Is my Motherboard/memory responsible for sudden shutdowns

Status
Not open for further replies.

indigo0086

Distinguished
Nov 23, 2011
8
0
18,510
Here's my setup:
ASUS P8Z68-V PRO LGA 1155 Intel Z68
Western Digital VelociRaptor WD6000HLHX 600GB
CORSAIR Professional Series HX750 (CMPSU-750HX) 750W ATX12V 2.3
G.SKILL Ripjaws X Series 16GB
Intel Core i5-2500K Sandy Bridge 3.3GHz (not overclocked)
HIS Radeon HD 6950 IceQ X Turbo 2GB (256bit) GDDR5

The problem is that several weeks after putting it together I noticed that during long gameplay sessions, and times without having restarted, my computer would suddenly shut down, then start up shortly after and I would look at the event viewer and see, "Kernel power failure". I've reproduced it to where if I play a game for a long period of time I would have reltemp on another monitor when it shuts down and it is well away from the TJ Max (about 50+ degress away) so I've ruled out the cpu. I recently replaced the power supply and I still experienced it, so that is ruled out. Only things now are the motherboard, memory and possibly the graphic card (though I've had the computer shut down when no games were playing (maybe flash videos and the like but I've had way more videos playing and it hasn't failed).

My question is how can I rule out these components without just replacing them outright, are there stress tests that I can perform on the components to rule them out or is it pretty much rma them each at a time if possible?

 
Solution
1) To test the ram, run memtest86+. Get the latest version which supports sandy bridge. You should get NO errors for acouple of passes.

2) To stress the cpu, you could run prime95. Specify rounding checking. Run it long enough to get the temperature rise stabilized at it's max.
Again, you should get NO errors.

3) to stress the graphics card, run furmark.

4) try furmark and prime95 to get everything going.

5) Look at the smart data for your hard drive. everything should be green. WD has some test programs available for their drives. Doubt this could be it.

6) Is it possible that you might have some sort of a short, possibly with the case? Try running everything out of the case.

7) Are there any bios updates for the...
1) To test the ram, run memtest86+. Get the latest version which supports sandy bridge. You should get NO errors for acouple of passes.

2) To stress the cpu, you could run prime95. Specify rounding checking. Run it long enough to get the temperature rise stabilized at it's max.
Again, you should get NO errors.

3) to stress the graphics card, run furmark.

4) try furmark and prime95 to get everything going.

5) Look at the smart data for your hard drive. everything should be green. WD has some test programs available for their drives. Doubt this could be it.

6) Is it possible that you might have some sort of a short, possibly with the case? Try running everything out of the case.

7) Are there any bios updates for the motherboard that might correct something. How about for the graphics card?

8) Do you have a discrete sound card? They can cause strange results.

9) Are you up to date on drivers, particularly graphics drivers?
 
Solution

Trixanity

Distinguished
Nov 15, 2011
17
0
18,510
While a Kernel Power Failure can probably be caused by lots of things, I had the same issue and it was the RAM. I noticed random reboots as well and on the Memtest it showed thousands of errors, so after testing each RAM module individually I discovered that one of the sticks was bad and that my system was rock solid with just one stick in place.

So as the others suggested. Do a Memtest and if any errors show up, try test each module individually to locate the problem. If there are no problems with the RAM, then move on to the CPU.

You can test the CPU (which has also been mentioned) with Prime95. Memtest should preferably be run for at least 2-3 passes to be sure and Prime95 for at least a couple of hours, but the longer the better. I ran mine for 6 hours. If there are any serious problems, the issue should arise within an hour of testing or if very serious within minutes. My RAM module produced nothing but errors, so it was fast to determine where the issue was.

Good luck!
 
Status
Not open for further replies.