CHIP DESIGN BASED ON GENETIC APPROACH

Gregor Papa - Jurij Šilc

   With the growth of circuits and higher testability demands there is a need to speed up the design to shorten the time-to-market delay. But ensuring this demands many parameters have to be taken into account. The most appropriate algorithms are based on an evolutionary approach, since they are able to investigate a wide range of possible solutions and to give a near optimal solution in a relatively short time. This paper presents a new Allocation-Based Scheduling (ABS) algorithm. This algorithm improves part of the chip design process - scheduling and allocation. The algorithm bases on a genetic scheduling algorithm, but due to its cost function it is able to produce the optimal schedule for the process of allocation. The algorithm is also compared with other algorithms through the evaluation of schedules of different input graphs. It turned out that the ABS algorithm made the best solutions in all tests and is therefore very appropriate for use in chip designing.

Keywords: evolutionary technique, high-level synthesis, scheduling, allocation

[full-paper]