Benchmark for special matrix-matrix multiplication

The goal is to find a faster implementation?

To compile and run the code, you can use CMake:

mkdir build && cd build && cmake .. 
make && ./benchmark