Software/Hardware Solutions for Information Processing in All Programmable Systems-on-Chip

Abstract

Algorithms in computer engineering and applied informatics often require extraction of data with some desired properties from large sets. Such tasks appear within different clustering algorithms in the scope of data mining, in classification of objects in accordance with given criteria, in knowledge acquisition obtained from controlled environments, in statistical analysis and in other areas. Many of these tasks involve widely used data processing techniques that are sort and search and for numerous practical applications, especially in real-time and high-performance systems, speeding-up is important. The paper suggests effective portable solutions that enable fast parallel information processing to be implemented in recently appeared all-programmable systems-on-chip that combine multi-core computations with programmable logic interacting through multiple high-performance interfaces. Acceleration is achieved with parallel networks for data sort created in programmable logic and enabling software running in multi-core processing units to be speeded-up which is demonstrated in numerous practical examples fully implemented and tested in commercial microchips.