Re-booting and lockup problem

[Markfw]

I think I know the answer, but I would love some opinions anyway.

So I have this P6T WS pro motherboard with an E5520 AT STOCK and 2 GTX260 video cards. Now first, I had a stick of memory go bad (Corsair), so I thought "well, lets set the memory to 1,5 volt instead of auto and see what happens. So it stabe a couple of days. Then I have EUE units on F@H or blue screens Looks like memory problems). Then it starts re-booting (Yes I have the auto-reboot turned off) 5 minutes after it boots. So I think "Maybe the PSU is getting weak" so I turn off the CPU client, and I just have 2 GPU clients (still a lot of load), goes 10 minutes, still reboots. So then I do ONE GPU client. Now its been 1/2 hour and still going.

So I think I need a new PSU with more watts. The old one is an OCZ (rebadged fortron) 700 watt >85 % efficiency. I think I need an 850-1000 Silverstone, Corsair or Seasonic, and >85% efficiency. What do you all think ??

If I need a new one What about this one ?
http://www.newegg.com/Product/Product.aspx?Item=N82E16817256057

 

[Markfw]

OK< so its been 24 hours with no problens with just the 2 GTX260 cards @ 100&#37;load, no problems.

Now to turn the CPU client back on. And it can't be a bad cpu or memory or motherboard, since all that is involved with the system being up and running with 2 video cards, right ? So if it does now (as in lockup or reboot) doesn;t that say PSU for sure ?

 

[Idontcare]

Hi Mark, I know you like to rely on F@H as an intrinsic stress-tester but unfortunately other than it indicating to you that something is not stable in your rig it doesn't really help you pinpoint and debug the instability all that well (beyond the random "let me change xyz and see if it gets any better " type debugging).

I wholly recommend you install Prime95 and run large FFT for 24hrs to verify your memory circuitry (ram, the controller, the uncore logic that is involved, etc) is stable. An error in large FFT is usually strong indication that you've isolated the instability issue to something in the memory path.

Likewise I recommend you run small FFT (never run BLEND, the output is useless in terms of diagnosing where the instability likely lies, and it is less rigorous than running Large followed by Small anyways) for 24hrs to just confirm you don't have an instability in one of your cpu cores.

Further, I recommend you download OCCT and run the full "Power Supply" test. Make sure under the Options menu you have selected for OCCT to make the graphs.

When the power-supply test is complete you will want to investigate those graphs for voltage/amperage swings throughout the duration of the test as a means of isolating whether the PSU is a walking wounded. OCCT defaults to a 1hr PSU test but for your situation I'd make that a 6hr test because you really want it to get warm (that's when walking wounded stuff goes unstable) while having the voltages documented.

If all that checks out then there really is no more software-based "free" diagnostics to do...you'll need to swap out the PSU with another one and see if the problem persists or goes away, rinse and repeat.

 

[Markfw]

OK, so I ran OCCT, and within 10 minutes it faield. Something about core 7 reported an error.

Bad CPU ?? thats odd.