MathsMark

This app performs benchmarks and stress tests by calculating a double-precision floating-point real number forward FFT on different sets of 1D input data (sine, square, triangle waves of different frequencies, random data, combinations of these and other functions).

The FFT is done by using the wonderful JTransforms library. (https://sites.google.com/site/piotrwendykier/software/jtransforms)
The native FFT is done by using the excellent ooura FFT library. (http://www.kurims.kyoto-u.ac.jp/~ooura/fft.html)
The Java code is partly derived from the native code so the implementations should be quite similar. This lets you compare the native performance on the system with the performance of the Java code on the virtual machine.

Further points:
- uses all available CPU cores
- lets you view the most relevant system information

This app also exists as PC application to compare x86 and ARM CPUs. You can get the application here:
http://stefan-lober.blogspot.de/2014/04/mathsmark-13.html