Retro68/gcc/newlib/libm/machine/spu/headers/floorf.h

124 lines
4.1 KiB
C
Raw Normal View History

/*
(C) Copyright 2001,2006,
International Business Machines Corporation,
Sony Computer Entertainment, Incorporated,
Toshiba Corporation,
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the names of the copyright holders nor the names of their
contributors may be used to endorse or promote products derived from this
software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS
IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef _FLOORF_H_
#define _FLOORF_H_ 1
#include <spu_intrinsics.h>
#include "headers/vec_literal.h"
/*
* FUNCTION
* float _floorf(float value)
*
* DESCRIPTION
* The _floorf routine round the input value "value" downwards to the
* nearest integer returning the result as a float. Two forms of the
* floor function are provided - full range and limited (integer)
* range.
*
* The full range form (default) provides floor computation on
* all IEEE floating point values. The floor of NANs remain NANs.
* The floor of denorms results in zero.
*
* The limited range form (selected by defining FLOOR_INTEGER_RANGE)
* compute ths floor of all floating-point values in the 32-bit
* signed integer range. Values outside this range get clamped.
*/
static __inline float _floorf(float value)
{
#ifdef FLOOR_INTEGER_RANGE
/* 32-BIT INTEGER DYNAMIC RANGE
*/
union {
float f;
signed int i;
unsigned int ui;
} bias;
bias.f = value;
/* If positive, bias the input value to truncate towards
* positive infinity, instead of zero.
*/
bias.ui = (unsigned int)(bias.i >> 31) & 0x3F7FFFFF;
value -= bias.f;
/* Remove fraction bits by casting to an integer and back
* to a floating-point value.
*/
return ((float)((int)value));
#else /* !FLOOR_INTEGER_RANGE */
/* FULL FLOATING-POINT RANGE
*/
vec_int4 exp, shift;
vec_uint4 mask, frac_mask, addend, insert, pos;
vec_float4 in, out;
in = spu_promote(value, 0);
/* This function generates the following component
* based upon the inputs.
*
* mask = bits of the input that need to be replaced.
* insert = value of the bits that need to be replaced
* addend = value to be added to perform function.
*
* These are applied as follows:.
*
* out = ((in & mask) | insert) + addend
*/
pos = spu_cmpgt((vec_int4)in, -1);
exp = spu_and(spu_rlmask((vec_int4)in, -23), 0xFF);
shift = spu_sub(127, exp);
frac_mask = spu_and(spu_rlmask(VEC_SPLAT_U32(0x7FFFFF), shift),
spu_cmpgt((vec_int4)shift, -31));
mask = spu_orc(frac_mask, spu_cmpgt(exp, 126));
addend = spu_andc(spu_andc(spu_add(mask, 1), pos),
spu_cmpeq(spu_and((vec_uint4)in, mask), 0));
insert = spu_andc(spu_andc(VEC_SPLAT_U32(0xBF800000), pos),
spu_cmpgt((vec_uint4)spu_add(exp, -1), 126));
out = (vec_float4)spu_add(spu_sel((vec_uint4)in, insert, mask), addend);
return (spu_extract(out, 0));
#endif /* FLOOR_INTEGER_RANGE */
}
#endif /* _FLOORF_H_ */