LONDON – Processor intellectual property licensor ARM Holdings plc has renewed research cooperation with the University of Michigan into ultra-low energy and sustainable computing.
The five-year, $5 million extension of an existing research partnership will run until 2015 and cover technology for ultra-low energy computing and applications areas including energy-efficient cloud computing; wearable medical and lifestyle devices; energy-efficient trusted computing; and ubiquitous sensor networks.
Krisztian Flautner, vice president of R&D at ARM (Cambridge, England), is a former student of Trevor Mudge, Bredt Professor of Electrical Engineering and Computer Science at the University of Michigan, who leads the research. It was this link that helped set up the first five-year $5 million research partnership between ARM and the University of Michigan that is now coming to its conclusion.
Professor Mudge graduated from the University of Reading in England in 1969 before coming to the University of Illinois to collect to further degrees and embark on an academic career in the United States.
Some of the technologies that UofM researchers are currently exploring are: many-core computing; power-aware computer architecture; near- and sub-threshold computing; chip-stacking; non-volatile memory and timing speculation. The last of these can be used to increase energy efficiency and improve manufacturing yield.
The first collaboration produced a number of benefits, ARM said. One is an energy management system that enables mobile phones to optimize battery usage automatically. Overall the first five-year research partnership produced more than 40 patents and numerous research publications, ARM added.
If ARM executives are smart they will not make support conditional on using ARM platforms. After all they are trying to find out what extra needs to be done to make their processor architectures better.
And I am sure that there are plenty of x86-driven notebooks and servers in use in the relevant department of the University of Michigan.
But at the same time, where a host architecture is required and is orthogonal to what is under research, such as software agents roaming over wireless sensor mesh networks, why not use ARM. Perhaps Flautner will be arranging a drop-off of several hundred Cortex-M3 based microcontrollers.
This sounds like a great partnership in an area of increasing importance. The challenge I see is that sponsored research may trigger academic tunnel vision: it may be directed towards utilizing the sponsor's technology and not encourage promising work that happens to utilize competing technologies.