already deleted, hope I'll be able to reboot and work normally.

- -- Cheers, Carlos E. R.

<snip> wie immer, vielen Dank! =) of course... excuse my ugly english =( Carlos A. Cheers=Prost=Salud =P~ I can't drink, sorry _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/

On Monday 23 October 2006 23:55, Peo Nilsson wrote:

On Tue, 2006-10-24 at 02:52 -0700, Steve Reilly wrote:

...

Is this the same driver you speak of that is installed when I did tiny-nvidia-installer --update?

907 frames in 5.0 seconds = 181.314 FPS 991 frames in 5.0 seconds = 198.052 FPS 995 frames in 5.0 seconds = 198.878 FPS 1000 frames in 5.0 seconds = 199.983 FPS 981 frames in 5.0 seconds = 196.103 FPS 992 frames in 5.0 seconds = 198.380 FPS

this is what I get on an athlon xp 2.2 ghz, although I have no complaints..... World of Warcraft and unreal run perfectly. You get 1300 FPS?? maybe were not using the same driver?

Athlon XP 2200+, Nvidia GForce 420 MX, nvidia driver 1.0-8774 gives:

(not full screen)

5757 frames in 5.0 seconds = 1151.400 FPS 6577 frames in 5.0 seconds = 1315.400 FPS 7433 frames in 5.0 seconds = 1486.600 FPS 7409 frames in 5.0 seconds = 1481.800 FPS 7412 frames in 5.0 seconds = 1482.400 FPS

AMD Athlon 64 X2 4800+ Dual-Core 1MB Per, eVGA GeForce 6800 GS CO 256MB DDR3/PCI-E/ (not full screen) 42936 frames in 5.0 seconds = 8587.200 FPS 52189 frames in 5.0 seconds = 10437.800 FPS 52224 frames in 5.0 seconds = 10444.800 FPS 45713 frames in 5.0 seconds = 9142.600 FPS 47248 frames in 5.0 seconds = 9449.600 FPS 50815 frames in 5.0 seconds = 10163.000 FPS 51996 frames in 5.0 seconds = 10399.200 FPS Jerome

On Monday 23 October 2006 23:55, Peo Nilsson wrote:

On Tue, 2006-10-24 at 02:52 -0700, Steve Reilly wrote:

...

Is this the same driver you speak of that is installed when I did tiny-nvidia-installer --update?

907 frames in 5.0 seconds = 181.314 FPS 991 frames in 5.0 seconds = 198.052 FPS 995 frames in 5.0 seconds = 198.878 FPS 1000 frames in 5.0 seconds = 199.983 FPS 981 frames in 5.0 seconds = 196.103 FPS 992 frames in 5.0 seconds = 198.380 FPS

this is what I get on an athlon xp 2.2 ghz, although I have no complaints..... World of Warcraft and unreal run perfectly. You get 1300 FPS?? maybe were not using the same driver?

Athlon XP 2200+, Nvidia GForce 420 MX, nvidia driver 1.0-8774 gives:

(not full screen)

5757 frames in 5.0 seconds = 1151.400 FPS 6577 frames in 5.0 seconds = 1315.400 FPS 7433 frames in 5.0 seconds = 1486.600 FPS 7409 frames in 5.0 seconds = 1481.800 FPS 7412 frames in 5.0 seconds = 1482.400 FPS

AMD Athlon 64 X2 4800+ Dual-Core 1MB Per, eVGA GeForce 6800 GS CO 256MB DDR3/PCI-E/ (not full screen) 42936 frames in 5.0 seconds = 8587.200 FPS 52189 frames in 5.0 seconds = 10437.800 FPS 52224 frames in 5.0 seconds = 10444.800 FPS 45713 frames in 5.0 seconds = 9142.600 FPS 47248 frames in 5.0 seconds = 9449.600 FPS 50815 frames in 5.0 seconds = 10163.000 FPS 51996 frames in 5.0 seconds = 10399.200 FPS Jerome ps 1280x1024 (SXGA) 24 bit color (full screen) 6659 frames in 5.0 seconds = 1331.800 FPS 6661 frames in 5.0 seconds = 1332.200 FPS 6791 frames in 5.0 seconds = 1358.200 FPS 6782 frames in 5.0 seconds = 1356.400 FPS 6790 frames in 5.0 seconds = 1358.000 FPS

On Tuesday 24 October 2006 19:18, Robert Lewis wrote:

Anders Johansson wrote:

...

On Tuesday 24 October 2006 08:54, Fred A. Miller wrote:

...

I've also noticed that the performace of the nVidia commercial driver stinks! I have a PCI Express 16X card with the Geforce FX 5300 chipset. In the past, it's proved to be very quick.......NOT now!! Below are my results with glxgears.......this is a P-4 3.2Ghz. box.

Try

chmod 666 /dev/nv*

What SUSE version are you running? On 10.1 I don't find a /dev/nv

There isn't one. I said nv*, the asterisk matters Also, they (the expression should match two files) are only there if you use the nvidia binary driver

* Anders Johansson [10-24-06 13:41]:

On Tuesday 24 October 2006 19:18, Robert Lewis wrote:

...

Anders Johansson wrote:

...

Try

chmod 666 /dev/nv*

What SUSE version are you running? On 10.1 I don't find a /dev/nv

There isn't one. I said nv*, the asterisk matters

Also, they (the expression should match two files) are only there if you use the nvidia binary driver

15:12 wahoo:~ > l /dev/nv* crw-rw---- 1 root video 195, 0 2006-09-06 23:08 /dev/nvidia0 crw-rw---- 1 root video 195, 255 2006-09-06 23:08 /dev/nvidiactl crw------- 1 root kmem 10, 144 2006-09-06 23:07 /dev/nvram -- Patrick Shanahan Registered Linux User #207535 http://wahoo.no-ip.org @ http://counter.li.org HOG # US1244711 Photo Album: http://wahoo.no-ip.org/gallery2

On Tuesday 24 October 2006 3:15 pm, Patrick Shanahan wrote:

* Anders Johansson [10-24-06 13:41]:

...

On Tuesday 24 October 2006 19:18, Robert Lewis wrote:

...

Anders Johansson wrote:

...

Try

chmod 666 /dev/nv*

What SUSE version are you running? On 10.1 I don't find a /dev/nv

There isn't one. I said nv*, the asterisk matters

Also, they (the expression should match two files) are only there if you use the nvidia binary driver

15:12 wahoo:~ > l /dev/nv* crw-rw---- 1 root video 195, 0 2006-09-06 23:08 /dev/nvidia0 crw-rw---- 1 root video 195, 255 2006-09-06 23:08 /dev/nvidiactl crw------- 1 root kmem 10, 144 2006-09-06 23:07 /dev/nvram

Interesting......I don't have /dev/nvram. Fred -- MickySoft, the ultimate corporate parasite.

On Tuesday 24 October 2006 6:57 pm, Anders Johansson wrote:

On Wednesday 25 October 2006 00:54, Fred A. Miller wrote:

...

...

crw-rw---- 1 root video 195, 0 2006-09-06 23:08 /dev/nvidia0 crw-rw---- 1 root video 195, 255 2006-09-06 23:08 /dev/nvidiactl crw------- 1 root kmem 10, 144 2006-09-06 23:07 /dev/nvram

Interesting......I don't have /dev/nvram.

nvram has nothing to do with nvidia, it stands for non-volatile ram. It's basically the EEPROM in your system, the CMOS

'Thought so, and wonder why the system didn't write the file to /dev. Fred -- MickySoft, the ultimate corporate parasite.

Carlos E. R. wrote:

The Tuesday 2006-10-24 at 19:08 -0400, Fred A. Miller wrote:

...

...

On Wednesday 25 October 2006 00:54, Fred A. Miller wrote:

...

...

crw-rw---- 1 root video 195, 0 2006-09-06 23:08 /dev/nvidia0 crw-rw---- 1 root video 195, 255 2006-09-06 23:08 /dev/nvidiactl crw------- 1 root kmem 10, 144 2006-09-06 23:07 /dev/nvram Interesting......I don't have /dev/nvram. nvram has nothing to do with nvidia, it stands for non-volatile ram. It's basically the EEPROM in your system, the CMOS 'Thought so, and wonder why the system didn't write the file to /dev.

Anything written there get erased on the next boot. I don't have any of those three files, by the way, even though I was testing nvidia the other day.

I just downloaded the latest nvidia driver. One thing I discovered is that they provide *excellent* documentation. There is a full manual (I think it is called 'readme' though) with lots of excellent stuff including which driver options to tweak to change the permissions of the device files, which the driver recreates each time it starts, IIRC. There's also a lot of information in that manual and in the knowledgebase and release notes on problems with specific hardware/software configurations and how to get around them. I really recommend browsing their site if you have problems. Cheers, Dave

On Oct 25, 06 17:40:40 +0200, Carlos E. R. wrote:

...

I just downloaded the latest nvidia driver. One thing I discovered is that they provide *excellent* documentation. There is a full manual (I think it is called 'readme' though) with lots of excellent stuff including which driver options to tweak to change the permissions of the device files, which the driver recreates each time it starts, IIRC.

There is even better documentation here:

file:///usr/share/doc/NVIDIA_GLX-1.0/html/index.html

This should be exactly the same content, just in HTML. Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ labs www.mshopf.de