Accelerating the computation of FLAPW methods on heterogeneous architectures

Legacy codes in computational science and engineering have been very successful in providing essential functionality to researchers. However, they are not capable of exploiting the massive parallelism provided by emerging heterogeneous architectures. The lack of portable performance and scalability puts them at high risk, ie, either they evolve or they are destined to be executed on older platforms and small clusters. One example of a legacy code which would heavily benefit from a modern redesign is FLEUR, a software for electronic structure calculations. In previous work, the computational bottleneck of FLEUR was partially reengineered to have a modular design that relies on standard building blocks, namely, BLAS and LAPACK libraries. In this paper, we demonstrate how the initial redesign enables the portability to heterogeneous architectures. More specifically, we study different approaches to port the code to architectures consisting of multi-core CPUs equipped with one or more coprocessors such as Nvidia GPUs and Intel Xeon Phis. Our final code attains over 70% of the architectures' peak performance, and outperforms Nvidia's and Intel's libraries. On JURECA, the large tier-0 cluster where FLEUR is often executed, the code takes advantage of the full power of the computing nodes, attaining 5× speedup over the sole use of the CPUs