I'm not familiar with any. Typically J performs well with arrays as you would expect. It usually performs as well or better than most C or FORTRAN coders as J probably avoids many of the indexing issues that are common before people go to a lot of trouble to optimize their code. Obviously, once C or FORTRAN are seriously optimized they would probably match or exceed J performance as the J engine is written in C. But most projects do not justify such serious optimization efforts. There is interpreter overhead for scalar processing in J that would not be in compiled code, but J people who are well versed in thinking in groups of data avoid most loops, particularly innermost loops, which is where the interpreter overhead becomes significant.
As you may have noticed in some code examples I sent you that I never had to resort to writing loops. The code is interpreted once, then it is executed by C code that Roger Hui and others have gone to a lot of trouble to write highly optimized. Of course, since J is close to mathematical notation, given a mathematical description, code can be developed in J quite rapidly. On Wed, Aug 5, 2009 at 12:35 PM, Jonathan <[email protected]> wrote: > Are there any relatively recent benchmarks of J performance vs. other > languages. I dont see J represented in the Computer Language Shootout > (although kudos to those who made a strong showing over at Rosetta Code). > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
