Bandicoot: A Templated C++ Library for GPU Linear Algebra

TL;DR

Bandicoot is a C++ library that enables GPU-accelerated linear algebra computations, offering significant speedups over CPU-based Armadillo, with a focus on ease of integration and compile-time optimization.

Contribution

It introduces a GPU-enabled linear algebra library that closely integrates with existing Armadillo codebases, utilizing template meta-programming for optimized performance.

Findings

Bandicoot achieves substantial speedups over Armadillo CPU computations.

The library maintains user-friendly interface similar to Armadillo.

Empirical results demonstrate effective GPU acceleration.

Abstract

We introduce the Bandicoot C++ library for linear algebra and scientific computing on GPUs, overviewing its user interface and performance characteristics, as well as the technical details of its internal design. Bandicoot is the GPU-enabled counterpart to the well-known Armadillo C++ linear algebra library, aiming to allow users to take advantage of GPU-accelerated computation for their existing codebases without significant changes. Exploiting similar internal template meta-programming techniques that Armadillo uses, Bandicoot is able to provide compile-time optimisation of mathematical expressions within user code, leading to more efficient execution. Empirical evaluations show that Bandicoot can provide significant speedups over Armadillo-based CPU-only computation. Bandicoot is available at https://coot.sourceforge.io and is distributed as open-source software under the permissive Apache 2.0 license.