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#include <clc/clcmacro.h> 25#include <clc/math/math.h> 26 27_CLC_OVERLOAD _CLC_DEF float exp2(float x) { 28 29 // Reduce x 30 const float ln2HI = 0x1.62e300p-1f; 31 const float ln2LO = 0x1.2fefa2p-17f; 32 33 float t = rint(x); 34 int p = (int)t; 35 float tt = x - t; 36 float hi = tt * ln2HI; 37 float lo = tt * ln2LO; 38 39 // Evaluate poly 40 t = hi + lo; 41 tt = t*t; 42 float v = mad(tt, 43 -mad(tt, 44 mad(tt, 45 mad(tt, 46 mad(tt, 0x1.637698p-25f, -0x1.bbd41cp-20f), 47 0x1.1566aap-14f), 48 -0x1.6c16c2p-9f), 49 0x1.555556p-3f), 50 t); 51 52 float y = 1.0f - (((-lo) - MATH_DIVIDE(t * v, 2.0f - v)) - hi); 53 54 // Scale by 2^p 55 float r = as_float(as_int(y) + (p << 23)); 56 57 const float ulim = 128.0f; 58 const float llim = -126.0f; 59 60 r = x < llim ? 0.0f : r; 61 r = x < ulim ? r : as_float(0x7f800000); 62 return isnan(x) ? x : r; 63} 64 65_CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, float, exp2, float) 66 67#ifdef cl_khr_fp64 68 69#include "exp_helper.h" 70 71#pragma OPENCL EXTENSION cl_khr_fp64 : enable 72 73_CLC_OVERLOAD _CLC_DEF double exp2(double x) { 74 const double R_LN2 = 0x1.62e42fefa39efp-1; // ln(2) 75 const double R_1_BY_64 = 1.0 / 64.0; 76 77 int n = convert_int(x * 64.0); 78 double r = R_LN2 * fma(-R_1_BY_64, (double)n, x); 79 return __clc_exp_helper(x, -1074.0, 1024.0, r, n); 80} 81 82 83_CLC_UNARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, double, exp2, double) 84 85#endif 86