World First: Supercomputer H1N1 Flu Virus Simulated…

Our understanding of the universe is expanding in all directions, expanding outwards as we see more and more of the cosmos and inwardly as we learn more about the nano world. In the nano world the tools for capturing the action shots are improving but there are still a number of destinations we are unable to explore. Organic organisms – for example, influenza virus – are too delicate for the standard tools of nano exploration. Powerful x-rays and laser illumination used in modern microscopes simply disintegrate such delicacies. In these situations scientists have been turning to simulations – computer models – for answers. Until now, studying viruses has been challenging in laboratories, this new technology not only allows for greater offline research, it belies the complexity associated with simulating billions of particles in the correct conditions to create such simulations

Dr. Ying Ren along with a team of researchers at the Chinese Academy of Sciences Institute of Process Engineering have also been searching for answers using simulators, in the process they have developed the Computational Microscope, a breakthrough in simulation technology.

Using the Mole-8.5 supercomputer the researchers were able to simulate a complete H1N1 virus at the atomic level. Giving scientists the ultimate sandbox like environment to study the virus and how it interacts with various substances, the computational microscope. The Mole-8.5 Supercomputer used to perform the simulation was developed through a close partnership between NVIDIA and IPE-CAS, integrating 2,200 NVIDIA Tesla Processors spread over 288 nodes this is a serious piece of number crunching equipment that draws enough power to run a small town.

“The Mole-8.5 GPU supercomputer is enabling us to perform scientific research that simply was not possible before,” said Dr. Ying Ren, assistant professor at IPE-CAS. “This research is an important step in developing more effective ways to control epidemics and create anti-viral drugs.”

This milestone in simulation, the ability to simulate an entire organism at the atomic level, is giving researches new tools to further their research. Virus’s and many other subjects that scientists need to study operate in the very small world of nano – a billionth of a meter -. With the added complication that chemical – organic and inorganic – reactions happen at such fast speeds that they are very difficult to study in the real world. Not only do they need to simulate hundreds of millions of atoms or radicals but they need to simulate them interacting trillions of times a second, the femtosecond time scale – quadrillionth of a second or millionth billionth of a second -. Scientists using the Mole-8.5 supercomputer achieved a time resolution of 1 femtosecond for 300 million atoms or radicals. At this resolution the Mole-8 was able to generate 770 Pico seconds – 0.00000000077 seconds – of simulation for each day of processing on the supercomputer. While this may seem like an incredibly small amount of time to simulate but it is enough to capture complete reactions in the virus world.
.

Image of a virus.

.
By turning to supercomputer power and simulation software able to fast forward, rewind or pause entire chemical reaction processes or in the case of the H1N1 virus reactions. Ranked 21st on the Top 500 Supercomputers list the Mole-8.5 contains over 7,000 Intel Xeon X5520 CPU’s, 15 terabytes of RAM and draws over 500Kw of power when in operation.

Part of the breakthrough in processing power that is allowing this work is the adoption of high end graphics cards as number crunchers with the other side of this breakthrough being the software that Mole-8.5 uses to run the simulation. The molecular dynamics simulation application has been especially written to take advantage of NVIDIA ‘s Tesla Processor. It isn’t enough to have massive amounts of leading edge hardware, the software must bring it all together and drive it The Tesla Processors provided by NVIDIA are a descendant of it’s graphics cards processor, with its processing power being turned to more general purpose calculation tasks.

The Mole-8.5 doesn’t just spend its day in the digging through the virtual world of the H1N1 virus either, it also generates turbulent flow simulations, weather models and fusion energy simulations. As with all supercomputers it’s time is in high demand.

Many of the worlds most powerful supercomputers are regularly put to work simulating the real world. From nuclear explosions, to medications they are all benefiting from this advancing field of computing. One of the greatest challenges when attempting to real world simulations is the moving target that is our understanding of life. As our understanding of the properties and mechanisms of life improves the complexity of the simulation has to increase to cater for the new understandings. If the ultimate goal of a simulator is to incorporate all of the properties of matter and create a 100% facsimile of life then it is reasonable to expect to know all of those properties first. Paradoxically as we are using these simulators to help us understand these properties, so there is 100% chance of not understanding everything before building a simulator. Which actually turns out to be the strength of using this kind of simulator, they are easily modified. As we test and improve our understanding we modify the simulator and then continue on testing and learning. Atomic level simulation is an important milestone for a public institution, giving researchers access to these new tools.

Just as with personal computers the march of technology and processing power for supercomputers continues like a drum beat, constantly evolving. With each step forward in processing power new opportunities become available. In this case the latest technology is behind Mole-8.5 reaching the atomic level milestone. Over the next few years the increases in processing power will increase the number of atoms and time scale will improve as will the software driving the simulations. New discoveries will be made and old theories confirmed thanks to the power of 2,200 high end gaming video cards working together.

source: CAS-IPE