GPU architecture is different enough from CPU architecture that you don't need 10s of GPUs to see a performance benefit over today's, say, 8 core CPUs. Lots of GPUs now give you a (relatively cheap) "supercomputer" -- look up nVidia's Tesla marketing mumbo jumbo. One GPU still gives you a 'heckuva job'.
>From Wikipedia's GPU page, speaking on modern general purpose GPUs: http://en.wikipedia.org/wiki/Graphics_processing_unit "Typically the performance advantage is only obtained by running the single active program simultaneously on many example problems in parallel using the GPU's SIMD architecture[11]. However, substantial acceleration can also be obtained by not compiling the programs but instead transferring them to the GPU and interpreting them there[12]. Acceleration can then be obtained by either interpreting multiple programs simultaneously, simultaneously running multiple example problems, or combinations of both. A modern GPU (e.g. 8800 GTX) can readily simultaneously interpret hundreds of thousands of very small programs." The first sentence, you can imagine, applies to some a lot of matrix work. There are BLAS libraries for some GPUs (e.g. CUDA BLAS). You can probably imagine having R use it. Ahmed El Zein has a poster about his presentation "Performance Evaluation of the NVIDIA GeForce 8800 GTX GPU for Machine Learning" that gives some more interesting info. -Mose On Tue, Nov 18, 2008 at 10:56 PM, Prof Brian Ripley <[EMAIL PROTECTED]> wrote: > On Tue, 18 Nov 2008, Emmanuel Levy wrote: > >> Dear All, >> >> I just read an announcement saying that Mathematica is launching a >> version working with Nvidia GPUs. It is claimed that it'd make it >> ~10-100x faster! >> http://www.physorg.com/news146247669.html > > Well, lots of things are 'claimed' in marketing (and Wolfram is not shy to > claim). I think that you need lots of GPUs, as well as the right problem. > >> I was wondering if you are aware of any development going into this >> direction with R? > > It seems so, as users have asked about using CUDA in R packages. > > Parallelization is not at all easy, but there is work on making R better > able to use multi-core CPUs, which are expected to become far more common > that tens of GPUs. > >> Thanks for sharing your thoughts, >> >> Best wishes, >> >> Emmanuel > > PS: R-devel is the list on which to discuss the development of R. > > -- > Brian D. Ripley, [EMAIL PROTECTED] > Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ > University of Oxford, Tel: +44 1865 272861 (self) > 1 South Parks Road, +44 1865 272866 (PA) > Oxford OX1 3TG, UK Fax: +44 1865 272595 > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
