by BehindJava

How does the Java 'for each' loop work

Home » java » How does the Java 'for each' loop work

In this tutorial we are going to learn about Java For Each loop working in detail.

The for-each loop, added in Java 5 (also called the “enhanced for loop”), is equivalent to using a java.util.Iterator—it’s syntactic sugar for the same thing. Therefore, when reading each element, one by one and in order, a for-each should always be chosen over an iterator, as it is more convenient and concise.

For-each

for (int i : intList) {
   System.out.println("An element in the list: " + i);
}

Iterator

Iterator<Integer> intItr = intList.iterator();
while (intItr.hasNext()) {
   System.out.println("An element in the list: " + intItr.next());
}

There are situations where you must use an Iterator directly. For example, attempting to delete an element while using a for-each can (will?) result in a ConcurrentModificationException.

For-each vs. for-loop: Basic differences

The only practical difference between for-loop and for-each is that, in the case of indexable objects, you do not have access to the index. An example when the basic for-loop is required:

for (int i = 0; i < array.length; i++) {
   if(i < 5) {
      // Do something special
   }  else {
      // Do other stuff
   }
}

Although you could manually create a separate index int-variable with for-each,

int idx = -1;
for (int i : intArray) {
   idx++;
   ...
}

…it is not recommended, since variable-scope is not ideal, and the basic for loop is simply the standard and expected format for this use case.

For-each vs. for-loop: Performance

When accessing collections, a for-each is significantly faster than the basic for loop’s array access. When accessing arrays, however—at least with primitive and wrapper-arrays—access via indexes is dramatically faster.

Timing the difference between iterator and index access for primitive int-arrays

Indexes are 23-40 percent faster than iterators when accessing int or Integer arrays. Here is the output from the testing class at the bottom of this post, which sums the numbers in a 100-element primitive-int array (A is iterator, B is index):

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 358,597,622 nanoseconds
Test B: 269,167,681 nanoseconds
B faster by 89,429,941 nanoseconds (24.438799231635727% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 377,461,823 nanoseconds
Test B: 278,694,271 nanoseconds
B faster by 98,767,552 nanoseconds (25.666236154695838% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 288,953,495 nanoseconds
Test B: 207,050,523 nanoseconds
B faster by 81,902,972 nanoseconds (27.844689860906513% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 375,373,765 nanoseconds
Test B: 283,813,875 nanoseconds
B faster by 91,559,890 nanoseconds (23.891659337194227% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 375,790,818 nanoseconds
Test B: 220,770,915 nanoseconds
B faster by 155,019,903 nanoseconds (40.75164734599769% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 326,373,762 nanoseconds
Test B: 202,555,566 nanoseconds
B faster by 123,818,196 nanoseconds (37.437545972215744% faster)

I also ran this for an Integer array, and indexes are still the clear winner, but only between 18 and 25 percent faster.

For collections, iterators are faster than indexes

For a List of Integers, however, iterators are the clear winner. Just change the int-array in the test-class to:

List<Integer> intList = Arrays.asList(new Integer[] {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100});

And make the necessary changes to the test-function (int[] to List, length to size(), etc.):

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 3,429,929,976 nanoseconds
Test B: 5,262,782,488 nanoseconds
A faster by 1,832,852,512 nanoseconds (34.326681820485675% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,907,391,427 nanoseconds
Test B: 3,957,718,459 nanoseconds
A faster by 1,050,327,032 nanoseconds (26.038700083921256% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,566,004,688 nanoseconds
Test B: 4,221,746,521 nanoseconds
A faster by 1,655,741,833 nanoseconds (38.71935684115413% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,770,945,276 nanoseconds
Test B: 3,829,077,158 nanoseconds
A faster by 1,058,131,882 nanoseconds (27.134122749113843% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 3,467,474,055 nanoseconds
Test B: 5,183,149,104 nanoseconds
A faster by 1,715,675,049 nanoseconds (32.60101667104192% faster)

[C:\java_code\]java TimeIteratorVsIndexIntList 1000000
Test A: 3,439,983,933 nanoseconds
Test B: 3,509,530,312 nanoseconds
A faster by 69,546,379 nanoseconds (1.4816434912159906% faster)

[C:\java_code\]java TimeIteratorVsIndexIntList 1000000
Test A: 3,451,101,466 nanoseconds
Test B: 5,057,979,210 nanoseconds
A faster by 1,606,877,744 nanoseconds (31.269164666060377% faster)

In one test they’re almost equivalent, but with collections, iterator wins.