35 lines
1.1 KiB
Groff
35 lines
1.1 KiB
Groff
[comment {-*- flibs -*- doctools manpage}]
|
|
[manpage_begin select_precision n 1.0]
|
|
[copyright {2008 Arjen Markus <arjenmarkus@sourceforge.net>}]
|
|
[moddesc flibs]
|
|
[titledesc {Select the precision for real variables}]
|
|
|
|
[description]
|
|
|
|
The auxiliary [term select_precision] module defines three parameters
|
|
by which you can select the precision for real variables and constants:
|
|
|
|
[list_begin bullet]
|
|
[bullet]
|
|
[term sp] - Single precision
|
|
|
|
[bullet]
|
|
[term dp] - Double precision (the real kind whose precision is larger
|
|
than that of single precision)
|
|
|
|
[bullet]
|
|
[term wp] - Working precision. If you consistently use [term wp] as the
|
|
kind for your real variables and constants, then switching between
|
|
single and double precision is merely a matter of setting [term wp] to
|
|
either [term sp] or [term dp] in this module and recompiling.
|
|
|
|
[list_end]
|
|
|
|
[strong Note:] One of the pitfalls in using double precision is that
|
|
literal constants, such as [term 1.2345678901234567890] are not
|
|
automatically interpreted as double precision. You need to specify the
|
|
kind explicitly: [term 1.2345678901234567890_wp] will define the
|
|
constant in the working precision.
|
|
|
|
[manpage_end]
|