OpenMP in VASP: Threading and SIMD

Abstract

The Vienna Ab initio Simulation Package (VASP) is a widely used electronic structure code that originally exploits process‐level parallelism through the Message Passing Interface (MPI) for work distribution within and across nodes. Architectural changes of modern parallel processors urge programmers to address thread‐ and data‐level parallelism as well to benefit most from the available compute resources within a node. We describe for VASP how to approach for an MPI + OpenMP parallelization including data‐level parallelism through OpenMP SIMD constructs together with a generic high‐level vector coding scheme. We can demonstrate an improved scalability of VASP and more than 20% gain over the MPI‐only version as well as a 2× increased performance of collective operations using the multiple‐endpoint MPI feature. The high‐level vector coding scheme applied to VASP's general gradient approximation routine gives up 9× performance gain on AVX512 platforms with the Intel compiler.

3D Porous Metallic Boron Carbide Crystal Structure with Excellent Ductility

“…We have carried out calculations on the basis of density functional theory [24] as implemented in the Vienna ab initio simulation package. [25] The projector-augmented wave [26] potential is used to study the electron-ion interactions with a cutoff energy of 520 eV. We have applied Perdew-Burke-Ernzerhof (PBE) functional [27] to deal with the exchange-correlation potential energy and hybrid Heyd Scuseria Ernzerhof (HSE06) functional [28] is used to calculate the accurate results of the electronic band structure.…”

Tuning the Properties of Tetracene‐Based Nanoribbons by Fluorination and N‐Doping

“…Calculations are based on density functional theory (DFT) as implemented in Vienna Ab initio Simulation package (VASP) . A vacuum space is introduced with a size of 13 Å for straight chain and 15 Å for zigzag and 9.33 Å for an armchair chain in a perpendicular direction to the chains for reducing the interactions with the periodic images.…”

Special Issue on Emerging Architectures in Computational Chemistry

“…Peng et al present their massively parallel implementation of CCSD(T) within MPQC and demonstrate excellent performance on a wide range of hardware, including hybrid systems. Wende et al describe their hybrid OpenMP + MPI implementation of VASP with special attention to efficient use of SIMD vector units and demonstrate impressive performance gains compared to MPI‐only and compiler‐ based vectorization.…”

“…Peng et al [4] present their massively parallel implementation of CCSD(T) within MPQC and demonstrate excellent performance on a wide range of hardware, including hybrid systems. Wende et al [5] describe their hybrid OpenMP + MPI implementation of VASP with special attention to efficient use of SIMD vector units and demonstrate impressive performance gains compared to MPI-only and compiler-based vectorization.Software developers are living in the exciting times. As we transition to post-Moore era, more open and rapid advances in hardware architectures will require development of new software solutions.…”