Quantum Mechanics/Molecular Mechanics Simulations on NVIDIA and AMD Graphics Processing Units

Abstract

We have ported and optimized the graphics processing unit (GPU)-accelerated QUICK and AMBER-based ab initio quantum mechanics/molecular mechanics (QM/MM) implementation on AMD GPUs. This encompasses the entire Fock matrix build and force calculation in QUICK including one-electron integrals, twoelectron repulsion integrals, exchange-correlation quadrature, and linear algebra operations. General performance improvements to the QUICK GPU code are also presented. Benchmarks carried out on NVIDIA V100 and AMD MI100 cards display similar performance on both hardware for standalone HF/DFT calculations with QUICK and QM/MM molecular dynamics simulations with QUICK/AMBER. Furthermore, with respect to the QUICK/AMBER release version 21, significant speedups are observed for QM/MM molecular dynamics simulations. This significantly increases the range of scientific problems that can be addressed with open-source QM/MM software on state-of-the-art computer hardware.

Drug Design in the Exascale Era: A Perspective from Massively Parallel QM/MM Simulations

“…As modern architectures make extensive use of heterogeneous nodes that combine multicore CPUs with GPUs, ,, achieving exascale will require coupling GPU-ready MM and QM software able to scale on many (≈10 2|3 ) such nodes. While a plethora of classical MD codes already exist that fully exploit GPUs, − including GROMACS used in MiMiC, full implementation for these architectures is still an ongoing process for DFT codes, ,,,, except for the TeraChem proprietary software. , This is arguably the main reason why serious endeavors to port first principle QM/MM MD interfaces to GPUs are appearing only now in the literature. , …”

“…56,57 This is arguably the main reason why serious endeavors to port first principle QM/MM MD interfaces to GPUs are appearing only now in the literature. 119,120 Strong scaling on heterogeneous nodes is actually the major challenge for molecular simulation. In force field based MD simulations, this is related to the relatively fixed size of the biological systems of interest 71 and the intrinsic seriality of the time evolution integration algorithms.…”

Drug Design in the Exascale Era: A Perspective from Massively Parallel QM/MM Simulations

“…As modern architectures make extensive use of heterogeneous nodes that combine mul-ticore CPUs with GPUs, 55,110,111 achieving exascale will require coupling GPU-ready MM and QM software able to scale on many (≈ 10 2|3 ) such nodes. While a plethora of classical MD codes already exist that fully exploit GPUs, [112][113][114][115] including GROMACS 71 used in MiMiC, full implementation for these architectures is still an ongoing process for DFT codes, 61,63,65,66,116 except for the TeraChem proprietary software. 56,57 This is arguably the main reason why serious endeavours to port first principle QM/MM MD interfaces to GPUs are appearing only now in the literature.…”

AmberTools

“…Amber has had a long tradition of QM/MM methods and implementations, with the most recent additions being the QUICK/sander QM/MM implementation in AmberTools23. ,− QUICK/sander has been extensively updated, and its performance has been significantly improved. QUICK, as distributed with AmberTools23, can also be used as a standalone QM program for single point calculations or geometry optimizations.…”