Anyone have 128GB working on the R6EO?

[n0xlf]

After many hours of tweaking every related setting to include the most likely ones (DRAM voltage/VCCIO/SA/increased current capability) I can not get 128GB of G.Skill F4-3600C17-16GTZKW working reliably. It fails memtest on the most basic of tests and hard locks almost instantly on the Prime 95 large FFTs test. I've tested all the way down to 2133 at 20-22-22-42 without success.

At this point I'm thinking either a (very) crappy memory controller on my 9980XE, or a BIOS incompatibility issue with my modules. Running the latest 0702 BIOS, but I also tested what was in my BIOS 2 bank, which was really old (like 0204 or something). I have a custom loop that includes cooling the memory, so heat likely isn't an issue.

I'm leaning toward a BIOS problem because the memtest/P95 results are pretty much the same at any settings between 2133-3200, meaning that they don't last any longer at more conservative settings. It won't POST above 3200 which is understandable for 128GB and certainly not my concern at the moment.

I left out a lot of details on everything I've tested just to keep this shorter, so please feel free to suggest anything.

[Super_Gnome]

I have the same CPU and mobo, and have 128GB of ram installed. I just plugged it in and it worked. My memory is different though, and listed in my specs. I checked your memory at the GSkill website and it is compatible with the CPU, so it should work.

I have had problems with memory, noob problems, like the sticks not being seated properly, but have never had problems because of using a few different kits in one machine. I know you can't rule that out, but when I encountered these problems I learned a few things.

First, have you tried using only a couple sticks, and has that worked? Have you tried other memory, if you have any that is compatible? Have you removed and reseated your sticks? Have you tested them two at a time? Has your rigged ever worked properly since you built it? Have you tried not adjusting any settings at all?

Maybe doing these things will get you some idea of what is going on. Good luck. Sorry if I did not read all of the above posts. Just my two cents.

Edit: Oh, fixes. Well then. Disregard the above then.

[n0xlf]

In this case it certainly appeared to be the mixing of kits, although that doesn't entirely rule out that subtimings for the 3600 set were the problem. Since I RMA'd two 3600 kits for one 3200 kit, kind of hard to say exactly what it was. As you mentioned, I have also never had an issue with mixing kits until this. It would be interesting to compare subtimings of identical 32/64/128 kits to see if there is actually a difference. Might be difficult to find though.

[Super_Gnome]

Yeah. Apparently mixing kits can cause problems if the batch was different due to slight material differences. Anyhow, it's nice to know GSkill helped you out. I recently unknowingly installed three kits of two GSkill sticks in an older rig. The computer initially didn't see two of the six sticks installed, and so I then madly hunted around for information, and learned that mixing kits of the same memory model has been problematic for some. In my case, it simply turned out I hadn't pushed all of the sticks all the way in, so the problem was easily solved.