[R] benchmarks

J. Andy Royle jar at oriole.er.usgs.gov
Wed Dec 2 14:43:03 CET 1998

While we're on the subject, I ran a few tests on several of our platforms 
here and although I don't claim that these are very sophisticated 
(or even general), I did find some results that were interesting to me 
(in determining my computing platform).  This is my summary:

(1) Matrix operation benchmark on a Sun Sparc 20:

     R : 69 seconds 
     S+: 43 seconds

(2) Looping benchmark doing lots of flops (same machine):

     R : 6 seconds  
     S+: 24 seconds

(3) Matrix operation benchmark on Pentium II - 400 running NT:

     R : 30 seconds
(4) Matrix operation benchmark on Pentium II - 400 running Linux:

     R : 11 seconds

I can't afford Splus on my PC so I can only compare R across OS's ,
but that was sufficiently useful for my purposes.  I found it interesting
that the performance was so different between NT and Linux, but I suspect
it has something to do with the preferred development platform.

The matrix operation benchmark basically involved generating 
a huge matrix of random numbers, doing  X'X and inverting that,
(and maybe something else, I forget).  The looping benchmark
did a small number of muliplications 10000 times or something.

I also did an I/O benchmark where S+ was 20 or 30% faster by
my recollection, but I think this was flawed.

J. Andy Royle, U.S. Fish and Wildlife Service -----------------------
Adaptive Management and Assessment Team : 11510 American Holly Drive, 
Laurel, MD 20708-4017. phone - 301-497-5673 : fax - 301-497-5706
email- jar at oriole.er.usgs.gov or andy_royle at fws.gov ----------------- 

r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list