Parallel Implementation of Density Functional Theory Methods in the Quantum Interaction Computational Kernel Program

Manathunga, Madushanka; Miao, Yipu; Mu, Dawei; Götz, Andreas W.; Merz, Kenneth M.

doi:10.1021/acs.jctc.0c00290

Cited by 38 publications

(57 citation statements)

References 55 publications

(86 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Given quadrature microbatches with a sufficiently small spatial extent, basis screening via Equation 10produces an approximately constant number of basis functions per quadrature batch, thus leading to an overall scaling that depends only on the number of quadrature points. There are several ways to obtain the quadrature batches (Stratmann et al, 1996;Burow and Sierka, 2011;Manathunga et al, 2020). In this work, we recursively subdivide the domain spanned by the quadrature points into cuboids until the number of quadrature points within each cuboid is below a certain threshold.…”

Section: Numerical Integration Of Molecular Integrandsmentioning

confidence: 99%

“…In this work, we have chosen this threshold to be 512 quadrature points. In practice, this partitioning scheme produces batches similar to the octree method of Manathunga et al (2020). However, rather than bisecting every domain into octants, cuboids that contain an atomic center are partitioned into 27 cuboids as shown in Figure 1.…”

Section: Numerical Integration Of Molecular Integrandsmentioning

confidence: 99%

“…The rationale behind this step is to avoid polluting the device memory with redundant copies of P j and V j . While Algorithm 3 could be implemented on the GPU, as has been discussed in the context of batch generation in related work (Manathunga et al, 2020), we do not explore such implementations in this work. To improve the performance of the CPU implementation of Algorithm 3, the loop around the atomic quadrature batches may be parallelized using shared memory parallelism schemes such as OpenMP.…”

Section: Distributed Load Balance In the XC Integrationmentioning

confidence: 99%

“…Of the required integrals, the electron repulsion integrals (ERIs) and the exchange-correlation (XC) potential are among the most costly and the most challenging to port to GPU architectures. Over the years, there has been a considerable amount of research devoted to porting implementations of Gaussian basis set KS-DFT to the GPU (Yasuda, 2008;Brown et al, 2010;Titov et al, 2013;Luehr et al, 2016;Kussmann and Ochsenfeld, 2017;Manathunga et al, 2020;Peters et al, 2020); however, the vast majority of this work has been centered around the evaluation and digestion of the ERIs in the construction of the Fock matrix Martinez, 2008, 2009a,b;Asadchev et al, 2010;Miao and Merz, 2013;Kalinowski et al, 2017;Kussmann and Ochsenfeld, 2017;Laqua et al, 2020). On the other hand, the XC potential has received much less treatment in the literature in this regard (Yasuda, 2008;Luehr et al, 2016;Manathunga et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

On the Efficient Evaluation of the Exchange Correlation Potential on Graphics Processing Unit Clusters

et al. 2020

View full text Add to dashboard Cite

The predominance of Kohn–Sham density functional theory (KS-DFT) for the theoretical treatment of large experimentally relevant systems in molecular chemistry and materials science relies primarily on the existence of efficient software implementations which are capable of leveraging the latest advances in modern high-performance computing (HPC). With recent trends in HPC leading toward increasing reliance on heterogeneous accelerator-based architectures such as graphics processing units (GPU), existing code bases must embrace these architectural advances to maintain the high levels of performance that have come to be expected for these methods. In this work, we purpose a three-level parallelism scheme for the distributed numerical integration of the exchange-correlation (XC) potential in the Gaussian basis set discretization of the Kohn–Sham equations on large computing clusters consisting of multiple GPUs per compute node. In addition, we purpose and demonstrate the efficacy of the use of batched kernels, including batched level-3 BLAS operations, in achieving high levels of performance on the GPU. We demonstrate the performance and scalability of the implementation of the purposed method in the NWChemEx software package by comparing to the existing scalable CPU XC integration in NWChem.

show abstract

Section: Numerical Integration Of Molecular Integrandsmentioning

confidence: 99%

Section: Numerical Integration Of Molecular Integrandsmentioning

confidence: 99%

Section: Distributed Load Balance In the XC Integrationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

On the Efficient Evaluation of the Exchange Correlation Potential on Graphics Processing Unit Clusters

et al. 2020

View full text Add to dashboard Cite

show abstract

“…DFT-based QM/MM calculations require utilization of supercomputer facilities and many CPUs. Recently, considerable efforts have been performed to develop a GPU-based DFT code [8,9,11,12,18,19]. It was implemented in the TeraChem program package [17].…”

Section: Introductionmentioning

confidence: 99%

Computational Characterization of the Substrate Activation in the Active Site of SARS-CoV-2 Main Protease

2020

JSFI

View full text Add to dashboard Cite

Molecular dynamics simulations with the QM(DFT)/MM potentials are utilized to discriminate between reactive and nonreactive complexes of the SARS-CoV-2 main protease and its substrates. Classification of frames along the molecular dynamic trajectories is utilized by analysis of the 2D maps of the Laplacian of electron density. Those are calculated in the plane formed by the carbonyl group of the substrate and a nucleophilic sulfur atom of the cysteine residue that initiates enzymatic reaction. Utilization of the GPU-based DFT code allows fast and accurate simulations with the hybrid functional PBE0 and double-zeta basis set. Exclusion of the polarization functions accelerates the calculations 2-fold, however this does not describe the substrate activation. Larger basis set with d-functions on heavy atoms and p-functions on hydrogen atoms enables to disclose equilibrium between the reactive and nonreactive species along the MD trajectory. The suggested approach can be utilized to choose covalent inhibitors that will readily interact with the catalytic residue of the selected enzyme.

show abstract

Achieving performance portability in Gaussian basis set density functional theory on accelerator based architectures in NWChemEx

et al. 2021

View full text Add to dashboard Cite

Parallel Implementation of Density Functional Theory Methods in the Quantum Interaction Computational Kernel Program

Cited by 38 publications

References 55 publications

On the Efficient Evaluation of the Exchange Correlation Potential on Graphics Processing Unit Clusters

On the Efficient Evaluation of the Exchange Correlation Potential on Graphics Processing Unit Clusters

Computational Characterization of the Substrate Activation in the Active Site of SARS-CoV-2 Main Protease

Achieving performance portability in Gaussian basis set density functional theory on accelerator based architectures in NWChemEx

Contact Info

Product

Resources

About