Empirical Installation of Linear Algebra Shared-Memory Subroutines for Auto-Tuning

The final publication is available at Springer via http://dx.doi.org/10.1007/s10766-013-0249-6The introduction of auto-tuning techniques in linear algebra shared-memory routines is analyzed. Information obtained in the installation of the routines is used at running time to take some decisions to reduce the total execution time. The study is carried out with routines at different levels (matrix multiplication, LU and Cholesky factorizations and linear systems symmetric or general routines) and with calls to routines in the LAPACK and PLASMA libraries with multithread implementations. Medium NUMA and large cc-NUMA systems are used in the experiments. This variety of routines, libraries and systems allows us to obtain general conclusions about the methodology to use for linear algebra shared-memory routines auto-tuning. Satisfactory execution times are obtained with the proposed methodology.Partially supported by Fundacion Seneca, Consejeria de Educacion de la Region de Murcia, 08763/PI/08, PROMETEO/2009/013 from Generalitat Valenciana, the Spanish Ministry of Education and Science through TIN2012-38341-C04-03, and the High-Performance Computing Network on Parallel Heterogeneus Architectures (CAPAP-H). The authors gratefully acknowledge the computer resources and assistance provided by the Supercomputing Centre of the Scientific Park Foundation of Murcia and by the Centre de Supercomputacio de Catalunya.Cámara, J.; Cuenca, J.; Giménez, D.; García, LP.; Vidal Maciá, AM. (2014). Empirical Installation of Linear Algebra Shared-Memory Subroutines for Auto-Tuning. International Journal of Parallel Programming. 42(3):408-434. doi:10.1007/s10766-013-0249-6S40843442