While all the tests on the previous page were easily executed, I ran into issue when playing one my favourite online game of the moment. Bad Company 2 is very sensitive regarding system stability. To give an example : The CL7-9-7-24 tests were all completed at 1.62Vdimm. Though Bad Company would occasionally dump me back at the desktop. A small bump in RAM voltage got me stable enough for hours of fragging fun.
So I had to push the rams a bit harder than on the previous page. Ideal RAM stress tools are HCI Memtes and HyperPi ( the latter being a multi threaded SuperPI variant). If there's an instability with the ram , either one or both of these stress tools will display an error message.
1600 CL6-7-6-21 1T at 1.57Vdimm
1866 CL6-9-6-24 1T at 1.6Vdimm
CL6-8-6-24 failed each time in the HCI Memtest, even at 1.65Vdimm it didn't pass. So better to settle with lower volts for daily usage.
2133 CL 7-9-7 24 1T at 1.65Vdimm
Not bad at all for a 105 euro kit. Ofcourse Overclocking results can vary. Maybe you can match the above results. Maybe you can even do better. It all will depend on the overclocking GODs and luck, as always, has to be on your side. I was not looking for maximumefficiency. Especially the CL7 runs at 2133Mhz took some time to achieve. Overal I'm really impressed what these Gskills could pull off with a mere 5-10% voltage increase.
If you don't want to go as high on voltages, the CL7-10-7-24 1T setting was rockstable at 1.575Vdimm with my sample kit.
My current Sandy Bridge setup is just stable at 106.5Bclock, The Dlx board went up to 107.5, but it's at Massman's place. Upping the Vdimm to 1.65 again, I slowly went up. First testing CL7-10-7-28. Walling and Blue screening at 2168Mhz. Loosening the timings to CL8-10-8-28 1T got me this nice Hyperpi 32M screenshot at 2272Mhz
A quick run on an AMD system gave me 2000Mhz CL7-9-7-28 1T Hyperpi 32M stable at 1.62Vdimm
Time to wrap it up...