Why does assigning an a value greater than an int to an int not throw an exception by default?

Question

I used ILSpy to reflect into the LINQ's Sum method and noticed that it just does a foreach with the checked keyword. But if an int has a defined maximum and you try to go over it, why doesn't it throw an error by default. Assuming you don't use Sum and you do your own foreach without checked, you will not get an exception, it will just give you a garbage value if you go over the maximum int value, but I don't see the reason why this is not just default behavior. If you need something bigger than an int, then don't use an int.

Answer 1

Checked operations are significantly slower than unchecked:

const int startVal = Int32.MaxValue - 1000000;
Stopwatch sw = new Stopwatch();

sw.Start();
int i = startVal;
while (i < Int32.MaxValue)
{
    unchecked
    {
        i++;
    }
}
sw.Stop();
Console.WriteLine("Unchecked: " + sw.ElapsedTicks + " ticks");

i = startVal;
sw.Restart();
while (i < Int32.MaxValue)
{
    checked
    {
        i++;
    }
}
sw.Stop();
Console.WriteLine("Checked: " + sw.ElapsedTicks + " ticks");

Result:

Unchecked: 241 ticks
Checked: 1992 ticks

So using checked results in a performance hit, and since overflow is relatively rare (who actually counts up to Int32.MaxValue?), C# uses unchecked by default.

Answer 2

I don't see the reason why this is not just default behavior

There may be costs involved in performing such checks - either checking of processor flags, or (if using a larger register internally) checking the upper contents of a register. And basic assignment can be an operation occurring inside the tightest of loops - the costs could be appreciable, and not everyone should have to pay.

So you get to choose whether you want those checks to occur.