Optimal Granularity of Parallel Test Generation on the Client -Agent- Server Model

This paper proposes a Client-Agent-Server model (CAS model) which can decrease the work load of the client by adding agent processors to the Client-Server model and presents an approach to parallel test generation for logic circuits on the CAS model. In this paper we consider the fault parallelism in which a cluster of faults will be allocated from the client processor to an agent processor and from an agent processor to a server processor for the CAS model. Hence we have to consider two granularities; one is the size of the cluster between the client and agents and the other is the size of the cluster between agents and servers. We formulate the problem of test generation for the CAS model and analyze the optimal pair of granularities in both cases of static and dynamic task allocation. Finally we present experimental results based on an implementation of our CAS model on a network of workstations using the TSCAS'89 benchnlark circuits. The experimental results are very close to the analytical results which confirms the existence of an optimal pair of granularities that minimizes the total processing time for benchmark circuits as well as analysis.

This paper proposes a Client-Agent-Server model (CAS model) which can decrease the work load of the client by adding agent processors to the Client-Server model and presents an approach to parallel test generation for logic circuits on the CAS model. In this paper, we consider the fault parallelism in which a cluster of faults will be allocated from the client processor to an agent processor and from an agent processor to a server processor for the CAS model. Hence, we have to consider two granularities; one is the size of the cluster between the client and agents, and the other is the size of the cluster between agents and servers. We formulate the problem of test generation for the CAS model and analyze the optimal pair of granularities in both cases of static and dynamic task allocation. Finally, we present experimental results based on an implementation of our CAS model on a network of workstations using the TSCAS'89 benchnlark circuits. The experimental results are very close to the analytical results which confirms the existence of an optimal pair of granularities that minimizes the total processing time for benchmark circuits as well as analysis.