| 1 |
Long double format |
| 2 |
================== |
| 3 |
|
| 4 |
Each long double is made up of two IEEE doubles. The value of the |
| 5 |
long double is the sum of the values of the two parts (except for |
| 6 |
-0.0). The most significant part is required to be the value of the |
| 7 |
long double rounded to the nearest double, as specified by IEEE. For |
| 8 |
Inf values, the least significant part is required to be one of +0.0 |
| 9 |
or -0.0. No other requirements are made; so, for example, 1.0 may be |
| 10 |
represented as (1.0, +0.0) or (1.0, -0.0), and the low part of a NaN |
| 11 |
is don't-care. |
| 12 |
|
| 13 |
Classification |
| 14 |
-------------- |
| 15 |
|
| 16 |
A long double can represent any value of the form |
| 17 |
s * 2^e * sum(k=0...105: f_k * 2^(-k)) |
| 18 |
where 's' is +1 or -1, 'e' is between 1022 and -968 inclusive, f_0 is |
| 19 |
1, and f_k for k>0 is 0 or 1. These are the 'normal' long doubles. |
| 20 |
|
| 21 |
A long double can also represent any value of the form |
| 22 |
s * 2^-968 * sum(k=0...105: f_k * 2^(-k)) |
| 23 |
where 's' is +1 or -1, f_0 is 0, and f_k for k>0 is 0 or 1. These are |
| 24 |
the 'subnormal' long doubles. |
| 25 |
|
| 26 |
There are four long doubles that represent zero, two that represent |
| 27 |
+0.0 and two that represent -0.0. The sign of the high part is the |
| 28 |
sign of the long double, and the sign of the low part is ignored. |
| 29 |
|
| 30 |
Likewise, there are four long doubles that represent infinities, two |
| 31 |
for +Inf and two for -Inf. |
| 32 |
|
| 33 |
Each NaN, quiet or signalling, that can be represented as a 'double' |
| 34 |
can be represented as a 'long double'. In fact, there are 2^64 |
| 35 |
equivalent representations for each one. |
| 36 |
|
| 37 |
There are certain other valid long doubles where both parts are |
| 38 |
nonzero but the low part represents a value which has a bit set below |
| 39 |
2^(e-105). These, together with the subnormal long doubles, make up |
| 40 |
the denormal long doubles. |
| 41 |
|
| 42 |
Many possible long double bit patterns are not valid long doubles. |
| 43 |
These do not represent any value. |
| 44 |
|
| 45 |
Limits |
| 46 |
------ |
| 47 |
|
| 48 |
The maximum representable long double is 2^1024-2^918. The smallest |
| 49 |
*normal* positive long double is 2^-968. The smallest denormalised |
| 50 |
positive long double is 2^-1074 (this is the same as for 'double'). |
| 51 |
|
| 52 |
Conversions |
| 53 |
----------- |
| 54 |
|
| 55 |
A double can be converted to a long double by adding a zero low part. |
| 56 |
|
| 57 |
A long double can be converted to a double by removing the low part. |
| 58 |
|
| 59 |
Comparisons |
| 60 |
----------- |
| 61 |
|
| 62 |
Two long doubles can be compared by comparing the high parts, and if |
| 63 |
those compare equal, comparing the low parts. |
| 64 |
|
| 65 |
Arithmetic |
| 66 |
---------- |
| 67 |
|
| 68 |
The unary negate operation operates by negating the low and high parts. |
| 69 |
|
| 70 |
An absolute or absolute-negate operation must be done by comparing |
| 71 |
against zero and negating if necessary. |
| 72 |
|
| 73 |
Addition and subtraction are performed using library routines. They |
| 74 |
are not at present performed perfectly accurately, the result produced |
| 75 |
will be within 1ulp of the range generated by adding or subtracting |
| 76 |
1ulp from the input values, where a 'ulp' is 2^(e-106) given the |
| 77 |
exponent 'e'. In the presence of cancellation, this may be |
| 78 |
arbitrarily inaccurate. Subtraction is done by negation and addition. |
| 79 |
|
| 80 |
Multiplication is also performed using a library routine. Its result |
| 81 |
will be within 2ulp of the correct result. |
| 82 |
|
| 83 |
Division is also performed using a library routine. Its result will |
| 84 |
be within 3ulp of the correct result. |
| 85 |
|
| 86 |
|
| 87 |
Copyright (C) 2004-2025 Free Software Foundation, Inc. |
| 88 |
|
| 89 |
Copying and distribution of this file, with or without modification, |
| 90 |
are permitted in any medium without royalty provided the copyright |
| 91 |
notice and this notice are preserved. |