The advent of the Mira supercomputer, along with more powerful Sequoia and K supercomputers, marks the first time that computers have enough computational power to simulate trillions of particles on the move, running a simulation of the Universe.
Mira can process quadrillions of operations every second, which is rare in the machine world. It will track trillions of particles as they move, expand and react to each other. The simulation will use the known laws of physics and will try to give researchers a deeper understanding of how the cosmos came to be.
Mira was built along IBM’s BlueGene technology and it is powered by 768,000 cores spread across 48 blade racks. It has 8 petaflops of processing power, and at its peak performance, Mira will be able to perform 10 quadrillion floating point operations per second, using up a petabyte of RAM.
Just like previous simulations, including the Bolshoi simulation, researchers at the Argonne National Laboratory are interested in learning about star formation and how entire galaxies expand, clump together and form filament structures as well as other cosmic superstructures. The scientists plan on creating a computer simulation of the early universe starting just after the Big Bang, simulating a period of roughly 12 billion years.
The first nascent galaxies flew apart from each other, while others merged and fell into the enormous clusters and filaments to form the deeper structure of the Universe. The simulation might also give indications about dark energy, which is supposed to make up the majority of the matter in the known Universe.
[via The Atlantic]