Green Super computing Comes of Age

Energy-efficient (green) supercomputing has traditionally been viewed as passé, even to the point of public ridicule. But today, it’s finally coming into vogue. This article describes the authors’ view of this evolution. In 2002 the Japanese Earth Simulator supercomputer shattered US domination of high-end computing (HEC) with a 400 percent increase in computational speed, as measured by the TOP500 List’s Linpack benchmark. Soon thereafter, the Council on Competitiveness, sponsored by the US National Science Foundation and the Department of Energy, conducted a study that found 97 percent of businesses could not compete (or exist) without HEC . These businesses viewed HEC as essential to economic competitiveness: “to out-compete is to out-compute.” Although HEC might elicit an image of a gargantuan supercomputer that occupies a huge amount of space and consumes an outrageous amount of power to solve large-scale scientific and engineering problems, it is clearly important to the business and enterprise community. But rather than referring to an HEC resource as being a supercomputer, the business and enterprise communities refer to HEC in general as large-scale data center computing; in this article, we refer to supercomputing as HEC, large-scale data center computing, or both. Until recently, such systems enjoyed a “free ride” on institutional infrastructures. However, with the annual costs of both power consumption and cooling projected to exceed annual server spending in data centers in 2007, institutions with data center supercomputers have become particularly sensitive to energy efficiency—so-called “green” issues. We explore those issues in this article and describe how the supercomputing industry has evolved from viewing power and cooling as a secondary concern to a primary design constraint. The Problem with Supercomputing Although supercomputers provide an unparalleled level of computational horsepower for solving challenging problems across a wide spectrum of fields—from scientific inquiry, engineering design, and financial analysis to national defense and disaster prediction—such horsepower usually comes at the expense of enormous power consumption, not only to run the supercomputer but also to cool it. This, in turn, results in extremely large electricity bills and reduced system reliability. Accordingly, the HEC research community started exploring green supercomputing as a way to achieve autonomic energy and power savings with little to no impact on performance. However, the notion of green supercomputing is still viewed as an oxymoron: a supercomputer summons up images of speed, a Formula One race car of computing, whereas green or energy-efficiency computing evo

Free download research paper