Persistent lock-ups with black screen... please help

BillTheButcher

BillTheButcher

BillTheButcher

Associate
Joined
2 Nov 2005
Posts
395
I've been trying various settings and posting some of the results on here but I can't seem to stabilise my new system. :(

Here's the hardware:
Asus P8Z68-V Pro
i7-2600k
2x 4GB Corsair Vengeance DDR3 1600
120GB OCZ Vertex 3 SSD (system drive)
2x 250GB WD Velociraptors RAID0 (storage)
Gainward GTX 590
Corsair AX1200 PSU
Creative Labs X-Fi Fatal1ty

This is the first time I've ever tried overclocking so I started off cautiously, following the guide on clunk.org. I manually set the multiplier in the BIOS to 40 and gradually increased the voltage until it stopped BSODing.

So no more 0x124 errors but now I'm getting a different kind of lock up. It doesn't matter what I'm doing: gaming, browsing the web, sending emails... The PC will suddenly stop responding. I can still move the mouse but everything elses freezes. If I Ctrl-Alt-Del I get a black screen but nothing else. The only way to get out is to hard reset the PC.

Sometimes this happens after a few minutes, sometimes it will be fine for a few days and I'll think it's stable but then it'll crash again.

There's nothing in Minidump following these crashes.

I've tried loading the BIOS defaults, disabling Turbo mode, loading the new 0651 beta BIOS... nothing seems to make any difference and I'm at my wits' end. :(

I thought maybe it was my 590 overheating (it's been hitting 90C in some DX11 games) but the PC crashed on me a few minutes ago and the 590 is idling at about 37C so it's not that.

I know this is probably a TL;DR but I don't know what else to try and I don't want to have to give up on the new build.

Any ideas? I'm happy to post any info that might help to troubleshoot this.

Thanks a lot for reading. :)
 
Sorry. :o

I've tried lots of different settings: speeds ranging from 3.4 up to 4.4GHz. I've tried setting the voltage manually starting at 1.12 up to 1.3. It stopped BSODing at about 1.28.

I've also tried leaving all voltage settings on auto and just changing the turbo multiplier.

And I've tried loading BIOS defaults and not changing anything.

Stability testing: IBT for a couple of hours and Prime95 for about 6 hours. Passed everything. It's never crashed during a stability test.
 
Have a look at the results others are getting i.e. what kind of volts are needed for different clocks.

Being methodical is a good idea. Go up 100 MHz, raise the volts until it does 5 passes IBT, record the numbers, repeat. That way you can find the sweet spot.

IBT needs to be done properly. Make sure you're using v2.51. The peak GFLOPS for your chip is 32 times the clock speed in GHz e.g. at 4.4 GHz it would be 32*4.4=141 GFLOPS. Then use enough memory in IBT to get the GFLOPS column up to at least 80% of the peak, e.g. 0.8*141=113 GFLOPS. Five passes like that I would consider somewhat stable.
 
I've just successfully completed 5 runs of IBT @ 4GHz with a GFlops highest reading of 103.29 which is just over 80% of the peak (32*4=128).

Values for the other 4 runs were 100.83, 101.99, 103.12 and 102.98.

Also recorded a new CPU temp record of 62C during the burns. :D
 
This is a problem with your OCZ vertex 3 send it back and switch it for another Brand...

overclocking Does not just freeze the PC and the mouse is continuing to work !!
 
joeyjojo said:
Sounds stable to me :) good work.

Let us know if you get more problems.
Click to expand...
Will do. Thanks for your help. :)

muriel said:
This is a problem with your OCZ vertex 3 send it back and switch it for another Brand...

overclocking Does not just freeze the PC and the mouse is continuing to work !!
Click to expand...
At first I thought it might be to do with the SSD so I took it out of the system and plugged my old OS drives back in (WD Velociraptors in RAID0) but the crashes still happened. So the problem is not with the SSD.
 
If you're just getting a black screen I'd suggest it's your GPU. I had similar situations recently with my GTX 580 which was "white screening" with most of the same errors you've expressed above. To test the stability of your GPU, I'd download Furmark to stress it. I replaced my 580 with a new card and hey presto, no problems.

Also, look at the temperatures you're running internally in your case. The 590 runs *very* hot - are you running the fan management software that came with it? If not, use the MSI afterburner application and run the fans on your card at full power for a long time, until the machine black screens or not - this may give you further clues...

Hope this helps. :)
 
Thanks. :)

The 590 gets up to a peak of just less than 90C on each GPU. This is quite high but below its recommended maximum of 97C. The fan is on automatic (via the nVidia control panel) and it's never been above 70%.

As I said above, I thought initially that it might be the GPU overheating but I had a black screen with both GPUs idling at 37C.

It's been on all weekend running all sorts of benchmarks and stress tests and I've done loads of gaming on it as well and *touch wood* I haven't had a black screen since Friday lunchtime.
 
Recommended maximum of 97C is *absolute limit*, and consistently running your GPU at anything approaching that will cook it in the long run. "You can drive a car with your feet if you really want to, but that don't make it a good fkn idea now, does it?" :D

Mid-high 70's are good operational temps for the Fermi architecture, and running your GPU much over this for consistent periods of time is BAD - this I know through personal experience... the black screen you experienced at idle might still be an indication of GPU errors - if there's damage that's been done, regardless of what you do now it might still be problematic, right?

Good luck. :)
 
Last edited:
Sounds about right for the idle temps. Just leave the machine run at it's default config (i.e. turn back the GPU fans) and run Furmark - this will rule out (at least) your GPU problems if any.
 
Basically you just want to simulate your machine at it's normal config and stress the individual components until you find something specific. Furmark will particularly stress GPU when you crank it up - Intel Burn Test (when run on maximum) will stress CPU/Memory.
 
You must log in or register to reply here.
Back
Top Bottom