For a multi-cpu design, you should take a look at the Propeller chip from Parallax. www.parallax.com. Its bascially 8 32-bit cpus with local and global ram on one single chip. They run upto 80Mhz! The best bit is that they only cost $15!!! The downside is that they need external ROM to store the program, but serial EEPROM is cheap enough.