Applications for arithmetic error codes in large, high-performance computers

Large, high-performance computers are too costly to allow full replication for fault detection and error correction in the communication and processing of numerical information. For this reason more cost-effective arithmetic error code applications offer an attractive alternative.