Even without any workaround, a performance loss of 67 MHz clock for a modern RISC processor is in the order of negligable at best and trivial at worst. And with the workaround, you don't even need to change the clock.
But you can bet it will feed many "PPC sucks!" rants from the x86 crowd.
Myself, I'll probably wait until PPC gets a bug as huge and embarassing as the Pentium float bug before going there.