Expect HP's BCS to crater in 1Q
Tukwila was supposed to be out in 4Q 2008 and wont make it into systems before July 2010.
Even then the only compelling value is twice the cores per floor tile. The performance per core is the same and its a complete fork lift replacement. I have no idea who would want this "upgrade" when software is priced per core not socket.
Last weeks memory price reduction by IBM will also help kill/stall any Itanium sales in 1Q as HP tries to figure out how to compete with their partitioning technology and "not for production" integrity virtual machine technology.