After 2 days of troubleshooting, I think I isolated the vid card [nope]

Ichinisan

Lifer
Oct 9, 2002
28,298
1,235
136
[edit]
Even though the vid card is messed up, it's not the cause of the infrequent reboots. Looks like I've got a few more days of troubleshooting to do...
[/edit]


PICTURE

Look at the picture. Would you suspect a faulty pixel pipeline? Vertex shader? Bad video memory? Hot GPU? The screenshot comes from an AGP GeForce FX 5200 and the problem follows the card when moved into a different computer.

I've been troubleshooting this machine for a customer for the last 2 days. It has been suffering random lock-ups and reboots while using Windows and Microsoft Office. I've already ruled out the memory, CPU, cooling, power supply, and IDE devices. I've updated the BIOS, Intel chipset/INF driver, video driver, cleared CMOS, etc. It still restarts and/or locks up occasionally. Sometimes it will seem like the problem has gone away for several hours of constant use, then it will come back unexpectedly and get stuck in an endless restarting loop again.

Finally, Windows was aware of a long string of crashes and submitted a few error reports after it stopped restarting. They all identified the video driver "stuck in a loop" and suggested downloading the latest version (I had already installed the latest from NVIDIA's site). I downloaded a few NVIDIA demos and they all looked fine except for the Ogre demo. When the Ogre is walking backward (the first few seconds of the demo), it's backside is a flickering, distorted, jumble of polygons. It then clears up and looks fine until the demo starts over again. I pulled the card out and put it in my system and I get the same results. I put a GeForce 6800 in his system and there was no problem.

I noticed that there were some annoying "MSI" tabs in Display Properties (the main dialog, not "Advanced" where the NVIDIA tabs should be). They kept coming back even when I removed all MSI and NVIDIA software, uninstalled the display adapter, restarted, and reinstalled the latest drivers from NVIDIA. I found instructions online to delete msicpl.* files and it worked like a charm. I uninstalled/reinstalled the NVIDIA drivers one more time just to make sure that they had not been modified by MSI's bullsh!t, but the 3D glitch remains.

MSI GeForce FX 5200
Motherboard - MSI (some board with flashy multi-colored lights on the north bridge HSF, integrated LAN, no integrated video)
Sound - Integrated (C-Media)
Memory - 1GB (2x512)
CPU - P4 2.8 GHz
 

CrispyFried

Golden Member
May 3, 2005
1,122
0
0
when overclocking the core of my 9600 I got stuff like that. wasnt heat or memory, just the core OC.

try underclocking it if you can, otherwise I would think the core is toast.
 

Ichinisan

Lifer
Oct 9, 2002
28,298
1,235
136
Thanks. That's what I needed to know. I'll see if there's any warrantee on this card.
 

Ichinisan

Lifer
Oct 9, 2002
28,298
1,235
136
Well, it looks like the system started rebooting again after I used it with no problems for an entire day (with my brand-new GeForce 6800 video card).

I have not yet been able rule out:
Hard Drive (performed surface scan, but there could be a problem with drive electronics)
Motherboard
CPU
Windows Installation

Eliminated:
Power Supply
Memory
Optical Drives
CPU cooling
Video Card
 

jayanath

Senior member
Jan 20, 2006
214
0
0
ur saying with a new card theres still problems.
undo any overcloks
or reinstall windows
 

TanisHalfElven

Diamond Member
Jun 29, 2001
3,512
0
76
reintall windows.
maybe your customer OCe'd then undid it. cuz i remember ocing my opteron 165 to 2.6 without increasing the voltage and that caused major problem with my windows install. a lot of programs stopped working and such.
 

pkrush

Senior member
Dec 5, 2005
468
0
0
Are you sure you can still rule out the power supply? A 6800 sucks a LOT more power than a fx 5200 does.
 

CraKaJaX

Lifer
Dec 26, 2004
11,905
148
101
I've had a similar problem on an old computer of mine. Long story short, I bought a 9800 Pro and put it in and started getting the exact same symptoms you're seeing here. Finally figured out that my PSU couldn't handle it, so I returned the 9800 and got a 9600. Problem solved.
 

Ichinisan

Lifer
Oct 9, 2002
28,298
1,235
136
Originally posted by: jayanath
ur saying with a new card theres still problems.
undo any overcloks
or reinstall windows

Originally posted by: tanishalfelven
reintall windows.
maybe your customer OCe'd then undid it. cuz i remember ocing my opteron 165 to 2.6 without increasing the voltage and that caused major problem with my windows install. a lot of programs stopped working and such.

When I first encountered it, the BIOS had a supervisor password. Suspecting that the original system builder overclocked it, I tried to check that early on in the troubleshooting process. I had to reset the CMOS to get into the BIOS setup utility, so any settings were lost anyway. Once I was in, I saw that the temps were fine and noticed that the "dynamic overclocking" feature is enabled by default. I loaded BIOS defaults (instead of "high performance defaults") and the feature was disabled. I went on to disable unused serial (COM) and printer (LPT) ports and tweak boot time by detecting IDE devices and adjusting boot order. I have tried changing settings like "system BIOS shadow" and the problem continues.

Due to the extremely unpredictable appearance of this problem, I doubt that it's a software problem.

Originally posted by: pkrush
Are you sure you can still rule out the power supply? A 6800 sucks a LOT more power than a fx 5200 does.

Knowing the symptoms, I came prepared with my Enermax power supply and tried it on-site before picking up the computer or installing my video card. I also reapplied thermal paste (Arctic Silver 5) to the CPU while I was there. This was all after I had the customer run Memtest86 overnight and it returned no memory errors. After ruling out memory, heat, and PSU, I had no choice but to bring it home and continue troubleshooting.
 

Ichinisan

Lifer
Oct 9, 2002
28,298
1,235
136
I'm betting MB too. Everyone place your bets...we'll know in a few days.

Possibilities:
-Motherboard
-CPU (not heat related)
-hard drive

I think I further ruled out memory after trying my Corsair DDR400 modules and getting the same thing. Is there some possibility that this MSI motherboard is picky about memory?