Related News

Computing Performance: Game Over or Next Level ?

Print
Last Updated on Monday, 18 March 2013

In the article "Computing Performance: Game Over or Next Level ?", Samuel Fuller, from Analog Devices, and Lynette Millett, from the National Research Council of the National Academy of Sciences, just overview the current gap between hardware technologies, going towards more and more parallelism degree in nearly all the areas of computing and the current practice of programming and writing software, still heavily influenced by the "sequential programming era". Among the research recommendations and challenges mentioned in the article, one can find the need for research on: novel programming models that "will enable efficient use of parallel systems by typical programmers as well as by experts"; system software for highly parallel systems, requiring a "rethinking of the canonical computing stack - applications, programming language, compiler, runtime, virtual machine, operating system, hypervisor and architecture"; architectures driven by application needs; power efficiency; and also general recommendations for reviewing the way programming is taught from the very beginning, with a final remark on the need for tools and mechanisms to "bring sequential programs into the parallel world", in order to leverage the current substantial investments in IT.

These arguments highlight once again the importance of the research themes investigated in the S(o)OS project. Read more in the original article.