Help me diagnose HW failure

[jcfougere]

Heya

I've had something plaguing me for some time. The symptom most often observed is application crashes, particularly games, but I've also had BSoD outside of games and in them.

I have a P5Q-E, Q9400, 4x2gb ddr2, 3 7200rpm sata3, a dvd-rw, ocz 700w stealthxtreme, 3 120mm and 1 80mm case fans.(lol @ firefox crash while typing just now)

Basically my system becomes extremly unstable when I have both 4870's in my pc. Initially I ruled out the memory, I reinstalled OS/games and made sure to update all drivers. I started narrowing in on my GPU's and the PSU. My original PSU went up in smoke, so I'm running a brand new replacement. I also switched from running a single 6870 to running the 2 4870's (owned all 3, couldn't sell the 4870's). I had 0 symptoms prior to the PSU change, but also prior to the GPU swap.

I obviously tried removing one of the 4870's, and for some time I had success, but I eventually had more crashes. Constant crashes in sc2, but strangely very few crashes in crysis2.

Recently I tried swapping which 4870 I ran in solo mode. I've been having good stability with the other card, running in the other pci-e slot. Today I added the second card, and crashes resumed.

I've checked my recommended wattage for my components, in most cases the 700w is ample power for this setup (used eXtreme PSCalc). Also of note, I ran these 2 4870's without issue prior to switching to the 6870, which was on the original PSU before replacement.

I'm leaning towards PSU, I've had new faulty power supplies in the past. It's strange that running one of the 4870's results in stability, and the other results in eventual crash. But also note, these are 2 different revisions of sapphire's 4870. The crashing card runs `30 amps idle, and the non crashing card runs `25 amps (same voltage though); 37.3amps and 40.6 amps at load, respectively.

I haven't bothered to do stress testing of both GPU's, in each pci-e slot to rule them out, because if it was the PSU, I'd probably get a crash either way.

Need some advise on this. I don't have anything better than a 535w spare to test with.

James.

EDIT: I should add that I've ran everything at stock to test stability with no success. OC or stock same symptoms and pattern.

 

[DigitalUK]

sounds like your power supply cant handle 2x4870's or is faulty, dont get hung up on your psu wattages ,amps is what matters. also make sure your bios is up to date.
maybe you could tell us what the BSOD error was.

 

[jcfougere]

Some memory related, some generic pointing to nothing. But again, swapped memory, and swapped DIMM slots, didn't affect crashes). Most of the time the crashes are application only and the system stays alive.

I didn't have any issues with the previous 700w PSU I was running(prior to exploding)...so yes I'm very suspicious of the new psu. Good thing I sold my multimeter few months ago...any way I can stress test the PSU while ruling out the GPU's?

 

[jcfougere]

HWmonitor voltage values are as follows:
3.3v=3.36 min 3.38 max
5v=4.99 min 5.02 max
12v=11.82 min 11.88 max

CPU 6.85amp min 34.94amp max
GPU#1 23amp min 53.5amp max
GPU#2 25.7amp min 56.2amp max

 

[DigitalUK]

HWmonitor voltage readings should be taken with a pinch of salt, ive seen 2v reading on 12v line with very good quality ,perfect working supply before.

 

[jcfougere]

Ok we won't rely on those. Short of testing a different PSU with the same voltage, anything else I can do to rule out other hardware?