FULL PRODUCT VERSION :
java version "1.5.0"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0-b64)
Java HotSpot(TM) Server VM (build 1.5.0-b64, mixed mode)

ADDITIONAL OS VERSION INFORMATION :
Linux RHAS 3.0

A DESCRIPTION OF THE PROBLEM :
Many monte carlo simulations involve making extensive use of the Math.nextGaussian() method. The performance of code based on this seems to have regressed sharply when running on Linux.

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
Compile the attached code sample then run on a linux machine. I tried running with the -server flag and a loop size (the arg passed to the application) of 50.

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
bash-2.05b$ $HOME/linux/j2sdk1.4.2_05/bin/java -server Gaussian 50
1358.7012578003387
Elapsed time = 20251 msecs

Elapsed time to run the test on JDK 1.4.2 was about 20 seconds.
ACTUAL -
bash-2.05b$ /home/murrap/linux/jdk1.5.0/bin/java -server Gaussian 50
-8949.363088040205
Elapsed time = 31138 msecs

Elapsed time to run the test on JDK 5 was about 31 seconds.

REPRODUCIBILITY :
This bug can be reproduced always.

---------- BEGIN SOURCE ----------
import java.util.Random;

public class Gaussian {
        public static void main(String[] args) {
                int loopSz = Integer.parseInt(args[0]);
                double deviation = 0.0;
                Random rnd = new Random();

                long startTime = System.currentTimeMillis();
                for (int i=0; i<loopSz; i++) {
                        double total = 0.0;
                        for (int j=0; j<1000000; j++) {
                                total += rnd.nextGaussian();
                        }
                        deviation += total;
                }
                long endTime = System.currentTimeMillis();
                System.out.println(deviation);
                System.out.println("Elapsed time = " + (endTime-startTime) + " msecs");
        }
}
---------- END SOURCE ----------

Release Regression From : 1.4.2_05
The above release value was the last known release where this
bug was known to work. Since then there has been a regression.
###@###.### 10/14/04 21:17 GMT