Maximal-performance scalable FFT library for accelerator-enhanced petascale computing

The Fast Fourier Transform (FFT), an efficient method for computing Discrete Fourier Transforms, is widely used in climate modeling, molecular dynamics, structural dynamics, astronomical computations, nuclear energy, and many other scientific applications. Representing one of the most important users of numerical routines, the Department of Energy (DOE) uses FFTs to solve the most-pressing challenges facing our scientific community. Most FFT software and numerical libraries scale inefficiently when targeted to a large number of processors. This makes the FFT a common strong bottleneck in petascale software. An industrial-quality scalable FFT numerical library that can work efficiently on FPGA/GPU-enriched heterogeneous supercomputers and that can deliver breakthrough speedups for FFT-intense