Industrial Applications of High-Performance Computing

179

Chapter 12

U.S. Industrial

Supercomputing at NCSA

Merle Giles, Seid Koric, and Evan Burness

12.1 INTRODUCTION

e beginnings of industrial supercomputing at the National Center for

Supercomputing Applications (NCSA) are rooted in a rising sentiment

during the 1980s that America had to maximize the benets found at

the intersection of science and industry to compete. National security

concerns borne out of World War II and the Cold War resulted in funding

for supercomputers at the Departments of Defense and Energy, leaving

open-science and open-engineering experts limited access to advanced

computational resources and expertise. ese concerns are described

more fully in Chapter 1.

CONTENTS

12.1 Introduction 179

12.2 Scientic Networks and Early Challenges with Data 181

12.3 NSF Award Funds Study of Simulation-Based Engineering

and Science (SBE&S) 182

12.4 Engineering-Centric Industrial Supercomputing 184

12.5 Blue Waters: An Unmatched Supercomputer for Achieving

Sustained Petaops 185

12.6 Engineering Simulation Performance Breakthroughs 187

12.7 Digital Manufacturing and Design Innovation Institute 193

12.8 Summary 193

References 193

180 ◾ Industrial Applications of High-Performance Computing

NCSA’s funding began in 1986 as an outcome of the 1982 Black Proposal

submitted by Dr. Larry Smarr and his colleagues (see Chapter1). Included

in the initial strategy was the establishment of the Industrial Partners

Program, later named the Private Sector Program (PSP). Since 1986, the

PSP has partnered with and provided consulting advice to nearly one-half

of the U.S. FORTUNE50,

®

in a range of sectors including manufacturing,

oil and gas, retail/wholesale/distribution, health care, pharmaceuticals,

agriculture, food, information technology, bioinformatics, transportation,

nance, and more. PSP’s participation in NDEMC (see Chapter 21) led to

an additional and continuing investment in experts trained in the use of

engineering applications in supercomputing environments.

A selection of large-company partners, both current and former, include

• Agriculture: ADM, FMC Corporation, Syngenta

• Finance: JPMorgan, Morgan Stanley

• Information and Communication Technology, including soware:

AT&T, Cray, Dell, IBM, Intel, Microso, Nokia-Siemens Networks,

American Airlines Sabre Systems, Schlumberger, Simulia Abaqus, Sony

• Insurance: Allstate, State Farm

• Manufacturers: Boeing, Caterpillar, Deere, Dow, Eastman Kodak,

Ford, GE, Honeywell, Motorola, Procter & Gamble, Rolls-Royce,

United Technologies

• Medical and Pharmaceuticals: Abbott Labs, Eli Lilly, Mayo Clinic

• Oil and Gas: BHP Billiton, BP, ExxonMobil Upstream Research,

Phillips Petroleum

• Retail, Wholesale, and Distribution: Kellogg, Sears

An expectation routinely heard from others is that large companies

have enough money and power to solve all technical challenges themselves.

Aer all, they have eets of PhDs and their own supercomputers, so why

should governments and national agencies bother with collaborating?

e answer is simple: large companies have enough expertise to know

what the problems are and what is not getting done with existing science

and engineering. In other words, it is these companies that push the very

boundaries of science and engineering in ways that no one else can.