I think you miss the point; Intel developed the Yonah chip by cobbling together two Pentium M cores to get bragging rights. Therefore, it was, like the quad, 70 percent faster than before at twice the power consumption. But then, Intel combined the two cores on the same substrate at a smaller process size in the Core 2 Duo chips at 25 percent faster than Yonah and at a 40 percent reduction in power. Hence, the Core 2 Duo chips were over twice as fast at the same power consumption of the Pentium M's.
I expect that the Quad core chips will follow the same pattern. The real gains in performance and power consumption is in the version after bragging rights have been taken. That is when the price falls too. So, just wait for it. It won't be long-- mid 2007.
"Your comment contains too many links and will not be added"
Oh, c'mon!
"Intel developed the Yonah chip by cobbling together two Pentium M cores to get bragging rights."
The Pentium-M Dothan was a single-core 32-bit processor, fabricated on a 90nm process. Core Duo aka Yonah is a dual-core 32-bit processor ("true" dual-core: 2 cores on a single die), manufactured on Intel's 65nm process tech.
"But then, Intel combined the two cores on the same substrate at a smaller process size in the Core 2 Duo chips"
"Please wait at least 30 seconds between posts (another 13 second(s))"
Oh, c'mon!
Core 2 Duo is also a "true" dual-core processor, but 64-bit, still produced with the same 65nm process. Laptop parts, known as Merom, are labelled T7xx and are currently running on a 667MHz front-side bus. The E6xx series are desktop parts, known as Conroe.
Same for the current Xeon, aka Woodcrest (51xx series).
The Core 2 Quad Kentsfield, which combines two Conroe chips, and Clovertown which combines two Woodcrest chips, won't be "true" quad-core processors, two dies will be grouped on a multi-chip module.
Bloomsfield, manufactured with the future 45nm process, will be a "true" quad-core processor.
"the two chip, quad-core processor feels a bit disingenuous, particularly when Intel starts crowing about being more power efficient"
Intel must serve different markets, Kentsfield is the successor of the Core 2 Extreme, targeting gamers and PC enthusiasts, and this particular market is all about bragging rights. Kentsfield won't be elegant, it won't be affordable, it won't be power efficient, but when Otellini says “You guys are misreading the market if you think people care what’s in the package,” I think he's right on the money. Pun 100% intended.
If you want power efficiency you will choose another chip or you will wait a little bit to get a low-power quad-core. On the long run power efficiency is still the main goal.
Hey Phil, this is the wrong keynote. You're buddy buddy with Intel, but go back to Apple right now! :-D
Carl:
There may be some marketing hype here, for sure. But by releasing the kludged quad core chip they create a platform that allows application developers to start writing multi-threaded code for consumer applications. Some apps are optimized for dual core now, but few if any are optimized for four. Developers won't pull the trigger until the hardware is there. By the time Quad cores (or more) are in demand because games and apps are pervasive, the kludgy quad will be replaced by the real deal.
The real question is whether multi-core applications outside of gaming and enterprise exist. I can tell you that you won't find out until the hardware is in the hands of some creative people.
Actually, I believe the quad-core concept is fine. Back in the 1980s, I led a team that wrote OSes and codes for 128 and 256-processor machines that scaled linearly with processors available. There's no lack of parallel processing opportunities, just a lack of good parallel programmers.
Heck, you can buy a PowerMac today with a quad Xeon -- two dual cores. And if Intel were clocking the faux-quad at the same frequency as the two dual-cores that power the PowerMac, they'd consume the same power. There's no free lunch just because you put two chips in the same package.
Yes, I get that Intel is trying to get the platform and socket going with their faux-quad in November. But let's not claim that it's some huge leap forward. Only when they actually ditch the need to drive the huge capacitive external wires will we see the real benefits and lower power consumption of a quad-core architecture -- and both Intel's and AMD's true quads will arrive within weeks of each other. It's a cute marketing ploy to prove that Intel is on the way back -- but it isn't really an honest quad.
Thanks for taking the time to comment!
Carl
Commenting by HaloScan
