A DataFlow course: Dataflow supercomputing
Veljko M. Milutinović · 2017
Summary form only given. The complete presentation was not made available for publication as part of the conference proceedings. This item describes a course that presents the DataFlow SuperComputing paradigm, defines its advantages and sheds light on the related programming model with hands-on coding experience. DataFlow computers, compared to ControlFlow computers, offer speedups of 20 to 200 (even 2000 for some applications), and power and size reductions of up to 1/20. However, the programming paradigm is different, and has to be mastered. The course explains the paradigm of programming in space, using Maxeler (a provider of multiscale dataflow computing systems) as an example, and gives an overview of ongoing research in the field. Examples include DataEngineering, DataMining, FinancialAnalytics, ImageProcessing, etc. The course also covers advanced DataFlow issues like: Compilation, OS, and methods for speed-up maximization, using tools like WebIDE and MaxIDE. DataFlow potentials are discussed through the notions introduced by four Nobel Laureates: Richard Feynman, Ilya Prigogine, Daniel Kahneman, and Andre Geim.