I know this type of process can't really take advantage of multi-processor support, but could something like Adapteva's Epithany processor be of benefit for code like this?
Oh, Bob, you know my answer. (One for me, please, one for meeee!) Though I don't frequent the boards nearly as much as I used to (family issues and other aspects of life keep intruding. ), I'm a 7800 supporter 'til the end.