BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20121114T173000Z
DTEND:20121114T174500Z
LOCATION:155-F
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Hardware reliability becomes a challenge with technology scaling. Silent Data Corruptions (SDCs) from soft-errors pose a major threat in commodity systems space. Hence significantly reducing the user-visible SDC rate is crucial for low-cost in-field reliability solutions.  This thesis proposes a program-centric approach to identify application locations that cause SDCs and convert them to detections using low-cost program-level error detectors. =0A=0AWe developed Relyzer to obtain a detailed application resiliency profile by systematically analyzing all application fault-sites without performing time-consuming fault injections on all of them. It employs novel fault pruning techniques to lower the evaluation time by 99.78% for our workloads.  Using Relyzer, we obtained and analyzed the comprehensive list of SDC-causing instructions. We then developed program-level error detectors that on average provide a much lower-cost alternative to a state-of-the-art solution for all SDC rate targets.  Overall, we provide practical and flexible choice points on the performance vs. reliability trade-off curves.
SUMMARY:Analyzing and Reducing Silent Data Corruptions Caused By Soft-Errors
PRIORITY:3
END:VEVENT
END:VCALENDAR