A genetic algorithm-based system for generating test programs for microprocessor IP cores

The current digital systems design trend is quickly moving toward a design-and-reuse paradigm. In particular, intellectual property cores are becoming widely used. Since the cores are usually provided as encrypted gate-level netlist, they raise several testability problems. The authors propose an automatic approach targeting processor cores that, by resorting to genetic algorithms, computes a test program able to attain high fault coverage figures. Preliminary results are reported to assess the effectiveness of our approach with respect to a random approach.

[1]  Christos A. Papachristou,et al.  Instruction randomization self test for processor cores , 1999, Proceedings 17th IEEE VLSI Test Symposium (Cat. No.PR00146).

[2]  Sujit Dey,et al.  DEFUSE: a deterministic functional self-test methodology for processors , 2000, Proceedings 18th IEEE VLSI Test Symposium.

[3]  Jacob A. Abraham,et al.  Test Generation for Microprocessors , 1980, IEEE Transactions on Computers.