1/*
2 * Copyright (c) 2014,2015 Advanced Micro Devices, Inc.
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a copy
5 * of this software and associated documentation files (the "Software"), to deal
6 * in the Software without restriction, including without limitation the rights
7 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
8 * copies of the Software, and to permit persons to whom the Software is
9 * furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice shall be included in
12 * all copies or substantial portions of the Software.
13 *
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
17 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
18 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
19 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
20 * THE SOFTWARE.
21 */
22
23#include <clc/clc.h>
24
25#include "math.h"
26#include "../clcmacro.h"
27
28_CLC_OVERLOAD _CLC_DEF float exp2(float x) {
29
30    // Reduce x
31    const float ln2HI = 0x1.62e300p-1f;
32    const float ln2LO = 0x1.2fefa2p-17f;
33
34    float t = rint(x);
35    int p = (int)t;
36    float tt = x - t;
37    float hi = tt * ln2HI;
38    float lo = tt * ln2LO;
39
40    // Evaluate poly
41    t = hi + lo;
42    tt  = t*t;
43    float v = mad(tt,
44                  -mad(tt,
45		       mad(tt,
46		           mad(tt,
47			       mad(tt, 0x1.637698p-25f, -0x1.bbd41cp-20f),
48                               0x1.1566aap-14f),
49                           -0x1.6c16c2p-9f),
50                       0x1.555556p-3f),
51                  t);
52
53    float y = 1.0f - (((-lo) - MATH_DIVIDE(t * v, 2.0f - v)) - hi);
54
55    // Scale by 2^p
56    float r =  as_float(as_int(y) + (p << 23));
57
58    const float ulim =  128.0f;
59    const float llim = -126.0f;
60
61    r = x < llim ? 0.0f : r;
62    r = x < ulim ? r : as_float(0x7f800000);
63    return isnan(x) ? x : r;
64}
65
66_CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, float, exp2, float)
67
68#ifdef cl_khr_fp64
69
70#include "exp_helper.h"
71
72#pragma OPENCL EXTENSION cl_khr_fp64 : enable
73
74_CLC_OVERLOAD _CLC_DEF double exp2(double x) {
75    const double R_LN2 = 0x1.62e42fefa39efp-1; // ln(2)
76    const double R_1_BY_64 = 1.0 / 64.0;
77
78    int n = convert_int(x * 64.0);
79    double r = R_LN2 * fma(-R_1_BY_64, (double)n, x);
80    return __clc_exp_helper(x, -1074.0, 1024.0, r, n);
81}
82
83
84_CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, double, exp2, double)
85
86#endif
87