In this chapter, you have been shown how to write your own parallel routines and make them callable directly from R programs. You have also learnt how to create your own suite of such parallel routines, and turn them into an R package that you can then reuse in other R programs. The SPRINT package has been introduced, and its architecture examined to show how you can organize your own such package, or instead, use the SPRINT package itself and include your own parallel routines within it.
Finally, the chapter has demonstrated how you can use such an MPI-based R package on a supercomputer to exploit hundreds, and potentially thousands, of cores to dramatically increase the performance of your R programs.
In the next chapter, we switch our attention from exploiting the world's most expensive supercomputers, to the admittedly much easier-to-access supercomputer lurking in your own laptop and desktop, the Graphics Processing Unit (GPU). We will explore how to make use of the GPU's particular...