BEAVERTON, Ore. Intel Labs' People and Practices Research group (PaPR) revealed its vision of a more human-centered technology future at a recent open house here. The lab showed such developments as the turbo boost feature in Intel's forthcoming Core i7 processor, algorithms for "pocket" supercomputers and concept designs for Atom-based smartphones.
PaPR applies the principles of sociology and forward-looking anthropology to predict what humans will require from tomorrow's architectures. "We are trying to understand what people actually do in order to better design technology," said Ken Anderson, an anthropologist at PaPR. "All our studies focus on how people live their lives, in order to better plan for [Intel's] architectural future."
For example, a turbo boost feature announced at the recent Intel Developer Forum will let the forthcoming Core i7 processors temporarily overclock one core to achieve bursts of speed during times of peak usage. "People are starting up tasks and switching to doing something else all the time, making it rather important to think about the need for an extra boost at certain moments of time," Anderson said. Turbo boost will be part of the next-generation 32-nanometer processors that Intel expects to deliver early next year.
Richard Beckwith, a research psychologist at PaPR, demonstrated pattern recognition algorithms that he claims could be scaled up to run on future pocket supercomputers. Today's educational pattern recognition programs for children, Beckwith said, will be scaled to adult size by the time the pocket supercomputer becomes a reality circa 2018.
"We have been looking at education," said Beckwith, with an eye toward devices "that will actually watch a child and be able to follow what they are doing, then give them hints and feedback to allow them to work through exercises at their own pace."
The group is wrapping up work now on a pattern recognition algorithm that will run on Intel's ClassMate PC, a low cost laptop-like computer for educational use whose internal camera has been modified so that it focuses on the user as tasks are performed. Teachers can configure the software to supervise students as they perform pattern-matching problems such as sorting coins. The system would provide verbal cues and encouragement to guide the students' progress. It is slated for testing in classrooms early next year.