Shouldn't higher clock frequency result in greater speed?

[logi-cat]

Running an Intel burntest with my 2500K clocked at 4.7GHz, which resulted in a speed of 126.2188 GFlops. When attempting 4.8GHz, the highest i saw was 122.600 GFlops. Any particular reason why? I'm a bit stumped.

 

[Trunks9486]

throttling? you are probably pushing it beyond what its stable.

 

[pgi947]

A decent CPU benchmark is more likely to show the gains of a higher clock speed than IBT. Try something like CINEbench or superPI. Or even good old 3dmark physics tests.

 

[logi-cat]

Using superPI i get a lower time when calculating PI, but when i was using the custompc benchmark suite (which include decoding, video editing, etc....) i get a higher score with the 4.7GHz then the 4.8GHz.

 

[JonSmith]

That might be suggestive of instability in the overclock.

 

[kitch9]

Error correction probably kicking in, your cpu is not stable at 4.8 simple as that.

 

[logi-cat]

when you say not stable what do you mean? Does it need more voltage or have i just hit the wall of performance. if i run 15 times with ibt it doesnt notice an error.

 

[Martini1991]

I'd have to go more with throttling than straight out instability.
Also, IBT isn't exactly the be all and end all of CPU performance scaling.

 

[logi-cat]

tried Cinebench, here's are the results:

4.7GHz - 7.32
4.8GHz - 7.59

 

[Jokester]

Assuming everything else being equal you've probably just hit the point where IBT is starting to show instability errors that error correction is kicking in to sort out with a knock on performance hit.

In the event viewer in Admin tools you'll see WHEA errors cropping up if this is the case.

 

[Martini1991]

Bump the Vcore a little more in IBT, see what happens, it could fine well be the error correction.

 

[logi-cat]

thanks for the reply everybody.