On Sat, Feb 10, 2007 at 03:24:47PM +0100, Juerd Waalboer wrote: > Gerard Goossen skribis 2007-02-10 15:25 (+0100): > > NONE*: 1.79s > > UTF-8: 1.86s > > UTF-16: 2.10s > > UTF-32: 2.90s > > Did you use a word oriented regex engine for the latter? The benchmark was attached at the bottom of the mail. It does a simple for-loop looking for the byte/word/dword. And yes there are much better ways to search through memory, but it demonstrates the impact of using memory. Gerard Goossen