To run these benchmarks, just change to the directory containing this
file and then invoke

  ./run binary1 binary2 binary3...

At the beginning of this ./run script, there is a line
  N="-n=3"
which means that every binary will be run three times for every test
and the average (and standard deviation) over these runs will reported.

If you only want a rough picture, you can change this to
  N="-n=1"
and slash the time required down to 33.3%.
