PDA

View Full Version : ARBvbo && NV && linux ... once again



vincoof
05-13-2003, 01:11 PM
Hi,

I've installed the latest NVIDIA drivers for Linux and I'm running into a problem with support for the ARB_vertex_buffer_object extension.

This time, every call to VBO functions crashes. No matter if I call glGenBuffersARB or glBindBufferARB it always crashes.

I'm using a GeForce4 MX 440 card on a Mandrake 9.0 system.

Anyone else tried the latest linux drivers with vbo ?


Note: I ran the simple_vbo executable which also sends a Segmentation Fault error.

satan
05-13-2003, 01:59 PM
Originally posted by vincoof:

Anyone else tried the latest linux drivers with vbo ?


Yeah, me.
Everything stayed the same as in my last post about this topic.
Therefor I came to the conclusion that the bug concerning vbo on Geforce 2/4 MX cards is not fixed.
So I am just waiting for the next driver release and hoping that it will be fixed.

Gorg
05-13-2003, 02:06 PM
I have a geforce 3 and everything runs absolutely great. Might just be a bug for your card.

Vincoof : It might sound silly, but are you sure the functions are properly initialized? I mean, I would suspect a driver bug if only one function was failing, but if all them are, that sounds quite fishy.

cass
05-13-2003, 02:30 PM
Sorry for the trouble. We've repro'd this internally at NVIDIA. The problem has been fixed, but it will take a little time for the updated driver to appear.

The simplest work-around is to use an NV2x or NV3x card if you have one available.

Thanks -
Cass

velco
05-13-2003, 10:34 PM
Originally posted by cass:

Sorry for the trouble. We've repro'd this internally at NVIDIA. The problem has been fixed, but it will take a little time for the updated driver to appear.

The simplest work-around is to use an NV2x or NV3x card if you have one available.


Works just fine on GF4TI with one annoying problem: it does not seem to be any faster than plain vertex arrays. (I'm 99% sure I'm using it correctly, my program has support for both VAO and VBO and with the very similar VAO it shows 100-200% speedup).

~velco

vincoof
05-13-2003, 10:51 PM
Thanks for the info cass.

I'm just a bit sad since now it works even worse than before, but I know I *just* have to be patient. Well, what else can we do anyway ? http://www.opengl.org/discussion_boards/ubb/smile.gif

Gorg: The problem does not seem to appear on GeForce3 and GeForce4 Ti cards. That's probably the reason why it works for your card.
The function pointers ARE initialize properly because I'm not the one who initializes them. Since the 43.49 drivers, the function pointers are already available just like glBegin is available for instance. Anyway I'm using other extensions like register combiners and vertex programs which work very well. Moreover the VBOs worked with the driver before (43.49) on my GeForce4MX under some circumstances.

Patience, patience...

satan
07-29-2003, 03:04 PM
Just for everyone who is interested the vbo bug on Geforce2 MX and 4 MX cards seems to be fixed in the new NVIDIA Linux drivers (4494). At least it works on my Geforce2 MX and I am just happy with the performance gain (I used vanilla VAs before).

thx

Mihail121
07-29-2003, 11:36 PM
Hi!I also have a GeForce4 MX and the VBO bugs are simply killing me.If i'm not wrong there is something wrong also with the NV extension but if they are gonna be soon fixed who cares :d.