1

I understand that math library integration is the main tweak that sets MRO apart from the CRAN R distribution, and gives it a performance advantage.

I wonder if MRO (Microsoft R Open) binaries still have a performance advantage over Homebrew R, given that both Homebrew R and Microsoft R link their binaries against Accelerate (MKL) math libraries in macOS.

1

In my experience MRO is faster when using data.table and large datasets.

However, in Ubuntu, OS X, etc I found more convenient to use CRAN version and compile everything to avoid crashes, crossed wires, etc.

  • 1
    data.table performance variations are due to the fact that it needs to be compiled with openmp enabled in order to benefit from multi-core architectures. You can enable openmp in the r config when compiling data.table from CRAN source as well. It will do the trick. But don't forget to disable it afterwards. – retrography Jul 7 '17 at 4:55

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .