The US Department of Energy's Oak Ridge National Laboratory (ORNL) has unveiled a computer capable of handling 200,000 trillion calculations per second (200 petaflops). Laying claim to the title of the world's most powerful supercomputer, Summit is eight times more powerful than ORNL's previous supercomputer, Titan, which came online in 2012 with a capacity of 27 petaflops.

Supercomputers have advanced so far and so fast that it's easy to forget that the computers called smartphones we carry around in our pockets could stroll past a state-of-the-art supercomputer of a generation ago without breaking a digital sweat. However, ORNL's Titan supercomputer was 200,000 times more powerful than a desktop computer, so why make one eight times more powerful than that?

That answer is that scientists are about the greediest consumers of computer power imaginable. Every time someone reaches the limits of contemporary computer technology, the results generated by those computers raise new questions that require computers many more times more powerful to provide the answers.

Whether it's physics, biochemistry, meteorology, engineering, or artificial intelligence, many fields demand more number crunching than before. Start with trying to model a simple column of air and the next thing you know you're trying to model the weather patterns over an entire continent. A simple nuclear reaction calculation can end up simulating the complexities of a thermonuclear warhead explosion. Recreating a single neuron can lead to making a virtual rat's brain.

Even more important, having access to such computing power has real international consequences, which is why there is constant competition to make bigger and better supercomputers in China, Taiwan, Japan, India, the United States, and Europe. Such computers are vital not only for science, but for weather forecasting, climate modeling, nuclear weapons development and safety, and complex engineering problems.

An evolution of the Titan's hybrid CPU–GPU design, Summit uses the IBM AC922 system made up of 4,608 servers. Each of these is composed of two 22-core IBM Power9 processors, as well as six NVIDIA Tesla V100 graphics processing unit accelerators interconnected with dual-rail Mellanox EDR 100Gb/s InfiniBand. Supporting this architecture is over 10 petabytes of memory and high-bandwidth pathways to handle data traffic.

The upshot is that Summit is the first computer able to handle scientific calculations requiring at least one billion billion calculations per second, known as exascale calculations. It can work on AI research, machine learning, and deep learning as well as questions of medicine, and high-energy physics.

ORNL says Summit will help the United States achieve what it calls a fully capable exascale computing ecosystem for broad scientific use by the year 2021. However, the first tasks for the system starting next year will be help with the Department of Energy's Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program.

Some of the early projects for Summit include studies of how supernovae create heavy elements, like gold and iron, atomic-level simulations of new materials, analysis of public health data to better understand cancer in the US population, and using AI and machine learning to better understand human diseases.

"Today's launch of the Summit supercomputer demonstrates the strength of American leadership in scientific innovation and technology development. It's going to have a profound impact in energy research, scientific discovery, economic competitiveness and national security," says Secretary of Energy Rick Perry. "I am truly excited by the potential of Summit, as it moves the nation one step closer to the goal of delivering an exascale supercomputing system by 2021. Summit will empower scientists to address a wide range of new challenges, accelerate discovery, spur innovation and above all, benefit the American people."

Summit was unveiled on Friday by Department of Energy Secretary Rick Perry.

Source: ORNL

View gallery - 3 images