Regarding speed: How many non-zero elements?
What is the size of your input matrices? How long does it take to read the matrices without doing any multiplication? Your test matrices seem small for big sparse matrices. This sort of thing could be very useful. On Sun, Jun 26, 2011 at 1:47 PM, Vincent Xue <[email protected]> wrote: > Hi. > > I was wondering how useful an in memory sparse matrix multiplier would > be for mahout. In my current project I needed to multiply many large > sparse matrices but submitting hundreds of jobs to the cluster creates > too much overhead. > > I wrote up an implementation of sparse matrix multiplication using > threads which can multiply a 30,000 x 48,0000 matrix by its transpose > in about 5 minutes using 16 cores. Granted this matrix is composed > mostly of 1s, and -1s, (with about 16 elements per row), is this > considered fast? I have seen that my implementation is much faster > than iterating though a matrix naively and would like some input to > whether or not my 5 minute benchmark is by skewed. > > Many thanks for the input, > Vincent >
