WideLM: Fitting many skinny linear models to a single data set

WideLM fits large numbers of linear models to a fixed design matrix and response. An Nvidia GPU is used to accelerate model fitting, with the CUDA API enabling parallel execution of multiple individual fits. The individual models must have modest size. The package implements a chunking interface to the GPU kernel, allowing the caller to stream model specifications. Hence the number of fits to be performed is essentially unlimited.

Version: 0.1-1
Depends: Rcpp
LinkingTo: Rcpp
Published: 2012-02-17
Author: Mark Seligman, with contributions from Chris Fraley.
Maintainer: Mark Seligman <Mark.Seligman at insilicos.com>
License: GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]
URL: http://insilicos.com/products/widelm
NeedsCompilation: yes
SystemRequirements: Nvidia's CUDA toolkit (release >= 4.1); Linux operationg system; gcc (release <= 4.5).
Materials: README
In views: HighPerformanceComputing
CRAN checks: WideLM results

Downloads:

Reference manual: WideLM.pdf
Package source: WideLM_0.1-1.tar.gz
OS X binary: not available, see check log.
Windows binary: not available, see ReadMe.