Intel R&D on slow boat to China

Building chips in China is one thing. Moving Intel's critical research and development to China isn't in the cards just yet.Photos: Intel's China plant
Written by Tom Krazit, Contributor
BEIJING--Intel may be plunging ahead with manufacturing plans in China, but its research-and-development arm isn't moving at quite the same pace.

Engineers at labs in Beijing and Shanghai are working on important research into areas such as parallel programming tools and terascale computing. But no processors are in development here as of yet, and it may be some time before that happens, said Justin Rattner, who, as Intel's chief technology officer, leads its labs.

Intel is in China this week for its semiannual Intel Developer Forum. The company usually holds these conferences in San Francisco, but it moved the operation overseas this year, in part because of cutbacks to the IDF budget and in part to highlight its growing operations in China, including the recent announcement that it plans to build chips in Dalian later this decade.

Intel's China plant

In addition to chip-packaging plants in Shanghai and Chengdu, Intel currently operates research labs in Beijing and Shanghai. Much of the work there is focused on software development, which might at first glance seem a bit out of sorts for the world's largest chipmaker.

But Intel actually has loads of software developers in China working on things like a replacement for the BIOS (basic input/output system) technology that boots an operating system and compiler technology that helps programmers get ready for processors with multiple cores, Rattner said in an interview before the conference.

Some of the work surrounding Intel's so-called terascale research--most recently showcased through its 80-core chip prototype--is also being done at the company's Beijing labs, Rattner said. For example, Chinese researchers are working on applications that can harness the power of 80 separate processing cores--something today's applications and programming techniques can't handle.

Striving for parity in education
Even though the company plans to build chips in China, consumers should not expect Intel's next processor design to emerge from the country, Rattner said.

Justin Rattner Justin Rattner

"The fab announcement in Dalian is certainly an indication that we're willing to do more in China, but we're trying to pace ourselves," he said.

Part of the problem is that the Chinese educational system isn't up to par with the American university system, Rattner said. The Ph.D. students whom Intel has hired in China don't have the same level of expertise that its United States-trained engineers have.

Several engineers have gone to the United States to learn, then have returned to China to work in Intel's labs, Rattner said, but until Chinese computer science programs are as good as those in the States, Intel's research expansion into China will probably be gradual, he said.

This is something Intel is working on, said Tan Wee Theng, president of Intel China. The company is spending a lot of time and money working with the local university education system on science and technology education.

When Intel announced plans to set up a plant in Dalian, it also set up a program with the local university to focus on electrical engineers and computer science, he said.

Intel has been more aggressive with research and development in India. Much of the design work for its 80-core processor prototype was done in Bangalore, and Intel India actually came up with a processor design, called Whitefield that, while it never saw the light of day, was based on design principles that Intel plans to incorporate into its Nehalem generation of processors next year.

And the role of Intel's Israeli chip design operation in turning around the company's fortunes this decade cannot be overstated.

In the earlier part of this decade, the company's Oregon research labs had led Intel down a development path that focused on fast single-core processors. But overheating concerns forced Intel to seek an alternative path to increasing performance, and that solution came from Intel's Israeli labs in the form of the Banias design--known to the world as the Pentium M.

Still, Rattner has been impressed with the work coming out of Intel's China operation. More and more work could be headed that way, but the above concerns--plus the export controls that prevent cutting-edge chipmaking technology from being taken into China--will keep the most critical projects out of the country for some time to come.

Editorial standards