SAN FRANCISCO--Intel is making an aggressive shift toward mobile computing and designing its next-generation core microprocessors with mobility in mind, according to David (Dadi) Perlmutter, executive vice president and general manager of the firm’s architecture group.
“It’s about being able to deliver great capabilities in whatever form factor,” Perlmutter told EE Times in an interview after his keynote at the Intel Developer Forum (IDF) this week, where much emphasis was placed on increasing performance, decreasing power envelopes and pushing new user experiences like gesture and voice recognition.
[ARM TechCon 2012, the largest ARM design ecosystem under one roof, is Oct. 30 - Nov. 1 in Santa Clara. Click here to learn more]
Last year, Intel re-organized its design teams, bringing together the R&D groups for both high performance (Core) and low-power (Atom) products. Perlmutter said Haswell—the company's next generation microarchitecture due to come to market in 2013—was probably the first to really take advantage of the skill sets of this new unified team.
“We’ve brought together the experts from both worlds, from high performance computing to extremely low power, low cost, high integration, and having these people sitting around the same table helps a lot,” he said.
The results Intel is already citing for Haswell, including a 20x reduction in idle power consumption over SandyBridge, certainly indicate that the merged R&D teams are working well together. But with Haswell's thermal design power (TDP) making so much progress over previous generations, it looks to some like the lines between Atom and Core may be blurring somewhat.
With Core coming down in power and Atom ramping up in terms of performance, does Perlmutter think Atom might someday disappear from Intel’s line-up? The answer, for now, is no, though Perlmutter admitted he would be loath to predict anything further than 10 years out.
“In order to get real optimization you have to have multiple power and performance points,” said Perlmutter. He added that keeping products at extremely low cost and running at close to zero power will always be trade-offs and limitations to how high performance can be increased.
But that's not to say the gap isn't closing somewhat between Atom and Core. “There’s going to be potentially some overlap, which is always good in the product plane,” he said.
Most important to Intel, said Perlmutter, is ensuring a consistent experience, no matter what levels of power and performance various Intel-powered devices offer.
“Different users will want different things [from their devices]. The overlap is helping us to keep the experience consistent and have multiple products that will run things the way consumers really want,” he said.
Check out the video of EE Times' interview with David Perlmutter below.
In other words Mr. Perlmutter is saying that by having multiple overlapping processors in the power plane, that Intel sales can lure price takers to next higher margin product's tied in sales bundle when prior margin leader's price dips below Intel marginal cost of producing the run down volume.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.