Performance Analysis of Signal Processing Algorithms Using Multi-Core DSP Platform

Signal processing applications in next generation radar and communication systems have enormous processing capacity requirements. Multi-core processor technologies have been widely used in order to meet these huge processing capacity needs. In this paper, edge detection, low pass filter, and fast Fourier transform algorithms has been developed on Texas Instruments TCI6630K2L system on chip device by efficiently utilizing the multi-core architecture. We present the performance results of these algorithms with respect to the number of DSP cores and the picture size. For all experiments, the image processing time decreases consistently when the number of DSP cores increases; however, the performance improvement rate varies according to the signal processing algorithm type and the picture size. Having the performance improvement rates independent of the picture size indicates that the processing power spent for data transmission between ARM and DSP is neglible compared to the processing power spent for signal processing algorithms.