Graphics Card Problems

Pc specs:
Processor: Core 2 Duo @ 1.86 GHz
Mbord: ASRock 4 Core Dual-VSTA
Memory: 2x512 MB PC3200
Graphics Card: BFG tech Geforce 6800 GT

Short Story:
Last year I was having some small bugs running some newer games, so I decided to try the newest beta drivers (don't have the exact version on hard). They fixed the problem, but every once in a while while running a game the graphics card would apparently fail and force a complete reboot (no BSOD, no error message, just an instant reboot). It wasn't often, maybe once during 2 hours of playing, so it didn't bother me.

About two weeks later the graphics card failed completely. Upon trying to start the system it would start for an instant and then immediately stop. After trying to start it once it wouldn't even try starting, to attempt to start it again I had to disconnect the power supply (presumably to clear the BIOS). Taking the graphics card out allowed the computer to start. So since the card was bad I was able to get the manufacturer to replace it. That card has worked ever since.

Skip forward to today. Last week I got Mass Effect, and it was completely unplayable without upgrading drivers. So I updated from 169 WHQL to the 175 WHQL drivers. This allowed me to play, but I started experienceing the same symptoms again. Now after only one week of playing, the card is similarly bricked again. I don't think I can get it replaced again, the warranty time has probably ran out.

So, my question is: what the hell? Is the BFG card series just a useless piece of junk? Is Nvidia putting out drivers that kill off your card? The card was running fine for months until I was forced to update drivers, then I lost it in under a week. I am notifying Nvidia of this problem but from my experience with their tech support the reply I get back will probably be useless, and fixing driver incompatabilites with cards years old probably won't be on top of their list.

Edit: I am running Windows XP SP1 btw, and I am sure my GPU was not overheating, the temps were topping off at around 60ish even under heavy load.

what are you using to report your temps? it maybe that the software or the heat sensor is lieing to you. 60c under load, on a new game is really low unless you live in the arctic circle and have all the windows open or are water cooling the gpu. also xp sp1 is a little old, i believe xp is up to sp3 which you can dl from M$ for free, unless you have a pirated version of xp.

I was using Rivatuner to report my temperatures. It is accurate as far as I know. My computer is in a well ventilated area and is in my basement which is pretty cool the entire year. I haven't seen the temp get over 63 C even after pushing my card to the limit in UT3.

I don't use the SP2 or 3 because it breaks many programs on my computer. I'll take the slight chance I may get a virus (even though I have proper protection and checker already, and I don't do stupid stuff) over having stupid stuff that pisses me off stare me in the face every time I boot my computer. Incidentally thats the same reason I don't have vista.

Try updating motherboard drivers, maybe a shot in the dark like.

I have the same board and ever since i installed my HD3850 i have had nothing but problems in certain games. It can be fine for ages, and then act up for a few days.

I have had it reboot on me too, but id recommend trying anyway.

I should have the newest drivers, they were last updated on the Asrock website in 2005, and I downloaded them last year. I can't check at the moment though because I haven't been able to steal another graphics card off a friend so I can try booting.

There is no reason for a computer to just start 'acting up' for a few days. Computers are electric currents running through transistors, they aren't people who start being annoying because they had a bad week.

I've decided to hold off on annoying Nvidia for the moment though, sent my complaint to BFG since its their card and I assume Nvida would be pretty prompt on fixing a problem if their drivers were affecting all nvidia cards. I have a 95% expectation of their reply being something along the lines of 'your warranty is up, we dont care' though.

Maybe a bad power supply? Unstable voltage from an underpowered/faulty PSU can cause reboots, shut-offs, and can ruin anything it's connected to, especially the biggest power-hog in a system: the vid card. I'm not saying that's definitely what it is, but it's worth looking into.

Thats what I would think, but I have run without any hiccups at all for about 7 months up until the day I installed the 175 drivers. Unless the PSU decided to fix itself in the meantime and then go bad again I don't think it could be the problem. I will test it though, can anyone enlighten me on how to do so? Hardware diagnosis isn't my thing.

My PSU is rated at 400W, that should be enough for anything less then a system with 2 SLI high end cards I think?

Model is Allied AL-A400ATX

400W should be ample for that setup.

What brand/model is your PSU? They vary somewhat...

I had a ton of issues with 175. The NVidia control panel crashes, GW ran at about 20 fps less than 169 drivers. My computer would also freeze up for several seconds and the screen would go wavy at times. Reverted back to 169 and everything has been fine. I sent a support question to NVidia but 0 response, same as I've gotten with NWN issues I've had for 1+ year.

Not as bad as killing your card though, aye?

At least my motherboard allows me to upgrade to a nice shiny PCI-E card, though anything new will be far out of my price range for a while. If there is any chance this problem is because of my motherboard or PSU or something though I sure want to fix it now before killing another expensive card.

EDIT: In the meantime, any suggestions on what card will give me the best price per performance ratio for around $100-150? I can handle PCI-e or AGP. Does ATI still have the crappy driver support they had a few years ago? If things are better on that end I am not biased between ATI or Nvidia either way.

EDITx2:Geforce 8800 GT for 130 after rebate, good buy for the price?

1 8800gt is awesome for guild wars. I can run the game at max with that card and still maintain above 60fps.

now for the bad news the card is rated for a 500watt psu. that came off of the box. modern components require more power. a few free temp programs that are easy to use are: core temp and everest altho i do not know how much i would put faith in the later but it is really all inclusive. both are easy to use, google will help you find them.

qft same boat.

The 8800GT is a good buy, yeah.

And with the new cards coming out soon here, I'm sure the prices are only going to drop more, I'm waiting to see.

Heh, I was perfectly fine with the performance of my 6800gt, even in mass effect where it supposedly only met the minimum requirements I was running in 1280x1024 resolution and it looked great. I am more interesting in future-proofing (even though we all know the idea behind that word is a joke in computers, but meh) for in the future for when Vista stops sucking and good games are coming out that won't support XP. I read the 8800GT is supposed to run crysis very well and crysis is gonna be the king of performance tests for a while I think.

I have read some reviews of the 8800GT, and they actually have whole systems running at below 300W. I have a techie friend whos gonna take a look at my 6800 and try to see what was screwing it up, and he says he has a 550W PSU I can buy off him for $20. No reason to chance it so I'll be doing this anyway.

My computer wasn't experienceing any of these control panel crashes or running slower because of 175. The symptoms were more like the graphics card overheating and failing, forcing a system reset. I just can't believe it has been overheating, rivatuner has never failed me before. It's just displaying the temperature sensor on the graphics card, its hard to screw that up. Now if the graphics card temperature sensor itself was bad...

May or may not be related, but that PSU is not highly thought of. It's a bog-standard PSU usually found in pre-builts. I've looked at reviews since 2003 - it's not a new model. If you look at the Newegg page, you can see that the favourable reviews were early on in its life, but every review recently is absolutely dire.

http://www.newegg.com/Product/Produc...OnlyMark=False

Did see a patchy report on Anandtech saying it had very suspect +3v figures - they stated at the time that it was "basically a re-badged 350W."

If you don't want the fastest board available I would go with ATI. I have 2 other computers that have had ATI cards for several years with 0 issues. The only time I had a problem they replaced my 9700 with a 9800. Also their drivers have been released about every month instead of every 6 months like NVidia. Also 29% of all Vista crashes are because of bad NVidia drivers. I would think twice before buying from a company with horrible driver support.

Actually you had the reviews sorted by score, not by date, when you sort by date you see the bad reports are pretty much widely spread out across its lifetime. I'll take a second look at the PSU though.

How does one test a PSU at this point though? Testing it when it isn't under load is pretty meaningless, and I can't very well load it right now with a bricked graphics card.

Some of my hesitation for getting another ATI card is I had a lot of trouble with their drivers as well. A month ago I was helping my brother out with his computer, and I had to update his drivers for his 9600 pro. It was god awful. I tried the newest verison, and it crapped out, then I started working backwards until the newest one that didn't fail to detect the card was the Catalyst 7.8 version (the most recent ones being 8.5, so you can see how far back I had to jump to get drivers that worked).

Can you give me a list of specific graphics card names along with their relative performance? I could care less about Vista compatability right now, I can't forsee myself using for at least a year. I hear Nvidia's drivers for Vista have much improved recently anyway (they were apparently pretty much useless when Vista came out though, I'll give you that).

Volt/Multi-meter? You could try your friends 550 Watt PSU and see if it still overheats your GPU. Could be that you just have bad/insufficient airflow or something. Leaving the side panel off will increase cooling a bit too but I'd don't recommend it as a perminent solution. Also temp sensors aren't allows very accurate or can become inaccurate.

The other two computers still using ATI cards are a server using X1300 Pro and a Multimedia Center using 9200. Neither are built for gaming at all. I do occasionally use the server to run GW which it does ok. If you want fps for that I can test it. The NVidia 8600gts I have runs above 60 fps at full detail and 2048x1536.

A multimeter alone doesn't load the PSU though. In this case where everything was running fine for months until finally dieing its a sure bet that if the PSU is bad its just barely bad, so the only way to find if its failing would be to test it under a full load.

I don't think I can test whether my GPU is overheating anymore, I am pretty sure its dead. I will try to swap it in to another computer though.

Well I would like something around the 8600-8800 level of nvidia cards. If ATI could get me something with about the same performance costing less I would go with them. Unfortunately comparing speeds in GW doesn't mean much, I was getting 60 fps at full detail at my max resolution of 1600x1200 with the 6800gt, so if possible can you give me relative performances in newer games? I really have no idea of the relative performances of the ATI cards, I stopped keeping up with them after I got my geforce 6800 a few years ago.

Figured out the problem. Tried the card out in another computer with a larger PSU, and it literally caught on fire and blew up. Apparently the power regulator on the card was gone, so when we hooked it up to power supply that could give what the card was wanting, it went up in flames.

I tested my powersupply and it appears to be just fine under load. My friend thinks the problem was that although my power supply was rated for 400W, it only supplied 17A at +12V, which according to him is quite low, and new graphics cards tend to require more and more on +12V. Either driver upgrades somehow caused the card to work harder and draw more power or it was a huge coincidence both times. Maybe the newer drivers would detect the power supply problem and shut down the system, where the older version just kept going.

My new power supply will give 47A at +12V so I should have more then enough there, but it makes me mad that I have no real way of telling what my new card draws, all I can find out is "requires x-Watt powersupply" which doesn't help much.

Told you that PSU was dodgy

(Ok, more of an educated guess, but still ^^)

I believe that card requires ≥18A on the 12v rail.

Not really the GPU's power regulator was missing/damaged so technically it was the GPU itself that was malfunctioning.

Absolutely correct.

Note to self: when clock hits 6.43 am, it's probably wise not to post.