MC SP
--------------------------------------
- Tests the MC solely, without touching the DRAM (you can even remove the dram while its testing) , same mc on different boards/dimms should give similar score
- Tests the FMax of an MC at a fixed low SA voltage. In the same way you test a core's fmax at a low voltage idling.
- Since there is an element of luck when testing FMax like you would when running for a cpuz record, the outcome will not be 100% the same everytime and score will shift a little.
- Temperature affects FMax but the effect is weighted and the maximum package temperature during the training is taken into account to generate an unbiased score.
there is a strong co-relation between MC FMax and DRAM OC just like Core FMax with Core Cinebench Freq.
a sample from a not so good board , no so good dram at 1.4v by a not so good overclocker (me!):
in other words, all things remaining the same, there is a high probability that a higher mc sp will net a higher dram oc.