The pedagogical value and importance of applicable computational intensive scientific kernels in parallel computing: a case study
Eduardo Colmenares, Heng Yuan Wu, Amy Knowles · Journal of computing sciences in colleges · 2017
Case studies constitute a fundamental cornerstone in the formation, reinforcement and consolidation of knowledge. A good case study not only should approach a realistic, and applicable problem, but it should also include multidisciplinary concepts that will be beneficial to the student not only during their academic years but also during their career. This paper presents and highlights key elements associated with a widely studied, used, and important scientific kernel, the Fast Fourier Transform (FFT). The focus is a complex single precision radix-2 FFT version of the algorithm, which forces the student to recall, or reinforce important concepts such as summations, signal symmetry, complex numbers, Euler's identity, radians, and signal decomposition among others. The problem under study is approached from three different perspectives, two of which correspond to a regular sequential solutions (not optimal, and optimal), and a parallel implementation on a Graphical Processing Unit (GPU).