Chipguy, I don't remember seeing as big of a jump from a Xeon DP to a Xeon MP as I see in the jump from Nehalem to Beckton.
That is because all previous Xeon MPs had the same
number of cores as the Xeon DP/desktop MPUs of its
generation and no more (and sometimes less) system
bandwidth
From what I see, there really isn't much excitement around Itanium anymore. Otherwise, I would see a renewed push to beat IBM POWER.
One of the managers in the Alpha design team once said
that the MPU game is like playing Russian roulette and
then waiting five years after pulling the trigger to see if
you killed yourself. The concept that illustrates is you fix
a multi year plan to design a new MPU and you execute to
it. You are on rails and have limited ability to make design
modifications to respond to changes in the market place
and actions by competitors. If four years in you discover
you are bringing a knife to market and your competitor is
about to release a handgun there isn't anything at all you
can do about it. If you cancel what you are doing then
you are resetting the clock to zero. If the product you are
about to release will be inadequate you certainly don't
want to cancel it and rely on your *previous* generation
product for another five years!
The plan to beat Power was laid in mid 2007 when Intel
and HP execs signed off on the requirements document for
Poulson and announced that IPF would skip 45 nm and go
straight from 65 nm to 32 nm. It will be a huge advance
over Tukwila (2 full process generations shrink, 3x more
cores, and a brand new microarchitecture replacing a 10
year old one). IBM will respond with a 32 nm follow on to
Power7. I guess we will just have to wait to see how the
competitors stack up. I am hoping Intel will present a few
Poulson papers at ISSCC 2011.
Market Data 
Markets 
AI
