Accelerating numerical applications using ESL methodologies

This thesis is about exploring the suitability of ESL methodology for numerical applications. Numerical computing is the foundation of digital signal processing and many science computing. These applications all have high demand for processing speed and precision. On the EDA side, ESL increasingly draws designer's attention in recent years. Therefore, addressing the demand of numerical application with ESL methodology becomes the motivation of this thesis. Three cases presented in this thesis are TLM, FFT and Smith–Waterman. In the first case, the thesis details the development of programmable and hard-wired TLM computing solutions from methodology to post synthesis result. The comparisons of these two solutions are also presented. Motivated by the outcomes of this comparison, it was determined to investigate the suitability of ESL further. Subsequently, other two cases (FFT and Smith–Waterman) are presented.