BSOD Issues

[GrayParrot]

Hey all, hoping you can lend me a hand as I'm going through a bsod nightmare.
My system has been crashing left and right after I've built it.

Here are the system specs:
- Intel i7 2600K (stock clocks)
- Asus Sabertooth P67
- MSI N580GTX 3GB Lightning
- Corsair Vengeance 16GB (4x4)
- Corsair HX 850

What I don't understand is that the system only seems to bsod at idle or below 10-15% load. I can run heavy heavy loads in Modo and not see an issue at all.

Another weird issue is that I ran Prime95 torture test and all eight cores stopped working after just a few minutes. I then restarted Prime95 and ran the test again and two of the cores stopped but the others remained running, was a short test though.

Here are some screen shots to give you an idea.

Thanks in advance for all the help.

 

[Malecoda]

Run memtest? Make sure you have the latest bios, and try clearing cmos.

 

[Tsuyoi]

If Prime95 is giving you errors, could be CPU issues.

On the other hand, you have the OCZ Vertex 3, which is infamous for BSOD. Do you have another HDD you can install Windows on to test?

 

[vroom_skies]

Thanks for the replies you two. I'm a friend of Gray in real life and am helping him go through the trouble shooting process.

We were running memtest86+ while the initial post was being written, hence no mention of it.
It turns out that not even one complete pass of the memory had gone by (about 60%) when there were four errors listed. That was enough to conclude that the memory is bad and needs replacing, so that is in the works.

However for all we know the bad memory could have been caused by the constant crashing. So I would love it if the memory was the only culprit, but I have a feeling that's wishful thinking.
I'll have to check on the latest bios, but I would think that's a far fetch, but worth a try. I have yet to fail safe the bios or clear the cmos, so well get around to that. However the replacement of the memory will come first.

Yes there were issues with prime95, although I'm pinning that on the memory for the time being. The first run of prime95 resulted in all eight threads being down in less then five minutes. The second run (right after) had two threads go within ten minutes, but the other six were still going strong after twenty minutes. So I would assume that was a memory issue.

Thanks for the info with the Vertex 3 bsoding; I didn't know about that. I have extra HDDs that we can use. Once we get the replacement ram well start off with a fresh install off of a fresh mechanical HDD.

Any ideas?
Thanks in advance.

 

[Paragon]

If that VIN6 voltage is the memory voltage, that could be your issue. With 4 sticks installed, I would nudge it up a tad and then rerun the test. 1.575 is the Intel stated max for SB. I would keep it below 1.55 and try 1.52 or 1.53 area and see. It seems to have helped some issues with my overclocking (gigabyte Z68 board). Have yet to run Prime but I was getting dropped workers before.

I was using CPUID HWinfo but the RPM was always off. I am now using HWInfo. You can configure to show what you want. It seems to read the voltages just fine off my board.
http://www.hwinfo.com/download64.html (scroll down and you will see the links. There is the zip version that does not need to be installed.. just extract and run.)

 

[Xeth]

Pretty easy to figure out what it is: Run memtest with 2 sticks only. Then run the other 2 sticks.

If both sets fail it's not memory related at all.
If one set fails it's probably bad memory, try to isolate it to one stick.
If none fail it's just the processor having a hard time driving 4 sticks. Bump the memory voltage.

 

[vroom_skies]

If that VIN6 voltage is the memory voltage, that could be your issue. With 4 sticks installed, I would nudge it up a tad and then rerun the test. 1.575 is the Intel stated max for SB. I would keep it below 1.55 and try 1.52 or 1.53 area and see. It seems to have helped some issues with my overclocking (gigabyte Z68 board). Have yet to run Prime but I was getting dropped workers before.

I was using CPUID HWinfo but the RPM was always off. I am now using HWInfo. You can configure to show what you want. It seems to read the voltages just fine off my board.
http://www.hwinfo.com/download64.html (scroll down and you will see the links. There is the zip version that does not need to be installed.. just extract and run.)

Thanks for the info. The voltages on the memory make a fair amount of sense since all of the blue screens have happened at idle or very low load. If the voltages are running at a lower level when at idle that would account for why there aren't any crashes at load. Although on the flip side you would assume the chips are getting full volts when running memtest...
I'll try upping the volts on the SB and see if that makes any difference.

Pretty easy to figure out what it is: Run memtest with 2 sticks only. Then run the other 2 sticks.

If both sets fail it's not memory related at all.
If one set fails it's probably bad memory, try to isolate it to one stick.
If none fail it's just the processor having a hard time driving 4 sticks. Bump the memory voltage.

I'm not following on how two sticks failing would be a sign of it not being memory related? It would be lesser power requirements for sure, but no sure sign that it's not faulty memory? For all we all four sticks could be damaged to an extent, although very unlikely. Or it could just be lack of power and none of the sticks could be damaged.

Thanks for the ideas you two, I'll let ya know how it goes.

 

[Tsuyoi]

I'm not following on how two sticks failing would be a sign of it not being memory related? It would be lesser power requirements for sure, but no sure sign that it's not faulty memory? For all we all four sticks could be damaged to an extent, although very unlikely. Or it could just be lack of power and none of the sticks could be damaged.

Thanks for the ideas you two, I'll let ya know how it goes.

Unless you bought the ram in one 4x4 pack, typically it's pretty unlikely for two individual sets of 2x4gb ram to fail simultaneously. You're looking at winning the lottery chances. If it was ram related, what usually happens is that one stick out of 4 is bad, so by swapping in combinations of pairs, or even trying single sticks, you can isolate the defective one. However, if there is consistent failure no matter what combination of sticks you use, chances are its MoBo or otherwise related.