Frying multiple graphics cards

[bjvanwash]

Hey everyone! I just joined Tom's Hardware and I have a problem...

I recently built a new computer and gave the remaining parts to a friend such as the PSU, MB, CPU, RAM, etc.

Now this friend of mine bought a "MSI NX8600GT-TD512EZ GeForce 8600 GT" video card for his computer and once it was installed with the drivers and everything it was working just fine. After about 2 weeks, the computer would just show a black screen at boot and nothing else. So we went out and bought another graphics card(same model) and the same thing happened.

I looked around inside the case and I did notice that the heatsink on the CPU was very, very loose. I reapplied thermal paste and all that stuff and got another graphics card(older than the previous one). Once it again it booted just fine for a couple of weeks and now it's back to the black screen again. 3 graphics cards in less than 3 months??? What's going on???

I have searched and searched to find a solution and just can't seem to figure out why these cards keep going out. There's no smoke/smell or anything on the card whatsoever. I have a feeling that it has something to do with the motherboard but I'm not positive considering I used the same MB, CPU, PSU, and RAM before building my new comp. Any help would be greatly appreciated.

Bare with me as the specs are only off the top of my head at the moment:

Gigabyte ga-965p-ds3 motherboard
Intel Core 2 Duo cpu (1.8?)
2GB DDR2 Ram
500-550 PSU
Win7 Ultimate x64

Thanks in advance!

 

[curtis_87]

WTF, weird. What is the brand of PSU ? Have you tried swapping that? I know this a needle in haystack thing, i had two comps, one PSU was old an not very good, when we got a power surge (and if you read the small print those bastards electric companies can get away without having to pay for anything that dies as a result) one PC fried completely other one was as good as gold but that did have a v nice psu,still good also after 11yrs unserviced. Also if it seems to be affecting just the graphics card, it could be the rails that the GPU is on..unless it doesnt need to have external power in which case the Mboard would be the next thing i point the crooked finger at....if you feel like risking another GPU for testing purposes I would set it all up and do a burn in test for at least 3 hours watching voltages and temp, it could be his case gets to hot when gaming or doing something else could be setting it off.

 

[bjvanwash]

Thanks for the reply. I haven't tried swapping the PSU since my current graphics card(Nvidia GeForce 8800GT) used to work just fine with it and thats a lot newer than the other ones. But maybe I'll give it a shot. As far as the "burn in test" I'm not sure exactly what you mean by that. I used to game on that computer all the time with no problems but my friend doesn't game at all. He has an xbox for that.

 

[curtis_87]

My definition of a "burn in test" is the following, give the computer more work to do than you will ever likely throw at it in normal usage for about 1 day and see if it falls over or anything shows unusual voltage characteristics or a given temp sensor gets too high.
The easy way to do this would be Sisandra Soft http://www.sisoftware.co.uk/, but they charge you for their program in some circumstances. The cheap and free way listed below.

1. Get speedfan to monitor the voltages and temperature on idle once installed. One of the tabs checks your SMART hdd status and lets you know about the health of your hdd against a database.
http://www.almico.com/sfdownload.php

2. Get something like "FurMark" this will be used to stress test the graphics card, ie let the program run and watch your graphics card get seriously loaded in terms of work to do, e.g Mass Effect 2 only peaks 70-80% load on my GPU but FurMark even in windowed mode gets the full 100%. I have also made a laptop GPU fall over using furmark after 20 seconds, it just overheated then the PC shutdown to cool off. If the GPU temp is the problem it will become apparent v quickly using this prog...
http://www.ozone3d.net/benchmarks/fur/


GPU-z should be alright to look at the load on the graphics card.
http://www.techpowerup.com/gpuz/



Ok now lets assume that the GPU per se is not the problem, you temps aren't the problem your cooling setup is adequate, you should have hopefully gone past the initial 5 mins mark if this is the case. PSU tend to take a while to warm up under usage so at about the 15min (possibly longer)mark so long as you are getting the PC to do as much as possible in terms of work you should either see voltage fluctuations or it might just turn off this to me would be a sign that the PSU is the issue because if it was the GPU itself it would have fallen much earlier under furmark.

If it doesn't keel over after 3 hours or so....hmm...back to the drawing board.

 

[wa1]

^+1
Try what he said above...