Frances E. Allen

International Business Machines Corporation


Election Year: 2010
Primary Section: 34, Computer and Information Sciences
Membership Type: Emeritus

Research Interests

Since joining IBM in 1957 with a Masters Degree in Mathematics, I have been working on the design and construction of compilers and programming languages for high performance computers. The goal is to bridge the gap between problem specifications and computer solutions so as to enable programmers and domain experts to be more productive and computer solutions to be faster and more precise. For more than 50 years advances in programming languages, software systems, computer designs and hardware components have enabled stunning advances in the problem-solving capabilities of high performance computing. However one of the main drivers of performance, the number of transistors/chip, hit a fundamental performance limit that was definitively identified in 2003. So how to continue achieving hardware-driven performance advances? The answer: multicores. These simpler, more flexible, hardware building blocks can be linked in multiple ways and performance achieved by parallelism. But how are problems specified so the performance potential of parallelism is utilized? The task of bridging this gap is considered the biggest problem Computer Science has ever faced! I agree. Having led a successful research project on automatic parallelization of programs, I am now giving lectures on the problem and possible solutions.

Powered by Blackbaud
nonprofit software