We consider the problem of partitioning of an electronic system into k partitioning blocks taking into account the minimisation of the number of off-chip wires, the area of substrate and the power dissipated in blocks. In the first part of the paper, the comparison of the partitioning results for evolutionary (EA) and Kernighan's-Lin's (K-L) algorithms for the chosen benchmarks [l] has been carried out. Theobtained results show the superiority of EA algorithm. In the second part, the EA has been applied for the constrained partitioning in which the limitations of substrate area and the power dissipated in each block has been considered.
INTRUDUCTIONThe design of contemporary electronic circuits (systems) is characterised by a greater diversity of system and subsystems' components, and requires an interdisciplinary analysis highly dependent upon physical and functional partitioning. The growing importance of partitioning methods in design of integrated circuits (ICs) is observed [2], [5]. Although, apart from Multi-Chip Module (MCM) there are solutions with a whole system on a chip (VLSI, ULSI and GSI -Giga Scale Integration), the optimal partitioning of an electronic system is still required [4]. Partitioning of electronic circuits belongs to the NP-complete problems, so partitioning algorithms are normally based on heuristics. They are finding suboptimal (local) solution in a reasonable computational time. The most widely used heuristics are Kernighan-Lin [8] and Fiduccia-Mattheyses [4] algorithms. However, these algorithms are not very suitable for solving real VLSI partitioning problemstheir performance is strongly dependent on the starting point and they are unable to handle any additional constraints in the efficient way. In such a case, evolutionary algorithms (EAs) seem to be a promising alternative as they have already turned out to be powerful tools for solving hard combinatorial problems (e.g. [lo]). A few EA based approaches to VLSI circuit partitioning have been already described in the literature [ 111. In this work we introduce another algorithm, based on the EA with heuristic recombination operators and adaptive penalty functions for constraint handling. The preliminary results show that our approach could be more efficient than the algorithms presented earlier.A contemporary complex electronic system can be partitioned in a variety of ways [71,[121,[141[151. Partition of an electronic system should be carried out to:provide testability and manufacturability, achieve system reconfigurability.fulfil a set of design constraints, Moreover, low power, low cost and high performance of the system play a very important role in the design of contemporary ICs.
PROBLEM FORMULATIONAn electronic circuit can be modelled by a graph G(V,E) in which a set of vertices V (lyl = n ) represents circuit elements and a set of edges E represents connections between elements. Our task is todivide Vinto k subsets (blocks) Vo, VI, ... , Vk.,, in such a way that an objective function is optimised, subject to some constrai...
