Introduction

This package performs microbenchmarking for determining the run time performance of aspects of the R programming environment and packages relevant to high-performance computation. The benchmarks are divided into three categories: dense matrix linear algebra kernels, sparse matrix linear algebra kernels, and machine learning functionality. The top-level benchmark functions covering the three categories are RunDenseMatrixBenchmark, RunSparseMatrixBenchmark, RunMachineLearningBenchmark.

Installation

The companion data package RHPCBenchmarkData contains the sparse matrix files needed by the sparse matrix benchmarking function.

Installation of the benchmarking and companion data packages is trivial with the use of the install.packages function

Examples

See the vignette named 'vignette' for a more detailed explanation of the package and additional examples. New benchmarks can be specified using the classes DenseMatrixMicrobenchmark, SparseMatrixMicrobenchmark, and ClusteringMicrobenchmark; see the vignette and the object documentation for each of these classes to learn how new microbenchmarks can be constructed.