Wasting energy no longer an option in race for speed: To add more support to Green500 founder Wu Feng's recent proposition that power has become as important as performance and speed in supercomputing design, even the mainstream media has picked up on the power-conscious supercomputer offerings from IBM. "As applications for supercomputers increase, so does the attention on this aspect [energy efficiency] of their use," said Dave Turek, VP IBM Deep Computing. According to Turek, IBM has driven advances in energy efficiency that include the Cool Blue technology portfolio and Project Big Green solutions (that are designed to reduce data center costs and energy use at the same time).
A faster passthrough
A high-throughput method that can use Cell/B.E., Elastic Cloud, and multi-GPU systems: This Slashdot mention notes that MIT graduate student Nicolas Pinto and researchers David Cox (Harvard) and James DiCarlo (MIT) assembled a 16-GPU system of NVIDIA 8x9800gx2s GPUs to help with real-time, human-level artificial vision systems development. They say the high-throughput method they're using can be used with other technologies such as the Cell Broadband Engine processor or Amazon's Elastic Cloud Computing services.