Accelerating VASP electronic structure calculations using graphic processing units

Abstract

We present a way to improve the performance of the electronic structure Vienna Ab initio Simulation Package (VASP) program. We show that high-performance computers equipped with graphics processing units (GPUs) as accelerators may reduce drastically the computation time when offloading these sections to the graphic chips. The procedure consists of (i) profiling the performance of the code to isolate the time-consuming parts, (ii) rewriting these so that the algorithms become better-suited for the chosen graphic accelerator, and (iii) optimizing memory traffic between the host computer and the GPU accelerator. We chose to accelerate VASP with NVIDIA GPU using CUDA. We compare the GPU and original versions of VASP by evaluating the Davidson and RMM-DIIS algorithms on chemical systems of up to 1100 atoms. In these tests, the total time is reduced by a factor between 3 and 8 when running on n (CPU core + GPU) compared to n CPU cores only, without any accuracy loss.

Colossal permittivity behavior and its origin in rutile (Mg1/3Ta2/3)xTi1-xO2

“…Calculations were performed using the PAW formalism with the VASP code 43, 44 . Periodic boundary conditions were employed on 3 × 3 × 4 supercells of the rutile crystallographic unit cell.…”

VASP hits the memory wall: Processors efficiency comparison

“…VASP code has got a GPU‐accelerated variant implemented in CUDA . This variant of VASP uses MPI as well and support multi‐GPU and multinode parallel execution.…”

“…However, the majority of them deploy only a fraction of the GPU theoretical performance even after careful tuning . With a lot of efforts, VASP has been ported to CUDA as well …”

GQSD: The program for the graphic processing units accelerated quantum scattering dynamics

“…Two approaches can be followed: new mathematical methods and parallel computing on latest hardware platforms. The GPU has hundreds or thousands arithmetic elements to perform massively parallel calculation, and has been used in many computational areas of theoretical chemistry, such as molecular dynamics, [5][6][7] quantum chemistry, [8][9][10] chemical kinetics, [11,12] and potential energy surface. [13] Recently, GPU has also been applied to quantum scattering calculations using both TID and timedependent wavepacket methods.…”