History, Trends, and Future Directions in High Performance Computing
University of Mannheim, Germany Lawrence Berkeley National Lab, Berkely, CA
OVERVIEW Chapter 1: Overview of the TOP500 project; purpose of the TOP500, how well does the list meet its purpose; Moore's Law, how well the TOP500 tracks not just Moore's Law per se, but the many other metrics of growth (CPU speed, memory size and speed, etc), and why perhaps TOP500 is good for some but not all of them Chapter 2: History of Supercomputer Statistics, Mannheim Supercomputer Statistics 1986 - 1992 Chapter 3: What are the TOP500, rules submission guidelines, quality control, how did the rules evolve, why no Strassen, mixed precision submissions; rules about vendor systems Chapter 4: LINPACK and the TOP500, pros and cons HISTORY AND TRENDS Chapter 5: Historic development based on the idea of Bell's Law Chapter 6: Architectures part of Market Place of HPC, update of the paper Chapter 7: The History of the TOP1 systems Chapter 8: TOP Sites Chapter 9: The most successful architectures in the TOP500 need to define first what is success Chapter 10: Papers about the TOP500 Chapter 11: Paper against the TOP500 Chapter 12: New Ranking metrics for Supercomputers Chapter 13: Alternative benchmarks in use in HPC Chapter 14: TOP500 website and database