Failed Prime95 and I haven't even OC'd my i7-3930k yet!

vidar5

Honorable
Jan 14, 2013
8
0
10,510
I ran the prime95 test on a new i7-3930k that hasn't even been overclocked yet and already some of the cores failed. A few of them failed because it said the error was 0.5 and it is expecting 0.4. I don't even know where to begin, do I need to try another stress testing program? Do I need to RMA my CPU...mobo...RAM? I appreciate any and all advice! Below I've posted information on the build and on the test run.

OS: Windows 7 Ultimate 64-bit
CPU: Intel i7-3930k
MoBo: ASRock X79 Extreme6
PSU: OCZ ZX Series 850W
RAM: 4x8GB Patriot Intel Extreme Master 240-Pin DDR3 SDRAM 1600

I ran the custom prime95 stress test under default settings plus the error and round checks.
 
Solution
Just because it runs at 60C doesn't mean it's going to be stable.

What you need to do is look into the voltage of the processor. I'm not absolutely certain about those 2011 chips like that, but I would imagine that it's not getting enough voltage and tossing up errors.

Ways to check...

1: Reset the BIOS to defaults and go back into Windows. Make sure you have all of the motherboard and hardware drivers installed. Then run Prime and see if you still have errors.

2: I'd next try running memtest386 if you're still getting the errors. A quick search on google will show you what to do to get memtest up and running. Test each stick of ram individually. If they all make it through the testing process, then it's time to really...

vidar5

Honorable
Jan 14, 2013
8
0
10,510


Sorry, I don't know why I left that information out. I am using the corsair H100 watercooler, cable management is very good (almost all of it is running in the back of the case), corsair 500r case (so a big one). I monitored the temps during the test and it never got much higher than 60 C, and yet some of the cores failed almost immediately. I've had experience with OC'ing and doing prime95 on another computer but I don't have enough experience to know what could be going on with such a high quality, right out-of-the-box cpu? :(
 

steddora

Honorable
Nov 13, 2012
686
0
11,160
Just because it runs at 60C doesn't mean it's going to be stable.

What you need to do is look into the voltage of the processor. I'm not absolutely certain about those 2011 chips like that, but I would imagine that it's not getting enough voltage and tossing up errors.

Ways to check...

1: Reset the BIOS to defaults and go back into Windows. Make sure you have all of the motherboard and hardware drivers installed. Then run Prime and see if you still have errors.

2: I'd next try running memtest386 if you're still getting the errors. A quick search on google will show you what to do to get memtest up and running. Test each stick of ram individually. If they all make it through the testing process, then it's time to really check that chip.

3: If the ram is good and the CPU is set to completely default settings. You need to check into the voltage. From what I see the average maximum voltage for the chip you have is 1.35v. So I'd leave it at stock and set a fixed voltage of about 1.3v and test it with Prime again with everything else set to default aside from the voltage. If the errors go away with that, then you need to RMA that chip.

Just remember this, if you're getting errors in Prime, it's not always the CPU's fault. I'd honestly point to the ram first thing if you're getting errors at completely stock settings.
 
Solution

vidar5

Honorable
Jan 14, 2013
8
0
10,510


Thank you for the response, I had previously ran memtest386 and got no errors after nearly a day. I tried your steps 1 and 3 today, step 1 still got me the errors so I moved on to step 3 which instead of giving me errors, gave me an immediate BSOD when I started the test X( I talked to Intel as well and they had me run their intel processor diagnostic tool, a.k.a piece of ***. Since it said my system passed and I am still able to use my computer intel says they wouldn't warranty the cpu, but since I bought the cpu less than 30 days ago, I can still RMA through newegg. I understand that I could just use the computer for now, but I'll never be able to overclock the cpu as it is and that was what I was planning on doing. So I guess I will do that soon unless anyone else has another idea. I'd also like to know if there was a way to make sure the problem isn't with the mobo or psu...

Ugh I don't want to have to clean the thermal grease off the cpu.
 

vidar5

Honorable
Jan 14, 2013
8
0
10,510
I will, thanks. I talked to Newegg and they set up a refund on condition I re-buy the i7-3930k this way I never have to be without a processor. I am really liking Newegg's customer service, I may end up buying all my parts from them instead of the vast majority of my parts from them, lol.
 


That is one of the reason I use them even though sometimes you can find the part slightly cheaper somewhere else.
 

steddora

Honorable
Nov 13, 2012
686
0
11,160


I've spent over $20,000 in the last seven or so years with them and never had any issue that wasn't fixed almost immediately. They are one of the best when it comes to customer service, shipping, and selection. I always suggest Newegg when it comes to someone ordering any electronics online. +55 to them. :)

As for the processor being a bum. That immediate BSOD could be two things. First it could be a lot of heat (doubtful if it immediately BSOD's on Prime without at least a few seconds) or the chip is just a really bad bin. Basically it seems like you got a chip that just barely passed their tests. It happens from time to time but I've been lucky with all of the intel processors I've had. As for AMD chips; I never got lucky with one of them aside from an Athlon XP 1700+ that made it to almost 2Ghz on air. :)

Let us know how the new chip differs! :)
 

vidar5

Honorable
Jan 14, 2013
8
0
10,510
Hey everybody guess what! I got the replacement CPU today and installed it and the results are exactly the same! Lucky me. I really don't know what's going on now. I am re-running some memtests and I am going to bring the RAM from my personal computer to work and see if it works there. If that doesn't work then the only parts left are the mobo and the psu and I don't think the psu is bad judging by it's voltages being within normal limits. How would you even test a mobo? It seems like the hardest computer part to pin the blame on!
 

alexnode

Distinguished
Apr 18, 2013
62
0
18,630


I have the same problem. with prime95 -windwows 8 pro all other tests normal and 48+ rendering hours from vray were perfect. But prime95 gets these rounding errors for 8 of the 12 threads. From the four that are working two are ok and two are not working! I did the blend test. temps maxed at 75c and voltage at v1.33
Is it possible that there is some incompatibility with prime95 and windows 8 ? The weird thing is that i have a problem with another machine that needs the m/b replaced and i said oh let's test my main workstation!
 

vidar5

Honorable
Jan 14, 2013
8
0
10,510


I don't know if it will be the same for you, but what I ended up finding out was that my problems were actually all being caused by the memory sticks. For some reason, the memory sticks were not able to handle the frequency and timings that they were rated for (1600MHz and 9-9-9-24...or something...it's been a while, lol). The motherboard automatically used these rated settings and somehow the memory passed MEMTEST86+. What I ended up doing was reducing the frequency to 1333MHz (I might have lowered the timings a bit too), and suddenly everything worked like it was supposed to.

See if that works for you!

 

alexnode

Distinguished
Apr 18, 2013
62
0
18,630
LOL ... I just did a new memtest86 and it showed errors i was about to write a correction to my post! I moved my kingston 16g to test another machine with a broken m/b and now I have two machines with problems !
I have to stay away from my hardware !
 

alexnode

Distinguished
Apr 18, 2013
62
0
18,630
So sorry but I don't remember what happened. Something was wrong with my motherboard settings. I managed to fix it. I also put another set of memory from another workstation then change it back to the original set of memory. Something along the way went wrong. I should have posted here when I fixed it.