Fragment Molecular Orbital Method Adaptations for Heterogeneous Computing Platforms

Abstract

Modern electronic structure calculations are characterized by unprecedented complexity and accuracy. They demand the full power of high-performance computing and must be in tune with the given architecture for superior efficiency. Thus, it is desirable to enable their static and dynamic adaptations using some external software (middleware), which may monitor both system availability and application needs, rather than mix science with system-related calls inside the application.Building on the successful usage of the NICAN middleware with the computational chemistry package GAMESS, the work described in this paper links NICAN with the fragment molecular orbital (FMO) method to augment FMO with adaptive capabilities. Specifically, its fragment scheduling is performed, both statically and dynamically, based on current conditions within a heterogeneous computing environment. Significant execution time and throughput gains have been obtained with static adaptations, while the dynamic ones prevented FMO to abort calculations due to the insufficient memory available at the runtime.

The fragment molecular orbital method: theoretical development, implementation in GAMESS, and applications

“…In GAMESS-US, generalized distributed data interface was developed, 120 in which each subsystem (fragments, pairs, or triples) is calculated on a group of central processing unit (CPU) cores. 121 The number of groups is specified for each of these subsystem kinds. For multilayer FMO, the number of groups also depends on the layer.…”

Electron-correlated fragment-molecular-orbital calculations for biomolecular and nano systems

“…208 In contrast, Katouda et al 209 performed the efficient FMO-MP2(RI)/6-31G* calculations for large influenza protein in only 20 minutes on the K-computer by using 12 288 nodes and 86 016 processor cores, and further 196 608 processors were successfully used. Along a different direction from the massive parallelism, Talamudupula et al 210 attempted a heterogeneous computing for the FMO-HF and FMO-MP2 calculations with GAMESS. The OpenFMO program has been developed 20,21 for the massively parallel FMO-HF calculations, where its design policy was oriented to the efficient usage of a few hundred thousands of processors.…”