Matrix-Benchmarks

Package to benchmark different matrix types and operations in Pharo

How to install the package?

1. Go to Playground (Ctrl+O+W) in your Pharo image.
2. Execute the following Metacello script (select it and press Do-it button or Ctrl+D):

EpMonitor disableDuring: [ 
	Metacello new
		baseline: 'MatrixBenchmark';
		repository: 'github://rakki-18/Matrix-Benchmarks/src';
		load ]

In order to supress load warnings, you may prefer

[ Metacello new
  baseline: 'MatrixBenchmark';
  repository: 'github://rakki-18/Matrix-Benchmarks/src';
  load ] 
on: MCMergeOrLoadWarning 
do: [:warning | warning load ]

Running the Benchmark GUI

MBApplication new start

If you prefer not to run the User-Interface, then you can run the benchmarks programatically using

Running Benchmarks

To run benchmarks on a particular operation for a particular matrix size:

MBOperationName runBenchmarks: matrixSize

To run benchmarks for a particular operation for a set of matrix sizes:

MBOperationName runAllBenchmarks

The range of the matrix sizes on which the benchmarks are run can be customized by:

MBOperationName runAllBenchmarksFrom: fromMatrixSize to: toMatrixSize by: aNumber

To see all the available operations that can be benchmarked:

MBAbstract operations

Examples

• On the operation - Addition:

MBAddition runAllBenchmarks.

• On the operation - Element-wise Multiplication

MBElementWiseMultiplication runBenchmarks: 200.

• On the operation - Accessing cells of a matrix for the matrix sizes {50,100,150,200}

MBAccess runAllBenchmarksFrom: 50 to: 200 by: 50 

Matrix Implementations covered

• Array2D
• PMMatrix
• PMSymmetricMatrix
• AIColumnMatrix
• AIRowMatrix
• AISymetricMatrix
• AIDiagonalMatrix
• PMJacobiTransformation
• PMLinearEquationSystem
• PMLUPDecomposition

Matrix Operations covered

• Building an empty Matrix
• Building a matrix filled with zeros
• Accessing matrix cells
• Addition of matrices
• Subtraction of matrices
• Matrix Multiplication
• Element-wise Multiplication
• Multiplication with a number
• Division with a number
• Comparing matrices
• Finding determinant
• Finding inverse
• Transpose of a matrx
• Trace of a matrix
• Finding eigenvectors
• Gaussian Elimination
• LU Decomposition
• QR Decomposition
• Rank factorization
• Singular Value Decomposition
• Strassen Multiplication

How to depend on the project?

If you want to add a dependency to your project, include the following lines into your baseline method:

spec
  baseline: 'MatrixBenchmark'
  with: [ spec repository: 'github://rakki-18/Matrix-Benchmarks/src' ].

Check out Baselines to learn more about baselines.

Running Benchmark UI

• Select the matrix operation.
• Select the matrix sizes parameters and execute
• Outputs the benchmark results and plots.