Stanford to Announce Parallel Computing LabCategory: Software
Posted: May 1, 2008 09:23AM
Stanford University is partnering with some of the world's largest hardware and graphics processor manufacturers to research and develop new methods for parallel processing. Stanford is poised to announce the Pervasive Parallelism Lab with support from major companies such as Intel, AMD, NVIDIA, Hewlett Packard, Sun Microsystems and IBM. With multi-core processors now the norm and chips with even more processor cores on the horizon, software that makes use of all the extra processing power has not kept pace. The latest processors, both CPUs and GPUs, have largely abandoned the megahertz race and have instead turned to adding additional processor cores to improve performance. All of the companies supporting the lab have a major stake in improving the ease with which programmers can develop software to take advantage of the extra cores. Parallel programming is more difficult than the serial programming approach most programs use today. Stanford intends to address this issue by developing new programming languages designed for parallel computing as well as creating hardware for multi-core processors. Stanford has already down significant work with the Brook programming language designed for parallel computing and has been able to apply this work to their Folding@Home project in the form of a GPU software client. In addition Stanford has also developed SMP Folding@Home clients for both Linux and Windows to make use of multiple processor machines.