What the hell???

[MariuxOC]

http://3dmark.com/compare/pcmv/490977/pcmv/490963

Everything is the same, exept I OC'd my CPU from 3.6 to 3.9Ghz...

[Jarnis]

The differences between the two tests are pretty much within margin of error.

PCMark tests common usage scenarios for desktop use and doesn't scale on most tests to 6 cores (just like most desktop tasks don't use 6 cores for anything) so the minor overclock doesn't affect things that much - you can see the difference in Gaming CPU test. The difference that pushes the 2nd score lower seems to be HDD related tasks; perhaps you had something running in the background on the 2nd test that accessed the HDD while benchmark was running and that influenced some of those test scores? Or perhaps the overclock had unintended side effects that influence storage or memory performance?

[MariuxOC]

The problem is the score is lower when i OCd my CPU by 300Hz...? Even if it would be the same i would be surpriced, but why the hell it's lower???

[Jarnis]

The problem is the score is lower when i OCd my CPU by 300Hz...? Even if it would be the same i would be surpriced, but why the hell it's lower???

First, there is an acceptable margin of error for FM products (it is 3-5% depending on the benchmark) so if the score is within that, it is just variance; you'd need to do several runs on both settings and average them out to get rid of this variance and be sure that the difference isn't explained by it.

Second, you should look at the various subscores and see where the difference comes from. As far as I can see, you got generally worse HDD related performance figures on the overclocked result, hence my suggestion that there may have been something in the background accessing the disk during benchmark on the overclocked run, causing the performance to take a dip.

Also the CPU clock rate is just a small part of PCMark test; just overclocking the CPU by a small amount (less than 10%) isn't going to cause major shifts in the score as memory and hard drive performance are also a large part of the score and video card also influences it. In other words; you changed one part of the performance equation slightly and got a result that is within the inherent error margin of the benchmark, indicating that it didn't really improve the performance of the system (which is true - 300Mhz CPU overclock won't make any noticeable difference in day-to-day use)