* Set max cpu to 1 for PPC until atomic functions are finished * We have atomic functions inline in the kernel and assembly code in libroot post-scheduler merge... isn't that a lot of duplication?