This preview shows page 1. Sign up to view the full content.
Unformatted text preview: 188 VLSI Test Principles and Architectures A powerful implication engine can have a significant impact on the performance of ATPG algorithms. Thus, much effort has been invested over the years in the efficient computation of implications. The quality of implications was improved with the computation of indirect implications in SOCRATES [Schulz 1988]. Static learning was extended to dynamic learning in [Schulz 1989] and [Kunz 1993], where some nodes in the circuit already had value assignments during the learning process. A 16-valued logic was introduced by Cox et al . [Rajski 1990], and reduction lists were used to dynamically determine the gate values. Chakradhar et al . proposed a transitive closure procedure based on the implication graph. Recursive learning was later proposed by Kunz et al . [Kunz 1994] in which a complete set of pair-wise implications could be computed. In order to keep the computational costs low, a small recursion depth can be enforced in the recursive learning procedure. Finally,small recursion depth can be enforced in the recursive learning procedure....
View Full Document
- Spring '08