PICSEL (28064)

INVENTION: Computer scaling techniques attempt to optimize a tradeoff between computer performance (speed) and power consumption. User studies have shown that such optimizations greatly improve if a user’s personal tastes are considered during usage. In response to this observation, the investigators have invented PICSEL (Perception-Informed CPU performance Scaling to Extend battery Life).

PICSEL is a dynamic voltage and frequency scaling (DVFS) technique that tailors computer speed and power consumption to a user’s personal taste during usage. It estimates user-perceived performance by measuring variations in the rate of change of a computer’s video output, and it optimizes performance and power consumption accordingly. To the best of our knowledge, no previous DVFS techniques consider user-perceived performance. User study results are presented below.

For more details on PICSEL, see A. Mallik, J. Cosgrove, R. Dick, G. Memik, P. Dinda, PICSEL: Measuring User-Perceived Performance to Control Dynamic Frequency Scaling, Proceedings of the 13th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS 2008), March, 2008. This paper can be viewed as a PDF

Applications: Power Management for Embedded Processors and for Computers in general.

Results: Extensive user studies show that system-level power consumption of our target laptop can be reduced by an average of 7.1% for a conservative approach (cPICSEL) and 12.1% for the aggressive version (aPICSEL) compared to the Windows XP DVFS scheme. Furthermore, CPU temperatures can be markedly decreased through the use of our techniques. User studies also reveal that the difference in overall user satisfaction between the more aggressive version of PICSEL and Windows DVFS is statistically insignificant, whereas the conservative version of PICSEL improves the users’ overall satisfaction when compared to Windows DVFS.

Northwestern seeks a licensing partner to commercialize this invention

Inventor(s): J. Cosgrove, R. Dick, P. Dinda, A. Mallik, G. Memik

Type of Offer: Licensing



Next Patent »
« More Computer Science Patents

Share on      


CrowdSell Your Patent