Quadruple precision
From Wikipedia, the free encyclopedia
In computing, quadruple precision (also commonly shortened to quad precision) is a computer numbering format that occupies four storage locations in computer memory at address, address+1, address+2, and address+3. A quad-precision number, sometimes simply a quad, may be defined to be an integer, fixed point, or floating point.
In the draft IEEE 754r proposed revised floating point standard, this 128-bit format is officially referred to as binary128. It is the fourth basic binary floating point interchange format, together with the 64-bit double precision, the 32-bit single precision, and the 16-bit half precision formats.
Contents |
[edit] Quadruple precision memory format
Sign bit: 1 Exponent width: 15 Significand precision: 112 (113 implicit)
The format is written with an implicit integer bit with value 1 unless the written exponent is all zeros. Thus only 112 bits of the fraction appear in the memory format, but the total precision is 113 bits (approximately 34 decimal digits, log10(2113)).
syyy yyyy yyyy yyyy xxxx xxxx xxxx xxxx … xxxx xxxx (112 xs)
[edit] Exponent encodings
Emin (0x0001) = -16382 Emax (0x7ffe) = 16383 Exponent bias (0x3fff) = 16383
The true exponent = written exponent - exponent bias
0x0000 and 0x7fff are reserved exponents 0x0000 is used to represent zero and denormals 0x7fff is used to represent infinity and NaNs
All bit patterns are valid encodings.
[edit] Quadruple precision examples in hexadecimal
3fff 0000 0000 0000 0000 0000 0000 0000 = 1
c000 0000 0000 0000 0000 0000 0000 0000 = -2
7fef ffff ffff ffff ffff ffff ffff ffff ~ 2.37946299071446353017 x 104932 (Max Quad)
3ffe 5555 5555 5555 5555 5555 5555 5555 ~ 1/3
By default, 1/3 rounds down like double precision, because of the odd number of bits in the significand. So the bits beyond the rounding point are 0101... which is less than 1/2 of a unit in the last place.
0000 0000 0000 0000 0000 0000 0000 0000 = 0 8000 0000 0000 0000 0000 0000 0000 0000 = -0
7fff 0000 0000 0000 0000 0000 0000 0000 = Infinity ffff 0000 0000 0000 0000 0000 0000 0000 = -Infinity
[edit] See also
- half precision – single precision – double precision – quadruple precision
- Floating point
- long double
