ALL the benchmarking software and games are run at least three times before taking any numbers from them. To make thing equal we tried our best to make the rest of the system run identically to measure the reviewed parts perform to their best. We believe that everyone should take advantage maximum capability of their products by modestly overclocking them to bring the best performance possible so, we run everything Overclocked. Though we didn’t push them or make tweaks any way that any regular users cannot do, we keep the overclocking to a basic level, which anyone can achieve easily. For a game benchmark, we use one particular section of the game which can repeatable and has a decent level of coverage of that game (including, action, long distant seen and other elements of that game). If any component didn’t perform as expected on any particular benchmark, we retest multiple time and sometimes cross check with the manufacturer or other 3rd party sources to confirm the results.