Graphics Card failing?

Boro

My system:
Intel Core2 Duo E6600 @2.4ghz
Gigabyte GA-965P-S3 mobo
OCZ 450w PSU (modular)
2gb DDR2 ram
Radeon x1900 512mb PCIe gfx
2xIDE, 1xSATA hdd
XP pro


Its been a good few years since I was seriously into building/tweaking so when I built this system a few years ago (when the components were current) I just wanted a stable system. I got that, it was a good game playing machine and rock solid.

Over the last year though games such as Mass Effect 2, Dragon Age, etc, started giving me GPU lockups (screen goes black, gpu recover starts) and then locking the system completely. I persevered because they were infrequent and tbh I wasnt entirely convinced it was not the games fault. I checked temperatures, voltages, drivers, cleaned out dustbunnies, reapplied thermal paste and all that. No problems.

Now in the last few days I have been getting much more frequent lockups as soon as I do anything 3d (such as playing quake live for example). I have been getting artifacts on reboots after such crashes. I have been tracking with SiSoft Sandra, ATITool, MBM5, RivaTuner and various programs like that and apart from GPU temperatures being a little high (70's) I still cant see much going wrong. One thing I have noticed though is all the voltage monitoring programs show the +12v line as producing 0.0v and occasionally rising to 0.2v when I do something like artifact testing through atitool.

Is it my PSU borked, my graphics card or both? Its seriously peeing me off at the moment anyway especially as i'm too broke to go out and get new components and this lot is almost certainly out of warranty. Help, comments and questions appreciated.

Boro

Any techies out there want to put me out of my misery?

Boro

nobody? not even a guess?

Ghost Train

I'd suspect the graphics card, but like you say could be the power supply although I would expect this to cause some other problems (not just when video load got high)

Can't think of any easy way to test other than swap out the parts and see if the problem goes away. You could take load off the power supply by disconnecting as many drives and devices as possible but leaving you enough connected to try and run something that causes the problem

I've had the same dilema before not knowing if the motherboard or power supply was the issue. Just had to guess which was more likely or start with the cheaper part and replace that

DonkeyStyle \o/

What you've already tried sounds pretty comprehensive, I'm not surprised there aren't replies

Artefacts after reboot sounds pretty bad though... would you try underclocking the card to see if that solves the stability problem?
It may well draw less power too, so still mightn't rule out the PSU, but it'd rule out software reasonably enough I think.
On a side note, when my PSU started having problems under heavy load, the whole machine would cold reboot.

The only time I've seen artefacts from a card was when I was overclocking (both core & memory).
Is there cooling for the ram chips on the gfx card? I don't know, but do you think the temperature being reported for the card is taken from the main gpu core, so that the memory might be overheating without being able to tell?

The weather has been heating up lately, so overheating sounds likely.

My guesswork anyway.

Boro

Thanks for the replies lads.

Software: I have had a stable system since it was built and no random crashes or anything like that. Drivers and OS all up to date. I have tried Omega as well as ATI drivers and multiple versions of each.

Heating: I dont believe it is an overheating issue. I cranked up the fan speed to 100% which kept the temperature pretty reasonable on the GPU and still had the issues. Occasionally I get the VPU recover after a black screen, mostly the system halt. This was also happening before it got warm though and the case is pretty well ventilated. The memory chips do not have heatinks.

Power: I have never had any system reliability issues apart from when the graphics card is in action. For example FM2010 runs just fine for long periods of time, movies play full screen no bother, etc. Just 3D games.

I tried underclocking the card while testing with ATITool for artefacts, I still got them.

Does all that rule out the mobo and psu do you reckon? Its just that I get strange readings from the PSU with the monitoring programs in terms of voltages, yet the system is rock solid apart from playing games?

ricman

have you got a spare graphics card you can put in, try older games ,demos from 2007.IT sounds like the graphics card to me.
download a program that measures the cpu, mobo, and graphics card tempeture ,than load up a game like mass effect 2.
you could set up an extra fan to cool the graphics card, then see does it still crash in 3d games.
try setting res to 800x 600, does it still crash.

Boro

ricman wrote: »

have you got a spare graphics card you can put in, try older games ,demos from 2007.IT sounds like the graphics card to me.
download a program that measures the cpu, mobo, and graphics card tempeture ,than load up a game like mass effect 2.
you could set up an extra fan to cool the graphics card, then see does it still crash in 3d games.
try setting res to 800x 600, does it still crash.

Thanks for the reply ricman but I dont have a spare. I have monitored temperatures and the card does not seem to be overheating. Additionally I have cranked the fan speed up to 100% (from the default of 27%) and the temperature has been stable.

Assuming it is the gfx card, does anyone have any cheap recommendations? I was thinking about a radeon 4650 or something as they seem quite cheap for reasonable performance. I dont have the cash to spend big bucks on a new card at the moment and im fed up of being restricted to 2d games!

Boro

As an aside, someone suggested this to me

http://www.youtube.com/watch?v=_taI-CkPD3M

any opinions?

Boro

For anyone interested, I binned the card and replaced it with a radeon 5500. Problem solved.

Naikon

I had an 8800GTS and it died a while back following weird lockups after the temp averaged around 90-100. Not good. If in doubt, replace the thermal compound on the heatsink.