Iterating performance: ArrayList, LinkedList, and Hashmap for N 1,000,000

Question

I have seen some websites discussing about the performance of ArrayLists, LinkedLists, and Hashmaps:
http://forum.java.sun.com/thread.jspa?threadID=442985
http://java.sun.com/developer/JDCTechTips/2002/tt0910.html
http://www.javaspecialists.co.za/archive/Issue111.html
http://joust.kano.net/weblog/archives/000066.html
If I understand it right an ArrayList in general is faster for accessing a particular element in the collection and I can't find some pro's of using a LinkedList.
My question is: If I only use a large collection with more than 1 million elements for iterating from begin to end (i.e. for loop), is it faster to use a linked list or a custom linked list instead of an arraylist (or hashmap)? Since you can iterate "directly" through a (custom) linked list, which is not possible with a arraylist?
Edited by: 9squared on Nov 23, 2007 1:48 PM

Answer 1

Thanks for the help, I wrote some code and tested it
import java.util.ArrayList;
import java.util.List;
import java.util.LinkedList;
public class TestTemp {
     public static void main(String[] args) {
          List<Node> a = new ArrayList<Node>();
          Node b = new Node("a");
          String[] c = new String[10000000];
          Node temp = b;
          for (int i = 0; i < 10000000; i++)
               a.add(new Node("a"));
               temp.next = new Node("b");
               temp = temp.next;     
               c[i] = "c";
          long tstart;
          tstart = System.currentTimeMillis();
          for (int i = 0; i < 10000000; i++)
               c[i] = "cc";
               if (i%200000 == 0)
                    System.out.println("Array " + i + ": " + (System.currentTimeMillis()-tstart));
          tstart = System.currentTimeMillis();
          temp = b;
          for (int i = 0; i < 10000000; i++)
               temp.next.text = "bb";
               temp = temp.next;
               if (i%200000 == 0)
                    System.out.println("LinkedList " + i + ": " + (System.currentTimeMillis()-tstart));
          tstart = System.currentTimeMillis();
          for (int i = 0; i < 10000000; i++)
               a.get(i).text = "aa";
               if (i%200000 == 0)
                    System.out.println("ArrayList " + i + ": " + (System.currentTimeMillis()-tstart));
public class Node {
     public String text;
     public Node next;
     public Node(String text)
          this.text = text;
}Here are some results in milliseconds, and indeed just iterating doesn't take very long
Elements     Linked     Arraylist     Array
200000     0     0     1
400000     5     13     5
600000     9     22     9
800000     14     32     12
1000000     20     42     16
1200000     25     52     19
1400000     31     63     23
1600000     37     72     26
1800000     42     82     30
2000000     47     92     33
2200000     51     101     37
2400000     56     112     40
2600000     60     123     44
2800000     65     134     47
3000000     69     143     51
3200000     73     152     55
3400000     78     162     59
3600000     84     175     63
3800000     103     185     67
4000000     108     195     70
4200000     113     207     74
4400000     117     216     78
4600000     122     225     81
4800000     127     237     85
5000000     131     247     88
5200000     136     256     92
5400000     142     266     97
5600000     147     275     101
5800000     153     286     107
6000000     159     298     113
6200000     162     307     117
6400000     167     317     121
6600000     171     326     125
6800000     175     335     128
7000000     180     346     132
7200000     184     358     136
7400000     188     368     139
7600000     193     377     143
7800000     197     388     147
8000000     201     397     150
8200000     207     410     154
8400000     212     423     157
8600000     217     432     162
8800000     222     442     167
9000000     227     452     171
9200000     231     462     175
9400000     236     473     178
9600000     242     483     182
9800000     249     495     185
10000000     257     505     189