PACT 2007: Keynotes

International Conference on 
Parallel Architecture and Compilation Techniques
Brasov, Romania
Sept 15-19, 2007

The Sixteenth International Conference on
Parallel Architectures and Compilation Techniques (PACT)
Brasov, Romania
September 15-19, 2007

PACT 2007 Keynotes

We have a stellar line-up of keynote speakers for PACT 2007, including researchers who defined the field in architecture, compilers, and software:

Compilers and Multi-Cores: A Performance Partnership or Missed Opportunity?
Fran Allen, IBM T.J. Watson Research Laboratory

9:00am, Tuesday, September 18, 2007

Multi-core computers are ushering in a new era of parallelism everywhere. As more cores (and parallelism) are added, the potential performance of the hardware will increase at the traditional rate. But will single applications be faster? How can such applications take advantage of all the parallelism? This talk will consider the potential role of compilers and languages in helping deliver the multi-core performance promise.

Fran Allen, IBM Fellow Emerita at the T. J. Watson Research Laboratory, specializes in compilers for high-performance computers. Allen's early work laid the foundations of the theory and practice of program analysis and optimization. Her later parallel translation project (PTRAN) continued the development of these methods to encompass whole program analysis and parallelization. Allen is a member of the National Academy of Engineering, the American Philosophical Society, and the American Academy of Arts and Sciences, and is the recipient of ACM's 2006 Turing Award.

Harnessing the Transformation Hierarchy
Yale Patt, The University of Texas at Austin

9:00am, Wednesday, September 19, 2007

The marketplace knows that process technology continues to provide twice the raw capability every couple of years. Moore's Law they call it. In ten years: more than 100 billion transisters on each silicon die, running on a 10GHz clock, at least. Yet, in the recent past, at least, the marketplace has not seen comparable performance. Part of the problem is clearly due to the fact that these transistors have not been harnessed optimally. I would argue that an avenue of attack that could offer great rewards is what I have been calling the Transformation Hierarchy. That is, problems are stated in natural language, but it is the electron at the substrate layer that is actually solving the problem. In between, we have the algorithm, the mechanical language, the ISA, the microarchitecture and the circuits. The more one layer is able to interact with the others, the more we can produce higher performing engines. The compiler and the microarchitecture have already started this symbiosis. It is time they bring the other layers into the act. In this talk, I will suggest some ways.

Yale Patt is a teacher at The University of Texas at Austin, where he also does research in microarchitecture and has consulted for microprocessor manufacturers for more than 30 years. He and his students have been responsible for a number of innovations which are now taken for granted in most high-end microprocessors. HPS, introduced at Micro-18 in 1985 was the first comprehensive microengine to introduce wide-issue, agressive branch prediction, speculative execution, out-of-order execution and in-order retirement to preserve precise exceptions. His two-level branch predictor, introduced at Micro-24 in 1991, has been adapted to just about every high end chip since Intel's Pentium Pro in 1995. He has earned the appropriate degrees from reputable universities and received more than enough awards for his research and teaching. More detail is available on

Write clean (parallel) code!
Bjarne Stroustrup, Texas A&M University

9:00am, Monday, September 17, 2007

This keynote articulates the ideal of raising the level of abstraction of parallel code for portability and portability. The focus is on the idea of "clean code" which includes concerns of direct expression of domain-specific ideas and maintainability. It also argues the need for integrating parallel code into "ordinary" code to avoid segregating the community using and supporting parallel code from the larger software development community.

Bjarne Stroustrup is the designer and original implementer of C++ and the author of "The C++ Programming Language". His research interests include distributed systems, design, programming techniques, software development tools, and programming languages. Dr. Stroustrup is the College of Engineering Chair Professor in Computer Science at Texas A&M University . Member of The National Academy of Engineering ; ACM Fellow, IEEE Fellow, AT&T Bell Laboratories Fellow, and AT&T Fellow . He is actively involved in the ANSI/ISO standardization of C++.