Skip to content

razorx89/nmfgpu4R

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

nmfgpu4R

R binding for the nmfgpu library

About

This package is a binding of the nmfgpu library for the R language. By default the package installs the relevant release of nmfgpu from github into the package directory. If you want to use a custom compiled version, then define the NMFGPU_ROOT environment variable. Please visit the nmfgpu project page or read the package documentation for further information.

Citation

@Article{koitka-friedrich:2016,
  author       = {Sven Koitka and Christoph M. Friedrich}, 
  title        = {{nmfgpu4R}: {GPU}-Accelerated Computation of the Non-Negative Matrix Factorization ({NMF}) Using {CUDA} Capable Hardware}, 
  journal      = {The R Journal},
  year         = 2016,
  volume       = 8,
  number       = 2,
  pages        = {382--392},
  month        = december,
  url          = {http://journal.r-project.org/archive/2016-2/koitka-friedrich.pdf}
}

Licence

This library is distributed under the terms of the General Public Licence Version 3 (GPLv3).

GPLv3 Logo

Installation

Prerequisites

  • Compiled version of nmfgpu
  • CUDA Toolkit (version depends on nmfgpu compilation)
  • CUDA capable computation device (minimum: Kepler 3.0)
  • Windows only: If you build from source, then you have to install RTools

Instructions

The latest stable version of nmfgpu4R can be installed from CRAN using

install.packages("nmfgpu4R")

When using devtools, the latest git version can be installed using:

install_github("razorx89/nmfgpu4R")

Known issues

  • Currently nmfgpu is only prebuild for Windows and Linux using the CUDA 7.5 toolkit. If you are using Mac OS X or CUDA 7.0 then you have to compile nmfgpu according to the installation instructions.
  • Installation/Loading of the package fails if the nmfgpu library cannot be loaded
    • Do you have created a NMFGPU_ROOT environment variable?
    • Do you have all required dependencies installed? (CUDA, Visual Studio Runtime on Windows platforms, ...)
    • Important: Loading of the nmfgpu library can fail even if it can be found in the filesystem. Such an error can occur if one of the required dependencies cannot be loaded (e.g. CUDA runtime or cuBLAS from the CUDA Toolkit)