Intel Corp (INTC)

[SemiconEng]

Smooth, I'm really afraid that the whole push to dual-core will fall flat on its face, as once again the critics bash it for not making a difference on "current" desktop applications. I just don't see what "killer app" is demanding dual-core, at least in the consumer market where such Centrino-like platform strategies are targeted. It's a different story when it comes to servers, of course, and I suspect Otellini will try and craft a new kind of platform strategy targeted toward business and corporations. But as for the consumer market, I don't see anyone really screaming for dual-core at the moment.
Tenchu


WAG: I think that those who are "claiming" that Smithfield won't be a good performer are wrong. In any case, I think that intel should focus on promoting the power reduction of dual core. What reduction you ask? Well, although I don't have "hands on" acess to the data numbers, I've recently seen reports that indicate there is an unusual effect of linking 2 cores together. As I understand it, due to transistor physics, when dual cores are linked, a small reduction in clock speed, results in a huge reduction in power, as well as an almost doubling in performance.

I should have asked the question, if they meant 1.8x performance on "optimized" applications, but it didn't occur to me until later. I would guess that "optimized applications" was the case, but what if they meant "all applications"? In any case, the point is, it appears that 2 cores linked, with a slight decrease in clock speed, results in significant power reduction, while still increasing performance, on at least some applications. Perhaps that's why, as The Register reported, that Smithfield appears to be 2 cores, while being introduced at lower clock speeds, while at the same time, intel is claiming significant increased performance, "in the same thermal budget"..... And if that's true, It wouldn't even be too far fetched to decrease the clockspeed a little further, for Dual Core Mobile.


http://www.theregister.co.uk/2004/12/14/intel_dual-core/
What all that will do for performance remains to be seen. Smith claimed by 2008, multi-core chips will offer ten times the performance of an "initial Pentium 4" processor, compared to the threefold gain today's HyperThreading chips provide. Next year, dual-core will see that improvement increase beyond 3x, but from we'll be well into 2006 and the 65nm node before multi-core gives a significant performance boost. Only then will dual-core offer a greater gain than HT alone will do by 2008.

Smith would only say that Smithfield contains "two execution cores", not whether they're on the same die or not. Separately, he admitted Smithfield's clock speed range would be lower than the top end of today's P4 CPUs, to ensure the new chips "operate in the same thermal budget". Intel's roadmaps put those speeds at 2.8-3.2GHz, well below today's 3.8GHz P4 570.