Efficient Fault Tolerance using Intel MPX and TSX

Oleksii Oleksenko, Dmitrii Kuvaiskii, Pramod Bhatotia, Christof Fetzer, Pascal Felber

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Hardware faults can cause data corruptions during computation, and they are especially harmful if these corruptions happen in data pointers. Existing solutions, however, incur high performance overheads, which is unacceptable for computeintensive applications. In this work, we present an efficient fault tolerance approach against hardware faults by exploiting the new extensions to the x86 architecture. In particular, we propose that Intel MPX can be effectively used to detect faults in data pointers, while Intel TSX can provide roll-back recovery against these corruptions. Our preliminary evaluation supports this hypothesis, and we estimate the average overhead to be roughly around 50%.
Original languageEnglish
Title of host publication46th Annual IEEE/IFIP International Conference on Dependable Systems and Networks
Number of pages2
Publication statusPublished - 3 Oct 2016
Event46th Annual IEEE/IFIP International Conference on Dependable Systems and Networks - Toulouse, France
Duration: 28 Jun 20161 Jul 2016
https://dsn-2016.sciencesconf.org/

Conference

Conference46th Annual IEEE/IFIP International Conference on Dependable Systems and Networks
Abbreviated titleDSN 2016
CountryFrance
CityToulouse
Period28/06/161/07/16
Internet address

Fingerprint Dive into the research topics of 'Efficient Fault Tolerance using Intel MPX and TSX'. Together they form a unique fingerprint.

Cite this