Utilizing Spares in Multichip Modules for theDual

Deening a dual role for spare processing elements (PEs) in reliability-challenged processing arrays is the major focus of the paper. The paper also explores a practical way to include reconnguration hardware in single-package arrays. The implementation of array processor systems may include spare PEs for fault tolerance. These systems typically require a host for fault diagnosis, while the healthy spares sit idle. It is proposed to utilize the idling spare PEs for purposes of fault diagnosis, giving the array the capability of self diagnosis. Fault tolerance must incorporate additional hardware for reconnguration, and existing plans have not found widespread use in single-package systems due to the extra cost and extra real estate. Multichip modules (MCMs) have the potential to ooer fault tolerance with no increase in primary circuit area. It is proposed to contain the reconnguration hardware in the active substrate of a silicon-based MCM. Further, the switches required for spares coverage can aid in the job of comparison based self-testing. We ooer a complete solution to fault-tolerant arrays in the sense that diagnosis, reconnguration and switching details are all addressed. In order to obtain reliable high speed operation, single-package array processor systems must employ fault tolerance. One way to tolerate faults is to commission redundant processors (spares) to cover (replace) faulty processing elements (PEs). A method must be in place to diagnose faulty nodes. We consider a method of fault tolerance which employs spares and provides a means of self diagnosis. The identiication of faults will be able to take place during normal operation and computational throughput will not be degraded. Most existing works consider diagnosis as a separate problem, generally administered by a central host. In such systems, it is common to nd healthy spare PEs sitting idle on the silicon real estate. By developing proper isolation schemes, the idling spares can be used to perform supplementary tasks or other \administrative" functions. Much research has been done in wafer scale integration (WSI) to obtain reliable, single-package arrays. Unresolved practical issues such as extra cost and extra chip area (for additional links and switches) have kept these solutions from nding widespread use in the eld. The multichip module (MCM) will make a good packaging candidate for fault tolerance because it has the potential to ooer reconnguration without increasing primary array area. Reconnguration hardware can be housed outside the array and can be fabricated independently. The technology of silicon substrates …