On the optimal code generation for signal flow graph computation
Bernhard Wess · 2002
An algorithm is proposed for translating signal flow graphs into optimized code for asymmetric target machines. It is based on the dynamic programming principle, and its complexity is linear with respect to time. The method permits the construction of very efficient compilers generating code for general-purpose digital signal processors. The efficiency of the algorithm is demonstrated by several results of a compiler translating digital linear time-invariant systems, specified by a signal flow graph, into code for the processor TMS32020.>