Andreas Recktenwald <a.recktenwald <at> mx.uni-saarland.de> writes: > another option if you're using Linux AND an Intel processor would be > linking R against Intel MKL (Math Kernel Library). Under Linux you can
You do not have to "link" R against MKL. One simply builds and links R against _any_ BLAS implementation: reference BlAS, Atlas, GotoBLAS, OpenBLAS or MKL. (This assumes that you compiled R with the shared library option). I have a package (and vignette / unfinished paper) on R-Forge which implements a testing and timing comparison framework for this, using the fact that this is in fact plug&play -- look for 'gcbd' if interested. Lastly, to come back to the OP's initial question, you are of course right. Armadillo calls the LAPACK / BLAS routines; so it ends up making the same call as R does. (Rcpp)Armadillo makes a lot of other transformations faster, but the core multiplication is about the same as both R and Armadillo (and everybody else, apart from Eigen) "just" call the specialised BLAS. As one should. Dirk ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
