A Case Study of the Reproducibility Issues in EigenExa

Abstract : In this article, we study the accuracy and reproducibility issues in EigenExa due to the non-associative of floating-point operations, rounding-off errors, dynamic thread scheduling, and different reduction trees, etc. By investigating the EigenExa’s algorithmic structure and the corresponding implementation, we find the origins of the non-reproducibility and demonstrate it on the numerical results. As a solution, we, at first, propose to apply the accurate and reproducible parallel reduction from the ExBLAS library. Moreover, we outline possible extensions to ExBLAS that would enhance further the numerical properties of EigenExa.
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01539365
Contributor : Stef Graillat <>
Submitted on : Wednesday, June 14, 2017 - 4:59:31 PM
Last modification on : Thursday, March 21, 2019 - 1:14:05 PM

Identifiers

  • HAL Id : hal-01539365, version 1

Citation

Roman Iakymchuk, Imamura Toshiyuki, Stef Graillat, Stefano Markidis, Erwin Laure. A Case Study of the Reproducibility Issues in EigenExa. EASC2016, Exascale Applications and Software Conference, Apr 2016, Stockholm, Sweden. ⟨hal-01539365⟩

Share

Metrics

Record views

99