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

[Tradesman]

Let us know how it goes

[GSKILL TECH]

If you believe the memory is out of spec, feel free to send them in for a replacement kit to see if another works better. Raising DRAM Voltage beyond the rated specifications can be considered a band-aid, but VCCSA is necessary when using high performance and/or high capacity kits.

Thank you
GSKILL TECH

[Tanquen]

If you believe the memory is out of spec, feel free to send them in for a replacement kit to see if another works better. Raising DRAM Voltage beyond the rated specifications can be considered a band-aid, but VCCSA is necessary when using high performance and/or high capacity kits.

Thank you
GSKILL TECH

I understand what you are saying and I?m not trying to be a **** but I?ve been given some misinformation and don?t feel that anyone should have to mess with voltages to get a kit to work when:
You?re not trying to over clock the CPU.
You?re not trying to over clock the RAM.
You?re not try to mix and match different RAM kits.
You?re not trying to run the RAM faster than the kit is rated.
You?re not trying to run the RAM faster than Intel supports for their CPU and chipset.
You?re not trying to run More DIMM sockets than Intel supports for their CPU and chipset.

If any of the above were true I?d be right with you but Intel seems to be conservative with everything else but not RAM? If I?m running the CPU at default settings it should just work. If I?m running the RAM at default settings it should just work. Intel says their CPU/chipset supports up to DDR3-1600, 8 DIMM sockets and up to 64GB of RAM but they messed up on the internal memory controller voltages and didn?t account for people using 8X8 DIMMs? It could be they have messed up other stuff but I usually feel the most confident about the CPU and less about the motherboard and RAM.

Anyway?

So far raising the VCCSA and VTT has only succeeded in increasing the CPU temperature a little. So the important thing to me now is to find the most likely cause. I?d say maybe the RAM is most likely then motherboard and then CPU or power supply but I don?t know. I don?t have a spare of any component to swap out. How long would an RMA take? Do you offer cross shipping in these situations?

[Tradesman]

If you RMA and the same thing continues, it might help narrow it down to mobo, CPU or PSU. Might check w/ the rma folks at rma@gskillusa.com, though I just dropped a note to GSkill Tech, may be able to provide the info for you Thurs.

[Tanquen]

Thanks for the help, I’ve just been dreading the RMA process and giving up the PC for a week or two.
I spoke with GSkill phone support today and one thing the guy mentioned was being sure to run Memtest86+ on a single DIMM at a time as when they are in dual, tri or quad channel you don’t really know that every single address in each DIMM has been tested.

I also saw some folks online talking about HCI MemTest and GoldMemory test and how they had some bad RAM that Memtest86+ showed no errors on but these other programs did.

I ran Memtest86+ on the one DIMM that PC had locked up with the other day. I ran it three times with no errors. GoldMemory crashed after two runs but I’m not sure what that means. Right now I’m trying HCI MemTest.

[Tradesman]

Keep us informed, imagine GSkill will be in tomorrow and did drop him a note to take a look

[GSKILL TECH]

It is a tricky problem. If individually each module locks up, it can be a problem with another piece of hardware. If only 1 or 2 modules lock up, it may be a problem with the memory.

I agree with you as well, this kit should be no problem for your hardware combination. But right now there is a strange problem, so we must explore all possibilities to determine the cause.

RMA typically takes a week. For cross-shipping, ask the RMA dept.

Thank you
GSKILL TECH

[Tanquen]

Ok, I’ve been testing the DIMMs one at a time and with the first one I did see one lockup but no errors in MemTest86+ overnight and then Prime95 overnight. The second DIMM had no lockups and no errors in MemTest86+ or Prime95.

Now with the third DIMM I’m seeing something new. MemTest86+ overnight was ok but the next night Prime95 had an error about three hours in and then the PC rebooted an hour or so after that. So I ran Prime95 again last night and while there was no error in the Prime95 log the PC had rebooted again after about 5 hours. In all other cases thus far the PC has been locked up or frozen. I’m going to run Prime95 again tonight and see if it happens again.

[Tradesman]

Let us know, I'll drop GSkill tech a note to take a look and see what he thinks

[Tanquen]

I’m feeling pretty good that something is different about this third DIMM. I started Prime95 at about 8pm and had an error at 1am. This time the PC locked up an hour later at 2am.

I’m tempted to go back to DIMM two and see what happens. It ran overnight but only once and had no errors and did not lockup. Were as DIMM three has locked up or rebooted three nights in a row.

[Tradesman]

May have gotten a bad stick, doesn't happen often, but on occasion......let us know what happens

[Tanquen]

Testing, testing and more testing.

Ok, so what I’ve found is that if I raise the VTT to 2.41V, the VCCSA to 2.48V and the DRAM voltage to 1.515V I can (thus far, running 8-10 hour tests overnight) eliminate Prime95 failures from any of the DIMMs installed individually but… If I install all 8 DIMMs together, I again get Prime95 errors. ???

I should be able to run all 8 DIMMs at DDR3-1600 with stock settings. I have to bump things up quite a bit to get a single DIMM to not throw errors and I don’t know what it would take to get all 8 running with out errors.

I was also finally able to set the timings manually (10-10-10-30) but that did not seem to change anything.

Sad, distraught, disappoint and not sure what to do next.

[Tradesman]

First I hope the VCCSA and VTT settings are bot 1.whatever instead of 2.whatever 2.anything is toooo high. Think its time for a RMA, as GSKIll Tech said if there's problems with all the sticks it prob something else (normally I look at CPU or mobo, particularly the socket for bent broken pins), but if it's only a single stick, maybe even 2 it may be the DRAM,, might see if you have or can borrow some other sticks, try that DRAM in the rig

[Tanquen]

Yes and oops. The VTT is 1.41V, the VCCSA is 1.48V and the DRAM voltage is 1.515V

That is true. I have only been testing the single DIMMs in the first socket. It could be that one of the other sockets is in worse shape and when putting all eight in I get the Prime95 errors. It just seemed that the fact that upping the voltages seemed to cure the Prime95 errors when using a single DIMM would mean it was not socket related.

I did get a DDR3-1600 2GB DIMM from a friend for a few days and I used its XMP profile. It used 1.6V for the DRAM voltage and different timings but it had no Prime95 errors at default settings but I did not have eight of them to try.

I went through most of the DIMMS at stock and all of them had Prime95 errors. A DIMM I’ve labeled DIMM 3 had errors every night until I upped the voltages. Now none of the ones I’ve tried after upping the voltages have errors in an overnight run of Prime95 but even with the voltages increased the first night I ran Prime95 overnight with all 8 DIMMs I had errors. Where the single DIMM (at stock settings) would sometimes have only one of the 12 threads fail when I put all eight back in with the voltages increased I had 4 or 5 threads fail.

I did try compressed air in the sockets and they seem ok to the naked eye. I could try a single DIMM in each socket but the manual says to use the first one if you are only using one DIMM.

I think its RMA time too but the problem is so random that I fear a swapped out motherboard or RAM will just do the same thing. Like maybe the motherboard was just not fully tested with this kit even though it’s listed on the GSkill site. I mean my friends DDR3-1600 2GB DIMM just works with its XMP profile selected but each of these 8 DIMMS needs the voltages increased? Like each one is bad or are the timings or DRAM voltage just wrong? Maybe they really need 1.6V but then they would need fans and so on.

Also, no one does cross shipping anymore so while I’ve long since given my money for this hardware have to send it back and wait. I’ve been thinking about buying an ASUS board and a different RAM kit from Newegg and then paying the shipping both ways and the 15% restocking fee just to find out if it the RAM or motherboard. It may have been the better answer over a month ago.

[Tradesman]

Getting errors on most of the sticks makes me think mobo or CPU (or on the outside GPU or PSU), did you have a chance to check the socket and the pins, or possibly thermal compound might have seeped in or gotten on the base of the CPU....just sort of hard to imagine multiple bad sticks....and I'd say the same about any brand of sticks....another possible would be if a friend or coworker has a same socket rig where you could swap CPUs and test....then there's always the dreaded take it to a shop....at that, I'd suggest a small independent shop....i.e. I've yet to run into anyone from the famed Geek Squad that really has any idea of what they are doing, that's just been my experience and that of some others who've used them who told me about it