EA-based Resynthesis: An Efficient Tool for Optimization of Digital Circuits

EA-based Resynthesis: An Efficient Tool for Optimization of Digital Circuits

Článek WoS

Since the early nineties the lack of scalability of fitness evaluation has been the mainbottleneck preventing the adoption of evolutionary algorithms for logic circuits synthesis.Recently, various formal approaches such as SAT and BDD solvers have beenintroduced to this field to overcome this issue. This made it possible to optimisecomplex circuits consisting of hundreds of inputs and thousands of gates. Unfortunately,we are facing another problem-scalability of representation. The efficiencyof the evolutionary optimization applied at the global level deteriorates withthe increasing complexity. To overcome this issue, we propose to apply the conceptof local resynthesis in this work. Local resynthesis is an iterative process basedon the extraction of smaller sub-circuits from a complex circuit that are optimizedlocally and implanted back to the original circuit. When applied appropriately, thisapproach can mitigate the problem of scalability of representation. Two complementaryapproaches to the extraction of the sub-circuits are presented and evaluated inthis work. The evaluation is done on a set of highly optimized complex benchmarkproblems representing various real-world controllers, logic and arithmetic circuits.The experimental results show that the evolutionary resynthesis provides betterresults compared to globally operating evolutionary optimization. In more than 85%cases, a substantially higher number of redundant gates was removed while keepingthe computational effort at the same level. A huge improvement was achievedespecially for the arithmetic circuits. On average, the proposed method was able toremove 25.1% more gates.

Since the early nineties the lack of scalability of fitness evaluation has been the mainbottleneck preventing the adoption of evolutionary algorithms for logic circuits synthesis.Recently, various formal approaches such as SAT and BDD solvers have beenintroduced to this field to overcome this issue. This made it possible to optimisecomplex circuits consisting of hundreds of inputs and thousands of gates. Unfortunately,we are facing another problem-scalability of representation. The efficiencyof the evolutionary optimization applied at the global level deteriorates withthe increasing complexity. To overcome this issue, we propose to apply the conceptof local resynthesis in this work. Local resynthesis is an iterative process basedon the extraction of smaller sub-circuits from a complex circuit that are optimizedlocally and implanted back to the original circuit. When applied appropriately, thisapproach can mitigate the problem of scalability of representation. Two complementaryapproaches to the extraction of the sub-circuits are presented and evaluated inthis work. The evaluation is done on a set of highly optimized complex benchmarkproblems representing various real-world controllers, logic and arithmetic circuits.The experimental results show that the evolutionary resynthesis provides betterresults compared to globally operating evolutionary optimization. In more than 85%cases, a substantially higher number of redundant gates was removed while keepingthe computational effort at the same level. A huge improvement was achievedespecially for the arithmetic circuits. On average, the proposed method was able toremove 25.1% more gates.

Cartesian genetic programming, Evolutionary resynthesis, Logic
optimization

Cartesian genetic programming, Evolutionary resynthesis, Logic
optimization

KOCNOVÁ, J.; VAŠÍČEK, Z.

2021

12.07.2020

1389-2576

Genetic Programming and Evolvable Machines

21

3

Spojené státy americké

287

319

33

https://www.fit.vut.cz/research/publication/12104/

s10710-020-09376-3