Towards an automatic generation of dense linear algebra solvers on parallel architectures - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Reports (Research Report) Year : 2014

Towards an automatic generation of dense linear algebra solvers on parallel architectures

Abstract

The increasing complexity of new parallel architectures has widened the gap between adaptability and efficiency of the codes. As high performance numerical libraries tend to focus more on performance, we wish to address this issue using a C++ library called NT2. By analyzing the properties of the linear algebra domain that can be extracted from numerical libraries and combining them with architectural features, we developed a generic approach to solve dense linear systems on various architectures including CPU and GPU. We have then extended our work with an example of a least squares solver based on semi-normal equations in mixed precision that cannot be found in current libraries. For the automatically generated solvers, we report performance comparison with state-of-the-art codes, showing that it is possible to obtain a generic code with a high-level interface (similar to Matlab) that can run either on CPU or GPU and that does not generate significant overhead.
Fichier principal
Vignette du fichier
RR-8615.pdf (951.25 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01075663 , version 1 (19-10-2014)

Identifiers

  • HAL Id : hal-01075663 , version 1

Cite

Marc Baboulin, Joel Falcou, Ian Masliah. Towards an automatic generation of dense linear algebra solvers on parallel architectures. [Research Report] RR-8615, Université Paris-Sud; INRIA. 2014, pp.20. ⟨hal-01075663⟩
408 View
324 Download

Share

Gmail Facebook Twitter LinkedIn More