View Single Post
Old 07-26-06, 05:11 AM   #326
Registered User
Join Date: Dec 2002
Posts: 7
Default Re: Trying to make headway into finding the Xid crashes source...

Originally Posted by energyman76b
Why are you so convinced that the Xid's are nvidias fault?

There are the manufacturers of the cards, and this guys can make A LOT of errors.

Examples? Not enough cooling by design, not correctly installed heatsink (both would lead to overheating and Xids... solveable by underclocking ... hmmm).
Bad caps. Bad routing on the PCB. Bad current supply. Low-cost-low-quality resistors.
This are all points that might lead to misbehaving cards - and this is all stuff that nvidia can not solve.
That's simply because everything works fine after ctrl-alt-backspace, rmmod nvidia, modprobe nvidia, startx.

My problem:
After using Blackbox, Mozilla, Licq, Dillo, xterm for about an hour, every OpenGL application results in 100% CPU in the X process and
NVRM: Xid: 8, Channel 00000002
on bad days followed by
NVRM: Xid: 13, 0000 02004a00 0000004a 00000800 8e0c31c6 00004000
I usually save everything and do a sync if I want to use OpenGL.
I then run glxinfo and if it doesn't finish in 2 seconds, I frantically hit ctrl-c.
If I wait too long, I have to ssh to my box and do a kill -9 on the X server.
If I then rmmod nvidia and restart X, OpenGL works fine.
When I periodically run OpenGL apps, the system remains stable.

This sucks.

My configuration:
ELSA Gloria Synergy II (TNT2, passively cooled)
Pentium 2 board with Intel 440LX/EX chipset
NVIDIA driver 1.0-7174 (happended with older versions as well)
AGP Rate: 2x
Fast Writes: Disabled
SBA: Disabled
denial is offline