Does the integration of Intel® Parallel Studio XE 2013 for Linux* and R lead to significant performance improvements?

Question

I am running some resource intensive computations in R. I use for-loops, bootstrap simulations and so on. I have already integrated the Intel® Math Kernel Library for Linux* with R and it seems that this has lead to a significant improvement in computation times. I am now thinking about integrating Intel® Parallel Studio XE 2013 for Linux* and R. This means passing the different compilers that ship with it to R:

(1) Will the integration of Intel® Parallel Studio XE 2013 for Linux* and R lead to significant performance improvements?

(2) Could you give some examples in which situations I would have a benefit?

Thank you!

Answer 1

Very rough order of magnitude:

• parallel / multi-core BLAS such as the MKL will scale sublinearly in the number of cores but only for the parts of your operations that are actually BLAS calls ie not for your basic "for-loops, bootstrap simulation and so on"

• byte-compiling your R code may give you up to a factor of two, maybe three

• after that you may need heavier weapons such as for example Rcpp which can give 50, 70, 90-fold speedups on code involving "for-loops, bootstrap simulation and so on" which why it is eg so popular with the MCMC crowd

• similarly, the Intel TBB and other parallel tricks will require rewrites of your code.

There is no free lunch.