Efficient Implementation of Many-Body Quantum Chemical Methods on the Intel® Xeon Phi Coprocessor

This paper presents the implementation and performance of the highly accurate CCSD(T) quantum chemistry method on the Intel^® Xeon Phi coprocessor within the context of the NWChem computational chemistry package. The widespread use of highly correlated methods in electronic structure calculations is contingent upon the interplay between advances in theory and the possibility of utilizing the ever-growing computer power of emerging heterogeneous architectures. We discuss the design decisions of our implementation as well as the optimizations applied to the compute kernels and data transfers between host and coprocessor. We show the feasibility of adopting the Intel^® Many Integrated Core Architecture and the Intel Xeon Phi coprocessor for developing efficient computational chemistry modeling tools. Remarkable scalability is demonstrated by benchmarks. Our solution scales up to a total of 62560 cores with the concurrent utilization of Intel^® Xeon^® processors and Intel Xeon Phi coprocessors.

I. Introduction

Many aspects of computational chemistry require accuracies that can only be met by a very limited class of methods that can properly account for the instantaneous interactions between electrons or correlation effects in molecules. In particular, high accuracies are needed to bridge the gap between theory and experiment in a profound way. Among the many methods that describe correlation effects, the coupled cluster (CC) formalism [2], [7]–[9], [39], [42] has evolved into a widely used and very accurate method for solving the electronic Schrödinger equation. Compared with other formalisms such as perturbative methods or approaches based on the linear expansion of the wavefunction, the main advantage of the CC formalism lies in the fact that the correlation effects are very efficiently encapsulated in the exponential form of the wavefunction. This enables one to describe the correlated motion of one or two (or more) electron pairs simultaneously. A simple consequence of this feature is the size-extensivity of the resulting energies, or equivalently a proper scaling of energy with the number of electrons. This feature is essential for describing complex chemical processes.