C++ attribute: assume (since C++23)

From cppreference.com
< cpp‎ | language‎ | attributes
 
 
C++ language
General topics
Flow control
Conditional execution statements
if
Iteration statements (loops)
for
range-for (C++11)
Jump statements
Functions
Function declaration
Lambda function expression
inline specifier
Dynamic exception specifications (until C++17*)
noexcept specifier (C++11)
Exceptions
Namespaces
Types
Specifiers
decltype (C++11)
auto (C++11)
alignas (C++11)
Storage duration specifiers
Initialization
Expressions
Alternative representations
Literals
Boolean - Integer - Floating-point
Character - String - nullptr (C++11)
User-defined (C++11)
Utilities
Attributes (C++11)
Types
typedef declaration
Type alias declaration (C++11)
Casts
Implicit conversions - Explicit conversions
static_cast - dynamic_cast
const_cast - reinterpret_cast
Memory allocation
Classes
Class-specific function properties
explicit (C++11)
static
Special member functions
Templates
Miscellaneous
 
 
Attributes
(C++14)
(C++17)
(C++20)(C++20)
assume
(C++23)
 

Specifies that an expression will always evaluate to true at a given point.

Syntax

[[assume( expression )]]
expression - expression that must evaluate to true

Explanation

Can only be applied to a null statement, as in [[assume(x > 0)]];. This statement is called an assumption. If the expression (contextually converted to bool) would not evaluate to true at the place the assumption occurs, the behavior is undefined. Otherwise, the statement does nothing. In particular, the expression is not evaluated (but it is still potentially evaluated).

The purpose of an assumption is to allow compiler optimizations based on the information given.

The expression may not be a comma operator expression, but enclosing the expression in parentheses will allow the comma operator to be used.

Notes

If the expression would have undefined behavior, or if it would cause an exception to be thrown, then it does not evaluate to true.

Since assumptions cause undefined behavior if they do not hold, they should be used sparingly. They are not intended as a mechanism to document the preconditions of a function or to diagnose violations of preconditions. Also, it should not be presumed, without checking, that the compiler actually makes use of any particular assumption.

Example

void f(int& x, int y)
{
    void g(int);
    void h();
 
    [[assume(x > 0)]]; // Compiler may assume x is positive
 
    g(x / 2); // More efficient code possibly generated
 
    x = 3;
    int z = x;
 
    [[assume((h(), x == z))]]; // Compiler may assume x would have the same value after
                               // calling h
                               // The assumption does not cause a call to h
 
    h();
    g(x); // Compiler may replace this with g(3);
 
    h();
    g(x); // Compiler may NOT replace this with g(3);
          // An assumption applies only at the point where it appears
 
    z = std::abs(y);
 
    [[assume((g(z), true))]]; // Compiler may assume g(z) will return
 
    g(z); // Due to above and below assumptions, compiler may replace this with g(10);
 
    [[assume(y == -10)]]; // Undefined behavior if y != -10 at this point
 
    [[assume((x - 1) * 3 == 12)]];
 
    g(x); // Compiler may replace this with g(5);
}

References

  • C++23 standard (ISO/IEC 14882:2023):
  • 9.12.3 Assumption attribute [dcl.attr.assume]

See also

marks unreachable point of execution
(function)

External links

1.  Clang language extensions doc: __builtin_assume.
2.  Clang attribute reference doc: assume.
3.  MSVC doc: __assume built-in.
4.  GCC doc: __attribute__((assume(...))).