Otellini: ARM servers 'ain't gonna work'
Intel boss says ARM chips still in short britches
Intel CEO Paul Otellini has a low opinion of ARM Holdings' efforts to crack the server market, an arena in which Chipzilla's processors are the dominant force.
"It ain't gonna work," Otellini told his audience at Intel's Investor Meeting 2011 in Santa Clara, California, on Tuesday.
When asked what ARM would have to do to become a player in the server space, Otellini was dismissive. "The short answer is the architecture has to grow up," he said.
"The expectations in servers are pretty significant: 64-bit, error correction, multiple parallelism, hyperthreading capabilities, highly parallel systems infrastructure," he said, rattling off a list of what he considered ARM deficiencies. "All that has to get built around a new architecture, and historically that's been very difficult."
ARM, though, seems ready for the challenge. During their own analysts' day, also on Tuesday, the company's VP of software alliances James McNiven said: "We think server ... is a good opportunity for ARM. We're looking to apply the lessons we've learned over several different ecosystems over many years to that ecosystem in servers," PCWorld reported.
From Otellini's point of view, the ARM question is déjà vu all over again. "This is so reminiscent of the RISC-CISC arguments of, gosh, twenty-five years ago," he said. "But then the argument was the other way around, it was that Intel can never possibly take on RISC architecture in the server space."
He then allowed himself a bit of gloating. "Well, we kind of proved that wrong," he said.
"And now they're using that argument against us, to say they're coming in. It ain't gonna work. There's a huge legacy here," he argued, using the same tack as his software boss Renée James did when detailing why she believed that ARM was not going to succeed in the PC market.
Defending his company's offerings against the new wave of ARM-based microservers coming from companies such as Calexa and ZT Systems, and being investigated by Dell, Otelli said: "The elements of computing that are predicated to this kind of environment would be well-served and better-served with Atom-based, many, many-core kinds of microservers."
Why? The "L-cord" yet again. "We can preserve the legacy of the software – no one has to rewrite it – and you still have the power-performance advantages of Intel."
Otellini asserts that one reason Intel will maintain its server advantage over ARM is experience. "Power-performance advantages are a function of the architecture – not just the instruction set, but the compute architecture that you embed over many, many years," he said, reminding his audience that "We've been doing server chips now since 1990."
Although ARM claimed at their analysts' day that they're "partnering and investing with thought leaders" to develop a software "ecosystem", Otellini sees Intel's own software offerings as a competitive advantage. "And the software environment, you know? Who does compilers for them? Who does the tuning? Who does the multithreading tools? Well, we've got many thousand people to do that. They have to find those people," he said.
Without citing specific example, Otellini said that potential microserver partners agree with him. "So as we talk to customers – and this question comes up at both the OEM and at the user level all the time – they're saying, 'We need more performance, and at certain parts of the market, your Atom-based servers are going to be just fine for us'," he claimed.
To our ears, the amount of time Otellini spent answering this one question – and, for that matter, the amount of energy that many Investor Meeting 2011 presenters spent dissing ARM in mobile devices, PCs, and servers alike – makes us wonder if Otellini and his compatriots aren't a wee bit uncomfortable with the successes and ambitions of Cambridge, UK's pride, ARM Holdings PLC. ®
Sponsored: Beyond the Data Frontier