ram Hard system freeze and Memtest86 error with XMP on

[Xen0m0rph]

I own a 16 GB DDR4 memory kit produced by G.Skill (F4-2800C16Q-16GRK) and i recently enabled the XMP profile in the motherboard's UEFI settings (my motherboard model is Asus Rampage V Extreme). Soon after that, about a week later, i noticed the first hard PC freeze - no BSOD, no crash log, just screen and sound freeze, after which i had to do a hard reset. 2-3 days later, another crash. Until now i experienced about 4 crashes in about 2 weeks. I wasn't doing anything special - watching a video, running Visual Studio, or even leaving my PC idle.

 

I read that this kind of behaviour is usually caused by faulty hardware, especially RAM. I stress tested my CPU with Prime95 and all went fine. Then, i ran Memtest86 for about 1 day and i got 1 error in 8 passes (~20 hours), as you can see in the attached photo:

 

 

This is another photo showing the UEFI screen with XMP enabled, not sure if the settings are correct though:

 

 

My question is, should i be worried about this error? Is there a correlation between the freezes and the error found by Memtest86? As i said, XMP was enabled but i don't think this is something that should push my RAM beyond the manufacturer's limits. Would disabling XMP help, or does the error reveal an irreversible hardware fault in the memory modules which manifests itself regardless of XMP?

 

 

Edited July 8, 2016 by Xen0m0rph

[manikyath]

i'm guessing you just have a bad stick, if you're running the sticks at their rated speeds you could RMA, if you're running them over the rated speeds you just tone them back.

 

what's happening here is that as you increase the speed of the sticks, one of them becomes slightly derpy, and starts throwing out errors. its up to you if you just run them at lower speed, or contact your retailer of choice about the issue.

[porina]

I've experienced multiple compatibility problems with higher speed ram (typically 3000+) and different mobos. If you turn XMP off, it'll run at whatever the SPD lets it, typically 2133 or 2400. This is fine and safe, but you're not getting the full performance of the ram you paid for.

 

I found that newer bios releases for mobo can help, so make sure you're up to date. Advanced uses might involve tweaking voltages but I'm not really experienced enough to guide further. In the end I just got different ram which happened to work better. Lucky dip.

 

Note I first had this problem on my main rig (Maximus VIII Hero), which started with G.skill Ripjaws 4 3333. Early on I had perhaps one detectable error per week (the software I run on it is memory hungry and comparable to prime95 running 24/7). With successive bios updates it reduced to an error rate of about once a month. Even reducing speed to 3000 didn't eliminate the errors. Replaced it with Ripjaws 5 3200 and there has only been one unexplained error since.

 

More recently I have Corsair LPX 3000 working fine at XMP in a MSI Z170A Pro Gaming. Move both CPU and ram into an Asus Z170I Gaming Pro, no boot at XMP, unless I drop the block to 2800. I think we have to remember even if the ram is rated for a high speed, the CPU (and mobo) together might not be up to it. Different manufacturers could vary there.

[Xen0m0rph]

I disabled XMP on my PC, i'll see in the following days if i get any hard freezes again.

 

Out of curiosity, i temporarily enabled XMP and ran Memtest again, this time choosing a much narrower address range to include the one where i got the original error. Now it went much faster, however even after several hundred passes i got no errors. I assume now it's not as simple as that to pin down faulty memory addresses.

 

I do have the latest BIOS updates and both the motherboard and the CPU (Intel Core i7 5930k) should be more than capable to handle XMP with any memory sticks. I'm pretty sure it's the memory itself who is at fault here.

 

Thanks for your suggestions.