Logic Synthesizer by the Transduction Method:Design of Logic Networks with Negative Gates by the Transduction Method.

Design of Logic Networks with Negative Gates by the Transduction Method

The transduction method has been described for the logic networks of NOR gates for the sake of simplicity, but it can be applied to logic networks of other types of gates, such as MOS logic gates and a mixture of AND gates and OR gates, tailoring its basic concepts (i.e., permissible functions and transformation). In this sense, it is important to understand what features different types of logic gates and consequently corresponding transistor circuits have in terms of logic operations and network trans- formations. In order to test the feasibility of design of logic networks with negative gates (MOS logic gates are negative gates) by the transduction method, a few synthesizers, called SYLON (an acronym for SYnthesis of LOgic Networks), were developed by modifying the transduction method [1,2,7,8,11,12].

Some SYLON logic synthesizers consist of a mixture of technology-dependent optimization and technology-independent optimization. Here, let us outline SYLON-REDUCE [7], a logic synthesizer which is of totally technology-dependent optimization and is more algorithmic, wherein a logic network is processed in its target technology throughout the execution of REDUCE. REDUCE reduces an initial network, using permissible functions, where in order to make each logic gate easily realizable as a MOS logic circuit, each logic gate throughout the execution of REDUCE is a negative gate that satisfies prespecified constraints on the maximum numbers of MOSFETs connected in series in each path and the maximum number of parallel paths. The reduction is done by repeatedly resynthesizing each negative gate. In other words, the outputs of some candidate gates or network inputs are connected to a gate under resynthesis and the connection configuration inside the gate is restructured, reducing the complexity of the gate and disconnecting unnecessary candidate gates or network inputs. The resynthesized cell is adopted if it has no more MOSFETs than the old gate and does not violate the constraints on the complexity (i.e., the specified maximum number of MOSFETs connected in series or the specified maximum number of parallel paths) otherwise, it is discarded, restoring the old gate. This resynthesis of each gate is repeated until no improvement can be done. Thus, the network transformation is done in a more subtle manner than the original transduction method. The result is a network where each gate still satisfies the same constraints on the complexity and contains no more MOSFETs than the corresponding gate in the original network and the connection configuration of the network may be changed.