Joe, I think he means that CPU communication to memory and I/O devices goes over the same bus (FSB), which is true.
Other than snoop cycles, that's not true. I/O devices access memory directly. These accesses don't consume FSB bandwidth. As for the snoops, they might impact FSB performance, but Blackford should reduce those snoops with a snoop filter.
Meanwhile, if I wanted to twist facts Dan-style, I'd say that HyperTransport "sux" because some of its bandwidth is consumed by I/O. But HyperTransport has more than enough bandwidth to spare, so that point is moot.
Tenchu
Market Data 
Markets 
AI
