String,StringBuffer,StringBuilder

用String,StringBuffer,StringBuilder测试一下速度,发现在Netbeans,Jcreator里执行结果有一定的差异
StringBuffer应该是线程安全的,带有Sychronized的,而StringBuilder是以牺牲线程安全为代价的,在单线程情况下
应该会比StringBuffer来得快,在Netbeans得到了想要的结果.而在Jcreator里却是StringBuffer来得快一些..
这是为什么???
下附代码和结果:
/*
* To change this template, choose Tools | Templates
* and open the template in the editor.
*/

package comparestrings;
//import java.util.*;
/**
*
* @author lynn
*/
public class CompareStrings {

    /**
     * @param args the command line arguments
     */
    public static void main(String[] args) {
        // TODO code application logic here
         String str = "abcdefghijklmnopqrstuvwxyz";
         StringBuffer strBuffer = new StringBuffer("abcdefghijklmnopqrstuvwxyz");
         StringBuilder strBuilder = new StringBuilder("abcdefghijklmnopqrstuvwxyz");
        long time1= System.currentTimeMillis() ;
        for(int i= 0 ;i<10000;i++)
            str = str+"abcdefghijklmnopqrstuvwxyz";
        long time2= System.currentTimeMillis() ;
        long t1 = time2 - time1;
        
        time1= System.currentTimeMillis() ;
        for(int i= 0 ;i<10000;i++)
            strBuffer.append("abcdefghijklmnopqrstuvwxyz");
        time2= System.currentTimeMillis() ;
        long t2 = time2 - time1;
        
        time1= System.currentTimeMillis() ;
        for(int i= 0 ;i<10000;i++)
            strBuilder.append("abcdefghijklmnopqrstuvwxyz");
        time2= System.currentTimeMillis() ;
        long t3 = time2 - time1;
         
        // System.out.println(str);
         System.out.println("use time: "+t1 +" \n\n**********************************");
         //System.out.println(strBuffer.toString());
         System.out.println("use time: "+t2 +" \n\n**********************************");
         //System.out.println(strBuilder.toString());
         System.out.println("use time: "+t3 );
    }

}
NetBeans运行结果:
run:
use time: 50953

**********************************
use time: 16

**********************************
use time: 0


Jcreator的运行结果:
----------------解决方案--------------------------------------------------------
use time: 73610

**********************************
use time: 0

**********************************
use time: 0
----------------解决方案--------------------------------------------------------
API：
例如，测试某些代码执行的时间长度：
   long startTime = System.nanoTime();
   // ... the code being measured ...
   long estimatedTime = System.nanoTime() - startTime;

结果：
use time: 75743151993

**********************************
use time: 7904077

**********************************
use time: 6222020
----------------解决方案--------------------------------------------------------
nb更专业点，jc本身bug就比较多。
直接命令行应该是最准的。

测这种东西没什么用，java不是c，这种速度上的优化对web工程影响不大。
最大的影响还是数据库读取方面的。
----------------解决方案--------------------------------------------------------