Already good answer there. Just add a benchmark result for StringBuffer and StringBuild performance difference use new instance in loop or use setLength(0) in loop.
The summary is: In a large loop
Very simple benchmark (I just manually changed the code and do different test ):
public class StringBuilderSpeed {
public static final char ch[] = new char[]{'a','b','c','d','e','f','g','h','i'};
public static void main(String a[]){
int loopTime = 99999999;
long startTime = System.currentTimeMillis();
StringBuilder sb = new StringBuilder();
for(int i = 0 ; i < loopTime; i++){
for(char c : ch){
sb.append(c);
}
sb.setLength(0);
}
long endTime = System.currentTimeMillis();
System.out.println("Time cost: " + (endTime - startTime));
}
}
New StringBuilder instance in loop: Time cost: 3693, 3862, 3624, 3742
StringBuilder setLength: Time cost: 3465, 3421, 3557, 3408
New StringBuffer instance in loop: Time cost: 8327, 8324, 8284
StringBuffer setLength Time cost: 22878, 23017, 22894
Again StringBuilder setLength to ensure not my labtop got some issue to use such long for StringBuffer setLength :-) Time cost: 3448