Harnessing the Power of Multi-GPU Acceleration into the Quantum Interaction Computational Kernel Program

We report a new multi-GPU capable ab initio Hartree–Fock/density functional theory implementation integrated into the open source QUantum Interaction Computational Kernel (QUICK) program. Details on load balancing algorithms for electron repulsion integrals and exchange correlation quadrature across multiple GPUs are described. Benchmarking studies carried out up to four GPU nodes, each containing NVIDIA V100-SXM2 type demonstrate that our is of achieving excellent high parallel efficiency. For representative medium large size protein/organic molecular systems, observed efficiencies remained above 82% Kohn–Sham matrix formation 90% nuclear gradient calculations. The accelerations A100, P100, K80 platforms also have realized higher than 68% in all tested cases, paving way large-scale electronic structure calculations with QUICK.