Dependable Computing on Inexact Hardware through Anomaly Detection.

Reliability Against Transient Faults;Quality Control in Approximate Computing;Computer Science;Engineering;Computer Science and Engineering

Khudia, Daya ShankerMars, Jason ;

University of Michigan

关键词: Reliability Against Transient Faults;    Quality Control in Approximate Computing;    Computer Science;    Engineering;    Computer Science and Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/113341/dskhudia_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Reliability of transistors is on the decline as transistors continue to shrink in size. Aggressive voltage scaling is making the problem even worse.Scaled-down transistors are more susceptible to transient faults as well as permanent in-field hardware failures.In order to continue to reap the benefits of technology scaling, it has become imperative to tackle the challenges risen due to the decreasing reliability of devices for the mainstream commodity market. Along with the worsening reliability, achieving energy efficiency and performance improvement by scaling is increasingly providing diminishing marginal returns. More than any other time in history, the semiconductor industry faces the crossroad of unreliability and the need to improve energy efficiency.These challenges of technology scaling can be tackled by categorizing the target applications in the following two categories: traditional applications that have relatively strict correctness requirement on outputs and emerging class of soft applications, from various domains such as multimedia, machine learning, and computer vision, that are inherently inaccuracy tolerant to a certain degree.Traditional applications can be protected against hardware failures by low-cost detection and protection methods while soft applications can trade off quality of outputs to achieve better performance or energy efficiency.For traditional applications, I propose an efficient, software-only application analysis and transformation solution to detect data and control flow transient faults. The intelligence of the data flow solution lies in the use of dynamic application information such as control flow, memory and value profiling. The control flow protection technique achieves its efficiency by simplifying signature calculations in each basic block and by performing checking at a coarse-grain level. For soft applications, I develop a quality control technique. The quality control technique employs continuous, light-weight checkers to ensure that the approximation is controlled and application output is acceptable. Overall, I show that the use of low-cost checkers to produce dependable results on commodity systems---constructed from inexact hardware components---is efficient and practical.