Wide Floating Point
Digit Ordering
In a WideFloat number, each digit is
Assume the digit packets of a wide number are declared as follows in
 |
unsigned int d[6]; |
For a WideInteger, d[0] would be the least significant digit, and d[5] the most significant digit.
For a WideFloat, d[0] would be the most significant digit, and d[5] the least significant digit.
For WideInteger these digits are to the left of the radix point in reverse
For WideFloat the first digit is to the left of the radix point, with the remaining digits following in sequence after the radix
In the following examples on this page, all digit packets that are not specified are assumed to be zero.
The number 0x200000030000004 is stored in a WideInteger as:
|
d[0]=4; d[1]=3; d[2]=2; |
That same number is stored in a WideFloat as:
|
d[0]=2; d[1]=3; d[2]=4; |
The WideFloat exponent will be 2, indicating that the radix point is two digit positions to the right compared to a standard normalized significand of
To store the number 1.0 in a WideFloat:
|
d[0]=1; |
The exponent will be 0. The significand is already in normal form in which the radix point is assumed to be between
Next we will store the number 1.5. That is 1½, which is
|
d[0]=1; d[1]=0x20000000; |
The exponent will be 0. The significand is already in normal form.
To understand why d[1] is 0x20000000, consider the polynomial with
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Now we will store the number 0.5, which is
|
d[0]=0x20000000; |
The exponent will be –1, indicating that the radix point must move one digit position to the left compared to the standard normalized significand of