A 16-bit unsigned value
For an unsigned type with precision p = 16, the value 240 has bits 4, 5, 6, and 7 set. Its maximum value is 2^16 − 1, or 65535; 65536 is the first value outside its range. section 5.7.1
Precision describes a property of a type, not the accuracy of a floating-point computation. C constrains the precision of its basic types but does not prescribe a single precision for each: the actual values depend on the implementation. section 5.2
For an integer type, precision p is the number of bits used to represent nonnegative values. Its maximum is 2^p − 1. For an unsigned type, the representable values run from zero to that maximum, and precision determines the type’s arithmetic. Limits such as UINT_MAX are available in <limits.h>. section 5.7.1
Precision also matters when an unsigned value is interpreted as a bit set. The possible bit positions run from 0 through p − 1, so a complement depends on p. In a left shift, bits that do not fit are omitted; the shift count must be less than the precision of the operand’s type. section 5.7.2 section 5.7.3
Shifting the 16-bit value
For a 16-bit unsigned shift operand, 240 << 2 gives 960. Shifting 240 left by 9 would place a bit at position 16, which does not fit; the remaining bits give 57344. section 5.7.3
In a simplified floating-point representation, precision p is the number of mantissa bits. The exponent bounds also depend on the type. For double, DBL_MANT_DIG, DBL_MIN_EXP, and DBL_MAX_EXP expose these properties; a typical DBL_MANT_DIG is 53. section 5.7.7
TAKEAWAY: A finite precision can cut off values during intermediate floating-point computations. For example, adding a sufficiently small value to a value greater than one can return the larger value unchanged. section 5.7.7
The word precision also appears in printf format specifications as the optional .PP component. That output-format component is distinct from the precision property of a type. section 8.3.4 section 5.2









