Java stack performance

L

light2

Recruit
I tried comparing the stack efficiency of a variety of programming languages when computing the Fibonacci sequence using the naïve recursive technique just for fun. The code is mostly the same in all languages; however, I'll provide a java version:
Java: 
public class Fib {
 public static int fib(int n) {
  if (n < 2) return 1;
  return fib(n-1) + fib(n-2);
 }

 public static void main(String[] args) {
  System.out.println(fib(Integer.valueOf(args[0])));
 }
}

So, using this approach with input 40, I obtained the following timings:
Java: 
C: 2.796s
Ocaml: 2.372s
Python: 106.407s
Java: 1.336s
C#(mono): 2.956s

They were taken on a twin core Intel processor on an Ubuntu 10.04 box with the versions of each language accessible in the official repository.

I understand that functional languages like ocaml suffer from the slowdown that results from treating functions as first order citizens, and I have no problem explaining CPython's running time because, as I follow this article, it's the only interpreted language in this test, but I was impressed by the java running time, which is half of the c for the same algorithm! Would you put this down to JIT compilation?

How would you interpret these findings?
 
Last edited by a moderator:
Yup , probably non optimised compiler option specified, like debug build instead of release.

Can you check and retest?
 
“JIT compilation” would only do the stuff that C complier is already doing ( given you use the correct compiler flags ). So there’s no way it could be the reason behind the (relatively) slow performance in you tests. As others mentioned, you might have to use the correct compilation flags for C to measure it’s true performance.
 
Back
Top