How can I diagnose this problem?

[andyt]

I have a P35C-DS3R MB with an E8200 CPU and a 512MB XFX 9600GT Alpha Dog XXX card. Nothing has been overclocked.

It's running Vista and for day to day use it's solid as a rock. However, these days whenever I do something that involves 3D, then after not very long, the PC just crashes. I get a black screen and the fans all kick in full. I can't do anything other than reset it but when I do it just boots up fine again and doesn't even complain of a bad shutdown. I've got Vista prepped to not reset after a BSOD so it doesn't seem to be one of those.

The symptoms are similar regardless of which of the 3 games I have that I'm playing (C&C, Crysis, HL2). I get anywhere between 1 and 15 minutes before it crashes. Sounds like temperature rising yeah? Doesn't seem to be though because Google Earth usually triggers it too, and I can monitor the CPU/GFX temperatures in a side window whilst playing with GE and they don't increase by much and usually plateu well before it crashes. I'm confident that it didn't used to do it but I'm also confident that it didn't start happening after any particular driver upgrade either.

So my issue is not so much what the problem might be, but any advice on diagnosing it in the first place! Thanks.

 

[aeiouandxyz]

What driver version are you running? Give Prime95 a run using small ffts and see if the system crashes. Prime95 stress the CPU more than any application will and will really test the stability of your system.

 

[swifty_morgan]

How much memory do you have in it ? Did you try and re-seat the processor heatsink and fan ? Is there enough smootch on it ? Do you have another video card you can try ? Are all the connections tight ? sata/power cables/etc. Did you try running it with the side cover off. The card may be getting too hot. Do you keep the dust out of everything ?
In the nvidia control panel... is it set to run on 1 monitor. Default I think it's set to multi-monitor. Is vsync off and is everything set to "let the application choose" when it comes to all settings ?
is there enough memory allocated to virtual memory ?

 

[roadrunner197069]

Sounds like a junky power supply to me.

 

[geofelt]

Did you plug in the 6 pin pci-e connector to the video card?
Is your PSU adequate; what is it?
It sounds like a power problem or a temperature problem.
Try running with the side cover off and a house fan directed at the innards to eliminate case cooling as an issue.

 

[arson94]

I'm going to agree with roadrunner, the several times I've seen this problem in the past, it's been because of the power supply. A couple of cases were fixed with a driver update but those were odd as they didn't blue screen either, but the majority were failing power supplies.

 

[bobbknight]

It's the PSU or the video card.

 

[andyt]

Cheers for the thoughts guys. The PSU is a 500w Silverstone unit (http://www.scan.co.uk/Product.aspx?WebProductId=696394), so it's not cheap junk. That said, I wondered about the PSU myself. I'll try messing around with the hardware a little, strip it down as much as possible, take out half the RAM etc.. and see if anything changes. Thanks.

 

[Priv]

Just run some general tests like prime95, memtest86+. So you can narrow it down to certain components.

 

[flyin15sec]

You might have the dreaded 1st gen reference Nvidia 9600GT black screen of death.

Google 9600GT Black Screen, you'll get a lot of hits with that problem.

If you have that problem, not much you can do. Downclocking your core and shader helped some. Others RMA the video cards for newer cards. The problem usually shows up with older games, so far no one has found the exact trigger for it, as it is random when it happens while gaming.

 

[andyt]

Bingo! It would indeed seem to be the problem. I've read a load of links, underclocked the card using RivaTuner and all is well. A bit gutting given that I paid extra to get the faster one in the first place. I might look into taking it back. Thanks very much for that info though!