floorf4_fast

NAME

floorf4_fast - return approximate floors of float elements

SYNOPSIS

Procedure call syntax:

#include <simdmath.h>

vector float floorf4_fast(vector float x);

Link with -lsimdmath

Inline call syntax:

#include <simdmath.h>

#include <floorf4.h>

vector float _floorf4_fast(vector float x);

Parameters
x	input vector

DESCRIPTION

These functions round the elements of the input vector downwards to the next integer value.

They provide floor computation for the entire input range of IEEE floating point numbers. The floor of NaN values remain NaN, and the floor of subnormal values become zero.

On the SPU floorf4_fast provides a limited range form which computes the floor of all floating-point values in the 32-bit signed integer range. Values outside this range get clamped to either 0 or MAX_INT. This mode is faster to compute, but has less range.

RETURN VALUE

The function floorf4_fast returns a float vector in which each element is defined as the largest integer value not greater than x.

ENVIRONMENT

SPU only

CONFORMING TO

SIMD Math library specification for the Cell Broadband Engine™ Architecture.

NOTES

Basis

ISO9899 (C99) floor functions.