kmcudaR: 'Yingyang' K-Means and K-NN using NVIDIA CUDA

K-means implementation is base on "Yingyang K-Means: A Drop-In Replacement of the Classic K-Means with Consistent Speedup". While it introduces some overhead and many conditional clauses which are bad for CUDA, it still shows 1.6-2x speedup against the Lloyd algorithm. K-nearest neighbors employ the same triangle inequality idea and require precalculated centroids and cluster assignments, similar to the flattened ball tree.

Version: 1.1.0
Depends: R (≥ 3.3.2)
Imports: Rcpp (≥ 0.12.9)
LinkingTo: Rcpp, RcppEigen
Suggests: testthat
OS_type: unix
Published: 2019-03-22
Author: Vadim Markovtsev, Charles Determan
Maintainer: Charles Determan <cdetermanjr at>
License: Apache License (≥ 2.0) | file LICENSE
NeedsCompilation: yes
SystemRequirements: CUDA 8.0 tookit, OpenMP 4.0 capable compiler
CRAN checks: kmcudaR results


Reference manual: kmcudaR.pdf
Package source: kmcudaR_1.1.0.tar.gz
Windows binaries: r-devel: not available, r-release: not available, r-oldrel: not available
OS X binaries: r-release: not available, r-oldrel: not available
Old sources: kmcudaR archive


Please use the canonical form to link to this page.