| 1 |
9 |
jlechner |
/* -------------------------------------------------------------- */
|
| 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 |
|
|
|
| 43 |
|
|
#ifndef _EXP2D2_H_
|
| 44 |
|
|
#define _EXP2D2_H_ 1
|
| 45 |
|
|
|
| 46 |
|
|
#include <spu_intrinsics.h>
|
| 47 |
|
|
|
| 48 |
|
|
|
| 49 |
|
|
/*
|
| 50 |
|
|
* FUNCTION
|
| 51 |
|
|
* vector double _exp2d2(vector double x)
|
| 52 |
|
|
*
|
| 53 |
|
|
* DESCRIPTION
|
| 54 |
|
|
* _exp2d2 computes 2 raised to the input x for each
|
| 55 |
|
|
* of the double word elements of x. Computation is
|
| 56 |
|
|
* performed by observing the 2^(a+b) = 2^a * 2^b.
|
| 57 |
|
|
* We decompose x into a and b (above) by letting.
|
| 58 |
|
|
* a = ceil(x), b = x - a;
|
| 59 |
|
|
*
|
| 60 |
|
|
* 2^a is easily computed by placing a into the exponent
|
| 61 |
|
|
* or a floating point number whose mantissa is all zeros.
|
| 62 |
|
|
*
|
| 63 |
|
|
* 2^b is computed using the polynomial approximation.
|
| 64 |
|
|
*
|
| 65 |
|
|
* __13_
|
| 66 |
|
|
* \
|
| 67 |
|
|
* \
|
| 68 |
|
|
* 2^x = / Ci*x^i
|
| 69 |
|
|
* /____
|
| 70 |
|
|
* i=0
|
| 71 |
|
|
*
|
| 72 |
|
|
* for x in the range 0.0 to 1.0.
|
| 73 |
|
|
*
|
| 74 |
|
|
*/
|
| 75 |
|
|
#define EXP_C00 1.0
|
| 76 |
|
|
#define EXP_C01 6.93147180559945286227e-01
|
| 77 |
|
|
#define EXP_C02 2.40226506959100694072e-01
|
| 78 |
|
|
#define EXP_C03 5.55041086648215761801e-02
|
| 79 |
|
|
#define EXP_C04 9.61812910762847687873e-03
|
| 80 |
|
|
#define EXP_C05 1.33335581464284411157e-03
|
| 81 |
|
|
#define EXP_C06 1.54035303933816060656e-04
|
| 82 |
|
|
#define EXP_C07 1.52527338040598376946e-05
|
| 83 |
|
|
#define EXP_C08 1.32154867901443052734e-06
|
| 84 |
|
|
#define EXP_C09 1.01780860092396959520e-07
|
| 85 |
|
|
#define EXP_C10 7.05491162080112087744e-09
|
| 86 |
|
|
#define EXP_C11 4.44553827187081007394e-10
|
| 87 |
|
|
#define EXP_C12 2.56784359934881958182e-11
|
| 88 |
|
|
#define EXP_C13 1.36914888539041240648e-12
|
| 89 |
|
|
|
| 90 |
|
|
static __inline vector double _exp2d2(vector double vx)
|
| 91 |
|
|
{
|
| 92 |
|
|
vec_int4 ix, exp;
|
| 93 |
|
|
vec_uint4 overflow, underflow;
|
| 94 |
|
|
vec_float4 vxf;
|
| 95 |
|
|
vec_double2 p1, p2, x2, x4, x8;
|
| 96 |
|
|
vec_double2 vy, vxw, out_of_range;
|
| 97 |
|
|
|
| 98 |
|
|
/* Compute: vxw = x - ceil(x)
|
| 99 |
|
|
*/
|
| 100 |
|
|
vxw = spu_add(vx, spu_splats(0.5));
|
| 101 |
|
|
vxf = spu_roundtf(vxw);
|
| 102 |
|
|
ix = spu_convts(vxf, 0);
|
| 103 |
|
|
ix = spu_add(ix, (vec_int4)spu_andc(spu_cmpgt(spu_splats(0.0f), vxf), spu_cmpeq(ix, spu_splats((int)0x80000000))));
|
| 104 |
|
|
vxf = spu_convtf(ix, 0);
|
| 105 |
|
|
vxw = spu_sub(vx, spu_extend(vxf));
|
| 106 |
|
|
|
| 107 |
|
|
/* Detect overflow and underflow. If overflow, force the result
|
| 108 |
|
|
* to infinity (at the end).
|
| 109 |
|
|
*/
|
| 110 |
|
|
exp = spu_shuffle(ix, ix, ((vec_uchar16) { 0,1,2,3, 0,1,2,3, 8,9,10,11, 8,9,10,11 }));
|
| 111 |
|
|
|
| 112 |
|
|
overflow = spu_cmpgt(exp, 1023);
|
| 113 |
|
|
underflow = spu_cmpgt(exp, -1023);
|
| 114 |
|
|
out_of_range = (vec_double2)spu_and(overflow, ((vec_uint4) { 0x7FF00000, 0, 0x7FF00000, 0 }));
|
| 115 |
|
|
|
| 116 |
|
|
/* Calculate the result by evaluating the 13th order polynomial.
|
| 117 |
|
|
* For efficiency, the polynomial is broken into two parts and
|
| 118 |
|
|
* evaluate then using nested
|
| 119 |
|
|
*
|
| 120 |
|
|
* result = (((((c13*x + c12)*x + c11)*x + c10)*x + c9)*x + c8)*x^8 +
|
| 121 |
|
|
* ((((((c7*x + c6)*x + c5)*x + c4)*x + c3)*x + c2)*x + c1)*x + c0
|
| 122 |
|
|
*/
|
| 123 |
|
|
p2 = spu_madd(spu_splats(EXP_C07), vxw, spu_splats(EXP_C06));
|
| 124 |
|
|
p1 = spu_madd(spu_splats(EXP_C13), vxw, spu_splats(EXP_C12));
|
| 125 |
|
|
x2 = spu_mul(vxw, vxw);
|
| 126 |
|
|
p2 = spu_madd(vxw, p2, spu_splats(EXP_C05));
|
| 127 |
|
|
p1 = spu_madd(vxw, p1, spu_splats(EXP_C11));
|
| 128 |
|
|
x4 = spu_mul(x2, x2);
|
| 129 |
|
|
p2 = spu_madd(vxw, p2, spu_splats(EXP_C04));
|
| 130 |
|
|
p1 = spu_madd(vxw, p1, spu_splats(EXP_C10));
|
| 131 |
|
|
p2 = spu_madd(vxw, p2, spu_splats(EXP_C03));
|
| 132 |
|
|
p1 = spu_madd(vxw, p1, spu_splats(EXP_C09));
|
| 133 |
|
|
x8 = spu_mul(x4, x4);
|
| 134 |
|
|
p2 = spu_madd(vxw, p2, spu_splats(EXP_C02));
|
| 135 |
|
|
p1 = spu_madd(vxw, p1, spu_splats(EXP_C08));
|
| 136 |
|
|
p2 = spu_madd(vxw, p2, spu_splats(EXP_C01));
|
| 137 |
|
|
p2 = spu_madd(vxw, p2, spu_splats(EXP_C00));
|
| 138 |
|
|
vy = spu_madd(x8, p1, p2);
|
| 139 |
|
|
|
| 140 |
|
|
/* Align the integer integer portion of x with the exponent.
|
| 141 |
|
|
*/
|
| 142 |
|
|
ix = spu_sl(ix, ((vec_uint4) { 20, 32, 20, 32 }));
|
| 143 |
|
|
vy = (vec_double2)spu_add((vec_int4)vy, ix);
|
| 144 |
|
|
|
| 145 |
|
|
/* Select the result if not overflow or underflow. Otherwise select the
|
| 146 |
|
|
* the out of range value.
|
| 147 |
|
|
*/
|
| 148 |
|
|
return (spu_sel(vy, out_of_range, (vec_ullong2)spu_orc(overflow, underflow)));
|
| 149 |
|
|
}
|
| 150 |
|
|
|
| 151 |
|
|
#endif /* _EXP2D2_H_ */
|
| 152 |
|
|
#endif /* __SPU__ */
|
