dburgess | 82c46ff | 2011-10-07 02:40:51 +0000 | [diff] [blame] | 1 | /* |
| 2 | * Copyright 2009 Free Software Foundation, Inc. |
| 3 | * |
| 4 | * This software is distributed under the terms of the GNU Affero Public License. |
| 5 | * See the COPYING file in the main directory for details. |
| 6 | * |
| 7 | * This use of this software may be subject to additional restrictions. |
| 8 | * See the LEGAL file in the main directory for details. |
| 9 | |
| 10 | This program is free software: you can redistribute it and/or modify |
| 11 | it under the terms of the GNU Affero General Public License as published by |
| 12 | the Free Software Foundation, either version 3 of the License, or |
| 13 | (at your option) any later version. |
| 14 | |
| 15 | This program is distributed in the hope that it will be useful, |
| 16 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 17 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 18 | GNU Affero General Public License for more details. |
| 19 | |
| 20 | You should have received a copy of the GNU Affero General Public License |
| 21 | along with this program. If not, see <http://www.gnu.org/licenses/>. |
| 22 | |
| 23 | */ |
| 24 | |
| 25 | |
| 26 | #ifndef F16_H |
| 27 | #define F16_H |
| 28 | |
| 29 | #include <stdint.h> |
| 30 | #include <ostream> |
| 31 | |
| 32 | |
| 33 | |
| 34 | /** Round a float to the appropriate F16 value. */ |
| 35 | inline int32_t _f16_round(float f) |
| 36 | { |
| 37 | if (f>0.0F) return (int32_t)(f+0.5F); |
| 38 | if (f<0.0F) return (int32_t)(f-0.5F); |
| 39 | return 0; |
| 40 | } |
| 41 | |
| 42 | |
| 43 | |
| 44 | /** A class for F15.16 fixed point arithmetic with saturation. */ |
| 45 | class F16 { |
| 46 | |
| 47 | |
| 48 | private: |
| 49 | |
| 50 | int32_t mV; |
| 51 | |
| 52 | |
| 53 | public: |
| 54 | |
| 55 | F16() {} |
| 56 | |
| 57 | F16(int i) { mV = i<<16; } |
| 58 | F16(float f) { mV = _f16_round(f*65536.0F); } |
| 59 | F16(double f) { mV = _f16_round((float)f*65536.0F); } |
| 60 | |
| 61 | int32_t& raw() { return mV; } |
| 62 | const int32_t& raw() const { return mV; } |
| 63 | |
| 64 | float f() const { return mV/65536.0F; } |
| 65 | |
| 66 | //operator float() const { return mV/65536.0F; } |
| 67 | //operator int() const { return mV>>16; } |
| 68 | |
| 69 | F16 operator=(float f) |
| 70 | { |
| 71 | mV = _f16_round(f*65536.0F); |
| 72 | return *this; |
| 73 | } |
| 74 | |
| 75 | F16 operator=(int i) |
| 76 | { |
| 77 | mV = i<<16; |
| 78 | return *this; |
| 79 | } |
| 80 | |
| 81 | F16 operator=(const F16& other) |
| 82 | { |
| 83 | mV = other.mV; |
| 84 | return mV; |
| 85 | } |
| 86 | |
| 87 | F16 operator+(const F16& other) const |
| 88 | { |
| 89 | F16 retVal; |
| 90 | retVal.mV = mV + other.mV; |
| 91 | return retVal; |
| 92 | } |
| 93 | |
| 94 | F16& operator+=(const F16& other) |
| 95 | { |
| 96 | mV += other.mV; |
| 97 | return *this; |
| 98 | } |
| 99 | |
| 100 | F16 operator-(const F16& other) const |
| 101 | { |
| 102 | F16 retVal; |
| 103 | retVal.mV = mV - other.mV; |
| 104 | return retVal; |
| 105 | } |
| 106 | |
| 107 | F16& operator-=(const F16& other) |
| 108 | { |
| 109 | mV -= other.mV; |
| 110 | return *this; |
| 111 | } |
| 112 | |
| 113 | F16 operator*(const F16& other) const |
| 114 | { |
| 115 | F16 retVal; |
| 116 | int64_t p = (int64_t)mV * (int64_t)other.mV; |
| 117 | retVal.mV = p>>16; |
| 118 | return retVal; |
| 119 | } |
| 120 | |
| 121 | F16& operator*=(const F16& other) |
| 122 | { |
| 123 | int64_t p = (int64_t)mV * (int64_t)other.mV; |
| 124 | mV = p>>16; |
| 125 | return *this; |
| 126 | } |
| 127 | |
| 128 | F16 operator*(float f) const |
| 129 | { |
| 130 | F16 retVal; |
| 131 | retVal.mV = mV * f; |
| 132 | return retVal; |
| 133 | } |
| 134 | |
| 135 | F16& operator*=(float f) |
| 136 | { |
| 137 | mV *= f; |
| 138 | return *this; |
| 139 | } |
| 140 | |
| 141 | F16 operator/(const F16& other) const |
| 142 | { |
| 143 | F16 retVal; |
| 144 | int64_t pV = (int64_t)mV << 16; |
| 145 | retVal.mV = pV / other.mV; |
| 146 | return retVal; |
| 147 | } |
| 148 | |
| 149 | F16& operator/=(const F16& other) |
| 150 | { |
| 151 | int64_t pV = (int64_t)mV << 16; |
| 152 | mV = pV / other.mV; |
| 153 | return *this; |
| 154 | } |
| 155 | |
| 156 | F16 operator/(float f) const |
| 157 | { |
| 158 | F16 retVal; |
| 159 | retVal.mV = mV / f; |
| 160 | return retVal; |
| 161 | } |
| 162 | |
| 163 | F16& operator/=(float f) |
| 164 | { |
| 165 | mV /= f; |
| 166 | return *this; |
| 167 | } |
| 168 | |
| 169 | bool operator>(const F16& other) const |
| 170 | { |
| 171 | return mV>other.mV; |
| 172 | } |
| 173 | |
| 174 | bool operator<(const F16& other) const |
| 175 | { |
| 176 | return mV<other.mV; |
| 177 | } |
| 178 | |
| 179 | bool operator==(const F16& other) const |
| 180 | { |
| 181 | return mV==other.mV; |
| 182 | } |
| 183 | |
| 184 | bool operator>(float f) const |
| 185 | { |
| 186 | return (mV/65536.0F) > f; |
| 187 | } |
| 188 | |
| 189 | bool operator<(float f) const |
| 190 | { |
| 191 | return (mV/65536.0F) < f; |
| 192 | } |
| 193 | |
| 194 | bool operator==(float f) const |
| 195 | { |
| 196 | return (mV/65536.0F) == f; |
| 197 | } |
| 198 | |
| 199 | }; |
| 200 | |
| 201 | |
| 202 | |
| 203 | inline std::ostream& operator<<(std::ostream& os, const F16& v) |
| 204 | { |
| 205 | os << v.f(); |
| 206 | return os; |
| 207 | } |
| 208 | |
| 209 | #endif |
| 210 | |