1- Optimize For Loops
Where-ever possible, count down to zero rather than up to n. For example, use
for (i = n-1; i >= 0; --i)
rather than
for (i = 0; i <>
The test is done every iteration and it's faster to test against zero than anything else. Note also that
++i
is faster than
i++
when it appears in the third part of the for loop statement.
2- Optimize If Statements
Factor out jumps. For example, use
bar();
if (condition)
{
undoBar();
foo();
}
rather than
if (condition)
{
foo();
}
else
{
bar();
}
Use a profiler and good judgement to decide if undoing the bar() operation is faster than jumping.
3- Optimize Switch Statements
Put the most common cases first.
Posts
2 comments:
i've heard ++i being faster than i++ is compiler specific.
and i've also heard if..else.. is better than logical if. the post says otherwise. and thinking of it i guess undoing would take more time anyway especially if it is a large operation being done in bar()
> Note also that
> ++i
> is faster than
> i++
yes, (sometimes - compiler specific) and only if "i" is an object (which has to be copied before i++ operation), not in your example - "i" is just an integer
> Optimize If Statements
How the hell did you invented this?
Compiler works (in assembly) like:
1.test if(~condition)
2.if passed skip {code}, else do nothing so the {code} will be executed
And if you got "else" statement only difference is one relative jump from end of if code to end of else code. One single assembler instruction.
Post a Comment