PDA

View Full Version : OpenGL causing Video Crashes...



IPvFletch
03-04-2009, 07:23 PM
I don't know what is going on but the last 2 days I have been experiencing what I believe to be OpenGL stability issues.

The symptom essentially is weird "shards" of my screen flickering and showing odd colors, usually purple+pink, for split-seconds. Sometimes it is a triangle, other times it is a line, or a trapezoid shape, or a combo of those. It is kind of wacky looking, like as if your LCD shattered right before you eyes, so lots of sharp lines and geometric shapes.

I first noticed this 2 nights ago while playing Quake Wars (ETQW). I have been playing ETQW for over a year on this PC, usually 2-3 hours a night (or more admittedly, I'm an addict, at least I can admit it though!) :)

Last night it was so bad Quake crashed and hung my entire system.

I rebooted and then Windows was crashing/locking up video-wise and thusly locking-up entirely moments later.

I let my PC cool off a bit thinking that was the issue. It didn't seem to matter. I let it cool and left it idling overnight, like I always do, and it was hung this morning.

I installed the new NVidia drivers this morning and it seemed fine all day but when I tried Quake a bit ago, it began to do the same thing. At one point trying to get Quake going last night, I managed to alt+tab out and I happened to see a popup (hidden by Quake's screen but I could see it in my system tray) that said something about Nvidia OpenGL. It looked like it was some error popup, but I couldn't get to it in time (I'd have to move it to my other monitor) because Quake right then locked up my whole system.

I also just downloaded FurMark v1.6.5 and ran that. Within 60 seconds, it began to do the same thing. It was not as bad and it didn't lock up my system, but I exited it about 3 minutes into the test, just in case.

So now I am almost convinced it is related to OpenGL, but specifically I am certain it is related to my video card/settings in some way.

NOTHING HAS CHANGED IN MONTHS in terms of video settings, quake settings, or even windows system or display settings.

Here are my specs, for reference. My system has not changed in about 8 months and like I said, Quake has been working great for months (800+ hours logged), up until now:

Windows XP SP2 - 32bit
Intel QX6850 3.0GHz (quad-core, not overclocked)
Asus P5Q-EM Motherboard
XFX 8800GTS 640MB (the "overclocked" GTS model)
NVidia 182.08 WHQL Driver for XP 32bit with stock settings
2 x Dell 22" Widescreen High-Color LCDs (1920x1200 each)
Several system fans
SATA 120GB HD
Internal IDE & External USB 500MB SimpleDisk both for Backups
(Yes I mirror my SATA drive each night alternating just in case)
3 x 1GB DIMMs (not cheapies, onboard heat sinks/blades)
Rosewill 500W PSU
Lots of USB peripherals
Keyboard, Mouse, Speakers, USB Mic, etc...

This is really annoying and I don't know if it is Hardware related (like a bad video card or bad memory), or if it is dust in my system (like my video card's fan), or if it is something else. My system is not too hot, even when gaming, and I checked my Nvidia GPU temp using NVidia Monitor (and also confirmed with FurMark just a moment ago) and it idles around 58-60C, and gets up to 65C when gaming or during the FurMark test. I've read reports that these GPUs can easily withstand 90C temperatures under load, and I'm coming nowhere near that. I busted out my can O' compressed air and took it to my PC internals, but that didn't seem to do anything (except make me sneeze a lot). I also ran my PC for a while with the case open. I think it actually made my system run hotter (according to Nvidia GPU temp) by 1 or 2 degrees, if you can believe that.. But still, no dice.

ANY IDEAS ON WHAT IT COULD BE OR HOW TO TROUBLESHOOT??

I could not find anything in the XP Event logs, and I can't find anywhere to troubleshoot or log NVidia, OpenGL, or Direct3D functions, so I don't know if it is related to that or if that would tell me more info or what...

Thanks in advance!!

IPvFletch
03-04-2009, 08:11 PM
I just read on nvidia.com's FAQ that this could mean a damaged GPU core or Video Memory... :( :(

Brolingstanz
03-05-2009, 02:05 AM
Don't suppose you have another card handy you could try... probably the only way to eliminate the hardware question.

Check BIOS settings for e.g. memory voltages to make sure they're in line (this bit me once), and that you're not overclocking anything. Make sure you have the latest BIOS from your mainboard vendor and that your drivers are up to date. (If you've tinkered with BIOS setting at all you can restore defaults from the main screen but watch your memory voltages!)

Install SP3.
Get the game's latest patch.
Completely uninstall video driver then reinstall it (perhaps to a version that was known to work for you).
Otherwise to eliminate the software question you can obviously format and rebuild - in the quest to leave no stone unturned and all that...

A few other things to try....
- Re-seat video card and memory sticks
- Go down to 2 memory sticks, then 1 (DDR works best in pairs of 2, btw)
- Go down to 1 monitor, keyboard and mouse - whittle system down to the nub
- If board has diagnostic lights/LED make a note of them during boot and check the manual for status.
- Try a different PSU (sudden hangs or trouble booting could be a sign of a PSU heading south). After a year of gaming that Rosewill may be on its last leg.


...

painterb
03-05-2009, 07:18 AM
Did you make sure to blow out your video card's fan and air channel with the compressed air? I've seen those get filled with dust bunnies to the point it was causing the GPU to overheat (which is what it sounds like is going on to me). Also, similar to what Brolingstanz suggested, take a look at your power supply voltages (if you can) and make sure they are within spec. Get a new one if needed.

IPvFletch
03-05-2009, 08:39 AM
Wow, I am just glad somebody replied to my thread. :)

I have heard mixed reviews on the Rosewill, and in fact I think on Anandtechforums, they made fun of me for using it. *shrug*

:)


I was also thinking it could be bad PSU, as that is likely the only element in my case which has variable output (i.e. power). I don't have another handy unfortunately. Maybe I will see about getting a new one - any brands you'd recommend?

I need another can-o-compressed air, but I will re-spray my Nvidia card/fan and see if that makes a diff. I doubt it though because I'm seeing this happen at ~61C, which is plenty safe for this card.

I will try 2 memory sticks. I know it was supposed to be 2, but I had 4 and ran Vista and, well, I jumped on the Vista-back-to-XP bandwagon (and couldn't be happier) and left them in there, but I was told XP only supports 2 but XP SP2 supports 3, so I plucked one. I will pluck another and see if that makes a difference. But honestly, I don't see how it could (unless of course one of them is bad) because it has been working fine like this (with 3GB) for ~6 months or more. I also want to do a memtest so I downloaded the ISO but need to reboot and devote some time to that.

The game has the latest patch already as do I have the latest drivers and there are no new firmwares out, or else I'd try them too.

I don't have a different video card (except on-board). I'm certain if I tried another video card, it would solve my problems. I know this because with a new video card, comes new drivers, and all sorts of changes, so I'm sure it would resolve the issue but it doesn't really help me in narrowing it down. Now if I had the exact same video card, that would be the ideal test, but I'm not in a position to dump a few hundred $$$ just to prove this theory -- at least not until I'm convinced it is possibly a bad card (which is soon becoming the case). Then again, Fry's has a great return policy on things like this, don't they? :) Then again, say it was my mobo, and say it broke this new test card, then I'm truly SOL (cuz Fry's tests HW before taking returns).

Is there any way to (stress) test my video memory???

Keep the ideas coming.. I obviously want to try the least intrusive options first, and then do some minor and then major surgery to get this working.

IPvFletch
03-05-2009, 12:42 PM
Ok, well it got worse this morning. It was fine all night and all morning and then suddenly it showed some pink in the shaded areas and my mouse had a black spotted box around it. My system then hung after some minimal mouse movement.

I rebooted and noticed the BIOS and Boot screen was now showing vertical lines, alternating pink and blue, and they were dashed lines.

I tried swapping out memory. No dice. I switched to my on-board VGA and of course that resolved it.

Still not sure what is wrong. Could it be a bad HDMI cable? Bad Video Card? I doubt it is bad drivers because it was working fine until one day it started doing this with no driver or other system changes...

I also monitored my PSU voltages as read from my BIOS:

CPU 1.27v
3.3v 3.36v
5v 5.112v
12v 11.84v

It is pretty consistent. Is that 12v too low?? Isn't the 12v used to power my PCIX Nvidia (XFX 8800 GTS) card?? Hrmmm...

I'm about to give up on this and go buy a new video card.. :(

ZbuffeR
03-05-2009, 01:41 PM
Did you follow painterb advice about dust and overheating of the GPU ?
It is probably too lake, but anyway. I had a fan on a video board that stopped working due to dust etc. I bought a new one, swapped with the old, and it went perfectly well after that.

Brolingstanz
03-05-2009, 02:11 PM
...

Ceramique your mosfets

...

IPvFletch
03-05-2009, 02:35 PM
How can I check the mosfets though? I'd have to disassemble this card, and I don't think it's worth it (I'll likely break it worse and then be further from finding out the truth).

Yes, I dusted the heck out of it. No dice. Now it is even showing more vertical lines, alternating RGB, about 1/4" apart, all dashed lines, about 10 total on the right of my display..

Anyways, I think you're right -- TOO LATE.

OK so doing some research on a new card.. :)

I found a Geforce 9600 with 768MB which I'm planning to get. It is not as nice as the 8800GTS but damn close. The thing is, I just noticed it says "requires 12v @ 26amp power". Well, I looked again and the 8800GTS I have even required 30amps!

I checked my PSU and it's a measly 18amp for the 12v rails.

So that means I've been WAY underpowering my Nvidia for some time now... I read elsewhere this could cause it to heat up excessively...

CRAP!

So I wonder if that is what actually happened or if it was just a bad card, or a faulty PSU, or worse, something else I still haven't identified???

BTW, to go along with the new video card, I also found a Corsair (awesome reputation) with a single 12v rail of like 41amps.. It should be more than enough...

WHAT DO YOU GUYS THINK??

I'm quite :( now...

But hopefully soon I will be :) again...