Artificial intelligence can be regarded as having been founded as a discipline in the mid-1950s, its essential mathematical underpinnings having been developed in the 1930s and 40s. From its founding, it has seen periods of extensive investment, enthusiasm, and progress, interspersed with periods of consolidation and re-evaluation.
But it is now the overwhelming consensus of industry and academia that AI has finally entered the mainstream of society and research. The evidence all around is completely compelling. To give just one example, in a recent report by a task force charged with developing an AI research strategy in France and Europe, mathematician and member of the French Parliament Cédric Villani, who, incidentally, cites as a fundamental influence the writings on AI of IU Distinguished Professor Douglas Hofstadter, writes that "its visionary nature makes artificial intelligence one of the most fascinating scientific endeavors of our time. It is," he continues, "no longer merely a research field confined to laboratories or to a specific application. It will become one of the keys to the future."1
Of particular significance and importance in this report was its recommendation for the acquisition of a supercomputer with an architecture designed specifically for AI, which exactly reflects the strategy IU has already implemented.
Indiana University has been a center of research in a number of areas of AI for many years, and in 2019, we established a formal AI initiative. The first phase of this was the announcement in June of IU's acquisition of its new Big Red 200 artificial intelligence supercomputer.
The second phase was the announcement of an extraordinarily generous gift of $60 million by Fred Luddy to support the establishment of an AI Center with a focus on digital health and medicine, to be based in what is now the Luddy School of Informatics, Computing, and Engineering. The groundbreaking for the building that will house this center will take place within the next few months. We expect that further phases of this initiative will be announced in the months ahead.
So it is fitting that today, on the exact date of the 200th anniversary of one of the nation’s leading research universities, we gather to dedicate Big Red 200, one of the centerpieces of IU's newest initiative in an area of rapidly growing global importance aimed at ensuring we will continue to lead in education and research as we begin our third century.
Supercomputing at Indiana University
There is no question that, in the 21st century, the supercomputer more broadly is one of the mainstays of both theoretical and applied science. In the areas of climate modeling, genome analysis, protein folding, molecular modeling, nuclear fusion research, and in many, many other disciplines, including in the social sciences and the humanities, scholars and scientists are working with hugely complicated calculations and vast data sets and seeking to solve problems that are becoming more and more complex. All of them benefit from—and, indeed, fundamentally require—advances in processing power.
For the past 20 years, Indiana University has pushed the boundaries of high-speed computing to support the world-class research being conducted by IU's scientists and scholars.
In 2001, we acquired the first university-owned supercomputer capable of calculations at the rate of one TeraFLOP—or one trillion mathematical operations per second.
In 2006, IU's Big Red supercomputer announced what was then the fastest academic supercomputer in the western hemisphere. Capable of calculations at the rate of over 20 TeraFLOPS, Big Red allowed our researchers to analyze and store massive amounts of data and to perform groundbreaking simulations of biological phenomena. It was also a tremendous resource for the state, as we collaborated with groups across Indiana, including the Indiana Economic Development Corporation, to expand the benefits of Big Red's power.
Many of you were with us in this very space in 2013 when we dedicated Big Red II, which was capable of calculations at a peak speed of one PetaFLOP, that is, a thousand trillion mathematical operations per second. Big Red II accelerated discovery and facilitated new research by hundreds of IU scientists and scholars right across the university, including in medicine, biology, physics, chemistry, astronomy, network science, sustainability science, global climate research, public health, and, of course, informatics, computer science, and engineering.
But, at nearly seven years old, Big Red II is obsolete, in terms of both speed and the specialized requirements of AI computations.
Big Red 200: Accelerating research and discovery
The supercomputer we dedicate today, Big Red 200, will ensure that IU remains at the forefront in the use of high-speed and data-intensive computing in support of some of the most vital and complex research in the world. It is the fastest supercomputer in the state of Indiana, and one of the fastest university-owned artificial intelligence-capable supercomputers in the nation. A recent article noted that "Supercomputers are the only machines that offer the tools and technologies that organizations will inevitably need as they embrace the next wave of AI growth."2 Big Red 200’s impressive AI capabilities will allow IU researchers to understand phenomena in the natural world that simply cannot be understood otherwise—and to make breakthrough discoveries in a wide range of research areas, including the early detection of breast cancer and neurodegenerative diseases such as Parkinson's and Alzheimer's.
Big Red 200 was originally planned to have a theoretical top speed of nearly six PetaFLOPS—six times faster than its predecessor. In fact, I am pleased to announce that it will be even faster still, as nearly two additional PetaFLOPS of computing speed are to be added due to some very new technology that will be incorporated, making Big Red 200 about one-third faster than originally announced.
As the program for today's event notes, it would take the entire population of the state of Indiana more than 28 years—performing one calculation per second around the clock—to perform the same number of calculations that Big Red 200 can perform in just one second.
Inscribed on Big Red 200’s decorative panel covers are the names of more than 300 academic disciplines that will benefit from its enormous computational capacity. It will play a major role in the recruitment of new faculty in many of these fields. And it will help add knowledge-intensive jobs in the state of Indiana, particularly in the state's growing life sciences economy.
An initiative such as this, of course, requires the cooperative efforts of many people.
University Information Technology Services have done a superb job in leading IU's acquisition of Big Red 200 and bringing it online. I especially want to recognize Brad Wheeler, who succeeded me as Vice President for Information Technology and CIO, and many other colleagues—especially Matthew Link, associate vice president for research technologies, Dave Hancock, director for advanced cyberinfrastructure, and Craig Stewart, Executive Director of the Pervasive Technology Institute, who has been a central and key figure in IU’s efforts in supercomputing going back over 20 years. We are grateful to all of them for their outstanding efforts.
Let me also express my thanks, through Charles Morreale, from whom we will hear in a few moments, to our partners at Cray Incorporated, a company that bears the name of one of the most legendary figures in supercomputing.
And I want to thank Vice President Tom Morrison and his staff at the IU Physical Plant who were involved in the installation of Big Red 200, including its critical cooling system.
Big Red 200, then, will return Indiana University to the forefront of academic high-speed computing and allow the university to continue to reap the benefits of its position as a leader in information technology.
The phenomenal success of the original Big Red and its successor, Big Red II, demonstrated that investing in high-speed computing returns dividends many times over in the form of scientific advances, research funding, and job creation.
Big Red 200 will allow us to build further upon this success as it serves as a driver of progress at the beginning of Indiana University’s third century of service to the state, the nation, and the world.
Thank you very much.
1. Cédric Villani, et. al, For a Meaningful Artificial Intelligence: Towards a French and European Strategy, (Conseil national du numérique, 2018).
2. "The Next Wave of AI Won’t Happen Without Supercomputing," CRAY, June 11, 2019, Web, Accessed January 13, 2020, URL: https://www.cray.com/blog/ai-wont-happen-without-supercomputing/.