Lines Matching defs:zero
88 * The exponent normally used for zero/subnormals is an invalid input. Such
89 * values are specified with the leading mantissa bit of zero and the lowest
91 * mantissa set to zero is a shorthand notation for the correctly signed
92 * floating point zero. Inf and NaN must be specified directly with an
132 static Float zero (int sign);
199 inline Float<StorageType, ExponentBits, MantissaBits, ExponentBias, Flags> Float<StorageType, ExponentBits, MantissaBits, ExponentBias, Flags>::zero (int sign)
237 // Repurpose this otherwise invalid input as a shorthand notation for zero (no need for caller to care about internal representation)
240 // Handles the typical notation for zero (min exponent, mantissa 0). Note that the exponent usually used exponent (-ExponentBias) for zero/subnormals is not used.
241 // Instead zero/subnormals have the (normally implicit) leading mantissa bit set to zero.
276 // Negative number, truncate to zero.
277 return zero(+1);
292 return zero(other.sign());
351 return zero(other.sign());