There is a report about performance of GPU-accelerated matrix computation with M/R.
http://mrcl.googlecode.com/svn/trunk/report/ (korean) In a nutshell, they performed matrix multiplication using Map/Reduce (block algorithm for distributed computing) and GPU acceleration technology. GPU technology was used for each local computation. And, it implying that no improvement was made, because we can see a obvious improvement of GPU acceleration when input is large. On Fri, Feb 12, 2010 at 1:35 PM, Edward J. Yoon <[email protected]> wrote: > I just found this project -- http://code.google.com/p/mrcl/ > > CUBLAS is a BLAS library ported to CUDA, which enables the use of fast > computing by GPUs without direct operation of the CUDA drivers. > > -- > Best Regards, Edward J. Yoon @ NHN, corp. > [email protected] > http://blog.udanax.org > -- Best Regards, Edward J. Yoon @ NHN, corp. [email protected] http://blog.udanax.org
