Lines Matching refs:IEEE
32 return U.IEEE.METHOD_CALL; \
56 definition of IEEE 754. */
60 matches the definition of IEEE 754. */
79 /* The IBM double-double semantics. Such a number consists of a pair of IEEE
92 operation. It's equivalent to having an IEEE number with consecutive 106
106 semPPCDoubleDoubleLegacy -> IEEE operations
110 legacy IEEE float.
1402 /* As specified in IEEE 754, since we do not trap we do not report
1769 /* If two numbers add (exactly) to zero, IEEE 754 decrees it is a
1975 /* If two numbers add (exactly) to zero, IEEE 754 decrees it is a
1984 to do in the latter case. The IEEE-754R standard says it is
2002 // [IEEE Std 754-2008 6.1]:
2013 // [IEEE Std 754-2008 6.2]:
2018 // [IEEE Std 754-2008 6.2]:
2025 // [IEEE Std 754-2008 6.2]:
2039 // [IEEE Std 754-2008 6.3]:
2278 standard doesn't require an inexact exception to be raised. IEEE
2935 stored. This implies the exponent will fall within the IEEE format
3111 // the actual IEEE respresentations. We compensate for that here.
4024 // IEEE-754R 2008 6.2.1: A signaling NaN bit string should be encoded with the
4029 /// IEEE-754R 2008 5.3.1: nextUp/nextDown.
4051 // IEEE-754R 2008 6.2 Par 2: nextUp(sNaN) = qNaN. Set Invalid flag.
4052 // IEEE-754R 2008 6.2: nextUp(qNaN) = qNaN. Must be identity so we do not
4794 new (&IEEE) IEEEFloat(std::move(F));
4814 return hash_value(Arg.U.IEEE);
4835 return U.IEEE.convert(ToSemantics, RM, losesInfo);
4839 auto Ret = U.IEEE.convert(semPPCDoubleDoubleLegacy, RM, losesInfo);
4840 *this = APFloat(ToSemantics, U.IEEE.bitcastToAPInt());