Parallella Community

Looks great, thanks.

Do you know where in the R's source would the math library be, and how to interface with it (in terms of required data structures, etc.)?

Yaniv

I would look into the R source tree somewhere here

R-2.15.1/src/modules/lapack/

The configure flags needed to use an external BLAS are probably something like

./configure --enable-BLAS-shlib --enable-R-shlib .....

(taken from this nice document)
http://www.rochester.edu/college/gradst ... g/BLAS.pdf

censix,
This is really awesome. R is a big deal! Do you know how much speed up we expect from an optimized BLAS implementation? (I am sure someone has already done the benchmark using GPUs). Next up..Octave.
Andreas

Andreas

good to know this is useful to others. It certainly is to me.
I personally dont have any benchmark numbers for GPU accelerated R, but am sure they are out there somewhere.
Octave yes ... or Python...we just need to find an appropiate Guru.

Soren
http://censix.com

Thank you for this guide. It's fantastic to see R running on Parallella (big yay)

However, R is a high memory demanding environment, especially when dealing with big chunk of data or large simulation. Do you have any suggestion on how to overcome this?

I wonder if in the future Parallella will support extendable memory banks?

Cheers

So, in order to effectively accelerate R on Parallella, we'd need a port for the BLAS library for the Epiphany. Good to know.

I'm no expert on this but it seems BLAS is a de facto standard and there are many implementations. R comes with a basic one bundled in. This article explains more, but pours cold water on the idea it is a universal panacea, as not all R routines can make use of the library. At that point you either start to rewrite R, or make use of its ability to call external stuff you've written separately.

There is some hope though.

Tim

Hi

all that is needed is a BLAS library that uses the epiphany, i.e. through OpenCL. In another post I have suggested that we could use the "viennacl" for this.

Some clarifications:
This BLAS library does in no way have to be specific to R !
In addition, with a few additional ./configure flags, R can be compiled in a way so that it uses an external (epiphany driven) BLAS library. In that scenario, again, R does not care how the BLAS library does the calcuations.

For comparison:

1) Lets say you have an nvidia graphics card and install the cuda/opencl capable dirvers for it. These drivers come with a library called "cuBLAS" which implements BLAS routines using the nvidia GPU though cuda.

2) Lets say you have an ati graphics card and install the opencl drivers. Then again, these dirvers come with a library called "libclamdblas" that implements BLAS routines using the ati GPU.

In both scenarios, 1) and 2) it is possible to configure R so that it will use 'cuBLAS' or 'libclamdblas' instead of its own built in BLAS library, therby using the GPUs to accelerate almost all basic R calculations!

So coming back to the original point. If someone manages to create a standard BLAS implementation of as many levels as possible (1-3) using the epiphany processor ... we have won the game! Like I said. Viennacl look like a very promising starting point.