Logo image
在邏輯合成中之可重疊函數分解
Thesis

在邏輯合成中之可重疊函數分解

洪麗雯
Masters, National Tsing Hua University
1999

Abstract

重疊函數分解邏輯合成編碼分離性非分離性相容級 overlappingfunctional decompositionlogic synthesisencodingdisjunctivenondisjunctivecompatible class
Functional decomposition is an effective technique to restructure logic networks.Previous researches have studied how to select variables in bound set, how to encode compatible classes, etc.However, most of previous works consider non-overlapping structures..Circuit structures with overlapping variable are more flexible and may have lower area costs than the non-overlappingstructures in some cases.Therefore, in this paper, we will study nondisjunctive functional decompositions for Boolean functions.We first propose an algorithm to select overlapping variables from the bound set of a function.After the overlapping variable selection algorithm is applied, the compatible classes are paired to assign don't cares.The objectives of the don't care assignment heuristic are to reduce the compatible classes and to increase the decomposabilityof the image function that will be decomposed in the next level.Finally, we proposed an algorithm to improve the compatible class encoding algorithm proposed in HYDE.Then, we apply this nondisjunctive functional decomposition technique to look-up table (LUT) based FPGA synthesis.The experimental results show that our overlapping approach can lead to better mapping results on some particular cases of benchmarks.

Metrics

1 Record Views

Details

Logo image