Sample code of using vrcp28pd with accuracy corrections.
- test.cpp (a.out): without corrections
- test2.cpp (b.out): with corrections
$ make
$ ./a.out
1.0028231394486751
0 01111111111 0000000010111001000001000110101110011010101001101100
1.0028231400601602
0 01111111111 0000000010111001000001000110111000111010111111001000
29
(snip)
1.1815988407141602
0 01111111111 0010111001111101010000101111100111011011110111000110
1.1815988415114642
0 01111111111 0010111001111101010000101111110101001000100000011000
29
$ ./b.out
1.0028231394486751
0 01111111111 0000000010111001000001000110101110011010101001101100
1.0028231394486751
0 01111111111 0000000010111001000001000110101110011010101001101100
52
(snip)
1.1815988407141602
0 01111111111 0010111001111101010000101111100111011011110111000110
1.1815988407141602
0 01111111111 0010111001111101010000101111100111011011110111000110
52