Photo courtesy of OLCF
Project director Buddy Bland stands by Titan, the hybrid-architecture Cray XK7 system at Oak Ridge Leadership Computing Facility.
This is a series of profiles on the directors of the user facilities funded by the Department of Energy (DOE) Office of Science. These are the scientists who lead a variety of research institutions that provide researchers with the most advanced tools of modern science including accelerators, colliders, supercomputers, light sources and neutron sources, as well as facilities for studying the nano world, the environment, and the atmosphere.
Meet the director:
The cellphone Buddy Bland carries today is a thousand times more powerful than the Cray supercomputer installed at Oak Ridge National Laboratory (ORNL) back in 1985. And he should know. Bland helped to stand up that Cray computer and every new supercomputer since, starting as a programmer/system administrator and advancing in 30+ years to the position of Oak Ridge Leadership Computing Facility's (OLCF) project director.
The director's background:
Bland first got his hands on a computer in the ninth grade. "Our high school got access to a computer at a local college. None of my teachers knew how to use it, so I taught myself to program.”
His programming skills blossomed at the University of Southern Mississippi, which he attended on a U.S. Air Force ROTC scholarship. Bland earned his bachelor's and master's degrees in computer science in an era of punch cards and machine intelligence programs that taught computers their first chess moves. After graduation, the Air Force ordered him to Albuquerque, New Mexico. It was a long drive and a far cry from home. "I was pretty sure that Albuquerque was not in another country,” laughs Bland, "but I didn't know much more than that going in.”
In Albuquerque, Bland managed the Air Force's first supercomputer, collaborating with Los Alamos/Sandia and Los Alamos National Laboratories and becoming very familiar with DOE supercomputers. As his Air Force duty drew to an end, Bland heard that the DOE lab in Oak Ridge, Tennessee was in the process of getting a Cray supercomputer and needed someone to run it. The job and the man were a perfect fit.
Starting with that 1985 Cray X-MP system, Bland has been there for the evolution of the Oak Ridge computing systems: Intels, IBMs, Compaqs, SGIs, and more Crays. The main supercomputer, a Cray XK7 capable of 27,000 trillion calculations per second, is called Titan and as of June 2016 is ranked third fastest in the world of high performance computing (HPC).
OLCF's supercomputer Titan is a Cray XK7 computer system, the United States' fastest and most powerful supercomputer dedicated to open scientific research. Titan is about the size of a basketball court (6,000 sq. ft.) and when running at its peak performance, it uses the amount of electricity that a town of 3-4,000 people would use.
The OLCF is part of the DOE's HPC mission – to discover, develop, and use computers to solve the biggest problems in science, industry, and government.
Historically, the U.S. has always had the fastest computers. But in 1994 Japan and the U.S. began to trade the lead back and forth. And for the last three years China has had the fastest computer in the world. "At the scale, where we are today – we all have machines with similar capability,” says Bland. But, "our user program is different than other countries. We make it available to scientists, free of charge to people who are publishing science results. Because of that, our systems have been enormously productive.”
Bland runs a very busy shop; the OLCF receives three times more applications for access to the supercomputers than time is available on the machine. A proposal comes in to address a particular problem; it can take months to years to run the series of jobs needed to run the simulations, analyze the runs, and get results.
Researchers then analyze the data from the runs and write research papers.
"Over the last five to 10 years, the rate of growth in the performance of computer processors has slowed, to the point that individual processors are not getting faster,” explains Bland. "For supercomputers to get faster, we have to run more and more processors in parallel. This is important not just because the science community wants a bigger machine to solve more complex problems. Much of our economy is driven by technology - more powerful phones, more realistic game systems and we are part of that demand. It drives our economy and keep us competitive around the world. This innovation economy is very important.”
"Eighty percent of my time is spent managing the center and the administrative aspects of my job,” says Bland. "Twenty percent of my time is on technology keeping myself up to date on the market place, meeting with staff on the technology side of the next big machines.”
"As an Office of Science lab, we want to share with the public. We get school groups, community groups, and groups from around the world,” says Bland, totaling 300-400 tour groups each year. Two or three times a month, Bland joins the tour to talk with the visitors. "I really like talking to community groups who don't know what a supercomputer is, what it does, and why they should care. As a trustee of the government's investment in science, it is my responsibility to tell the public what we do, and why it is important.”
In 2018, Titan will be replaced by Summit, an IBM/NVIDIA machine expected to reach 200 petaflops and deliver more than five times the computational performance of Titan's 18,688 nodes, using only approximately 4,600 nodes.
Calling ORNL a research lab with computers is like calling the Appalachian Trail a long walk. "Think of a supercomputer as a tool to amplify what people can do on their own,” says Bland. "We use the world's most powerful computers to help solve some of the world's most difficult problems: energy, security, climate change, personalized medicine, geophysical science – earthquake predictions and assessment - and basic science questions like the origin of the universe and matter; why do stars explode?”
In addition to basic research, the OLCF also works with industry doing applied research. A trucking company wanting to increase fuel efficiency approached the OLCF scientists about improving the aerodynamics of its long-haul trucks. OLCF worked with the company, using a computer code borrowed from NASA that models air flow around vehicles. The company put a retrofitted kit on the market to streamline trucks, at a price that would pay for itself with fuel savings within one year of operation.
Best advice for a future director at the OLCF:
Bland gives credit for the successes and research triumphs to the OLCF staff. He shares a story about the time ORNL applied, but was told it could not become a supercomputing host facility. The major barrier was that ORNL didn't have the necessary infrastructure; a supercomputing user facility had to have a fiber optic cable connection to the rest of the country. He recalls that "some of the problems first seem to be insurmountable.” Bland's solution: "Get the right people together, who have a vision, who won't take ‘No' for an answer.” By the late 1990s, Oak Ridge created the necessary connections, teaming up with the Tennessee Valley Authority to link its facility to the fiber optic cable already in place crossing the country between Atlanta and Chicago.
"The people,” says Bland, "are what make everything work.”
Intel co-founder Gordon Moore predicted that computer components per circuit would double every year, revised to every two years, and revised again to double every 18 months. Moore was right; the performance of top HPC systems did doubled every year until 2004. But now this rate of growth is slowing and has a foreseeable end.
Speeding up computers requires research in many new directions. To overcome the heat-limitation of today's silicon chips, researchers are designing new microprocessors with materials such as graphene and molybdenum disulfide. Alternative architectures – changing from computation centered to a data movement centered designs – can reduce run times. And radically new forms of computation – neuromorphic and quantum – are being developed.
In Fiscal Year 2015, the OLCF and other user facilities welcomed over 32,000 researchers - from academic, industry, and government laboratories in all fifty states and the District of Columbia – to perform new scientific research. For details on the Office of Science User Facilities, go to: http://science.energy.gov/user-facilities/.
The Office of Science is the single largest supporter of basic energy research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information please visit http://science.energy.gov.
Sandra Allen McLean is a Communications Specialist in the Office of Science, firstname.lastname@example.org.