G.Skill Ripjaws 4 16GB 2666MHz (F4-2666C15Q-16GRR), Overclocking Ambrosia!

Testing & Methodology

Please note that this represents a complete retest of all kits on a different motherboard and almost completely different machine than previous RAM reviews.

RAM testing can be affected by a lot of factors: background programs eating RAM bandwidth, any unforeseen background programs sneaking in, OS maintenance activities, that sort of thing. We turn off all background programs, clear the cache and prefetch, make sure Windows indexing is turned off, and have Windows defender disabled. In short, we turn off everything that we can to ensure that nothing interferes with the tests we are running. Each test is run three times and the average of those three runs is reported here.

Memory can be a little dicey to test, and you can never make everyone happy with testing memory. What we shoot for is the highest OC you can get with the timings at the factory rating. Then we might loosen the timings a little and overclock with a voltage jump, but only to the point that OCing produces positive results. At times you can get a sweet 500MHz OC, but then you look at the SuperPI numbers and the score tanked to the floor. At the point where we hit max safe voltage and show improvement in scores, we are fine. When we hit speeds that are higher frequencies, but scores are dropping, then it’s time to back up and find the last speed where we got productive increases in SuperPi scores and call that the Max OC.

For benchmarks on the charts we generally run the rated speed on any Intel 2.0 XMP, which is where the vast majority of end users will be. The overclocking fun is left in the overclocking section and we test real world XMP speeds.

Test Rig