For university I benchmark the influence of certain refactorings (e.g., inline class and replace inheritance with delegation) on performance. Although I use HotSpot 5 and 6 in interpreted mode (-Xint), the differences in performance between the refactored and the not refactored versions are very small (few milliseconds at best).
Therefore, it would be interesting what is the effect on a device with scarce hardware resources (e.g., Qualcomm MSM72000, 400 MHz processor). Unfortunately, I can only conduct the benchmarks on my laptop (Intel Core2 Duo T9400 2,53GHz, 4gb ram,).
Is there any way to estimate the performance influence on a device with scarce hardware resources in a "scientifically sound manner" (i.e., I cannot use rough estimations like "such a device is about 100 times slower"). Papers, Books, Websites, and so on are welcome. Thanks for your help!
(I am aware that micro-benchmarking is often misleading!)