Compiling R package in parallel on multiple platforms - r

I am developing a new R package using Rcpp.
We reached the point were compile times become significantly long.
So I was wondering how to compile an R package in parallel.
We develop on Linux, OSX and Windows for max compatability and so far I was only able to answer my question for Linux (sudo MAKE="make -j8" R CMD INSTALL package).
Can some one tell me how to do the same thing on a Windows and OSX system?
thanks
Cedric

Related

Incompatible packages with MRO on anaconda

I have some complicated code in R that uses the neuralnet library for some computations.
Sorrily, I'm new to R and I'have less than a week to obtain some results using the existing code, which take quite a while in the processors I have to my disposal.
My idea is to implement the code via microsoft R open (MRO), which could accelerate the computations, but I haven't been able to install the neuralnet library via anaconda (I prefer anaconda because it's simple and allows to create environments easily). The installation goes with "Solving environment" forever.
Is there a way to install these libraries to be compatible with MRO on anaconda?? should i desist of using anaconda for this task?

mxnet packages with Rstudio on Virtual machine

I want to install r studio on virtual machine to work with mxnet package.
Here is how to configure Rstudio on microsoft azzure linux virtual machine:
http://moresi.de/posts/2016-04-02-setting-up-r-studio-server-on-microsoft-azure.html
And there is how to install mxnet packages for linux:
https://mxnet.incubator.apache.org/get_started/install.html
My question is: If I will install R-studio and Mxnet package following the step in this 2 tutorial will I be able to use mxnet package from R Studio workspace?
Another question is: It is possible to install all availible r packages in this configuration from R studio workspace?
I know that maybe this is a trivial question, but i am completly new to microsoft azzure, and spend a lot of time looking for an answer on web. Its seems that there wasn't any exemple of using mxnet packages with R on cloud.I would be very grateful even for some tips : )
Yes and no. Yes, you can run mxnet through Rstudio server, I currently do this on my AWS linux instance and login via the browser.
No, you cannot install everything from Rstudio and will have to ssh in and compile/install mxnet and a few other things. Unless you want the CPU mxnet instead of the GPU verion, that might be precompiled. The GPU installation process is a nightmare, I wouldn't recommend unless you have a moderate level of linux experience.

Install old version of R on Windows 10

How do I install an old version of R on a window 10 64 PC.
I am currently running R 3.2.3 with RStudio without any problems. However a package I want to use (GSIF) only works on R2.15. I have found that it is possible to easily get RStudio to work with different R versions and I have found and downloaded the old version of R I want ( R-2.15.1-w.tar.gz) from the CRAN archives. My problems is cannot work out how to install the R-2.15.1-w.tar.gz package on a windows machine or in fact even if it is the right package for windows. All the instruction I can find seem to relate to Linux.
Any guidance greatly appreciated.
You can hold down the Control key during the launch of RStudio you can cause the R version selection dialog to display at startup.
https://support.rstudio.com/hc/en-us/articles/200486138-Using-Different-Versions-of-R

R doMPI installation not working - requires building from source?

I'm trying to install the doMPI package in R.
Apparently there are no binaries available for the 3.x version?
Do i need to build it from source?
http://cran.r-project.org/web/packages/doMPI/
<>
The goal is to run parallel processing with caret on a windows machine.
CRAN doesn't build binaries of doMPI for Mac OS X or Windows because it depends on the Rmpi package, and it doesn't build binaries for Rmpi because it depends on MPI libraries which don't come by default on those platforms. Some people have suggested that I declare Rmpi to be a suggested package to work-around this issue, but in fact, doMPI really does depend on Rmpi, so it always seemed like an odd thing to do. The way I see it, if you're able to build Rmpi from source, you'll have no trouble building doMPI from source.
So yes, you have to build it from source, but the bigger problem is to build Rmpi from source, unless you're using a Linux distribution like Debian that distributes both Rmpi and doMPI as binary deb packages.
But if you just want to run caret in parallel on a Windows machine, the normal solution is to use the doParallel package using a PSOCK cluster. People have trouble with that as well, but at least installation of the packages is easy since there are binary packages available for doParallel on CRAN.

Compiling R packages for a memory-profiling configuration

Say I have two R installations. Same version but one built for Ubuntu Linux (locally) with memory profiling and the other without. Do I need to compile the installed packages for each separately?
Short answer is 'Nope' as packages are unaffected by this optional feature in the R engine.
If you have particular questions concerning R use on Debian and Ubuntu, come to the r-sig-debian list.

Resources