Overclock.net › Forums › Components › Memory › Memtest86+ errors... possibly caused by CPU OC?
New Posts  All Forums:Forum Nav:

Memtest86+ errors... possibly caused by CPU OC?

post #1 of 8
Thread Starter 
Is it possible for memtest86+ to display errors due to overclocking the CPU?

Currently using 12GB of memory with four sticks (2x4 GB 2x2 GB). Ran memtest for 6 hours with a very slight OC on my Phenom II X4 955 (3.2ghz to 3.4ghz on stock voltage) and it came up with several errors.

I've heard using 4 sticks of ram can be detrimental to CPU overclocks due to higher stress on the IMC. Would this cause memtest to fail? But the OC was so small...

I only ran memtest because I was getting BSODs in Windows, seemingly randomly (after OCing). I haven't had any problems when everything was running at stock, but I only got the 2x4GB of memory recently so they might be defective.

Going to run memtest all night for now.
Edited by Genome852 - 8/27/11 at 5:57pm
post #2 of 8
No, memtest only shows errors directly related to the IMC and the sticks themselves. Also, running memtest with ALL sticks in the IMC at the same time will NOT display correct/positive results. It is best to test your sticks individually for around 8 runs tops.

Also, with that BSOD, what was the error code? I'd analyze it for you but I cannot since I have an Intel rig, and it won't display correct information.
post #3 of 8
Thread Starter 
The error code was 4E (pfn list corrupt).

I just ran memtest for 12 hours using stock settings and got 0 errors this time. Are you sure the small overclock on the CPU (using multiplier) had nothing to do with it? The first time I ran it (6 hours) it had 30+ errors with the CPU at 3.4 ghz.

Also, if I run each stick through memtest individually and get no errors, but start getting errors with all of them in at once (AND with the CPU overclocked), would this relate to how more sticks of ram and overclocks on the CPU both put more work on the IMC, which could lead to system instability and errors in memtest? Then again I'm new to this and am just regurgitating what others have said.

I will try and test the sticks individually later.
Edited by Genome852 - 8/27/11 at 10:45am
post #4 of 8
Testing the sticks individually is the only way to gain positive results. Doing it with other sticks in is just wasting time, unfortunately. Also, running Memtest is only testing the sticks (or stick in this case) that you have seated, nothing else. So if an overclock is unstable due to a lack of core voltage or something, it will not show any errors because it's ONLY testing the memory.

Another thing that's very important to note is that Memtest WILL show errors if the timings/voltages are INCORRECT. I've had multiple issues in the past with DDR2 not running at factory advised timings/voltages, so I had to resort to "loosening" the timings to be stable.

However, for now I'd just recommend running one stick at a time for at least 8 passes and making sure no errors arise, that way you know it's not memory. After you've tested your memory, if by any chance you're still crashing at stock settings, we can move on and consider it to be software and use Driver Verifier to pinpoint which specific driver is causing this BSOD.
Edited by pjBSOD - 8/27/11 at 1:29pm
post #5 of 8
Is your CPU a C3 or C2 revision?
post #6 of 8
Thread Starter 
Quote:
Originally Posted by xd_1771 View Post
Is your CPU a C3 or C2 revision?
CPU-Z says C2.

Quote:
Originally Posted by E-Peen View Post
However, for now I'd just recommend running one stick at a time for at least 8 passes and making sure no errors arise, that way you know it's not memory. After you've tested your memory, if by any chance you're still crashing at stock settings, we can move on and consider it to be software and use Driver Verifier to pinpoint which specific driver is causing this BSOD.
The computer has yet to crash at stock settings, it has only crashed when the CPU is OC'd (I have never touched RAM timings, they're where they should be at rated specs) - which is strange since the BSOD indicates a memory error. It may just be luck that I haven't crashed at stock settings, though.

Also, you're saying memtest is completely useless with multiple sticks of RAM, as in the error reports are 100% unreliable? So having several errors when the CPU was at 3.4 and none when it was at 3.2 was just pure coincidence?
Edited by Genome852 - 8/27/11 at 3:24pm
post #7 of 8
Thread Starter 
Ran memtest individually on the 4gb sticks, they check out fine (6 and 8 passes).

I just got a BSOD (IRQL not less or equal), but the only ram I'm using is the 2x4 GB set which didn't have errors in memtest.

Everything is on stock settings.

I also noticed something strange. Memtest would recognize the 4GB sticks as running at 200 MHz (DDR 400), while it would recognize the 2GB sticks as 666 MHz (DDR 1333).
Edited by Genome852 - 8/28/11 at 12:01pm
post #8 of 8
C2 revision will have problems running 4 DIMMs at above around 1450Mhz. Expect a need to use more CPU-NB voltage and DRAM voltage.

Memtest can be pretty weird sometimes...
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Memory
Overclock.net › Forums › Components › Memory › Memtest86+ errors... possibly caused by CPU OC?