Single-precision floating-point format

In computer science, single precision is (english single precision or even single) a name for a floating-point format, which occupies a storage unit in the computer. Thus, the exact details depend on the computer manufacturer and its memory architecture. Especially for microprocessors with bytewise memory of the IEEE 754 standard was designed, the ( 32-bit) requires 4 bytes for this number format. The term is not reserved for floating point numbers; it is also applicable for integer formats.

An IEEE -754 number is accurate to 7-8 digits in the decimal applicable.

In numerical mathematics is, for example, totals scalar the need to store intermediate results in more detail, in order to calculate a final result with sufficient accuracy. Therefore, there are at most computer architectures a number format double precision, at least more memory, but often really provides twice as much memory.

Since programming languages ​​such as C treat single precision secondary principle, the double memory consumption today plays only a minor role and also falls the higher computation time into weight hardly produced the same numerical problem as for the simple double precision. Therefore, 754-2008 fourfold exact number formats have been introduced in the revision IEEE.

For special numeric tasks eg in computer graphics and for educational purposes, shorter number of formats exist as single precision.

More accuracy classes

  • Half accuracy or minifloats
  • Double precision
  • Quadruple precision
  • Floating-point number
  • Numerical Mathematics
  • Computer arithmetic
298953
de