Linked by Thom Holwerda on Tue 20th Nov 2007 16:46 UTC, submitted by Luis
AMD AnandTech takes a look at AMD's new Phenom. "If you were looking for a changing of the guard today it's just not going to happen. Phenom is, clock for clock, slower than Core 2 and the chips aren't yet yielding well enough to boost clock speeds above what Intel is capable of. While AMD just introduced its first 2.2GHz and 2.3GHz quad-core CPUs today, Intel previewed its first 3.2GHz quad-core chips. We were expecting Intel to retain the high end performance crown, but also expected AMD to chip away at the lower end of the quad-core market - today's launch confirms that Intel is still the king of the quad-core market."
Permalink for comment 285611
To read all comments associated with this story, please click here.
RE: low yields
by butters on Tue 20th Nov 2007 22:07 UTC in reply to "low yields"
butters
Member since:
2005-07-08

I look at the AMD/Intel landscape from a different perspective... Intel is the natural dominator in the x86 market. The only way that AMD can compete favorably is if Intel makes a huge strategic blunder. That happened with NetBurst, and it's unlikely to happen again anytime soon. Intel will always be 12-18 months ahead in process technology, they'll always have more fab capacity, and they'll always have more R&D resources to throw around.

AMD has consistently made wise design decisions to mitigate their inherent disadvantages. The Athlon K7/8 rested on two essential pillars: the on-die memory controller (IMC) and the serial, point-to-point system bus (HyperTransport). Intel didn't have them, and they still don't, because they don't need them. Because of their process lead, they can shove double the cache onto their chips, making them less vulnerable to bus performance.

This will change next fall with Nehalem, which will all of a sudden catch them up to AMD in terms of design sophistication. It will feature an IMC, a system bus called QuickPath (formerly CSI), and a unified L3 victim cache--just like Barcelona/Phenom. AMD needs to stay at least a year ahead of Intel on design because they're always a year behind on execution.

That's going to get a lot more difficult for AMD next year. I don't know how they're going to continue to remain competitive against Nehalem, especially after the tick-tock cadence takes this architecture down to 32nm in 2009. Intel has completely recovered from its NetBurst misstep, and they're catching up to AMD's design lead.

Then there's the battle between Fusion and Larrabee, where, IMHO, Intel has the better strategy. By treating graphics as a first-class citizen of the ISA, they're developing a true MPU with pipeline integration rather than just die/bus integration. Combined with HyperThreading (to which AMD has no answer), Intel could dispatch ALU and SP instructions in parallel through the same core.

AMD looks all kinds of screwed in their head-to-head competition with Intel. They desperately need to find a way to keep making x86 processors, but ones that don't compete directly with Intel's. Because what you see right now, with Barcelona up again Penryn, is the best it's going to be for AMD from here on out. Intel's going to pulverize them next year, and their new friends in Abu Dhabi are going to want them out of Intel's bread-and-butter markets.

Reply Parent Score: 5