A Bayesian reliability analysis of neutron-induced errors in high performance computing hardware

Curtis B. Storlie; Sarah E. Michalak; Heather M. Quinn; Andrew J. DuBois; Steven A. Wender; David H. DuBois

doi:10.1080/01621459.2013.770694

A Bayesian reliability analysis of neutron-induced errors in high performance computing hardware

Curtis B. Storlie, Sarah E. Michalak, Heather M. Quinn, Andrew J. DuBois, Steven A. Wender, David H. DuBois

Quantitative Health Sciences

Research output: Contribution to journal › Article › peer-review

6 Scopus citations

Abstract

A soft error is an undesired change in an electronic device's state, for example, a bit flip in computer memory, that does not permanently affect its functionality. In microprocessor systems, neutron-induced soft errors can cause crashes and silent data corruption (SDC). SDC occurs when a soft error produces a computational result that is incorrect, without the system issuing a warning or error message. Hence, neutron-induced soft errors are a major concern for high performance computing platforms that perform scientific computation. Through accelerated neutron beam testing of hardware in its field configuration, the frequencies of failures (crashes) and of SDCs in hardware from the Roadrunner platform, the first Petaflop supercomputer, are estimated. The impact of key factors on field performance is investigated and estimates of field reliability are provided. Finally, a novel statistical approach for the analysis of interval-censored survival data with mixed effects and uncertainty in the interval endpoints, key features of the experimental data, is presented. Supplementary materials for this article are available online.

Original language	English (US)
Pages (from-to)	429-440
Number of pages	12
Journal	Journal of the American Statistical Association
Volume	108
Issue number	502
DOIs	https://doi.org/10.1080/01621459.2013.770694
State	Published - 2013

Keywords

Accelerated testing
Cox proportional hazards
Gaussian process
Mixed effects
Neutron beam
Silent data corruption
Stochastic search variable selection

ASJC Scopus subject areas

Statistics and Probability
Statistics, Probability and Uncertainty

Access to Document

10.1080/01621459.2013.770694

Cite this

@article{31c867f519c84bea907a0eb7005508c9,

title = "A Bayesian reliability analysis of neutron-induced errors in high performance computing hardware",

abstract = "A soft error is an undesired change in an electronic device's state, for example, a bit flip in computer memory, that does not permanently affect its functionality. In microprocessor systems, neutron-induced soft errors can cause crashes and silent data corruption (SDC). SDC occurs when a soft error produces a computational result that is incorrect, without the system issuing a warning or error message. Hence, neutron-induced soft errors are a major concern for high performance computing platforms that perform scientific computation. Through accelerated neutron beam testing of hardware in its field configuration, the frequencies of failures (crashes) and of SDCs in hardware from the Roadrunner platform, the first Petaflop supercomputer, are estimated. The impact of key factors on field performance is investigated and estimates of field reliability are provided. Finally, a novel statistical approach for the analysis of interval-censored survival data with mixed effects and uncertainty in the interval endpoints, key features of the experimental data, is presented. Supplementary materials for this article are available online.",

keywords = "Accelerated testing, Cox proportional hazards, Gaussian process, Mixed effects, Neutron beam, Silent data corruption, Stochastic search variable selection",

author = "Storlie, {Curtis B.} and Michalak, {Sarah E.} and Quinn, {Heather M.} and DuBois, {Andrew J.} and Wender, {Steven A.} and DuBois, {David H.}",

note = "Funding Information: Many people contributed to the success of the Triblade beam-testing experiments. The authors thank Joe Abeyta, Chuck Alexander, Ben Bergen, Derek Bingham, Sean Blanchard, Ann Borrett, Henry Brandt, James Campa, Randy Cardon, Nathan DeBardeleben, Tom Fairbanks, Parks Fields, Alan Gibson, Gary Grider, Josip Loncaric, Pablo Lujan, Alex Malin, Andrea Manuzzato, Fred Marshall, David Modl, Andrew Montoya, John Morrison, Bill Rust, Andrew Shew-maker, Manuel Vigil, Bob Villa, Cornell Wright, and the reviewers of the IEEE TDMR article documenting some more of these results, and apologize for any inadvertent omissions from this list. This work was funded by Los Alamos National Security, LLC (LANS), operator of the Los Alamos National Laboratory under Contract No. DE-AC52-06NA25396 with the U.S. Department of Energy. This article is published under LA-UR 11-04750.",

year = "2013",

doi = "10.1080/01621459.2013.770694",

language = "English (US)",

volume = "108",

pages = "429--440",

journal = "Journal of the American Statistical Association",

issn = "0162-1459",

publisher = "Taylor and Francis Ltd.",

number = "502",

}

TY - JOUR

T1 - A Bayesian reliability analysis of neutron-induced errors in high performance computing hardware

AU - Storlie, Curtis B.

AU - Michalak, Sarah E.

AU - Quinn, Heather M.

AU - DuBois, Andrew J.

AU - Wender, Steven A.

AU - DuBois, David H.

N1 - Funding Information: Many people contributed to the success of the Triblade beam-testing experiments. The authors thank Joe Abeyta, Chuck Alexander, Ben Bergen, Derek Bingham, Sean Blanchard, Ann Borrett, Henry Brandt, James Campa, Randy Cardon, Nathan DeBardeleben, Tom Fairbanks, Parks Fields, Alan Gibson, Gary Grider, Josip Loncaric, Pablo Lujan, Alex Malin, Andrea Manuzzato, Fred Marshall, David Modl, Andrew Montoya, John Morrison, Bill Rust, Andrew Shew-maker, Manuel Vigil, Bob Villa, Cornell Wright, and the reviewers of the IEEE TDMR article documenting some more of these results, and apologize for any inadvertent omissions from this list. This work was funded by Los Alamos National Security, LLC (LANS), operator of the Los Alamos National Laboratory under Contract No. DE-AC52-06NA25396 with the U.S. Department of Energy. This article is published under LA-UR 11-04750.

PY - 2013

Y1 - 2013

N2 - A soft error is an undesired change in an electronic device's state, for example, a bit flip in computer memory, that does not permanently affect its functionality. In microprocessor systems, neutron-induced soft errors can cause crashes and silent data corruption (SDC). SDC occurs when a soft error produces a computational result that is incorrect, without the system issuing a warning or error message. Hence, neutron-induced soft errors are a major concern for high performance computing platforms that perform scientific computation. Through accelerated neutron beam testing of hardware in its field configuration, the frequencies of failures (crashes) and of SDCs in hardware from the Roadrunner platform, the first Petaflop supercomputer, are estimated. The impact of key factors on field performance is investigated and estimates of field reliability are provided. Finally, a novel statistical approach for the analysis of interval-censored survival data with mixed effects and uncertainty in the interval endpoints, key features of the experimental data, is presented. Supplementary materials for this article are available online.

AB - A soft error is an undesired change in an electronic device's state, for example, a bit flip in computer memory, that does not permanently affect its functionality. In microprocessor systems, neutron-induced soft errors can cause crashes and silent data corruption (SDC). SDC occurs when a soft error produces a computational result that is incorrect, without the system issuing a warning or error message. Hence, neutron-induced soft errors are a major concern for high performance computing platforms that perform scientific computation. Through accelerated neutron beam testing of hardware in its field configuration, the frequencies of failures (crashes) and of SDCs in hardware from the Roadrunner platform, the first Petaflop supercomputer, are estimated. The impact of key factors on field performance is investigated and estimates of field reliability are provided. Finally, a novel statistical approach for the analysis of interval-censored survival data with mixed effects and uncertainty in the interval endpoints, key features of the experimental data, is presented. Supplementary materials for this article are available online.

KW - Accelerated testing

KW - Cox proportional hazards

KW - Gaussian process

KW - Mixed effects

KW - Neutron beam

KW - Silent data corruption

KW - Stochastic search variable selection

UR - http://www.scopus.com/inward/record.url?scp=84890078418&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84890078418&partnerID=8YFLogxK

U2 - 10.1080/01621459.2013.770694

DO - 10.1080/01621459.2013.770694

M3 - Article

AN - SCOPUS:84890078418

SN - 0162-1459

VL - 108

SP - 429

EP - 440

JO - Journal of the American Statistical Association

JF - Journal of the American Statistical Association

IS - 502

ER -

A Bayesian reliability analysis of neutron-induced errors in high performance computing hardware

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this