INDUSTRY

The Texas Advanced Computing Center (TACC) at The University of Texas at Austin has hired Dr. Victor Eijkhout as a Research Scientist in the center's High Performance Computing (HPC) group. TACC is well known for its HPC-related research and development as well as its support activities. The center operates numerous HPC systems supporting both local and national users. Eijkhout will bring leadership and technical expertise to TACC's HPC research, development, and support activities, which help to advance research programs at the university and in the national academic research community. TACC's HPC group facilitates the connection of these communities to TACC resources and to the National Science Foundation (NSF) TeraGrid, the largest computational infrastructure for scientific research. "Victor's deep expertise in producing scalable software for high-performance computing systems will be of tremendous value to the center as we deploy increasing large systems to help our users tackle the most challenging research problems," says Dr. Jay Boisseau, director of TACC. Prior to joining TACC, Eijkhout was for seven years a Research Professor at the University of Tennessee at Knoxville in the Innovative Computing Laboratory (ICL) directed by Dr. Jack Dongarra. There Eijkhout was the lead principal investigator for SALSA (Self-Adapting Large-Scale Solver Architecture), a software project funded by NSF that assists applications seeking optimal linear and nonlinear system solvers; and Project SANS (Self-Adaptive Numerical Software), optimizing software at different levels in relation to the execution environment. "I am delighted to be part of the HPC group because its collaborative environment includes a wide range of interdisciplinary projects in numerical linear algebra, parallel computing, and grid computing," Eijkhout says. "The staff is highly regarded in developing libraries, formalizing interfaces between applications and actual grid technology, and several other important areas of HPC technical expertise." Eijkhout received his doctoral degree in mathematics from the University of Nijmegen (Netherlands) in 1990. He has carried out research, taught, and published on such topics as numerical linear algebra; adaptive systems for numerical computation; parallel computing; performance optimization and evaluation; and use of aspect-oriented programming and common components in scientific computing. In addition, Eijkhout is the author of Sparsebench, a benchmark for kernels in iterative methods for linear systems; the very popular and widely used Templates book, focusing on finding solutions for linear systems; and TeX by Topic, a comprehensive discussion of the deeper foundations of TeX as a programming language.