- Version: 0.2-0
- License: BSD 2-Clause
- Author: George Ostrouchov
- Project home: https://github.com/RBigData/launchr
- Bug reports: https://github.com/RBigData/launchr/issues
Provides ease of use and configuration components for controlling a distributed set of R instances on a cluster from a remote R session (e.g. running on a laptop with RStudio). The intent is to provide an easy-to-use interactive SPMD-style development platform for distributed parallel programming in R. Other parallel approaches (multicore, multithreading, and GPU) are compatible with a launchr() distributed server session, which provides a means of specifying the number of R sessions per node, potentially leaving cores available for further parallel shared-memory applications.
Leave BIG data on BIG platforms and control its analysis using R codes on your laptop.
A default script for launching on a PBS-managed cluster is constructed. A means of editing the script by an R function allows for custom modifications necessary for local configurations. An example modification is provided in the inst/demo directory. Please let us know about your configurations to help us provide an easier experience for everyone.
For understanding how port forwarding works, see stackexchange.
When producing mass graphics in parallel on the remote server file system, use \code{grDevices::dev.off()} to bypass a modified \code{dev.off()} that intends to hijack graphics back to the client.
A development version is maintained on GitHub:
remotes::install_github("RBigData/launchr")