Ive heard that Java is slow, and is poorly suited to numerical computation. Are there faster C/Fortran implementations of the algorithms in the toolkit?
A This question is a bit difficult to answer. Performance comparisons between languages are fraught with difficultly, and religious wars seem to break out over the issue of which is faster, and by how much. We consider that Java is roughly as fast as C and Fortran for most numerical tasks. On some specific tasks (in particular some Linear Algebra tasks) that require non-jagged arrays or peculiar memory management, Java may perform significantly worse, but we do not believe that the algorithms implemented in the toolkit would run much faster if written in another language. We are happy to live with the possibility of a small drop in performance for the other benefits that Java has. If you are skeptical, try out the toolkit yourself and see if it is fast enough for your needs.
Related Questions
- Ive heard that Java is slow, and is poorly suited to numerical computation. Are there faster C/Fortran implementations of the algorithms in the toolkit?
- Are server-based calculations like ASP.NET or Java faster than client-based calculations like HTML or Flash?
- What is a Java Evolutionary Computation Toolkit?
