Quantum chemistry codes must make efficient use of parallel computing resources in order to reduce execution time.  This is true for simulating both large and small molecular systems, as parallel hardware is now unavoidable.  Parallelism in hardware exists at multiple levels, with SIMD parallelism (vectorization) being particularly difficult to exploit for many types of scientific codes.  In quantum chemistry, the computation of electron repulsion integrals is a component that suffers from very low computational efficiency because of its inability to exploit SIMD parallelism provided by the processor cores due to the recursive nature of the algorithms.  We will discuss the vectorization of the Obara-Saika recurrence relations in a new software library called Simint, which has been designed from scratch to efficiently utilize SIMD on x86-based and other current CPUs.

Vectorization and high performance quantum chemistry software

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.