G.Skill PI DDR3-2000 C7
SuperPI, and to a lesser extent, Cinebench, both show sensitivies to memory subsytem tweaks, so we’ll see if memory tweaks result in synthetic score increases as a result of performance boosts here.
SuperPi is a simple benchmark that is very popular among overclockers, and is a good indication of raw CPU calculating speed, and shows senstivities to memory subsystem tweaks as well. Here we calculate PI to 1 million digits and record the time it takes to complete the run.
Here we see SuperPI scaling rather equally through timings and bandwidth. The performance is very close between the competitors, although you can see the G.Skill modules manage almost a tenth of a second improvement on average until the upper limit where the gap lessens. They will also let you overclock slightly higher than the Patriot modules, and using the memory as the limiting factor here we managed a speedy 10.78 second SuperPi result at the maximum overclock.
Cinebench is a program that renders a high resolution scene, complete with complex lighting and shaders, and can approximate what workstation users would perform on a regular basis. Cinebench can be run in single-thread or multi-threaded setup for testing CPU performance and hyperthreading effectiveness, and we’ll run multi-threaded for this test.
Cinebench shows a minor but discernible difference here across the board, with the G.Skill PI modules pulling ahead at all speeds and timings. When we look at what is possible for maximum overclock potential here in a total system result, the increases are staggering, allowing us to push the CPU to almost 3.9GHz and an an excellent 31% performance increase in Cinebench.