Test Scheduling of Interposer-Based 2.5-D ICs Using Enhanced Differential Evolution Algorithm

Abstract

Interposer-based 2.5-dimensional integrated circuits (2.5D ICs) are seen as an alternative choice and they are rising as a precursor toward 3D integration. However, as the number of dies embedded in the interposer increases, the efficient test of 2.5D ICs becomes more difficult. In the design of test wrapper and test scheduling, both the test-time and the hardware cost have to be take into account. This paper presents an innovative differential evolution algorithm with dynamic subpopulations and adaptive searching strategy for the optimization of 2.5D IC test scheduling and hardware cost control. The whole population are partitioned into subpopulations dynamically using affinity propagation based clustering algorithm. In the subpopulations, a new mutation scheme which is controlled automatically by fitness values and distances between individuals is also presented. Parallelism among subpopulations and the proposed adaptive mutation and rotation crossover strategy can increase the speed of evolution without losing population diversity. Test wrapper scan chain balance design and the test scheduling algorithm, which combine the DE variant algorithm show an excellent performance in optimization ability comparing with the integer linear programming formulation (ILP) and some other configurations. It can make a good balance between the hardware cost and the test-time cost.