Association for Computing Machinery
Shrinking feature sizes as a result of technology scaling have led to an increased hardware susceptibility to soft errors. Soft errors may cause spurious bit flips in the underlying hardware that may then propagate through the software layer and finally jeopardize software correctness. Extensive reliability-increasing research has been conducted at hardware-level. Hardware-level soft-error mitigation methods typically incur significant area, performance, and power overhead. Software-level reliability techniques have evolved to provide further improved system reliability and may be used in addition to hardware techniques.