Jump to content


Photo

Book 3 Crashing


  • Please log in to reply
109 replies to this topic

#101 Olstyle

Olstyle

    Vestrum Crier

  • Istrum
  • PipPipPipPip
  • 597 posts
  • LocationGermany

Posted 15 July 2015 - 16:12

If you used a GHZ-Edition BIOS on a normal card you might have degraded your GPU and RAM by giving too high voltages on a card which wasn't made for those.
If you didn't change the BIOS you now have a prime example of why some didn't trust the GHz-Editions in the first place and could still hope for warranty.

#102 blackknighti30

blackknighti30

    Rubber Ducky

  • Istrum
  • Pip
  • 23 posts
  • LocationCharlotte, NC USA

Posted 15 July 2015 - 16:18

As I mentioned above, it is a GHZ edition and it came with those clocks out of the box. I updated to a newer version of the bios for that exact card shortly after I got it and it has worked flawlessly until the last couple months. I did look on Gigabyte's website for Warranty information and it is still under warranty as it is a 3 year warranty. I'll see what they say, but underclocking it to what were the non-GHz edition speeds will probably work as well. I want it to work the way they sold it to me though.



#103 Olstyle

Olstyle

    Vestrum Crier

  • Istrum
  • PipPipPipPip
  • 597 posts
  • LocationGermany

Posted 15 July 2015 - 16:25

I don't expect it to last very long on the lower clocks either since degrading isn't something which just stops(especially since it still runs high voltages on the GPU), that's why I proposed using the warranty.

#104 blackknighti30

blackknighti30

    Rubber Ducky

  • Istrum
  • Pip
  • 23 posts
  • LocationCharlotte, NC USA

Posted 15 July 2015 - 16:31

I don't expect it to last very long on the lower clocks either since degrading isn't something which just stops(especially since it still runs high voltages on the GPU), that's why I proposed using the warranty.

Ok, I guess I had misunderstood. I thought you were under the impression that I put a bios that wasn't designed for it on it. I have a RMA submitted with Gigabyte to see what they can do. They may replace it if they still have more, or maybe they will send something newer, but who knows.

 

Already RMA'd this card once since I got it within the first few days after I bought it due to bad artifacts with stock clocks.



#105 blackknighti30

blackknighti30

    Rubber Ducky

  • Istrum
  • Pip
  • 23 posts
  • LocationCharlotte, NC USA

Posted 25 July 2015 - 03:12

So, I just got the update today that they ran through several tests and found nothing wrong with my graphics card. I have run my old 4870x2 in the meantime while my card was shipped to CA, and it worked very well for pretty much the same games I was testing, but I did have a crash in Fallout New Vegas after a prolonged amount of time playing that caused a full reboot, so I guess the issue is something else. Maybe I'll try underclocking something else to see if it helps.



#106 Olstyle

Olstyle

    Vestrum Crier

  • Istrum
  • PipPipPipPip
  • 597 posts
  • LocationGermany

Posted 25 July 2015 - 07:16

I assembled a List of Test-Tools a few pages back:

Both your problems sound a lot like hardware errors. To test the different part I suggest the following:
Multiple instances of Memtest for RAM
http://hcidesign.com/memtest/
Lynx or Prime 95 for CPU accuracy (the burntest only tests heat)
http://www.xtremesys...npack-interface
http://www.mersenne.org/download/
Furmark for GPU(you might have to rename the exe for the driver not to stop it from rendering full speed)
http://www.ozone3d.net/benchmarks/fur/

E.g. broken RAM already happend twice to me.

Since underclocking of the GPU helped and the only other component which is directly affected by this is the power supply I would first look into this.
If it got problems with the load, running prime+furmark should mean nearly instant reboot.

#107 blackknighti30

blackknighti30

    Rubber Ducky

  • Istrum
  • Pip
  • 23 posts
  • LocationCharlotte, NC USA

Posted 25 July 2015 - 15:00

So, I ran the burn test again on "very high" last night and it handled just fine. Then I started memtest86 and have been running 4 passes (1 cpu active default) and it just started the 4th pass and has already found 1314 errors. It's been going for 8.5 hours so far. I'm starting to think I found my problem though.

Power supply is a 1250W OCZ model that is only 2 years old. I'm nowhere near it's limits power wise. Not really sure how to test it other than to monitor the voltages, which have been very stable.

#108 khh

khh

    Harbinger of the Balance

  • Moderator
  • 7122 posts
  • LocationNorway

Posted 25 July 2015 - 15:38

When I've seen power supplies fail they've caused the system to refuse booting. I've never seen them fail partially.

April Ryan is my friend,
Every sorrow she can mend.
When I visit her dark realm,
Does it simply overwhelm.


#109 Olstyle

Olstyle

    Vestrum Crier

  • Istrum
  • PipPipPipPip
  • 597 posts
  • LocationGermany

Posted 25 July 2015 - 15:51

So, I ran the burn test again on "very high" last night and it handled just fine. Then I started memtest86 and have been running 4 passes (1 cpu active default) and it just started the 4th pass and has already found 1314 errors. It's been going for 8.5 hours so far. I'm starting to think I found my problem though.

Everything above 0 errors is a defect...
But this doesn't at all match your reports of more stable gaming while underclocking the GPU.

#110 blackknighti30

blackknighti30

    Rubber Ducky

  • Istrum
  • Pip
  • 23 posts
  • LocationCharlotte, NC USA

Posted 25 July 2015 - 17:32

Ok, so I think I might have discovered the issue. So, in my Antec P280 case, there is a filter on the bottom of it where the power supply goes that is supposed to have the power supply fan sitting over, but I have carpets in my living room. I've never had a problem before, but I decided to flip the power supply 180 degrees with the fan facing up instead. The case allows you to mount it either way. I just played New Vegas for about 2 hours or so straight with absolutely no problems. I guess the power supply was just getting too hot, without getting enough air through it. It still doesn't explain the errors in memtest86, but I'm going to try running it again now with 4 CPUs activated and see if it still gets errors.






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users