Induction Variable Optimizations


for(j=1;j<5;++j)
 fprintf(stdout,"%d\n",j*3);

Addition is cheaper than multiplication, so the compiler can convert this fragment to the one bellow.


int s = 0;
for(j=1;j<5;++j)
{
 s += 3
 fprintf(stdout,"%d\n",s);
}

A temporary variable has been introduced in order to eliminate the multiplications.

It's not nearly as costly as you think it is. Stop trying to microoptimize useless stuff like this at the cost of clarity. A good enough C compiler will already handle this nicely, for example, multiplication by constants is optimized by some compilers, here's one possible optimization for reg*3 on x86:

8D0440          LEA EAX,DWORD PTR DS:[EAX+EAX*2]
Which is faster than your add by 3, especially if the variable is not in a separate register.

tl;dr: Stop microoptimizing, a good compiler will take care of it.

>>2
I was describing an optimizing process used by good enough compilers, not programmers.

The latter produces a different result to the former, moron.

>>4
Can't read for shit

I am a professor in Angola. I know what I am doing.

(^_SAGE)

for(j=1;j<15;j+=3)
 fprintf(stdout,"%d\n",j);

>>2
When we're talking about clarity, why would the latter be less readable than the former (ignoring the missing semicolon)?

fprintf(stdout
IHBT

>>9
The first is clear in that it will print the multiples of 3 from 1 to 5 from the first look at the code, while the second requires you to infer that it's generating then printing each multiple of 3 from 1 to 5. The difference is small, but the first is slightly more clear, if not by much.

-funroll-loops
*VROOM VROOM*

>>11
No-one ever proposed that anyone should write the second form, only that it is a transformation a compiler can perform.

^{Back to /b/, ``GNAA Faggot''}