Udaya B. Kogalur
2018-May-03 02:32 UTC
[R-pkgs] randomForestSRC 2.6.0 - Our Fastest Build Yet!
Dear useRs: The latest build of randomForestSRC (2.6.0) includes significant improvements in speed when a number of user options including general ensembles, forest weights, importance and partial plots are in play. We managed to remove some bottlenecks that were occurring between the OpenMP threads in larger-n and larger-p situations. This was especially pronounced in survival families. As per the previous release, we believe the binaries provided by CRAN for Windows, MacOS, and Linux support OpenMP out-of-the-box. If you are running a serial version of our package, we strongly recommend an upgrade path that will allow you to run the parallel version. This upgrade path may be as simple as installing the CRAN binaries. But, on some systems, it may also involve a scenario where you will need to install an OpenMP capable C-compiler such as GCC, and then compile and install randomForestSRC from source. Documentation and more can be found at: https://kogalur.github.io/randomForestSRC/ Thank you. Udaya Kogalur [[alternative HTML version deleted]]