CoreMark of MPC5607B: size optimization runs faster than speed optimization?!

[Spete]

Hello Guys,

I'm a student at university of apllied science in Fulda/Germany.

For my Bachelor-Thesis i've executed the CoreMark at MPC5607B with CodeWarrior IDE 5.9.0.

I've chosen several optimization settings at settings... -> Code Generation -> Global Optimizations for size and speed optimization.

At every position, the size optimized application runs FASTER than the speed optimized app.

Does anybody have an idea what causes this fact?

It seems to me, that the check boxes "Faster Execution Speed" and "Smaller Code Size" are interchanged.

In hope for an informative answer

Peter

[Spete]

Does no one has an idea?

Or do you need more Informations?

Best regards

Peter

[stanish]

Hi Peter,

Actually you might get such results since both types of optimization uses different techniques e.g. loop unrolling in general should speed up the application, but in case there is a cache enabled (which is probably not your case)  you might get a different result then expected. 

If you attach the CW project  I can try to analyze it on my side and let you know.

Stanish

[Spete]

Hi Stanish,

it would be great, if you have a look on the project (Coremark2.zip). Maybe you can find the root cause.

Thanks

Peter