Communications of the ACM
Berkeley Discusses Progress in Parallel Programming
University of California, Berkeley researchers recently discussed their progress in finding new parallel programming models for multicore processor architectures. Parallel Computing Lab professor Kurt Keutzer described seven applications researchers have produced with the new Pallas parallel methodology, and noted that work has started on a more general parallelism framework called Copperhead that could be used by a broader spectrum of programmers.
The Pallas approach involves graduate students implementing complex algorithms from domain experts. They start by producing software architectures specific to the algorithms that maximize their employment of computational and structural patterns, then map those architectures on to parallel processors.
The Copperhead framework being co-developed with Nvidia concentrates on producing rapid executable code for data-parallel applications. It will interoperate with both Nvidia's Cuda and OpenCL environments.
From EE Times
View Full Article
Abstracts Copyright © 2010 Information Inc., Bethesda, Maryland, USA 
No entries found