Development and tuning framework of master/worker applications

• Main Authors: Anna Morajko, Eduardo César, Paola Caymes Scutari, José Gabriel Mesa, Genaro Costa, Tomás Margalef, Joan Sorribes, Emilio Luque Fadón
• Format: Article
• Language: English
• Published: Postgraduate Office, School of Computer Science, Universidad Nacional de La Plata 2005-10-01
• Series: Journal of Computer Science and Technology
• Subjects: dynamic tuning; performance analysis; performance model
• Online Access: https://journal.info.unlp.edu.ar/JCST/article/view/859

Parallel/distributed programming is a complex task that requires a high degree of expertise to fulfill the expectations of high performance computation. The Master/Worker paradigm is one of the most commonly used because it is easy to understand and there is a wide range of applications that match this paradigm. However, there are certain features, such as data distribution and the number of workers that must be tuned properly to obtain adequate performance. In most cases such features cannot be tuned statically since they depend on the particular conditions of each execution. In this paper, we show a dynamic tuning environment that is based on a theoretical model of Master/Worker behavior and allows for the adaptation of such applications to the dynamic conditions of execution. The environment includes a pattern based application development framework that allows the user to concentrate on the design phase and makes it easier to overcome performance bottlenecks.