The best and fastest answer I have so far, with a fixed latency (9 instruction cycles ?)takes 9 instructions and 8 bytes of look up.

I have a little proof, but it is too small to fit in this margin......


Rules:
1.) Latency is fixed. A solution where the delay of the output depends on the input is not allowed.
2.) Data comes in on P0 and is put out on P1 and P2.
3.) Fastest executing wins.
4.) 'C' programmers need not apply.

5.) Prize is non-existent.

Steve