Hardware Support for Accurate and Reliable Numerical Computations
Roundoff error and catastrophic cancellation in floating point computations can quickly lead to inaccurate results. The research objectives of this project are to design, develop, and evaluate hardware support for avoiding such errors. The project concentrates on hardware and instruction set modifications to support variable-precision arithmetic and interval arithmetic. This includes the development of algorithms and hardware designs for accurate and reliable computing, and the evaluation of cost, performance, and power consumption of the resulting hardware.Education in computer architecture and processor design is being addressed by the development of courses, by encouragement of undergraduate and graduate research, and by outreach to middle school and high school students.
This material is based upon work supported by the National Science Foundation under Grant No. 9703421. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.