How does the world’s fastest supercomputer work?

The majority of us think a computer is fast if it can run “LEGO Lord of the Rings” or a YouTube video of an English bulldog on a skateboard. For scientists working on complicated problems, 158 billion calculations per second are not nearly enough for a PC with an i7 processor.

This is why researchers are so excited about the Cray Titan fastest supercomputers from Oak Ridge National Laboratory (ORNL). It was six months before the Titan was unveiled in October 2012 that Lawrence Livermore National Laboratory’s Sequoia Blue Gene/Q machine held the title of fastest computer in the world.

What is the speed of the Titan? The theoretical top speed is 27 petaflops, which may not sound impressive until you realize it translates into 27,000 trillion calculations per second. It’s hundreds of thousands of times faster than your top-of-the-line PC. Titan, however, won’t fit on a desktop; it occupies an area as large as a basketball court.

The speed of Titan makes it a fantastic tool for solving really complicated problems involving huge amounts of data. It will be used by scientists to run detailed simulations of the Earth’s climate, which may yield ideas on how to limit global warming. It may also help design super-efficient internal combustion engines and solar panels and speed up the testing of new drugs by running biological simulations. Scientists could use Titan to simulate the breaking of bonds that hold molecules together, giving them new insights into one of nature’s most important processes.

The Titan is important not just because it is extremely world’s fastest supercomputer, but because it pioneers a new type of supercomputer design that could lead to even faster machines in the future. By building machines with thousands and thousands of central processing units, or CPUs, and then breaking down the calculations that they want to perform into smaller pieces that could be parceled out to all of those CPUs, scientists have achieved higher and higher speeds for years. This approach has the drawback of consuming an enormous amount of electricity due to all the CPU chips. To accelerate computations, the Titan pairs each of its 18,688 CPUs with a graphics processing unit, or GPU – a similar chip used in gaming computers. GPUs consume less power than CPUs, so their machines are faster, but also a lot more energy efficient than their predecessors.

As a result of the Titan, researchers envision exascale computers – machines that are thousands or more times more powerful than today’s most powerful supercomputers