A well conditioned checksum scheme for algorithmic fault tolerance

Daniel L. Boley, Franklin T.Luk

Research output: Contribution to journalArticlepeer-review

5 Scopus citations

Abstract

The weighted checksum scheme has been proposed as a low-cost fault tolerant procedure for parallel matrix computations. To guarantee multiple error detection and correction, the chosen weight vectors must satisfy some very specific properties about linear independence. However, previous weight generating methods that fulfill the independence criteria have troubles with numerical overflow. We will present a new scheme that generates weight vectors via Chebyshev polynomials to meet the requirements about independence and to avoid the difficulties with overflow.

Original languageEnglish (US)
Pages (from-to)21-32
Number of pages12
JournalIntegration, the VLSI Journal
Volume12
Issue number1
DOIs
StatePublished - Nov 1991

Keywords

  • Algorithm-based fault tolerance
  • Berlekamp-Massey algorithm
  • Chebyshev polynomials
  • Lanczos algorithm

Fingerprint Dive into the research topics of 'A well conditioned checksum scheme for algorithmic fault tolerance'. Together they form a unique fingerprint.

Cite this