UTF-8-decode and encode are basically the only things that could be done by a single instruction, are common enough for implementing this instruction to make sense, yet still don't exist in x86. I wonder why.
Every time I have worked for a company that had a tight relationship with Intel I've gotten the opportunity to meet with them and ask for chips features. I always ask for encoding / decoding Unicode. So far I've been ignored for 15 years or so.
