What's a good CPU benchmark?

[buchtis]

As per title.

Want to try my 955 at stock (3.2) and then at 4.2 to see if a 1Ghz overclock actually has any performance gains. Already have Heaven + STALKER benchmark installed for the GPU but I have no idea about decent CPU benches.

Thanks.

 

[Gashman]

encoding benchmarks tend to give a good indication of processor performance, something like x264 or Handbrake, that sort of thing. for figures, such as your arithmetic performance or floating point performance, Sandra is a decent tool for that, there will be an improvement mind.

did you increase the frequency of the CPU-NB as well...?

 

[Cooper]

As per title.

Want to try my 955 at stock (3.2) and then at 4.2 to see if a 1Ghz overclock actually has any performance gains. Already have Heaven + STALKER benchmark installed for the GPU but I have no idea about decent CPU benches.

Thanks.

Cinebench.

Scales almost exactly with the frequency increase.

 

[mmj_uk]

Cinebench.

Scales almost exactly with the frequency increase.

Agreed, excellent CPU benchmark for both individual core and maximum theoretical multithreaded performance.

 

[N00bbuilder79]

There is a thread here.
http://forums.overclockers.co.uk/showthread.php?t=18322106

 

[buchtis]

Excellent thanks guys, will get downloading Cinebench now. Is it O.K if I download the latest version from their website or does it have to be 11.5 like in this thread?

did you increase the frequency of the CPU-NB as well...?

Not sure, will have a look now. If not, what would I increase it to and at what volts?

 

[buchtis]

Here's mine with clocks in sig etc.:

OpenGL - 65.68 FPS

CPU - 4.85 PTS

Would I be right in thinking that my 955 is bottlenecking my 2 6870's?

Edit:

Stock:

 

[GTA767]

fritz chess banchmark

 

[Gashman]

Cinebench is bizarre with this chip, it does not at all scale like you'd expect with frequency. for example this chip scores 5.05 for multi-threaded and 0.89 single-threaded at its stock speeds, take it up-to 4GHZ and it all goes weird, the ratio between single-thread and multi-thread changes, at stock speeds the MP ratio is 5.70X, but as clock speed increases that changes, surely that indicates something bizarre that Bulldozer gets faster, as it gets faster?

the ratio for multi-cored set-up shouldn't change as frequency increases should it, the ratio should surely remain the same?