Informatics, TU Vienna

The Tensor Contraction Engine

A Domain-Specific Approach to Synthesizing High-Performance Codes for Quantum Chemistry

Abstract

The accurate modeling of the electronic structure of atoms and molecules in quantum chemistry involves computationally intensive tensor contractions over large multidimensional arrays. We are developing a compiler, the Tensor Contraction Engine (TCE), that translates a high-level specification of such computations into efficient, parallel code tailored to the characteristics of the target architecture. This talk provides an overview of the TCE and its domain-specific language. We will discuss several optimization components, including an approach to generating GPU code.

Biography

Gerald Baumgartner studied computer science at the Universität Linz and earned his Ph.D.~at Purdue University. He then taught at Ohio State University and is now associate professor at Louisiana State University. He teaches primarily in the areas of programming languages and compilers. His research interests include compiler optimizations, the design and implementation of domain-specific and object-oriented languages, cloud computing, and testing tools for embedded systems. He is the architect of the TCE, and his extension to the C++ programming language has been publicly available as part of the GNU C++ compiler, Version 2.8.

Note

This talk is organized by the Compilers and Languages Group at the Institute of Computer Languages.
Tea at the library of E185/1, Argentinierstr. 8, 4th floor (central) after the talk.