Memtest

[RocketJump]

Lately I've been getting crashes in some games and decided to run memtest to see if RAM might be the issue. I ran Memtest once and out of 4 passes, there was 1 error. I ran it a second time for another 4 passes and there were no errors at all. What would have caused the error the first time and not the second? Should I run it a few more times? Lastly, should I run it with each stick individually? 

[nick name]

I can't tell you what caused the error, but it absolutely indicates instability.  So if you're running your RAM at stock specs then you should test each stick individually to find the bad one.  However, you're gonna need to test for longer periods than you have already.  

 

And RAM errors can cause GPU driver crashes so it seems like you're on the right track.  

[RocketJump]

28 minutes ago, nick name said:

I can't tell you what caused the error, but it absolutely indicates instability.  So if you're running your RAM at stock specs then you should test each stick individually to find the bad one.  However, you're gonna need to test for longer periods than you have already.  

 

And RAM errors can cause GPU driver crashes so it seems like you're on the right track.  

I suspect it might have something to do with XMP. So let's say I run it multiple times today and come up with no errors, could there have been something that triggered that first error incorrectly? I did overclock my 8700k and it's been stable, but for my RAM I just set the xmp profile in BIOS and kept it as is. I might have to manually OC the RAM.

[nick name]

2 minutes ago, RocketJump said:

I suspect it might have something to do with XMP. So let's say I run it multiple times today and come up with no errors, could there have been something that triggered that first error incorrectly? I did overclock my 8700k and it's been stable, but for my RAM I just set the xmp profile in BIOS and kept it as is. I might have to manually OC the RAM.

It could be your CPU overclock then.  I would assume it's that above bad RAM at this point.  

[RocketJump]

9 minutes ago, nick name said:

It could be your CPU overclock then.  I would assume it's that above bad RAM at this point.  

Hmm, I ran 20 hours of stress tests on it and it didn't crash. I'll try bumping the voltage a little and see if it's better. These aren't hard crashes I'm getting, just crashes to desktop

[nick name]

A CPU overclock may prove stable in the sense of running, but can cause RAM instability.  You might just need a little more voltage or worst case -- dial back the overclock a hair.  

[RocketJump]

49 minutes ago, nick name said:

A CPU overclock may prove stable in the sense of running, but can cause RAM instability.  You might just need a little more voltage or worst case -- dial back the overclock a hair.  

Going to go ahead and bump the voltage up, I'm only running at 4.7GHz.

[nick name]

3 minutes ago, RocketJump said:

Going to go ahead and bump the voltage up, I'm only running at 4.7GHz.

Then it sounds like a voltage issue.  I'm not an Intel overclocker, but I can google.

 

Quote

VCCIO CPU 1.05V Voltage: This rail is for the IO transceivers within the CPU. Its primary impact is on memory stability, though it usually does not need to be increased as much as the System Agent voltage. Setting this rail ~0.05V lower than the System Agent Voltage is often sufficient to stabilize memory frequency.

Also:

 

https://www.overclockers.com/forums/showthread.php/792485-Overclocking-8700k-VCCIO-and-system-agent-voltage