9. Decimal floats

The IEEE-754-2008 standard defines 32-, 64- and 128-bit formats for decimal floats. The 32-bit format is used for storage. Decimal floats are used for management and accounting applications requiring precise decimal calculations. This explains why the standard is rarely implemented in hardware. IBM processors since Power6 and mainframes since IBM Systemz9 implement this format. Recent processors include Fujitsu's Sparc64 X/X+ processor, which implements both 32- and 64-bit formats. Intel uses a software library for decimal floats.

The floating decimal format, shown in figure 14 , is defined by (s,q,C) where :