Level 1 Parallel RTN-BLAS: Implementation and Efficiency Analysis - Archive ouverte HAL Access content directly
Conference Papers Year : 2014

Level 1 Parallel RTN-BLAS: Implementation and Efficiency Analysis

(1) , (1) , (1)
1

Abstract

Modern high performance computation (HPC) performs a huge amount of floating point operations on massively multi-threaded systems. Those systems interleave operations and include both dynamic scheduling and non-deterministic reductions that prevent numerical reproducibility, i.e. getting identical results from multiple runs, even on one given machine. Floating point addition is non-associative and the results depend on the computation order. Of course, numerical reproducibility is important to debug, check the correctness of programs and validate the results. Some solutions have been proposed like parallel tree scheme [1] or new Demmel and Nguyen's reproducible sums [2]. Reproducibility is not equivalent to accuracy: a reproducible result may be far away from the exact result. Another way to guarantee the numerical reproducibility is to calculate the correctly rounded value of the exact result, i.e. extending the IEEE-754 rounding properties to larger computing sequences. When such computation is possible, it is certainly more costly. But is it unacceptable in practice? We are motivated by round-to-nearest parallel BLAS. We can imple-ment such RTN-BLAS thanks to recent algorithms that compute correctly rounded sums. This work is a first step for the level 1 of the BLAS.
Vignette du fichier
(2014-09-24) SCAN2014.pdf (3.26 Mo) Télécharger le fichier Fichier principal
Vignette du fichier
scan2014.pdf (83.74 Ko) Télécharger le fichier
Format : Presentation
Origin : Files produced by the author(s)
Origin : Files produced by the author(s)

Dates and versions

lirmm-01095172 , version 1 (27-06-2015)

Licence

Attribution - NoDerivatives - CC BY 4.0

Identifiers

  • HAL Id : lirmm-01095172 , version 1

Cite

Chemseddine Chohra, Philippe Langlois, David Parello. Level 1 Parallel RTN-BLAS: Implementation and Efficiency Analysis. SCAN: Scientific Computing, Computer Arithmetic and Validated Numerics, Sep 2014, Wurzburg, Germany. ⟨lirmm-01095172⟩
369 View
290 Download

Share

Gmail Facebook Twitter LinkedIn More