About 2 weeks ago, I built a sandy bridge rig using a 2600k and a gigabyte z68x-ud3p-b3 mobo (F6 bios) The rig had 8gb ram and was not currently overclocked (plan on doin later) The ram is http://www.scan.co.uk/products/8gb-...3-14900-(1866)-non-ecc-cas-9-10-9-27-xmp-150v I have had no problems running the pc at stock using the xmp profile on the ram, and as I have been busy I have not had time to overclock etc I have been using the rig on a daily basis and decided it needed more ram (for vmware etc) so I ordered another kit of the same stuff. When it arrived, I saw that it was the same 'version number' as my current ram, so installed it into the remaining dimm slots. After booting windows, my sound wasnt working and got the general xxx has stopped responding on a load of my apps. So I figured it was cause all 4 dimms were populated it needed a slight voltage increase. Looking at the xmp profile it shows it sets the voltage for memory at 1.5 and vtt at 1.2 - I have put these all the way upto 1.65 and 1.3 in different steps but am having no luck getting all 4 dimms working. I have tried taking out the old dimms and running a memtest on the new ones - and using the standard xmp profile and voltage, everything seems to work fine. I am going to test running all 4 at a lower speed (1600) at 1.5 and vtt at 1.2 - after this im running low on ideas - anythin else I can try?
An overnight test shows it failed on the 'failsafe' options in bios - I will double check the options used and test again accordingly ... looks like I may have gotten bad ram
Not sure if this is at all relevant as I am running AMD but I had to drop from 1T to 2T to get four sticks of ram to run. Mine wouldn't even boot with it set to 1T. It's a quick thng you could try.
They were set at 2T - running a longer memory test with just the new sticks now All sticks pass the memtest86 v4.2 with no problems - just it fails in the windows HCI memory test with all 4 in (and progs like superpi produce errors occasionally - not always) - it is always the HCI memory test that catches it - usually within the first 20%
I will try running at 1333 with a higher voltage and see if that works - the new memory seems to pass a 150% test with memtest