Method half round_error

include/half_float/half.hpp:2417–2418

Maximum rounding error in ULP (units in the last place).

		/// Maximum rounding error in ULP (units in the last place).
		static HALF_CONSTEXPR half_float::half round_error() HALF_NOTHROW
			{ return half_float::half(half_float::detail::binary, (round_style==std::round_to_nearest) ? 0x3800 : 0x3C00); }

		/// Positive infinity.
		static HALF_CONSTEXPR half_float::half infinity() HALF_NOTHROW { return half_float::half(half_float::detail::binary, 0x7C00); }