Speaker

R. Clint Whaley (University of Texas, San Antonio)


Title

ATLAS Version 3.8 : Overview and Status (PDF, 246Kbytes)


Abstract

This paper describes the widely-used ATLAS (Automatically Tuned Linear Algebra Software) project. ATLAS is an instantiation of a paradigm in high performance library production and maintenance, which we term AEOS (Automated Empirical Optimization of Software); this style of library management has been created in order to allow software to keep pace with the incredible rate of hardware advancement inherent in Moore's Law. This paper overviews the basics of what ATLAS is and how it works, highlights some of the recent improvements available in version 3.8.0 (the newest stable release of ATLAS, released in summer 2007), as well as discussing some of the current challenges and future work.


Short biography

R. Clint Whaley is an assistant professor in the Computer Science Department of the University of Texas at San Antonio. He received his PhD in Computer Science in December 2004 from Florida State University in the area of optimizing compilers, his MS in Computer Science in May 1994, from the University of Tennessee at Knoxville in the area parallel programming, and his BS in Mathematics in May of 1991. He was a full-time researcher at the University of Tennessee at Knoxville as a Research Associate from May 1994 to June 1999, and as a Senior Research Associate from June 1999 to December 2001. He was a Post-doctoral researcher and adjunct at Florida State University from January 2005 through June 2005, and has been an assistant professor at the University of Texas at San Antonio since.

His research interests include empirical optimization, optimizing compilers, high performance computing, computer architecture and parallel computing. Dr. Whaley has been involved in the production of several widely-used research and optimizations frameworks. In particularly, he contributed to the design and development of ScaLAPACK and the BLACS, which are still the defacto standard for solving distributed memory dense linear algebra problems. As a full time researcher, his research into empirical code optimization lead to the development of the well-known ATLAS (Automatically Tuned Linear Algebra Software) project. He has since extended the idea of empirical tuning into a more general compilation framework, known as iFKO (iterative Floating Point Kernel Optimizer). .