|Name:||Tutorial 2: Dense Linear Algebra Libraries for High Performance Computing|
|Time:||Sunday, June 17, 2012
9:00 AM - 1:00 PM
Hamburg University, East Wing
|Speakers:||Jack Dongarra, University of Tennessee & ORNL|
|Jakub Kurzak, University of Tennessee|
|Hatem Ltaief, KAUST Supercomputing Laboratory|
|Abstract:||Today, a desktop computer with a multicore processor and a GPU accelerator can already provide a TeraFlop/s of performance. This tremendous computational power can only be fully utilized with the appropriate software infrastructure. Most often a major part of the computational effort in scientific and engineering computing goes towards solving linear algebra sub-problems. This tutorial surveys the state-of-the-art numerical libraries for solving problems in dense linear algebra.
The tutorial consists of three parts. The first part provides a brief historical look at the development of dense linear algebra libraries, from LINPACK, to LAPACK, to ScaLAPACK. The second part focuses on the PLASMA project (Parallel Linear Algebra Software for Multicore Architectures). Finally, the third part discusses GPU acceleration issues and the MAGMA project, and also ongoing efforts in linear algebra software for distributed memory machines (the DAGuE/DPLASMA projects).
PLASMA is a software library designed to achieve high efficiency on multi-socket multicore systems. Currently, PLASMA offers fast routines for solving linear systems, least square problems, symmetric eigenvalue problems, and singular value problems. PLASMA drastically outperforms its academic predecessors, such as LAPACK, and also outperforms vendor libraries, such as MKL, ACML, or ESSL, on top-of-the-line desktop and server-size systems.