Unstable system with Ripjaws Z Series 64GB (8 x 8GB) F3-12800CL10Q2-64GBZL

[Tanquen]

Ok, so I should be slowing raising the VCCSA and/or the VTT by .01V. One and then the other or just the VCCSA and what is the cut off? I don’t want to break something. Also, the BIOS gives the option of offset or fixed. Which should I use?

Is there a program that will show me the current VCCSA voltage? The ASRock utility just shows the Offset of +0V.

[Tradesman]

I prefer fixed and I'd stay at 1.25 or lower on both

[Tanquen]

I don’t know, those are the defaults settings. Everything is set to auto. I’m not trying to overclock I just want my new PC to not randomly crash. As instructed, last night I bumped up the VCCSA to 1.204V as that was the next setting above the default. The Intel Burn Test still had an error on the second run. With the highest stress level and the most RAM it takes around 2 hours to run once. So tonight I was going to bump up the VTT.

It looks like the Core Voltage is moving from .864V to 1.27V when the Core speed goes from 1.2GHz to 3.5GHz.

[Tradesman]

I'd go ahead and try the VTT up a pinch, could play with loadline calibration, but I'd get stable first and .8 to 1.2 isn't a bad fluctuation with the Freq diff being, at 1.2GHz, you are basically at idle, and power savings are kicking in to take the voltage down to .8

[Tanquen]

Not sure what that other guy was on about and why his post is gone now but why would one play with loadline calibration? What is wrong with the default setting?

I bumped the VTT up one notch from 1.202V to 1.215V and the Intel Burn Test was still running this morning. I’ll need to run it and Prime95 a few more times to be sure as it was totally random. I hope this will get rid of the random lockups. Originally I was able to run Prime95 overnight with no errors the first few days after putting the system together. I’m worried the RAM or CPU is failing. Maybe I did not run the tests long enough.

If this is indeed the fix what would be the cause? ASRock has the wrong default VTT, the CPU is bad/weak or the memory has a weak DIMM.

[Tradesman]

The thread was deleted by the member. Each individual system has to be set up for itself, these days tolerances are so tight you could build a half dozen 'identical' high end systems (all use the same exact model of components and might possibly have 6 different voltage setup for stability.

[Tanquen]

Ok, it still fails Prime95. Should I just keep bumping the VCCSA then test and then the VTT and test and so on until I get to 1.25V on each. I was wondering if I could bump both and then test. I don’t really think I should be doing this to get the system stable at default settings. I mean increasing the voltages is not good for the CPU.

I feel like I’m using over clocking techniques to try and get the RAM to run at its stock settings. DDR3-1600 is the Intel standard for the X79 boards, it’s not like I’m trying to run them DDR3-2400 or something.

[Tradesman]

I agree, know the VCCSA normally needs the increase, but that sounds high for VTT, have you checked with ASRock on their BIOS status, this should all work under XMP, unless they aren't ready for prime time with it

[Tanquen]

GSkill lists my motherboard as compatible with this kit. Maybe they did not really test it, I don’t know. When I was originally told it should have a XMP 1.3 profile I contacted ASRock and they told me the same thing that GSkill support told me. That the XMP 1.3 profile will only be on kits that are rated to go higher than DDR3-1600. I think their BIOS should then use the XMP 1.2 profile by default with the AUTO setting I instead have to manually select it but the DRAM timing settings that then show up in the BIOS do match the GSkill packaging. GSkill support did ask that I try setting the DRAM timings manually to the timings that are on the RAM packaging but they already match when I select the XMP 1.2 profile.

[Tradesman]

I'll drop GSkill a line and see if they can help.

[Tanquen]

Ok, I just tried to manually set the DRAM timings but the BIOS won’t let me. There is no disable for the XMP setting and I tried AUTO and STANDARD but whenever I go to the DRAM timings the tCL, tRCD, tRP and tRAS settings all say AUTO and when I click on them nothing happens. ARG!

I’m just guessing that those are the 10, 10, 10, 30 settings that are on the RAM and the box they came in. Again they seem to be set correctly when I select the only available profile.

The other suggestion from GSkill support was to test one DIMM at a time. I can try this but it will take a day or two just to be sure one is stable. Sometimes I can get a Prime95 fault in 15 minutes and other times it can run overnight with no error. It’s the same with the Intel Burn in Test. My concern is that the instability may only show when in Quad Channel mode as the CPU, RAM or motherboard are just enough out of spec to cause the issue only when all 8 DIMMs are installed.

All that being said, that is where I’m headed. My thought was to test each in its individual socket so as to test the sockets as well.

[Tanquen]

Ok, I’ve now officially been driven crazy. Last night I left one DIMM in and reset the BIOS only setting the XMP 1.2 profile. It ran Prime95 overnight with no error but then I went to copy a file and while browsing the C drive it locked up. I’ve never had this with a new build and I’m getting very disheartened. I’m going to keep running the Intel Burn in Test and Prime95 to see if this one DIMM will end up showing the same errors it has when all 8 were installed. Do I have two different issues? Maybe the motherboard is bad. Without a BSOD how do you track down the cause of the lockups?

[GSKILL TECH]

Try a second module to see if it has that same problem.

If it does, it may be a BIOS/motherboard problem.

If it does not, the first module may be bad causing problems. If that is the case, you can send the memory in for RMA exchange and a replacement kit should work better.

Thank you
GSKILL TECH

[Tanquen]

I got a GSkill TECH email telling me that:
“The VCCSA is the memory controller voltage.”
and that:
“Intel only supports DDR3-1333 CL9 1.50V, so when you have memory better than this or many more modules, this voltage needs to be higher to be able to be stable with the additional stress.”

I don't think that is correct and this was after asking about what VCCSA and VTT did and why I need to increase them.

I can test another single DIMM but right now I’ve got all eight back in and am trying the VTT at 1.228V and the VCCSA to the next step up from 1.204V. I forget now what it was as there Windows utility only shows the VCCSA offset not the current VCCSA voltage. It’s not failed yet but it can take a few days.

I also feel changing voltages is a band-aid to mask the RAM, CPU or motherboard being out of spec.

This stuff was not cheap and I did everything I could make sure everything matched and should work.

[Tanquen]

I got the error in Proime95 again with VCCSA at 1.210V and VTT at 1.228V. Now trying VCCSA at 1.215V and VTT at 1.241V.