While introducing myself to Modern C++ & its new features introduced in C++11 & C++14, I have completely neglected this keyword constexpr. Initially, I was confused about when to use const vs constexpr in C++ & how this constexpr works & differ with const. So, I have studied this from different sources & here is the consolidation of it:

Primitive constexpr Variables

1
2
3

int varA = 3;
const int varB = 5;
constexpr int varC = 7;
  • All of the above variable having a value which is known at compile time. varA is a normal scenario while varB & varC will not take further value or assignment. varB & varC are fixed at compile time if we have defined them like above.
  • But, varB is not the right way(in some situation) of declaring the constant value at compile time. For example, if I declare them as follows:
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12

int getRandomNo()
{
  return rand() % 10;
}

int main()
{
    const int varB = getRandomNo();       // OK
    constexpr int varC = getRandomNo();   // not OK! compilation error

    return 0;
}
  • Value of varB would not anymore compile time. While statement with varC will throw compilation error. The reason is constexpr will always accept a strictly compile-time value.

constexpr Functions

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

constexpr int sum(int x, int y)
{
    return x + y;
}

int main()
{
    const int result = sum(10, 20);     // Here, you can use constexpr as well
    cout << result;
    return 0;
}
  • constexpr specifies that the value of an object, variable and a function can be evaluated strictly at compile-time. And an expression can use in other constant expressions.
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22

+--------------------------------+-----------------------------------+
|   int result = sum(10, 20);    |  const int result = sum(10, 20);  |
+--------------------------------+-----------------------------------+
|  main:                         |     main:                         |
|  ....                          |     ....                          |
|  ....                          |     ....                          |
|  ....                          |     ....                          |
|        subl    $20, %esp       |           subl    $20, %esp       |
|        subl    $8, %esp        |           movl    $30, -12(%ebp)  | <-- Direct 
|        pushl   $20             |           subl    $8, %esp        |     result 
|        pushl   $10             |           pushl   $30             |     substitution
|        call    _Z3sumii        |           pushl   $_ZSt4cout      |
|        addl    $16, %esp       |           call    _ZNSolsEi       |
|        movl    %eax, -12(%ebp) |     ....                          |
|        subl    $8, %esp        |     ....                          |
|        pushl   -12(%ebp)       |     ....                          |
|        pushl   $_ZSt4cout      |                                   |
|        call    _ZNSolsEi       |                                   |
|  ....                          |                                   |
|  ....                          |                                   |
|  ....                          |                                   |
+--------------------------------+-----------------------------------+
  • If you observe above code, you can see that when you catch result as const or constexpr, call to the function sum is not there in assembly rather compiler will execute that function by itself at compile time & substitute the result with function.
  • By specifying constexpr, we suggest compiler to evaluate the function sum at compile time.

constexpr Constructors

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14

class INT
{
    int _no;

public:
    constexpr INT(int no) : _no(no) {}
    constexpr int getInt() const { return _no; }
};
int main()
{
    constexpr INT obj(INT(5).getInt());
    cout << obj.getInt();
    return 0;
}
  • Above code is simple & self-explanatory. If it isn’t to you, then play with it here.

const vs constexpr in C++

  • They serve different purposes. constexpr is mainly for optimization while const is for practically const objects like the value of Pi.
  • const & constexpr both can be applied to member methods. Member methods are made const to make sure that there are no accidental changes by the method. On the other hand, the idea of using constexpr is to compute expressions at compile time so that time can be saved when the code is running.
  • const can only be used with non-static member function whereas constexpr can be used with member and non-member functions, even with constructors but with condition that argument and return type must be of literal types. You read about more limitations here.

Where to Use What?

  • Where you need a value not often & calculating it would be a bit complex, then that is the place you need constexpr. Otherwise, things are fine with an older buddy const. For example, Fibonacci number, factorial, etc.
1
2
3
4
5
6

constexpr unsigned int factorial(unsigned int n)
{
    return (n <= 1) ? 1 : (n * factorial(n - 1));
}

static constexpr auto magic_value = factorial(5);
  • Often programmer would suggest using constexpr instead of a macro.
  • Sometimes you have an expression, that evaluated down to a constant, while maintaining good readability and allowing slightly more complex processing than just setting a constant to a number. For example:
1
2
3
4
5

template< typename Type > 
constexpr Type max( Type a, Type b ) 
{ 
    return a < b ? b : a; 
}

Its a pretty simple choice there but it does mean that if you call max with constant values, it is explicitly calculated at compile time and not at runtime.

  • Another good example is converting units like
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17

using ull = unsigned long long;

constexpr ull operator"" _KB(ull no)
{
    return no * 1024;
}

constexpr ull operator"" _MB(ull no)
{
    return no * (1024_KB);
}

int main()
{
    cout << 1_KB << endl;
    cout << hex << 1_MB << endl;
}

Here you can use constexpr.