Made two simple modifications to phatk.cl, speed went from 131 Mh/s to 136 Mh/s on Radeon 5750 @ 800Mhz. Be sure to use -v -w 128 in Flags (I use -v -w128 -f60 -s 0.001 for desktop use, gets about 125 with no lag).
Optimized 2 inline functions and replaced all "mod 8" with "bitwise-and 7".
Just replace with this copy for 3% boost. Let me know if you get more out of it! I'd appreciate any feedback.
https://dl.dropboxusercontent.com/u/205653/bc/phatk.clDonations welcome :) 17jYVen8KDPYCAE2nMkB5BbiFjkZ1f7kU8