N-Queens on 50g (RPL language)

[Claudio L.]

Applying two simple improvements, one in the instruction execution loop of newRPL (moved some unnecessary things out of the loop), and then replacing memmove() with a version that moves words instead of bytes (which I should've done from the beginning), produced a nice speed improvement between 10 to 25%.
These improvements are nothing special and apply to all programs running on newRPL.
The original went from 226 ms to 198 ms, while Werner's went from 141 ms to 106 ms, being the one most impacted by the improvements.
I'll update the times in the original post with the new marks for recordkeeping.