Announcement of New Rensselaer Supercomputer, AMOS
“Rensselaer At Petascale”
Shirley Ann Jackson, Ph.D.
President, Rensselaer Polytechnic Institute
Rensselaer Polytechnic Institute, Troy, NY
Thursday, October 3, 2013
Amos Eaton would be very proud today of the school he helped to establish with Stephen Van Rensselaer, as we celebrate the arrival of yet another key element of the unique computational ecosystem that is coalescing at Rensselaer Polytechnic Institute.
This ecosystem is a combination of remarkable infrastructure, crucial partnerships, and towering talent that is allowing Rensselaer to lead the way in a new era of computationand beyond that, into a new era of computationally-intensive discovery and innovation in every field.
With our Advanced Multiprocessing Optimized System, or AMOS, Rensselaer now enters the petascale computing era. Earlier in the year, our 16 racks of IBM Blue Gene/Lwhich already represented one of the most powerful university-based supercomputers in the countrywere replaced by two racks of IBM Blue Gene/Q. Now, we have moved to five racks, for a supercomputer with a combined processing power of 1.1 petaflopsor a quadrillion floating point operations per second. That is nearly 150,000 calculations per second for every man, woman, and child on earth. In unscientific terms, a huge number.
Our new supercomputer now vaults into the top four most powerful university-based supercomputers in the United States and the top 12 university-based supercomputers in the world. It is the most powerful university-based supercomputer in New York Stateand throughout the northeast. In fact, it is the only supercomputer with such capacity at an American private university.
AMOS honors the academic father of Rensselaer Polytechnic Institute, Amos Eatonan excellent name for our truly super supercomputer. We have to credit this marvelous moniker to alumnus Mark Wojcik of the Class of 1991 and our student Benjamin Vreeland, who won our naming contest.
The speed and power of AMOS is important. Immense calculations, that previously might have taken weeks or months, now can be done during a coffee break. However, speed does not begin to describe what is special about this machine.
It also is massively networked and interactive within itself, allowing not only extremely large questions to be answered, but also extremely complex ones that require hundreds of thousands of simultaneous tasks orchestrated towards a single end.
The degree of high-performance internal networking, along with the advanced parallel storage of the supercomputer, make it an extremely useful tool for this era of petabyte-scale data generation and analysisand a cornerstone of The Rensselaer Institute for Data Exploration and Applications, or The Rensselaer IDEAwhich we launched this past spring.
The Rensselaer IDEAwhich is bringing together Rensselaer talents in high-performance computing, cognitive computing, web science, data analytics, and visualization and immersive technologiesis intended to maximize the ability of our researchers and students in every field to harness the expanding possibilities for discovery and innovation in this data-driven, web-enabled, supercomputer-powered, globally-interconnected world.
I am going to say more about the significance of the horsepower and balance of our new supercomputer in a moment.
First, however, I want to point out that our supercomputer namesake, Amos Eaton, a botanist and geologist, was a zealous collector and analyzer of dataand conducted significant geological and agricultural surveys across New York State. He also was an important innovator in scientific education, insisting that Rensselaer students learn by doingby collecting data in the field, by experimenting in the laboratory, and by preparing lectures for their peers and professors. In other words, he pioneered the studio classroom approach we still use here today. Amos Eaton also believed in data-driven discovery and innovation.
AMOS, the machine, is a meaningful part of that approach today. At Rensselaer, we are very unusual in allowing large numbers of undergraduates to learn, and to innovate on such powerful supercomputers. It helps us to attract some of the most brilliant incipient young computer scientists in the United Statesand to prepare them to shape the future. It allows our emerging engineers, scientists, artists, architects, and business students to do remarkable things.
I want to thank John Kolb, our Vice President for Information Services and Technology and Chief Information Officer, and Professor Chris Carothers, the Director of our high-performance computing center, as well as their teams, for laying the groundwork for the arrival of AMOS.
Many of you were here when we announced earlier this year that Watson was enrolling at Rensselaer. Watson is the remarkable IBM cognitive computing system that, in 2011, used its natural language processing abilities and inference engines to vanquish the best human champions at Jeopardy! There was some question at the time as to whether Watson would have a roommate.
Today, I can answer, yes: AMOS and Watson will room together at our high-performance computing center. This is only fitting, as Watson is named for the founder of IBM, Thomas J. Watson, Sr. Rensselaer and IBM like having systems named for our founders residing together, because Rensselaer and IBM both are pioneers in supercomputingas well as partners in this endeavor dating back nearly two decades to the IBM SP2.
In more recent years, our IBM counterpart in these explorations has been Dr. E. John Kelly III, who is director of IBM Researchworldwideas well as a Rensselaer alumnus and a distinguished member of the Rensselaer Board of Trustees. Dr. Kelly recently was elected a member of the National Academy of Engineering for his extraordinary contributions to the United States semiconductor industry. I want to thank him for his extraordinary contributions to Rensselaer.
The IBM partnership with us was the linchpin that enabled Rensselaer to launch our high-performance computing center, the Computational Center for Nanotechnology Innovations, in 2006, with support from New York State.
We have come a long way since then. Research on the supercomputer has expanded well beyond nanotechnology into fields that range from biotechnology to medicine to energy. Therefore, today, I am pleased to announce that we are renaming our high-performance computing center the Center for Computational Innovationsor CCI.
The unique combination of AMOS and Watson in the CCI will help Rensselaer further achieve the overarching goal we have set for ourselves as we approach the 200th anniversary of our founding, with The Rensselaer Plan 2024: to be a transformative force with global impact. We are creating the tools at the CCI that will allow us to address the grand global challenges surrounding food, water, and energy security; human health, climate impacts, national security, and the allocation of scarce natural resourceson a scale that will improve many, many lives.
I mentioned a new era in computation. Let me briefly describe the CCI’s place in it:
Thus far, supercomputing has been most useful for simulations, and the speed of AMOS will make them faster. We have the talent here to take maximum advantage of that speed. This year, Rensselaer scientists led by Professor Carothers, joining forces with colleagues at the Lawrence Livermore National Laboratory, set a new supercomputer simulation speed record, decisively smashing the old record of handling 12 billion events per second with 504 billion events per second.
We also have researchers across many fields ready to use AMOS to change the world. In fact, they already have begun. As just one among many examples, Professor of Physics, Constellation Chair, and Head of the Department of Physics, Applied Physics & Astronomy, Dr. Angel Garcia is using massively parallel molecular dynamics simulations to investigate protein and RNA foldingin part, to better understand the molecular pathology of diseases such as Alzheimer’s, so we can develop better treatments for them.
With the ability of AMOS to balance the needs of both computer- and data-intensive applications, we will be able to do such modeling and simulation, at the petascale level. The volume, velocity, and variety of data now being generated are far outrunning humanity’s capacity to turn it into useful information. Facebook, for example, has said that it is warehousing half a petabyte of new data every single day. The sources of data now range from scientific platforms such as the Hubble Space telescope, to teenagers posting vacation photos on Instagram, and increasingly include the systems of our physical universeas our electric meters, our cars, and our running shoes are equipped with sensors, and are networked digitally.
Within this flood of data are the seeds for an enormous amount of discovery and innovation, if only we can find them.
The stunning internal networking and storage capacity of AMOS will enable us to help develop the use of supercomputers in data analytics. Towards that end, a team led by Professor Petros Drineas has been awarded a $1 million grant from the National Science Foundation Division of Information & Intelligent Systems to explore new strategies for mining and extracting knowledge from petabyte-scale data using supercomputing.
We also will develop powerful synergies in the combination of AMOS and Watson. IBM sent Watson to Rensselaer, so that it could take classes from us, and extend its capabilities through our research. Rensselaer IDEA Director Jim Hendler and his team will move Watson from memory-based artificial intelligence, which is limited to information programmed into the computer’s memory, to knowledge-based artificial intelligenceor an ability to access the entire world of information on the Web, to infer from it, and to make decisions based on it in real-time.
Ultimately, Professor Hendler and his team will use their experience with Watson to expand the possibilities of agent-based modelingi.e. computational models that can simulate the effects, on a system, of the decision-making of individualsto the scale of AMOS. Their eventual intent is to run a Watson cognitive computing system on AMOS.
Because the great global challenges are social as much as they are physical, the human implications of merging cognitive computing with supercomputing are enormous. Events as different as the global financial crisis of 2007-2008, and the Fukushima nuclear meltdown in Japan in 2011, make clear the degree to which we are subject to intersecting vulnerabilities with cascading consequenceswhich can not only affect, but be affected by human actions and decision-making.
Giving AMOS the capacity to do agent-based modeling will allow us to perform economic modeling that could, for example, alert us that, offered certain kind of mortgages, homeowners in certain cities will overextend themselves, setting off a domino effect that will shake world financial markets. We will be able to consider the problem of how human dynamics affect the evacuation of a major city following a disaster, and save many lives. For the first time, we will be able to design a smart electric grid that takes into account, in a very detailed way, the various individual human responses to price incentives.
In short, we are developing an utterly remarkable computational ecosystem that will allow us to move beyond data mining to more sophisticated predictive analytics, and to prescriptive analytics. In other words, we will be able to understand, to describe, and, therefore, to shape our world with a new degree of precision.
Ultimately, Rensselaer scientists and engineers intend to do for the supercomputer what the graphical user interface did for the desktopand to make that tool broadly available to people who are not programmersso that business students, architects, and agroecologists, for example, can use it to change the world.
The arrival of AMOS is truly momentous for Rensselaer. Given our remarkable people, programs, platforms, and partners, we will make it momentous for the world at large.
Source citations are available from the division of Strategic Communications and External Relations, Rensselaer Polytechnic Institute. Statistical data contained herein were factually accurate at the time it was delivered. Rensselaer Polytechnic Institute assumes no duty to change it to reflect new developments.