Architecture Design for Soft Errors

CHAPTER 6

Fault Detection via Redundant Execution

6.1 Overview

Fault detection via redundant execution is a common form of fault detection that has been used for decades. Unlike error coding techniques—described in the previous chapter—that detect faults using redundant information in storage bits or logic units, the techniques described in this chapter detect faults by comparing outputs from redundant streams of instructions. Typically, fault detection via redundant execution can provide greater fault coverage across a processor chip compared to error coding techniques on individual hardware structures. This is because the same redundant execution technique can cover multiple hardware structures, unlike in many error coding implementations ...