Decimal64

Decimal64 is a decimal floating-point format defined by the IEEE 754 standard as part of the decimal floating-point family. It uses 64 bits to represent a decimal number with up to 16 significant digits and a decimal exponent, enabling direct decimal arithmetic with predictable rounding.

In decimal64, a finite value is represented by a sign, a coefficient (significand) with up to 16 decimal digits, and an exponent that scales the coefficient by powers of 10. The precision (p) is 16, and the exponent range is from -383 to +385. The encoding supports the usual floating-point special values, including positive and negative infinities and NaNs, along with standard rounding modes defined by the IEEE 754 specifications.

Decimal64 is one of three decimal floating-point formats standardized by IEEE 754: decimal32, decimal64, and decimal128. It provides exact decimal representation for many values that would not be exact in binary floating-point, reducing rounding and representation errors in decimal-centric computations, such as financial applications.

Implementation can be in hardware through decimal floating-point units or in software libraries; support varies by platform. Adoption has been gradual but present in numerical computing environments, databases, and some programming languages and processors that implement IEEE 754 decimal arithmetic.

See also: decimal32, decimal128, IEEE 754, decimal arithmetic, financial computing.

---