Instruction Level Duplication and Comparison for Data Error Detection: a First Experiment

Abstract

In this paper, we present a new software approach Instruction Level Duplication and Comparison (ILDC) for data error detection. We implemented ILDC in five different case studies. To validate our proposed technique, we measured its fault detection ratio and execution time overhead. Next, we compare it to the following two existing techniques: overhead reduction (VAR3+) and software-implemented fault tolerance (SWIFT), respectively. The results show that ILDC detects more errors than VAR3+ and SWIFT at a lower overhead.