Sisoft Sandra benchmark on video card and card goes kaboom.

[codejunk]

Long time reader, first time poster here....

Okay, here's the issue. New build, except power supply. Power supply was previously known to be working fine in a similar system. 500w, no stability issues (say with occt or prime 95 etc).

System specs are or were as follows:

Intel i7 920 cpu (don't overclock anything currently anyway)
ASUS P6T motherboard
500w psu (unsure of brand but it wasn't one of the best - had it for years and it's now gone for two reasons, see below)
3x2GB DDR 3 ram (kingston performance series)
2 x 1tb hdd, 1 x 320gb hdd, 1 x bd-r (all sata).
And it was a radeon hd 4890 (brand: xfx)

What happened is this:

I ran sisoft sandra's gpu calculation benchmark (I think that's what it was called). About maybe 30-50% through the test, I heard a quite loud bang and several loud cracks. Power shut off immediately and my UPS changed status too. UPS appears fine (though a bit worried too about that). I unplugged as quickly as I could and turned off power supply too. Removed the power supply.

I had a spare PSU that had enough power to test bare components (mostly to see if cpu and mobo would try to post - thankfully it did; beep etc). And nothing else was damaged, either, and I'm actually writing this on the same box (different video card now though, and new psu).

So, question is this:

Sisoft Sandra itself says that if there are weak or faulty components, they can fail. Which makes sense. But, anyone ever see anything like this with any component, or even the specific: xfx video card. Recommended psu was 500 or greater watt psu. And as far as I know, it was in okay condition. Also, pretty sure that not enough power will equate to stability issues, rather than what I experienced.

Of course, if the psu happened to die it could have taken out anything with it, but the fact i was doing a gpu benchmarks makes me wonder what really was the cause. Granted, the psu did die in the process and I now have an 850 w (thermaltake) but I am set back to my old geforce 8800gts 640mb (which isn't bad, but I was enjoying the better performance out of the 4890) ...

One more thing. I see no burn marks on the motherboard. And also no melting or anything. Right now my current video card is in the second PCI-E x16 slot. Thing is, I wonder if the first slot is okay. Bit worried about that.

Any ideas or help would be appreciated. If I missed something, or left something out, my apologies - let me know and I'll correct. Then again, it's 4:27 and been up since 3am, but that's another matter entirely.

Thanks.

 

[jaguarskx]

Unknown PSU is the problem. You should buy a good quality 500w PSU (or higher wattage). My name brand recommendation in order of my preference:

Seasonic
Power PC & Cooling (Some made by Seasonic)
Corsair (Some made by Seasonic)
Enermax
Antec (some made by Seasonic)
OCZ
Fortron Source (FPS Group)

 

[codejunk]

Interesting though I had a fear about that. Shame. The thing that didn't make sense to me is that it was during a GPU benchmark. However, knowing that the PSU went out too - I can only imagine how it would be it. Just thankful it take anything else out (which surprised me also).

Could it be that when I was benchmarking the GPU the psu was also stressed and that caused the issue ? If that's the case then it makes perfect sense to me. But being I'm better at troubleshooting software (being into programming and stuff) than I am hardware ...

As for PSU I have a thermaltake now (which afaik is a good brand also) of 850w. Modular and so on. Pretty happy with it now.

Does stressing a component like that actually stress the power too ? Oh, I guess it would: if it's being stressed it might need more power. Did not even think of that! So I guess benchmarking/stressing components is probably safe now.

Anyway, thanks for feedback!

Edit: what is your view on the pci-e slot it was in ? I guess hard to say but how would you approach it ? Only thing i can think of is getting a cheap non video pci-e card and testing it out. But they recommend (they = asus) that video card goes in first slot and of course at the moment it's in the second.

 

[jyjjy]

It's rare but when a PSU dies it can go in spectacular fashion, like what you described. Often using a GPU for intensive calculations actually uses more power than gaming so that + sketchy brand PSU + a few years old + specs lower than recommended for the card = bad news. Now that you've replaced it though I wouldn't foresee any problems. As for the PCIe slot give it a try, there's no reason not to.

 

[jaguarskx]

Stressing a component causes it to consume more power placing a greater strain on the PSU.

 

[codejunk]

Yep. I didn't think of that beforehand. It makes sense though.

And as for rarity, I would agree unless you count me. I think of any component ever in the system I've had the worst lucky with power supplies. Of course thinking about it, with the exception of completely new systems, I've had to take psu's in 'emergency situations' ie not the best brands. No more!

I'll give the pcie slot a try later on today. I was only thinking it could possibly damage the card if the slot was damaged.

I knew I should have replaced the psu and was even considering it but after the upgrade I didn't have the money off hand.. .but I had no choice so finally did it. Kind of stupid and I knew it was ... but well it happens.

Thanks for responses.

Cheers

 

[jyjjy]

Nah, the slot will just either work or not, can't possibly damage the card.

 

[codejunk]

Right. Thanks for info. I had mixed ideas about it.

Appreciate the advice/ideas.