| 1 |
148 |
jeremybenn |
/* -------------------------------------------------------------- */
|
| 2 |
|
|
/* (C)Copyright 2001,2008, */
|
| 3 |
|
|
/* International Business Machines Corporation, */
|
| 4 |
|
|
/* Sony Computer Entertainment, Incorporated, */
|
| 5 |
|
|
/* Toshiba Corporation, */
|
| 6 |
|
|
/* */
|
| 7 |
|
|
/* All Rights Reserved. */
|
| 8 |
|
|
/* */
|
| 9 |
|
|
/* Redistribution and use in source and binary forms, with or */
|
| 10 |
|
|
/* without modification, are permitted provided that the */
|
| 11 |
|
|
/* following conditions are met: */
|
| 12 |
|
|
/* */
|
| 13 |
|
|
/* - Redistributions of source code must retain the above copyright*/
|
| 14 |
|
|
/* notice, this list of conditions and the following disclaimer. */
|
| 15 |
|
|
/* */
|
| 16 |
|
|
/* - Redistributions in binary form must reproduce the above */
|
| 17 |
|
|
/* copyright notice, this list of conditions and the following */
|
| 18 |
|
|
/* disclaimer in the documentation and/or other materials */
|
| 19 |
|
|
/* provided with the distribution. */
|
| 20 |
|
|
/* */
|
| 21 |
|
|
/* - Neither the name of IBM Corporation nor the names of its */
|
| 22 |
|
|
/* contributors may be used to endorse or promote products */
|
| 23 |
|
|
/* derived from this software without specific prior written */
|
| 24 |
|
|
/* permission. */
|
| 25 |
|
|
/* */
|
| 26 |
|
|
/* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND */
|
| 27 |
|
|
/* CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, */
|
| 28 |
|
|
/* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF */
|
| 29 |
|
|
/* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE */
|
| 30 |
|
|
/* DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR */
|
| 31 |
|
|
/* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, */
|
| 32 |
|
|
/* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT */
|
| 33 |
|
|
/* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; */
|
| 34 |
|
|
/* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) */
|
| 35 |
|
|
/* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN */
|
| 36 |
|
|
/* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR */
|
| 37 |
|
|
/* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, */
|
| 38 |
|
|
/* EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */
|
| 39 |
|
|
/* -------------------------------------------------------------- */
|
| 40 |
|
|
/* PROLOG END TAG zYx */
|
| 41 |
|
|
#ifdef __SPU__
|
| 42 |
|
|
#ifndef _POWF4_H_
|
| 43 |
|
|
#define _POWF4_H_ 1
|
| 44 |
|
|
|
| 45 |
|
|
#include <spu_intrinsics.h>
|
| 46 |
|
|
#include <vec_types.h>
|
| 47 |
|
|
|
| 48 |
|
|
#include "exp2f4.h"
|
| 49 |
|
|
#include "log2f4.h"
|
| 50 |
|
|
|
| 51 |
|
|
/*
|
| 52 |
|
|
* FUNCTION
|
| 53 |
|
|
* vector float _powf4(vector float x, vector float y)
|
| 54 |
|
|
*
|
| 55 |
|
|
* DESCRIPTION
|
| 56 |
|
|
* The _powf4 function computes x raised to the power y for the set of
|
| 57 |
|
|
* vectors. The powf4 function is computed as by decomposing
|
| 58 |
|
|
* the problem into:
|
| 59 |
|
|
*
|
| 60 |
|
|
* x^y = 2^(y*log2(x))
|
| 61 |
|
|
*
|
| 62 |
|
|
*/
|
| 63 |
|
|
static __inline vector float _powf4(vector float x, vector float y)
|
| 64 |
|
|
{
|
| 65 |
|
|
vec_uint4 y_exp;
|
| 66 |
|
|
vec_uint4 y_mantissa;
|
| 67 |
|
|
vec_uint4 mant_shift;
|
| 68 |
|
|
vec_uint4 y_is_int;
|
| 69 |
|
|
vec_uint4 y_is_odd;
|
| 70 |
|
|
vec_uint4 x_sign_bit;
|
| 71 |
|
|
vec_uint4 zero = (vec_uint4)spu_splats(0);
|
| 72 |
|
|
vec_uint4 bit0 = (vec_uint4)spu_splats(0x80000000);
|
| 73 |
|
|
vec_int4 error = spu_splats(-1);
|
| 74 |
|
|
vec_float4 out;
|
| 75 |
|
|
|
| 76 |
|
|
y_exp = spu_and(spu_rlmask((vec_uint4)y, -23), 0x000000FF);
|
| 77 |
|
|
|
| 78 |
|
|
/* Need the implied bit in the mantissa to catch
|
| 79 |
|
|
* y = 1 case later
|
| 80 |
|
|
*/
|
| 81 |
|
|
y_mantissa = spu_or(spu_sl((vec_uint4)y, (unsigned int)8), bit0);
|
| 82 |
|
|
|
| 83 |
|
|
x_sign_bit = spu_and((vec_uint4)x, bit0);
|
| 84 |
|
|
|
| 85 |
|
|
/* We are going to shift the mantissa over enough to
|
| 86 |
|
|
* determine if we have an integer.
|
| 87 |
|
|
*/
|
| 88 |
|
|
mant_shift = spu_add(y_exp, -127);
|
| 89 |
|
|
|
| 90 |
|
|
|
| 91 |
|
|
/* Leave the lowest-order integer bit of mantissa on the
|
| 92 |
|
|
* high end so we can see if the integer is odd.
|
| 93 |
|
|
*/
|
| 94 |
|
|
y_mantissa = spu_sl(y_mantissa, mant_shift);
|
| 95 |
|
|
|
| 96 |
|
|
y_is_int = spu_cmpeq(spu_andc(y_mantissa, bit0), 0);
|
| 97 |
|
|
y_is_int = spu_and(y_is_int, spu_cmpgt(y_exp, 126));
|
| 98 |
|
|
|
| 99 |
|
|
y_is_odd = spu_and(spu_cmpeq(y_mantissa, bit0), y_is_int);
|
| 100 |
|
|
|
| 101 |
|
|
out = _exp2f4(spu_mul(y, _log2f4(spu_andc(x, (vec_float4)bit0))));
|
| 102 |
|
|
|
| 103 |
|
|
/* x < 0 is only ok when y integer
|
| 104 |
|
|
*/
|
| 105 |
|
|
out = spu_sel(out, (vec_float4)error,
|
| 106 |
|
|
spu_andc(spu_cmpeq(x_sign_bit, bit0), y_is_int));
|
| 107 |
|
|
|
| 108 |
|
|
/* Preserve the sign of x if y is an odd integer
|
| 109 |
|
|
*/
|
| 110 |
|
|
out = spu_sel(out, spu_or(out, (vec_float4)x_sign_bit), y_is_odd);
|
| 111 |
|
|
|
| 112 |
|
|
/* x = anything, y = +/- 0, returns 1
|
| 113 |
|
|
*/
|
| 114 |
|
|
out = spu_sel(out, spu_splats(1.0f), spu_cmpabseq(y, (vec_float4)zero));
|
| 115 |
|
|
|
| 116 |
|
|
return(out);
|
| 117 |
|
|
}
|
| 118 |
|
|
|
| 119 |
|
|
#endif /* _POWF4_H_ */
|
| 120 |
|
|
#endif /* __SPU__ */
|
