Applications of Performance Benchmarking to the Development of Signal Processing Systems Based on Personal Computer Technology
Robert Inkol, C.G. Wilson, Mathieu Eidus · 2006
Highly optimized libraries of vector and matrix math functions, such as the Intel integrated performance primitive (IPP) libraries, can be used to quickly and economically construct high performance signal processing systems built up around personal computer technology. This paper describes a portable benchmarking software suite that has been developed for benchmarking the performance obtainable with IPP implementations of common signal processing algorithms, including the fast Fourier transform (FFT) and finite impulse response (FIR) filter. The results presented provide useful insights into system design choices concerning algorithms, software architecture and processors