Linus Torvalds (torvalds@linux-foundation.org) on 2/24/09 wrote:

>Heh. Not likely. I's just an admission that it sounds like
>Intel put more Itanium-specific resources on the dog than
>I thought they did. I still think it's a totally worthless
>architecture.
>

You would be surprised.
Itanium has 2 independent design teams working in paralel ( Fort Collins - former HP and Hudson - former DEC Alpha ).

As a matter of fact , x86 has also 2 design centers : Oregon and Haifa, Israel. ( 3 if you consider Austin , the low power Atom line )

Until now , Hudson's star has yet to shine. They developed QPI and the original Tukwilla , canceled in 2004. I'm sure that they are eager to show they are still in business, after all they've build a jewel for its time, the Alpha series.


>I'm just not seeing the market accepting Itanium. Quite the
>reverse. The non-HP players are dropping, and the x86
>competition is getting stronger.
>

Non HP players are NEC,Hitachi and Fujitsu. These behemoths have global reach and although they started slow, Itanium is becoming their de facto high end chip. Fujitsu still builds Sparc CPUs, given the diminishing returns this won't last forever.In fact, if they were smart they would acquire a QPI license and transition SPARC to IPF servers. ( like HP did for PA-RISC in order to ease transition and costs )


>So while Tukwila certainly improves things, the thing is,
>Core i7 improved a lot more. And it was already
>ahead.
>

True.