F3-12800CL9Q-16GBZL in Rampage IV Formula stabilty

[Finalizer]

Motherboard: Asus Rampage IV Formula (x79)
BIOS: 1305
Memory modules: [ RipjawsZ ] F3-12800CL9Q-16GBZL (4Gx4)
CPU: Intel i7 3820 Sandy Bridge E

Hi all,
I've got a question regarding stablity with these components.
I am NOT overclocking the processor, only using XMP mode for the memory.
Everything else is stock clocks and voltages.
I chose XMP in BIOS, mode 1. DDR3-1600, 9-9-9-24-2T (as per module design)

When I first built this system 2 weeks ago, I had BSODs popping up while trying to install Windows 7. A quick test with Memtest86 revealed memory errors all over the place, in the hundreds.

I bumped up the DIMM voltage AB and CD from the stock 1.500 to 1.530 in 0.005 increments until Memtest86 reported 0 errors for a complete loop.
Windows 7 installation worked perfectly after that.
Fast forward to last night when I finally had some time to play a game (Tribes Ascend).
The game crashed to the desktop 3 times with application errors.
Thinking it was related to using SLI, I disabled SLI for the game in the Nvidia driver profile settings. (I have a GTX 690)
I played a little bit more and it did not crash.
Then, I decided to run Memtest86 again, just to make sure main memory was still OK.
Turns out, it wasn't. At around 9% through the test, errors started appearing.
I bumped the RAM voltage to 1.535 and retested. Still some errors, but fewer of them at the 9% mark.
Bumped RAM voltage to 1.540 and retested. Only 2 errors found at the 9% mark.
Bumped RAM voltage to 1.545 and retested. 0 errors found.
I left Memtest86 running all night long. 0 errors found in 7 hours of testing (I didn't note how many loops it was, but probably at least 5 full tests).

My questions are:
Is it normal to have to bump the RAM voltage in order to get the system stable? I mean, the stated normal voltage for these modules is 1.500V.
(Keep in mind this is an X79 board with all 4 DIMM slots occupied (Formula only has 4, Extreme has 8)).
I did not bump VCCS or any other voltages. They are all on Auto. Should I have to play with those due to the higher memory voltage, or leave them alone?
Should I try changing other memory timings and bring the DIMM voltage back down to 1.500, or is a 0.045V bump not a big deal?
Just want to make sure this won't shorten the life of the RAM or memory controller on the CPU as this will be a 24x7 system.

Thanks for the help!

[Tradesman]

Do you have the latest BIOS? 1305? They've been working on memory compatibility (can be found here)

http://www.asus.com/Motherboards/Int...mula/#download

[Finalizer]

As stated in my original post, BIOS is 1305.

[Tradesman]

Missed that, sorry...could be a bad stick or could be BIOS not setting things correctly....I see where you ran memtest, but was this with all sticks or each stick by itself? I'd try a few passes on each stick individually (maybe 4 or so and see if any pop up errors, if so then it's prob a bad stick...if no errors with each individually then I'd guess possibly that the advance timings are off or the mobo or PSU isn't regulating voltages properly

[Finalizer]

I have not tried each stick individually, so that is worth a shot.
What voltage should I use when trying 1 stick at a time? 1.500?
The power supply is a Seasonic X-850 Gold.
Thanks Tradesman. Maybe I can learn what each stick needs to run with no errors and then report back.

[Tradesman]

Yes, just a few passes at stock timings and voltages, really just looking to see if one is bad, running multiple sticks can be odd, sometimes it will give errors other times it can come back fine, no errors, but you might actually have a bad stick.... If they all test fine, then it narrows things down a bit and can look more towards voltages/timings/ possibly even CPU/MC

[Finalizer]

UPDATE:
After 9 hours, I have tested all 4 sticks individually at stock VDIMM 1.500V and XMP profile timings (DDR3-1600) and 9-9-9-24-2T as per G.Skill memory specifications for these modules.
Each stick was tested with Memtest86+ 4.20 from a bootable USB flash drive.
Each stick was installed into DIMM slot D1, as recommended by the Asus R4F manual when using 1 stick.
Each test consisted of 4 passes through Memtest86+ with the "Std" (default) test. (Over 2 hours for each stick)

Results:
Stick 1: PASSED
Stick 2: PASSED
Stick 3: FAILED with 2 errors: 1 error in pass 0, 1 error in pass 1.
Stick 4: PASSED

I'm shocked that the bad stick did not have "hundreds" of errors, as I was seeing hundreds when running all 4 modules at stock 1.500V.

[Tradesman]

Sounds like time for an exchange/rma...if within the return/exchange time limit of your re/e-tailer can go through them or contact rma@gskillusa.com

[Finalizer]

Thanks Tradesman. RMAing with Newegg now.

[Tradesman]

Let us know when you get the new, we'll be waiting to hear

[Finalizer]

2012-06-01 UPDATE:
Got the replacement memory yesterday from Newegg.
Installed all 4 modules and verified XMP profile was still selected with 1.500V in BIOS.
Tested overnight with Memtest86+ 4.20.
After 12 hours and 7 full loops through Memtest86+, zero (0) errors were found!
Awesome! Finally, my system is stable.

Thanks for the support, Tradesman.

[Tradesman]

No problem, glad all is better, Enjoy!

[Finalizer]

Whelp, looks like I spoke too soon.
After running Memtest86+ overnight, I left the machine on all day today.
Mostly idle, some web stuff, nothing special.
I went to start a game after work and heard terrible screeching from the speakers.
I quit out of the game menu, and Windows told me the sound mixer crashed (Appcrash).
I booted into Memtest86+, and sure enough, thousands of errors are now showing up.
Tested each module individually and one module had over 13000 errors after only 4% through the test.
This sucks. No amount of voltage helps either. I went up to 1.555V and it didn't get better.
This set of memory is worse than the other set I just shipped back to Newegg.

Now I have to RMA this set of RAM too.
I think I will look for another module type or brand, as it seems clear my Formula doesn't like this stuff.

[Tradesman]

Sort of sounds like a Memory Controller or may PSU or mobo issue

[Finalizer]

So, you mean to say that something (PSU, mobo, memory controller) is destroying the memory module?
The first set of RAM had the bad DIMM installed in B1.
This second set of RAM had the bad DIMM installed in A1.