Vector Operations for Accelerating Expensive Bayesian Computations – A Tutorial Guide

Many applications in Bayesian statistics are extremely computationally intensive. However, they often inherently parallel, making them prime targets for modern massively parallel processors. Multi-core and distributed computing is widely applied the community, however, very little attention has been given to fine-grain parallelisation using single instruction multiple data (SIMD) operations that available on most CPUs. In this work, we practically demonstrate, standard programming libraries, utility of SIMD approach several topical applications. Using C language, show can improve single-core floating point arithmetic performance by up a factor 6× compared scalar code more than 25× with optimised R code. Such improvements multiplicative any gains achieved through multi-core processing. We illustrate potential accelerating computations provide reader techniques exploiting processing environments.