GPU acceleration of local and semilocal density functional calculations in the SPARC electronic structure code

TL;DR

This paper introduces a GPU-accelerated version of the SPARC electronic structure code, significantly speeding up density functional theory calculations and reducing computational resources needed.

Contribution

The authors develop a modular GPU-accelerated implementation for the SPARC code, achieving substantial speedups while maintaining modularity and efficiency.

Findings

Up to 6x speedup with GPU acceleration

Reduced time to solution to under 30 seconds for large systems

Significant resource savings for DFT calculations

Abstract

We present a GPU-accelerated version of the real-space SPARC electronic structure code for performing Kohn-Sham density functional theory calculations within the local density and generalized gradient approximations. In particular, we develop a modular math kernel based implementation for NVIDIA architectures wherein the computationally expensive operations are carried out on the GPUs, with the remainder of the workload retained on the CPUs. Using representative bulk and slab examples, we show that GPUs enable speedups of up to 6x relative to CPU-only execution, bringing time to solution down to less than 30 seconds for a metallic system with over 14,000 electrons, and enabling significant reductions in computational resources required for a given wall time.