Skip to content
Charles Determan edited this page Mar 9, 2016 · 1 revision

Default Install

If you already have an OpenCL SDK installed you may already be good to go. You can try the basic installation method:

install.packages('gpuR')

This is will search your systems default library and include paths to find the OpenCL library and C++ API header.

Custom Install

If they are not in the standard locations (such as a user specific install) you can optionally define two environmental variables to explicitly point the package.

If in R (or using Rstudio) you can use the Sys.setenv function to define your environmental variables. The two are OPENCL_LIB (32 or 64 with Windows) and OPENCL_INC referring to the library and include directories respectively.

For example, on a Linux system where the AMD SDK has been installed you would do something like the following:

Sys.setenv(OPENCL_LIB = "/opt/AMDAPPSDK-3.0/lib/x86_64/")
Sys.setenv(OPENCL_INC = "/opt/AMDAPPSDK-3.0/include/")
install.packages("gpuR")