Computer scientists from Frankfurt’s Goethe University and the Karlsruhe Institute of Technology (KIT) have set a new world record in energy-efficient data processing. The new system improved on the work of a Stanford University team by a factor of three to four. Ph.D. candidates Johannes Singler and Andreas Beckmann used Intel Atom processors for the data processing system rather than server processors that use more power. Their lower processing power compared to server systems was compensated by the usage of highly efficient algorithms. Moreover, the team used solid state disks, which are faster and use less power, rather than hard drives. The data processing system sorted data amounts of 10 GB, 100 GB and 1 TB, respectively, consisting of datasets with 100 bytes each. When sorting 1 TB of data, the system only spent 0.2 kWh. “In the long run, many small, power-efficient and cooperating systems are going to replace the so far used, heavy weighted ones,” says professor Peter Sanders from KIT, a supervisor of Singler and Beckmann.
View Full Article

Intel researchers have developed a prototype chip that operates in a low-power yet error-prone mode, but can detect and correct its errors. The researchers say this approach is 37 percent more power efficient compared with running in conventional mode and offers comparable performance. To compensate for the errors that occur while running at low voltage, Intel has developed a strategy known as resilient circuits. The prototype chip runs at low voltage, and when an error occurs, a calculation is done at high voltage to correct it. “When you have to correct an error, and reexecute a process more slowly, there is a tiny penalty,” says Intel’s Wen-Hann Wang. However, laboratory tests have shown the chip can either save 37 percent on power consumption, or operate 21 percent faster at a given power level. “They push it as close to the danger zone as they can, and things sometimes go bad, and they correct for it, which is very clever,” says Rice University professor Krishna Palem.

View Full Article