Conversions from floating-point types

A floating-point value that's converted to another floating-point type undergoes no change in value if the original value is representable exactly in the result type. If the original value is numeric but isn't representable exactly, the result is either the next greater or next lower representable value. See Limits on floating-point constants for the range of floating-point types.

A floating-point value that is converted to an integral type is first truncated by discarding any fractional value. If this truncated value is representable in the result type, the result must be that value. When it isn't representable, the result value is undefined.

Microsoft Specific

Microsoft compilers use IEEE-754 binary32 representation for float values, and binary64 representation for long double and double. Since long double and double use the same representation, they have the same range and precision.

When the compiler converts a double or long double floating-point number to a float, it rounds the result according to the floating-point environment controls, which default to "round to nearest, ties to even." If a numeric value is too high or too low to be represented as a numeric float value, the conversion result is positive or negative infinity according to the sign of the original value, and an overflow exception is raised, if enabled.

The result of a conversion to an integral type smaller than long is the result of converting the value to long, and then converting to the result type.

For conversion to integer types at least as large as long, a conversion of a value that is too high or too low to represent in the result type may return any of the following values:

  • The result may be a sentinel value, which is the representable value farthest from zero. For signed types, it's the lowest representable value (0x800...0). For unsigned types, it's the highest representable value (0xFF...F).

  • The result may be saturated, where values too high to represent are converted to the highest representable value. Values too low to represent are converted to the lowest representable value. One of these two values is also used as the sentinel value.

  • For conversion to unsigned long or unsigned long long, the result of converting an out-of-range value may be some value other than the highest or lowest representable value. Whether the result is a sentinel or saturated value or not depends on the compiler options and target architecture. Future compiler releases may return a saturated or sentinel value instead.

END Microsoft Specific

The following table summarizes conversions from floating types.

Table of conversions from floating-point types

From To Method
float char Convert to long, then convert long to char
float short Convert to long, then convert long to short
float int Truncate at decimal point. If result is too large to be represented as int, result is undefined.
float long Truncate at decimal point. If result is too large to be represented as long, result is undefined.
float long long Truncate at decimal point. If result is too large to be represented as long long, result is undefined.
float unsigned char Convert to long, then convert long to unsigned char
float unsigned short Convert to long, then convert long to unsigned short
float unsigned Truncate at decimal point. If result is too large to be represented as unsigned, result is undefined.
float unsigned long Truncate at decimal point. If result is too large to be represented as unsigned long, result is undefined.
float unsigned long long Truncate at decimal point. If result is too large to be represented as unsigned long long, result is undefined.
float double Represent as a double.
float long double Represent as a long double. In MSVC, long double and double have the same representation.
double char Convert to float, then convert float to char
double short Convert to float, then convert float to short
double int Truncate at decimal point. If result is too large to be represented as int, result is undefined.
double long Truncate at decimal point. If result is too large to be represented as long, result is undefined.
double unsigned char Convert to long, then convert long to unsigned char
double unsigned short Convert to long, then convert long to unsigned short
double unsigned Truncate at decimal point. If result is too large to be represented as unsigned, result is undefined.
double unsigned long Truncate at decimal point. If result is too large to be represented as unsigned long, result is undefined.
double unsigned long long Truncate at decimal point. If result is too large to be represented as unsigned long long, result is undefined.
double float Represent as a float. If the double value can't be represented exactly as float, loss of precision occurs.
double long double The long double value is treated as double.

Conversions from long double follow the same method as conversions from double.

See also

Assignment conversions