Datatype Conversions

Support for converting floating-point numbers to fixed-point and fixed-point numbers to floating-point. More...

Overview

Support for converting floating-point numbers to fixed-point and fixed-point numbers to floating-point.

These intrinsics can generate exceptions, for more information you can go here.

Note

These intrinsics do real conversions. No bit reinterpretation takes place.

Functions
float	fix2float (int n, int sft)
	Fixed-point to Floating-point conversion with scaling. More...
float	fix2float (int n)
	Fixed-point to Floating-point conversion without scaling. More...
int	float2fix (float n, int sft)
	Floating-point to Fixed-point conversion with scaling. More...
int	float2fix (float n)
	Floating-point to Fixed-point conversion without scaling. More...

Function Documentation

float fix2float	(	int	n,
		int	sft
	)

Fixed-point to Floating-point conversion with scaling.

Parameters

n	Integer input value
sft	Binary point of input value. Range [-32:31].

Example:

If the input value has 8 fractional bits, then 0.5 in fixed point will be expressed as 128. To convert this value to float use the following code, which stores 0.5 in floating-point variable b:

int   n = 128;
float b = fix2float(n,8);

float fix2float

(

int

n

)

Fixed-point to Floating-point conversion without scaling.

Parameters

n	Integer input value

int float2fix	(	float	n,
		int	sft
	)

Floating-point to Fixed-point conversion with scaling.

Parameters

n	Floating point input value
sft	Binary point of output value. Range [-32:31].

Returns

round(n * pow(2.0, sft)); // round to nearest integer

Example:

The input value is 0.5 and the output shall be of format Q24.8 (meaning 8 fractional bits out of 128 bit integer output word). To convert this value to float use the following code, which stores 128 in fixed-point variable a:

float   n = 0.5;
int a = float2fix(n,8);

Note

To achieve better performance, consider using float2fix(v8float v, int sft) instead

int float2fix

(

float

n

)

Floating-point to Fixed-point conversion without scaling.

Parameters

n	Floating point input value