Post-Increment vs Pre-Increment

There is a well known C++ BKM for preferring pre-increment (++i) over post-increment (i++). The rationale is that when post-incrementing and object, the object must increment itself and then return a temporary containing its old value. While this makes perfect sense for complex types (e.g. STL iterators) I’ve always wondered if it holds true for built-in types such as int. Let’s find out.

First let’s create a trivial example to examine:

#include <iostream>

using namespace std;

int main()
{
	// post-imcrement
	for(int i=0; i<10; i++)
	{
		cout << "[" << i << "]" << endl;
	}

	// pre-imcrement
	for(int i=0; i<10; ++i)
	{
		cout << "[" << i << "]" << endl;
	}

	return 0;
}

Now using the Visual Studio 2008 C++ compiler, let’s take a look at the generated assembly code. First let’s keep things simple and disable compiler optimizations (/Od):

	// post-imcrement
	for(int i=0; i<10; i++)
004114D3  mov         dword ptr [i],0 
004114D5  jmp         main+30h (4114E0h) 
004114D7  mov         eax,dword ptr [i] 
004114DA  add         eax,1 
004114DD  mov         dword ptr [i],eax 
004114E0  cmp         dword ptr [i],0Ah 
004114E4  jge         main+86h (411536h) 
	{

<... snip ...>

	// pre-imcrement
	for(int i=0; i<10; ++i)
00411536  mov         dword ptr [i],0 
0041153D  jmp         main+98h (411548h) 
0041153F  mov         eax,dword ptr [i] 
00411542  add         eax,1 
00411545  mov         dword ptr [i],eax 
00411548  cmp         dword ptr [i],0Ah 
0041154C  jge         main+0EEh (41159Eh) 
	{

The relevant parts of the assembly code are highlighted. As you can see, there is no difference between the generated code for post-increment and pre-increment.

Now, with compiler optimizations enabled (\O2) and the relevant parts highlighted:

	// post-imcrement
	for(int i=0; i<10; i++)
00401008  xor         esi,esi 
0040100A  lea         ebx,[ebx] 
	{
<... snip ...>
0040104A  inc         esi  
0040104B  cmp         esi,0Ah 
0040104E  jl          main+10h (401010h) 
	}

	// pre-imcrement
	for(int i=0; i<10; ++i)
00401050  xor         esi,esi 
	{
<... snip ...>
0040108C  inc         esi  
0040108D  cmp         esi,0Ah 
00401090  jl          main+52h (401052h) 
	}

Again there is no difference in the generated assembly between post-increment and pre-increment.

So, based on this quick experiment, there is no difference between pre-increment and post-increment for int types when using the Visual Studio 2008 C++ compiler with optimizations enabled or disabled.

Note that this doesn’t change the BKM to prefer pre-increments over post-increments when using integer types. The above experiment was only conducted with a single compiler – other compilers may behave differently and so it is best to err on the side of caution and use pre-increments where possible. Also, the use of pre-increment adds some future proofing – if the type of the variable being incremented changed to a more complex type then pre-increment then becomes the optimal choice.