Just found this article at Linux Today about a mem paging bug in Athlon/Duron cpu's, it's the same as the win2k agp bug from 18 months ago and it only effects custom kernels optimised for these cpu's. The full story and a temporary fix can be had at gentoo.org. I've applied the fix and when I get the time I'll be having another crack at nvidia's 1.0-2314 drivers to see if it fixes the random lockups that I still get.

I have an AMD system. This bug has been preventing me from changing my voodoo card for my nvidia card.

I ended up with a pentium3 slack box running the nvidia.

Now I can go with my Radeon on the AMD


thinks for the info

it's now officially called the 'INVPLG' bug - apparently, it's not really an athlon bug - more like really faulty documentation between manufacturers. what bugs me about this is that it mainly affects desktop users - newbies - and god knows how easy it is to scare them off. i just responded to what i think was the first case of someone with this bug, but god knows how many bad nvidia installations we've had in this forum that could have been affected by this.

Anybody read the Linux Journal from Nov 2001 about putting together a high-end linux box?

I picked up the issue cuz I'm gonna be putting a new box together soon and I wanted some tips.

Anyway, I was gonna go AMD this time (like they did in the article) and probably ATI for video. The article basically said why buy hardware from vendors who don't support the open source community (they recommended not buying nvidia cards). They chose an ATI Radeon because it is well-supported in the open-source community.

Has this fact changed since the article?

When I buy the components for my new box, I don't want to be buying into any problems....

Not sure if this post really belongs here. I just got to thinking about the article when I read the nvidia comments in this thread.

well... i'd been under the impression that nvidia's support of linux wasn't half-bad. they give us ready-written drivers which are half-open source, and half-binary. they choose to keep some source closed... but at least they have some support out there. i've seen more unsolved ati problems on this board than nvidia probs.

and, again, the bug wasn't nvidia's fault - technically, it was the agp-gart.

Mine for a start......adding mem=nopentium to lilo has resolved the random X lock ups I was getting, I'm back using NVagp x4 again and I'm going to reinstall the -2314 drivers on the weekend and see if they'll work this time.

I have my nvidia on my slack box with a pentium, It works great. 3D games work really well.

I have experienced the same problem as Aussie has, and I just applied the fix mentioned, let's see what happens, so far I don't see any drop-outs in performance. I have 850 MHz of course Athlon socket A with 512 MB of SDRAM and GeForse256 ELSA ERAZOR X video card.

Well, then why all the nvidia questions/problems all over the discussion boards?

This may be getting off the track of this thread but:

I have an nvidia geforce2, but I can just forget about trying a different rez. If I try a diff rez, well my computer is just screwed from then on... windows don't display right, mouse might dissapear, just cant do anything. Have to hit reset.

I have a PIII 1Ghz, not an AMD. I have mandrake 8.1.

Now, I can't believe it's such a problem just trying a different rez. I mean, what is it in windoze - a couple of mouse clicks? I change my rez in linux, trashes everything, and time for the reset button.

Its either linux, or nvidia, or mandrake, or maybe just me...

Could be me, still very much a noob.

What version of X do you have installed, what version of the nvidia drivers and download nv_check.sh (left click) and run it to check for conflicting openGL libs.

Ok, thanks, I'll check on that later. I'm at work all day now.