[R-sig-hpc] was: 'Experience using gputools-package' on the r-sig-finance list

[Vince Fulco]

*This is a response to a post on GPUs on the R-sig-finance list.  Dirk
E. appropriately requested the topic be swung over here...

##
<Original edited question from Gero Schwenk>
Now my question: Does anybody have experience using this package or GPU-
resp. parallel-processing for exploration? Or do you use other
environments, resp. approaches?

<Brian Peterson's highly edited response>
I know firms in finance that are making extensive use of different GPU
architectures.  They are *all* doing a lot of low level C programming to
do it, using the API directly in many cases, or reference
implementations of linear and matrix algebra packages tuned for the GPU
they've chosen.  I appreciate the approach if you have the resources to
engage in it.
##

Low level C is not necessarily required for the GPU.  Besides Josh
Buckner's phenomenal early strides, Andreas Klockner at Brown has done
extensive work with PyCuda, Nicolas Pinto and his students @ MIT has
done yeoman's work bringing educational tools to the fore and blending
CUDA/PyCUDA and select CUDA developers are pretty far along building
out Thrust, a C++ like library.  These make life a whole lot easier
for those either not used to programming closer to the metal or
getting there progressively.  That having been said, I am partial to
an R path as I am sure many are.

http://mathema.tician.de/software/pycuda

http://sites.google.com/site/cudaiap2009/

http://code.google.com/p/thrust/


HTH, V.

-- 
Vince Fulco, CFA, CAIA
612.424.5477 (universal)
vfulco1 at gmail.com

 A posse ad esse non valet consequentia

“the possibility does not necessarily lead to materialization”