Reproducible, Accurately Rounded and Efficient BLAS - LIRMM - Laboratoire d’Informatique, de Robotique et de Microélectronique de Montpellier
Conference Papers Year : 2016

Reproducible, Accurately Rounded and Efficient BLAS

Abstract

Numerical reproducibility failures rise in parallel computation because floating-point summation is non-associative. Massively parallel and optimized executions dynamically modify the floating-point operation order. Hence, numerical results may change from one run to another. We propose to ensure reproducibility by extending as far as possible the IEEE-754 correct rounding property to larger operation sequences. We introduce our RARE-BLAS (Reproducible, Accurately Rounded and Efficient BLAS) that benefits from recent accurate and efficient summation algorithms. Solutions for level 1 (asum, dot and nrm2) and level 2 (gemv) routines are presented. Their performance is studied compared to Intel MKL library and other existing reproducible algorithms. For both shared and distributed memory parallel systems, we exhibit an extra-cost of 2× in the worst case scenario, which is satisfying for a wide range of applications. For Intel Xeon Phi accelerator a larger extra-cost (4× to 6×) is observed, which is still helpful at least for debugging and validation steps.
Fichier principal
Vignette du fichier
REPPAR16.pdf (313.96 Ko) Télécharger le fichier
Origin Files produced by the author(s)
Loading...

Dates and versions

lirmm-01280324 , version 1 (29-02-2016)
lirmm-01280324 , version 2 (28-07-2016)

Identifiers

Cite

Chemseddine Chohra, Philippe Langlois, David Parello. Reproducible, Accurately Rounded and Efficient BLAS. Euro-Par: Parallel Processing Workshops., Aug 2016, Grenoble, France. pp.609-620, ⟨10.1007/978-3-319-58943-5_49⟩. ⟨lirmm-01280324v2⟩
309 View
546 Download

Altmetric

Share

More