Simulation Tools For Fixed Point Dsp Algorithms And Architectures
K. B. Cullen, Guenole C. M. Silvestre, Neil Hurley · Zenodo (CERN European Organization for Nuclear Research) · 2007
This paper presents software tools that convert the C/Cµ floating point source code for a DSP algorithm into a fixedpoint simulation model that can be used to evaluate the numericalperformance of the algorithm on several different fixed pointplatforms including microprocessors, DSPs and FPGAs. The tools use a novel system for maintaining binary point informationso that the conversion from floating point to fixed point isautomated and the resulting fixed point algorithm achieves maximum possible precision. A configurable architecture is used during the simulation phase so that the algorithm can produce a bit-exact output for several different target devices.