Its much more than 10% - Erik (kholm) did a quick speed comparison and posted the results but I cant find it ... the comparison was something like the following (for the 749 byte string):
Winning code: 2.5 seconds Fast code: 30 msecs
Its no big mystery or anything - the fast code is fast because it only needs to perform two passes of the data, and one pass of the "alphabet" 0-255 ... no matter how big the string is.
![[Frown]](images/icons/frown.gif){kind=icon}
... the comparison was something like the following (for the 749 byte string):
