On 80MHz RISC system, that runs in ~2s. So, even on a 10MHz system, random
run should not make user wonder if a system hanged. Of course, for more
precise measurement, more loops should be used, e.g.:
pystone_lowmem.main(1000)
There're 2 arrays - one dimensional of 25 elements, and 2-dimensional
of 25x25 elements. Number of iterations stays the same, it's just
consecutive iterations may access the same array elements. So, while
the result of calculations may differ, the timing should stay about the
same (well, lowmem version will be a bit slower, as array index division
by 2 is used in several places).