Girlfriends setup... memtest errors in dual channel

[Jackther]

Hey guys, a bit of a strange one at least for me personally.

I'll list the specs then get on with the problem...

CPU: 3570K
Motherboard: Gigabyte G1 Sniper M3
RAM: Mushkin Blackline 1600MHz 1.65v 2x4GB sticks
SSD: Vertex 3 60GB
Wireless card: ASUS PCE-N15
GPU: AMD XFX 7870

So the issue is that the RAM runs absolutely fine individually when testing one stick, she gets no memtest errors when testing one stick in each DIMM slot.
When they're paired and running in any slot in any configuration on the DIMM slots, it throws up about 12,000'ish errors in Memtest.

The system boots fine with both sticks in, but randomly restarts with and without any BSOD's.
I've helped her setup the BIOS properly with the correct voltages for the RAM, timings and speed, which hasn't changed anything with the errors.

Again, running the system with one RAM stick seems to work fine, running them in a pair gives her random restarts with and without BSOD's.
I'm out of ideas, she doesn't have another system to test all her stuff in, so I'm hoping it's just the RAM and not the Motherboard.


She's currently taking out the wireless card, as she had posting issues when it was in one of the PCIe slots, so I'm kind of hoping it might be either that causing the issues or just the RAM.

I'm unable to help her with other BIOS settings like CPU voltages an stuff, as I currently have an ASUS board and I don't want to risk her frying the CPU, Motherboard etc with me telling her to do a certain thing in a setting that sounds similar to mine.

Anyone have any ideas what the problem could be, or any other advice she could use?

*edit*

[CAT-THE-FIFTH]

Try running the RAM at 1333MHZ at 1.5V to 1.58V and see if you have the same issue.

[kalniel]

Suspect the memory controller on the CPU needs a bump in voltage. On the old nahalem systems the voltage needed to be within a certain margin (0.5 or so) of the RAM voltage. Might be different for these chips, but that'd be where I'd target my testing.

I'm also surprised it's OK to run 1.65V ram with the CPU - I know sandy bridge chips are much happier with 1.5V Ram.

[Jackther]

I thought the same too with the voltage, this is my first jump into Intel after 10+ years, as both me and the girlfriend got new PCs, hers being a 3570K setup and mine being a 3770K setup.

So, I'm watching her screen right now running memtest with both sticks in, she removed the wireless card and so far so good, which I find a little bizarre but I've seen similarly weird things happen before.

She also ran ArmA 3 before she restarted to run memtest, as that game usually crashed before, as well as Saints Row 2 & The Third.

I'm hoping it's just a wireless card giving issues and not the motherboard itself... my fingers are crossed.

*edit* 25 minutes in an still so far so good, I think I may get her some RAM sticks that run at 1.5v, just for personal concerns as I'm not really a fan of running RAM at those voltages.

[kompukare]

I'm also surprised it's OK to run 1.65V ram with the CPU - I know sandy bridge chips are much happier with 1.5V Ram.

Pretty sure the spec is 1.5V ±5% which is 1.575V max. This is an answer from Intel but the linked spec doesn't seem to mention the voltage:
http://communities.intel.com/thread/30798

Not sure what kind of buffering / voltage transformation the memory controller does and if they have some way of making those parts more tolerant but the main CPU must be a lot more fragile at 22nm than SB was 32nm.

Even without direct control over DDR3 voltages in the BIOS, just running them at 1333MHz might force them down to 1.5V. Unless using the onboard HD4000 there's next to no performance difference (~1%?) to be had from RAM speed.

[Jackther]

So we've done further tests, with and without the wireless card in, memtest passed both runs with the BIOS settings at optimized defaults, RAM running at 1333 MHz and the voltage set to 1.5v

Now that we booted back into Windows, programs instantly started crashing, Skype as well as the catalyst display drivers.

We've gone back into memtest and it's now running with errors again, no settings had been changed from optimized defaults.

This is quite confusing, I'm again, out of ideas as to what could be causing the problem, the CPU voltage is set to auto with optimized defaults.

These errors just seems to be completely random no matter what settings we've tried, both setting the RAM to the correct timings/voltage and setting the BIOS to optimized defaults.

*update* Just attaching a screenshot of CPU-Z & Hw Monitor incase anyone wanted to see in my main post.

[richw]

I had a similar sounding problem just before xmas after replacing a motherboard & PSU on a AM3 setup along with a RAM upgrade - long story short but it was a memory timing issue but I don't know any more than that as the 'customer' took the PC to a shop as I couldn't go any further (lack of spare parts to swap out, my PC's still run DDR2). The RAM was replaced but don't know what with.

It confused me as the problems kept coming and going depending on what you did - often I could make the problem disappear and it then would reappear less than 24 hours latter.

[Jackther]

It's a weird one, but I think shes narrowed it down, she dropped the speed down to 1333MHz and 1.5v which seems to play ball with no problems.
As soon as she bumps up the speed to 1600MHz, it throws a hissy fit, so for now it seems to be running okay, the system has been running for 24+ hours.

I've ordered her some 1600MHz 1.5v corsair sticks instead, so hopefully it's just down to the RAM itself, as she seems a little bit set back by it, just needs reassurance that it isn't her fault