AMD's Barcelona core had a native quad-core architecture and an integrated memory controller too. AMD describes Nehalem as "the sincerest form of flattery".
Intel’s new Xeon W5580
is getting a phenomenal amount of praise at the moment, but Intel’s chief CPU rival, AMD, claims that it’s not even remotely scared of the chip’s Nehalem EP architecture. To demonstrate the point, AMD’s senior vice president and chief marketing officer, Nigel Dessau, wrote a mock interview with himself on his blog
yesterday, in which he discussed the reasons why he thinks AMD has nothing to worry about.
The interview kicks off with the following introduction:
Interviewer: Wow, you people must be really scared.
Interviewer: Nope? Ok, how about really, really scared?
Me: Nope. Nope. No, really we’re not.
Justifying the company’s lack of concern over Nehalem, Dessau points out that “Over the last ten or so years, we have traded performance leadership with Intel something like six times. Remember 2006? We heard some of these same statements from Intel then, about an eighty percent performance advantage and never losing another benchmark to AMD again with Woodcrest. That gap closed quickly.”
Dessau also points out that the best-selling chips in the server market aren’t the fastest and most expensive ones, saying that “more than ninety percent of what we sell is not our fastest part. The market for the fastest part is always small and in this economy it’s likely even smaller.”
Dessau admits that Intel’s new architecture is faster than what AMD currently has to offer, and even refers to it as a “nice job”
, but he also says that Intel’s “overhaul”
of its server architecture is unlikely to be popular with businesses looking to save money and avoid disruption.
As well as this, Dessau dismisses performance benchmarks as an important factor for businesses making a buying decision. When his mock interviewer asks if the benchmarks of Nehalem count for anything, he replies: “Only if you run your business on benchmarks.”
Dessau also notes the current lack of a Nehalem-based four-processor Xeon platform at the moment.
Meanwhile, AMD’s director of global platform and product communications, John Taylor, also blogged
his opinion about Nehalem. Taylor criticised the Nehalem EP’s use of DDR3 memory, which he says “draws more power and comes at a price premium.”
While the use of DDR3 memory is good news for PC enthusiasts looking to build a performance PC that can also be used for gaming, Taylor points out that it’s not a good technology for servers that usually have large amounts of memory.
AMD’s bloggers also took the opportunity to point out that there’s nothing essentially new about Nehalem’s core design. It’s a native quad-core chip with an integrated memory controller, which AMD has had since it introduced the Barcelona core in 2007. Dessau says that it’s an architecture that “some might call a copy, at least as far as the architecture is concerned,”
and also refers to it as the “Opti-clone”
. The title of the blog post: “The Sincerest Form of Flattery”
also refers to Intel’s use of ideas that AMD originally pioneered.
Is AMD right to dismiss Intel’s Nehalem EP architecture because of its cost, and do most businesses really care more about value than performance? Let us know your thoughts in the forums