lockdown source

[brew]

What is the purpose of i ^= i? Why not just i = 0?

Because it's more efficient then MOV (takes 7 less cpu cycles) and it doesn't have to store a constant value of 0.

[Yegg]

That depends on the compiler you're using. I created some binary files from a simple C source I just wrote:

Code Select Expand

int main () {
    int i;

    i ^= i;
}

produces

Code Select Expand

00000000  8D4C2404          lea ecx,[esp+0x4]
00000004  83E4F0            and esp,byte -0x10
00000007  FF71FC            push dword [ecx-0x4]
0000000A  55                push ebp
0000000B  89E5              mov ebp,esp
0000000D  51                push ecx
0000000E  83EC10            sub esp,byte +0x10
00000011  C745F800000000    mov dword [ebp-0x8],0x0
00000018  83C410            add esp,byte +0x10
0000001B  59                pop ecx
0000001C  5D                pop ebp
0000001D  8D61FC            lea esp,[ecx-0x4]
00000020  C3                ret

and

Code Select Expand

int main () {
    int i;

    i = 0;
}

produces

Code Select Expand

00000000  8D4C2404          lea ecx,[esp+0x4]
00000004  83E4F0            and esp,byte -0x10
00000007  FF71FC            push dword [ecx-0x4]
0000000A  55                push ebp
0000000B  89E5              mov ebp,esp
0000000D  51                push ecx
0000000E  83EC10            sub esp,byte +0x10
00000011  C745F800000000    mov dword [ebp-0x8],0x0
00000018  83C410            add esp,byte +0x10
0000001B  59                pop ecx
0000001C  5D                pop ebp
0000001D  8D61FC            lea esp,[ecx-0x4]
00000020  C3                ret

You'll notice the two are identical. Perhaps there is some optimization argument gcc should be given?

[brew]

Wow, that's pretty gay. Some optimization that is. From now on i'm going to turn off optimizations for Visual C++. Is this a good idea?

[Yegg]

I don't know much about Visual C++. It sounds like a good idea to keep the optimizations on unless you're a pretty advanced guru with the language and software.

[iago]

Optimizing C code like that is almost always stupid. If you're going to do that, you'd might as well start expanding your loops out.

Seriously, let the compiler/optimizer do what it's good for.

[Antarctica]

Anything that will work for vb6?

[UserLoser]

Optimizing C code like that is almost always stupid. If you're going to do that, you'd might as well start expanding your loops out.

Seriously, let the compiler/optimizer do what it's good for.

[Newby]

Wow, that's pretty gay. Some optimization that is. From now on i'm going to turn off optimizations for Visual C++. Is this a good idea?

You motherfucker, I thought I had posted today because I use this avatar in other places. You confused me you bastard.

[brew]

Wow, that's pretty gay. Some optimization that is. From now on i'm going to turn off optimizations for Visual C++. Is this a good idea?

You motherfucker, I thought I had posted today because I use this avatar in other places. You confused me you bastard.

....huh...?

Seriously, let the compiler/optimizer do what it's good for.

I say its not doing it's job good enough.

[l2k-Shadow]

How come you're using while (i < 5) i thought you said != is way more efficient.

I know you feel all leet coding in C and everything but let it go man, little things like these make no difference, let the compiler do it's work.

[Yegg]

How come you're using while (i < 5) i thought you said != is way more efficient.

I know you feel all leet coding in C and everything but let it go man, little things like these make no difference, let the compiler do it's work.

Don't always just "let the compiler do the work".  It's educational to learn how certain parts of the compiler are actually done. brew, don't expect that those kinds of small details will improve your applications at all, because typically they won't, but it's still fun to learn little things like that. Look more into it.

[raylu]

How come you're using while (i < 5) i thought you said != is way more efficient.

I know you feel all leet coding in C and everything but let it go man, little things like these make no difference, let the compiler do it's work.

Of course! There is no middleground between pure ASM and...well, every time I think I've seen the worst solution possible, I learn something new...but you get the point.

[brew]

How come you're using while (i < 5) i thought you said != is way more efficient.

I know you feel all leet coding in C and everything but let it go man, little things like these make no difference, let the compiler do it's work.

Of course! There is no middleground between pure ASM and...well, every time I think I've seen the worst solution possible, I learn something new...but you get the point.

eh... MASM
I don't feel very leet coding in C...
But now when I go back to finish a project in VB i'm totally disgusted at how dumbed down it is.

[Kp]

That assembly looks grossly unoptimized.  I could have done that in two instructions:

Code Select Expand


xor eax, eax
ret

brew: considering how concerned you are about manual optimization, I find it a little surprising you have not even considered security.  Read up about buffer overflows, look at how many patches Microsoft has to issue because their programmers do not understand buffer overflows, then look at your code again.  If you still think it is OK, then I will pick it apart and explain what is wrong.

[iago]

Don't always just "let the compiler do the work".  It's educational to learn how certain parts of the compiler are actually done. brew, don't expect that those kinds of small details will improve your applications at all, because typically they won't, but it's still fun to learn little things like that. Look more into it.

The problem is, if you make your code less readable but execute a fraction of no time faster, you lose. You're typically better off keeping your code readable rather than efficient.

That being said, efficient algorithms are important. If you bubblesort/insertionsort 100000000 items or quicksort/shellsort 100000000 items, it matters very little whether each instruction is fast or slow, the quicksort will always be faster. Algorithm choice is important in many cases.