NAME
rsqrt(), rsqrtf(), rsqrtl(), rsqrtw(), rsqrtq() — reciprocal square root functions
SYNOPSIS
#include <math.h>
double rsqrt(double x);
float rsqrtf(float x);
IA-64 ONLY:
long double rsqrtl(long double x);
extended rsqrtw(extended x);
quad rsqrtq(quad x);
DESCRIPTION
rsqrt()
returns the reciprocal of the non-negative square root of
x.
The expression
a * rsqrt(b)
is intended as a higher performance alternative to
a / sqrt(b).
The two expressions are comparably accurate, but do not compute
exactly the same value in every case. For example,
a * rsqrt(a*a + b*b)
can be just slightly greater than 1, in rare cases.
rsqrtf()
is a
float
version of
rsqrt();
it takes a
float
argument and returns a
float
result.
IA-64 ONLY:
rsqrtl()
is a
long double
version of
rsqrt();
it takes a
long double
argument and returns a
long double
result.
rsqrtw()
is an
extended
version of
rsqrt();
it takes an
extended
argument and returns an
extended
result.
rsqrtq()
is equivalent to
rsqrtl()
on HP-UX systems.
USAGE
To use these functions compile either with the default
-Ae
option or with the
-Aa
and the
-D_HPUX_SOURCE
options.
To use (for IA-64 systems)
rsqrtw()
or
rsqrtq(),
compile also with the
-fpwidetypes
option.
To use any of these functions, make sure your program includes
<math.h>,
and link in the math library by specifying
-lm
on the compiler or linker command line.
For more information, see the
HP-UX Floating-Point Guide.
RETURN VALUE
If
x
is ±0,
rsqrt()
returns ±INFINITY and raises the divide-by-zero exception.
If
x
is +INFINITY,
rsqrt()
returns +0.
If
x
is NaN,
rsqrt()
returns NaN.
If
x
is less than zero,
rsqrt()
returns NaN and raises the invalid floating-point exception.
rsqrt()
raises the inexact exception whenever a rounded result does not equal
the mathematical result.
ERRORS
No errors are defined.
STANDARDS CONFORMANCE
These functions are not specified by any standard.
Printable version
