Interesting benchmark

Question

recently i've been playing around with different ways to do things to optimize my program, and i came across an interesting discovery. i'm not a master at making benchmarks, but i think this should be ok.
time = System.nanoTime();
          for (int i=0;i<10000;i++) {
               float x = p.x;
          System.out.println(System.nanoTime()-time);
          for (int i=0;i<10000;i++) {
               float x = posX[i];
          System.out.println(System.nanoTime()-time);what i get was
229000
426160
the first one is for when you have an array of objects, and get their values, and the second is just a float array. what i found interesting was that the first way was 2 times faster.
Edited by: kotsoft on Jan 2, 2008 8:24 PM
Edited by: kotsoft on Jan 2, 2008 8:27 PM
Edited by: kotsoft on Jan 2, 2008 8:27 PM                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           

Answer 1

kotsoft wrote:
recently i've been playing around with different ways to do things to optimize my program, and i came across an interesting discovery. i'm not a master at making benchmarks, but i think this should be ok.
time = System.nanoTime();
          for (int i=0;i<10000;i++) {
               float x = p.x;
          System.out.println(System.nanoTime()-time);
          for (int i=0;i<10000;i++) {
               float x = posX[i];
          System.out.println(System.nanoTime()-time);what i get was
229000
426160
the first one is for when you have an array of objects, and get their values, and the second is just a float array. what i found interesting was that the first way was 2 times faster.Because you don't reset time to the current time before starting the second loop.