Failing Memtest86 Tests 4, 5, 6, 7!

[shmoo]

Hey,

I put together a new system and I've been having some pretty nasty instability issues. Here's the RAM I'm using:

https://www.newegg.com/Product/Product.aspx?Item=N82E16820232328&cm_re=3200C16Q-_-20-232-328-_-Product

I haven't changed anything in BIOS, it's running at 1.35v by default. BTW this memory is officially supported by my GIGABYTE X99P-SLI.

Memtest86 has been going for about an hour and at the moment I've had 32 errors on test 4, 19 on 5, 39 on 6, and 6 on test 7.

Should I just RMA these sticks, or is there any tweaking I should try? Should I up the voltage a tiny bit? I don't want to play around with pulling sticks here and there and re-testing - My giant cpu cooler and my tiny case makes this take a lot of time, scrapes, thermal paste, alochol and coffee filters, and frusteration.

 

[RaDiKaL_]

You could try bumping the voltage to the next possible value, if that doesn't help try enabling/disabling its XMP profile if it has one, otherwise it seems you need to RMA them.

 

[InvalidError]

If you are getting recurring single-bit errors at a handful of addresses, those addresses have bad/weak/intermittent bits and no amount of DRAM voltage fudging will make those reliable beyond reasonable doubt. I'd have them replaced.

 

[shmoo]

If you are getting recurring single-bit errors at a handful of addresses, those addresses have bad/weak/intermittent bits and no amount of DRAM voltage fudging will make those reliable beyond reasonable doubt. I'd have them replaced.

K. Would you suggest getting another one of these kits? G.skill is usually held in high regard, and these kits often have good reviews, however this has me leaning a bit more towards a Corsair Vengeance LPX kit. It's 400mhz slower though, about the same price...

 

[InvalidError]

All the major memory vendors/manufacturers have lifetime warranty on their memory and perform within a few % between similarly spec'd modules. I wouldn't worry too much about it, bad cells slipping through QC happen and that's why I run memtest86 for at least 10 hours (unless I see errors sooner) whenever I get new DIMMs. Back in 2010, I had a bad bit on a DIMM and it took 4 hours of memtest86 to get my first fail. By the end of my overnight run, I had four errors, all on the same bit at the same address of the same test pattern.