Rev 1031 | Rev 1657 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 829 | cejka | 1 | /* |
| 2 | * Copyright (C) 2005 Josef Cejka |
||
| 3 | * All rights reserved. |
||
| 4 | * |
||
| 5 | * Redistribution and use in source and binary forms, with or without |
||
| 6 | * modification, are permitted provided that the following conditions |
||
| 7 | * are met: |
||
| 8 | * |
||
| 9 | * - Redistributions of source code must retain the above copyright |
||
| 10 | * notice, this list of conditions and the following disclaimer. |
||
| 11 | * - Redistributions in binary form must reproduce the above copyright |
||
| 12 | * notice, this list of conditions and the following disclaimer in the |
||
| 13 | * documentation and/or other materials provided with the distribution. |
||
| 14 | * - The name of the author may not be used to endorse or promote products |
||
| 15 | * derived from this software without specific prior written permission. |
||
| 16 | * |
||
| 17 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR |
||
| 18 | * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES |
||
| 19 | * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. |
||
| 20 | * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, |
||
| 21 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT |
||
| 22 | * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
||
| 23 | * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
||
| 24 | * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
||
| 25 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
||
| 26 | * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
||
| 27 | */ |
||
| 28 | |||
| 29 | #include<sftypes.h> |
||
| 30 | #include<common.h> |
||
| 31 | |||
| 874 | cejka | 32 | /* Table for fast leading zeroes counting */ |
| 33 | char zeroTable[256] = { |
||
| 34 | 8, 7, 7, 6, 6, 6, 6, 4, 4, 4, 4, 4, 4, 4, 4, \ |
||
| 35 | 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, \ |
||
| 36 | 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, \ |
||
| 37 | 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, \ |
||
| 38 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \ |
||
| 39 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \ |
||
| 40 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \ |
||
| 41 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, \ |
||
| 42 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \ |
||
| 43 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \ |
||
| 44 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \ |
||
| 45 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \ |
||
| 46 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \ |
||
| 47 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \ |
||
| 48 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, \ |
||
| 49 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 |
||
| 50 | }; |
||
| 51 | |||
| 52 | |||
| 53 | |||
| 829 | cejka | 54 | /** Take fraction shifted by 10 bits to left, round it, normalize it and detect exceptions |
| 1653 | cejka | 55 | * @param cexp exponent with bias |
| 829 | cejka | 56 | * @param cfrac fraction shifted 10 places left with added hidden bit |
| 1653 | cejka | 57 | * @param sign |
| 829 | cejka | 58 | * @return valied float64 |
| 59 | */ |
||
| 1031 | cejka | 60 | float64 finishFloat64(int32_t cexp, uint64_t cfrac, char sign) |
| 829 | cejka | 61 | { |
| 62 | float64 result; |
||
| 63 | |||
| 64 | result.parts.sign = sign; |
||
| 65 | |||
| 66 | /* find first nonzero digit and shift result and detect possibly underflow */ |
||
| 67 | while ((cexp > 0) && (cfrac) && (!(cfrac & (FLOAT64_HIDDEN_BIT_MASK << (64 - FLOAT64_FRACTION_SIZE - 1 ) )))) { |
||
| 68 | cexp--; |
||
| 69 | cfrac <<= 1; |
||
| 70 | /* TODO: fix underflow */ |
||
| 71 | }; |
||
| 72 | |||
| 834 | cejka | 73 | if ((cexp < 0) || ( cexp == 0 && (!(cfrac & (FLOAT64_HIDDEN_BIT_MASK << (64 - FLOAT64_FRACTION_SIZE - 1)))))) { |
| 74 | /* FIXME: underflow */ |
||
| 75 | result.parts.exp = 0; |
||
| 835 | cejka | 76 | if ((cexp + FLOAT64_FRACTION_SIZE + 1) < 0) { /* +1 is place for rounding */ |
| 834 | cejka | 77 | result.parts.fraction = 0; |
| 78 | return result; |
||
| 79 | } |
||
| 835 | cejka | 80 | |
| 834 | cejka | 81 | while (cexp < 0) { |
| 82 | cexp++; |
||
| 83 | cfrac >>= 1; |
||
| 84 | } |
||
| 835 | cejka | 85 | |
| 86 | cfrac += (0x1 << (64 - FLOAT64_FRACTION_SIZE - 3)); |
||
| 87 | |||
| 874 | cejka | 88 | if (!(cfrac & (FLOAT64_HIDDEN_BIT_MASK << (64 - FLOAT64_FRACTION_SIZE - 1)))) { |
| 834 | cejka | 89 | |
| 835 | cejka | 90 | result.parts.fraction = ((cfrac >>(64 - FLOAT64_FRACTION_SIZE - 2) ) & (~FLOAT64_HIDDEN_BIT_MASK)); |
| 91 | return result; |
||
| 92 | } |
||
| 93 | } else { |
||
| 94 | cfrac += (0x1 << (64 - FLOAT64_FRACTION_SIZE - 3)); |
||
| 834 | cejka | 95 | } |
| 96 | |||
| 97 | ++cexp; |
||
| 98 | |||
| 829 | cejka | 99 | if (cfrac & (FLOAT64_HIDDEN_BIT_MASK << (64 - FLOAT64_FRACTION_SIZE - 1 ))) { |
| 100 | ++cexp; |
||
| 101 | cfrac >>= 1; |
||
| 834 | cejka | 102 | } |
| 829 | cejka | 103 | |
| 104 | /* check overflow */ |
||
| 105 | if (cexp >= FLOAT64_MAX_EXPONENT ) { |
||
| 106 | /* FIXME: overflow, return infinity */ |
||
| 107 | result.parts.exp = FLOAT64_MAX_EXPONENT; |
||
| 108 | result.parts.fraction = 0; |
||
| 109 | return result; |
||
| 110 | } |
||
| 111 | |||
| 1031 | cejka | 112 | result.parts.exp = (uint32_t)cexp; |
| 829 | cejka | 113 | |
| 114 | result.parts.fraction = ((cfrac >>(64 - FLOAT64_FRACTION_SIZE - 2 ) ) & (~FLOAT64_HIDDEN_BIT_MASK)); |
||
| 115 | |||
| 116 | return result; |
||
| 117 | } |
||
| 834 | cejka | 118 | |
| 874 | cejka | 119 | /** Counts leading zeroes in 64bit unsigned integer |
| 120 | * @param i |
||
| 121 | */ |
||
| 1031 | cejka | 122 | int countZeroes64(uint64_t i) |
| 874 | cejka | 123 | { |
| 124 | int j; |
||
| 125 | for (j =0; j < 64; j += 8) { |
||
| 126 | if ( i & (0xFFll << (56 - j))) { |
||
| 127 | return (j + countZeroes8(i >> (56 - j))); |
||
| 128 | } |
||
| 129 | } |
||
| 130 | |||
| 131 | return 64; |
||
| 132 | } |
||
| 133 | |||
| 134 | /** Counts leading zeroes in 32bit unsigned integer |
||
| 135 | * @param i |
||
| 136 | */ |
||
| 1031 | cejka | 137 | int countZeroes32(uint32_t i) |
| 874 | cejka | 138 | { |
| 139 | int j; |
||
| 140 | for (j =0; j < 32; j += 8) { |
||
| 141 | if ( i & (0xFF << (24 - j))) { |
||
| 142 | return (j + countZeroes8(i >> (24 - j))); |
||
| 143 | } |
||
| 144 | } |
||
| 145 | |||
| 146 | return 32; |
||
| 147 | } |
||
| 148 | |||
| 149 | /** Counts leading zeroes in byte |
||
| 150 | * @param i |
||
| 151 | */ |
||
| 1031 | cejka | 152 | int countZeroes8(uint8_t i) |
| 874 | cejka | 153 | { |
| 154 | return zeroTable[i]; |
||
| 155 | } |
||
| 156 | |||
| 157 | /** Round and normalize number expressed by exponent and fraction with first bit (equal to hidden bit) at 30. bit |
||
| 158 | * @param exp exponent |
||
| 159 | * @param fraction part with hidden bit shifted to 30. bit |
||
| 160 | */ |
||
| 1031 | cejka | 161 | void roundFloat32(int32_t *exp, uint32_t *fraction) |
| 874 | cejka | 162 | { |
| 163 | /* rounding - if first bit after fraction is set then round up */ |
||
| 164 | (*fraction) += (0x1 << 6); |
||
| 165 | |||
| 166 | if ((*fraction) & (FLOAT32_HIDDEN_BIT_MASK << 8)) { |
||
| 167 | /* rounding overflow */ |
||
| 168 | ++(*exp); |
||
| 169 | (*fraction) >>= 1; |
||
| 170 | }; |
||
| 171 | |||
| 172 | if (((*exp) >= FLOAT32_MAX_EXPONENT ) || ((*exp) < 0)) { |
||
| 173 | /* overflow - set infinity as result */ |
||
| 174 | (*exp) = FLOAT32_MAX_EXPONENT; |
||
| 175 | (*fraction) = 0; |
||
| 176 | return; |
||
| 177 | } |
||
| 178 | |||
| 179 | return; |
||
| 180 | } |
||
| 181 | |||
| 182 | /** Round and normalize number expressed by exponent and fraction with first bit (equal to hidden bit) at 62. bit |
||
| 183 | * @param exp exponent |
||
| 184 | * @param fraction part with hidden bit shifted to 62. bit |
||
| 185 | */ |
||
| 1031 | cejka | 186 | void roundFloat64(int32_t *exp, uint64_t *fraction) |
| 874 | cejka | 187 | { |
| 188 | /* rounding - if first bit after fraction is set then round up */ |
||
| 189 | (*fraction) += (0x1 << 9); |
||
| 190 | |||
| 191 | if ((*fraction) & (FLOAT64_HIDDEN_BIT_MASK << 11)) { |
||
| 192 | /* rounding overflow */ |
||
| 193 | ++(*exp); |
||
| 194 | (*fraction) >>= 1; |
||
| 195 | }; |
||
| 196 | |||
| 197 | if (((*exp) >= FLOAT64_MAX_EXPONENT ) || ((*exp) < 0)) { |
||
| 198 | /* overflow - set infinity as result */ |
||
| 199 | (*exp) = FLOAT64_MAX_EXPONENT; |
||
| 200 | (*fraction) = 0; |
||
| 201 | return; |
||
| 202 | } |
||
| 203 | |||
| 204 | return; |
||
| 205 | } |
||
| 206 |