A hybrid supercomputer capable of 10 to 100 petaflops, or a quadrillion calculations per second, can support the Materials Genome Initiative, says Jeongnim Kim of Oak Ridge National Laboratory.

The initiative aims to accelerate understanding of the fundamentals of materials, providing practical information that innovators can use to develop new products and processes.

Kim is the principal author of QMCPACK, an application code widely used for large-scale simulations of molecules, solids and nanostructures.

The code uses a technique called Quantum Monte Carlo to predict electronic structures.