Replies to post #14403 on Intel Corp (INTC)

[chipguy]

wbmw, There are very few PC applications that can take advantage of a processor like Cell.

The statement was unspecific enough to mean "server market share". What would you think about that?

Well big blue is aggressively pushing foward with
overlapping development of POWER5+, POWER6,
and POWER7. I think that tells us more about what
IBM really thinks of Cell's chances in the server market
than a 1000 BS press releases.

BTW, I think the leap from video game console to Mac
is a lot more realistic than to pSeries. I haven't seen
any signs at all that Apple is interested in Cell, have
you?

[wbmw]

Keith, Re: The statement was unspecific enough to mean "server market share". What would you think about that?

Same deal.

Let's say you take the subset of server applications that favor highly threaded environments with an emphasis on floating point math. I guess you'd have the HPC market, right? Now if IBM really wants to push Cell into this market, what would it give them, relative to what it would take away from Blue/Gene and Power5?

IBM doesn't need a brand new HPC solution, that's why Cell makes no sense outside of game consoles, and even then, I'm skeptical.

This is slightly off topic, but I'm beginning to think that Microsoft's X-Box2 with tri-core PowerPC CPUs may have a better chance than Sony's Cell implementation, which will require an entirely new development paradigm and make it *very* difficult to get close to max theoretical performance. Meanwhile, the PPC ISA is fairly well used already, and optimizing for three cores ought to be easy. I think the first generation of XB2 games will look better than the first gen PS3 games, and perhaps the second and third gen, too. I bet the Sony tech demos will look cool (they always do), but making the best use of the hardware won't come easily (or inexpensively, though it's fair to say that Sony will be carrying most of the burden to develop the firmware/middleware.)