Performance and Energy Analysis of OpenMP Runtime Systems with Dense Linear Algebra Algorithms

João V.F. Lima (1) , Issam Raïs (2) , Laurent Lefèvre (2) , Thierry Gautier (2)

1 UFRGS - Universidade Federal do Rio Grande do Sul [Porto Alegre]
2 AVALON - Algorithms and Software Architectures for Distributed and HPC Platforms

In this paper, we analyse performance and energy consumption of four OpenMP runtime systems over a NUMA platform. We present an experimental study to characterize OpenMP runtime systems on the three main kernels in dense linear algebra algorithms (Cholesky, LU and QR) in terms of performance and energy consumption. Our experimental results suggest that OpenMP runtime systems can be considered as a new energy leverage. For instance, a LU factorization with concurrent write extension from libKOMP achieved up to 1.75 of performance gain and 1.56 of energy decrease.

Domains: Distributed, Parallel, and Cluster Computing [cs.DC]

João V.F. Lima, Issam Raïs, Laurent Lefèvre, Thierry Gautier. Performance and Energy Analysis of OpenMP Runtime Systems with Dense Linear Algebra Algorithms. SBAC-PAD 2017 - International Symposium on Computer Architecture and High Performance Computing, Oct 2017, Campinas, Brazil. pp.1-6, ⟨10.1109/SBAC-PADW.2017.10⟩. ⟨hal-01666343⟩