Does this code,
for (byte b = 0; b < 100; b++)
{
//some code
}
run faster than this code?
for (int b = 0; b < 100; b++)
{
//some code
}
-
7Messure time and check it out. I think it would be faster than posting question here.Antoniossss– Antoniossss2013-10-22 13:18:10 +00:00Commented Oct 22, 2013 at 13:18
-
Duplicate question: stackoverflow.com/questions/229886/…jcklie– jcklie2013-10-22 13:18:42 +00:00Commented Oct 22, 2013 at 13:18
-
5the answer is a big no.user2511414– user25114142013-10-22 13:22:00 +00:00Commented Oct 22, 2013 at 13:22
-
1@reindeer This is not a duplicate of that question. That asks anout memory, this asks about speed.Raedwald– Raedwald2013-10-24 07:27:23 +00:00Commented Oct 24, 2013 at 7:27
Add a comment
|
3 Answers
No, not at all; if anything, it will be slower, because the underlying hardware generally has instructions for working with the native "int" type (32-bit two's complement integer) but not for working with 8-bit signed bytes.
answered Oct 22, 2013 at 13:17
Ernest Friedman-Hill
81.9k11 gold badges158 silver badges192 bronze badges
4 Comments
Holger
Not only the underlying hardware, even the Java bytecode itself has an instruction set for working primarily with
int. Using an int in the for loop means a single iinc instruction will do the loop’s increment but when using a byte a four instruction sequence of iload, iconst_1, iadd, i2b, istore is required.
Hussain Akhtar Wahid 'Ghouri'
i calculated time difference by System.nanoTime() , the result deviates both sides , btw , you just earned a fan by your answer
Holger
Did I say “four”? Even five instructions ;-)
Always use an int data type as the loop index variable whenever possible because it is efficient when compared to using byte or short data types. because when we use byte or short data type as the loop index variable they involve implicit type cast to int data type.
check this http://www.precisejava.com/javaperf/j2se/Loops.htm
2 Comments
Suresh Atta
Good point +1. But when you are pasting from some other link, quote them. I done that now.
Kayaman
That advice is worthless, since JIT will definitely compile the loop into an efficient form.
The correct way to optimize code is to profile and analyze it to find hotspots, i.e. places that take most of the time. These can be (and often are) database or network operations, or a poor algorithm operating on a massive amount of data. When hotspots are found, they can be further analyzed to find out ways to make them faster (modify database queries, use a cache etc.)
Optimization is not deciding whether to use a byte instead of an int, or whether to use a StringBuilder instead of a StringBuffer. This is known as micro-optimization, which is worthless. Optimization is always context dependent and cannot be performed with general "use X instead of Y" or "always do Z" rules.