[R] Java vs. C speed

Peter Muhlberger p.muhlberger at ttu.edu
Fri May 11 17:59:51 CEST 2007

I tried writing a textual comparison program in R, but found that it
is too slow for my purposes.  I need to make about 145 million
comparisons of the word patterns in pieces of text.  I basically
compare vectors that contain count data on a multitude of words and
find ones that are similar to others (145million X 2 %in% comparisons,
a couple million multinomial estimates).  The best I could do in R
would take about 9 hours of computation time (a matrix solution bogs
down because of the size of the matrix:  10k by 17k; a looping
solution takes too long).  I've been looking into whether a Java or C
routine would be the best alternative.  Seems to be quite a debate on
the web about which is faster, and I don't know how much of that
debate is up to date.  Any impressions out there regarding whether
Java or C would be faster for this application and by how much?  I'd
have to learn quite a bit to implement either so I'd rather just work
on one.

More information about the R-help mailing list