Optimization of synthesis process directed at FPGA circuits with the usage of non-disjoint decomposition

The paper presents the methods of searching for non-disjoint decomposition using BDD. Non-disjoint decomposition is a generalization of a classic decomposition model. Its main advantage is the reduction of necessary logic blocks needed to implement bound blocks. The essence of non-disjoint decomposition is based on searching for input variables that may fulfill the role of bound functions. The authors present the methods of searching for these variables in one or multi-root BDD diagrams. DekBDD enables to carry out non-disjoint decomposition on benchmark circuits. In the unit, entitled 'Experimental results', the number of necessary LUT blocks needed to implement multioutput function with the usage and without using non-disjoint decomposition, was determined for selected benchmarks. Moreover, the number of bound functions and the number of found 'switching' variables (non-disjoint decompositions) were given in the table. This attitude enables to determine the degree of optimization of a gained logic structure thanks to non-disjoint decomposition.