softfloat.h revision 1.1
1 2 /* 3 =============================================================================== 4 5 This C header file is part of the SoftFloat IEC/IEEE Floating-point 6 Arithmetic Package, Release 2a. 7 8 Written by John R. Hauser. This work was made possible in part by the 9 International Computer Science Institute, located at Suite 600, 1947 Center 10 Street, Berkeley, California 94704. Funding was partially provided by the 11 National Science Foundation under grant MIP-9311980. The original version 12 of this code was written as part of a project to build a fixed-point vector 13 processor in collaboration with the University of California at Berkeley, 14 overseen by Profs. Nelson Morgan and John Wawrzynek. More information 15 is available through the Web page `http://HTTP.CS.Berkeley.EDU/~jhauser/ 16 arithmetic/SoftFloat.html'. 17 18 THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort 19 has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT 20 TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO 21 PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY 22 AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE. 23 24 Derivative works are acceptable, even for commercial purposes, so long as 25 (1) they include prominent notice that the work is derivative, and (2) they 26 include prominent notice akin to these four paragraphs for those parts of 27 this code that are retained. 28 29 =============================================================================== 30 */ 31 32 /* 33 ------------------------------------------------------------------------------- 34 The macro `FLOATX80' must be defined to enable the extended double-precision 35 floating-point format `floatx80'. If this macro is not defined, the 36 `floatx80' type will not be defined, and none of the functions that either 37 input or output the `floatx80' type will be defined. The same applies to 38 the `FLOAT128' macro and the quadruple-precision format `float128'. 39 ------------------------------------------------------------------------------- 40 */ 41 /* #define FLOATX80 */ 42 /* #define FLOAT128 */ 43 44 /* 45 ------------------------------------------------------------------------------- 46 Software IEC/IEEE floating-point types. 47 ------------------------------------------------------------------------------- 48 */ 49 typedef unsigned int float32; 50 typedef unsigned long long float64; 51 #ifdef FLOATX80 52 typedef struct { 53 unsigned short high; 54 unsigned long long low; 55 } floatx80; 56 #endif 57 #ifdef FLOAT128 58 typedef struct { 59 unsigned long long high, low; 60 } float128; 61 #endif 62 63 /* 64 ------------------------------------------------------------------------------- 65 Software IEC/IEEE floating-point underflow tininess-detection mode. 66 ------------------------------------------------------------------------------- 67 */ 68 extern int float_detect_tininess; 69 enum { 70 float_tininess_after_rounding = 0, 71 float_tininess_before_rounding = 1 72 }; 73 74 /* 75 ------------------------------------------------------------------------------- 76 Software IEC/IEEE floating-point rounding mode. 77 ------------------------------------------------------------------------------- 78 */ 79 extern int float_rounding_mode; 80 enum { 81 float_round_nearest_even = 0, 82 float_round_to_zero = 1, 83 float_round_down = 2, 84 float_round_up = 3 85 }; 86 87 /* 88 ------------------------------------------------------------------------------- 89 Software IEC/IEEE floating-point exception flags. 90 ------------------------------------------------------------------------------- 91 */ 92 extern int float_exception_flags; 93 enum { 94 float_flag_inexact = 1, 95 float_flag_underflow = 2, 96 float_flag_overflow = 4, 97 float_flag_divbyzero = 8, 98 float_flag_invalid = 16 99 }; 100 101 /* 102 ------------------------------------------------------------------------------- 103 Routine to raise any or all of the software IEC/IEEE floating-point 104 exception flags. 105 ------------------------------------------------------------------------------- 106 */ 107 void float_raise( int ); 108 109 /* 110 ------------------------------------------------------------------------------- 111 Software IEC/IEEE integer-to-floating-point conversion routines. 112 ------------------------------------------------------------------------------- 113 */ 114 float32 int32_to_float32( int ); 115 float64 int32_to_float64( int ); 116 #ifdef FLOATX80 117 floatx80 int32_to_floatx80( int ); 118 #endif 119 #ifdef FLOAT128 120 float128 int32_to_float128( int ); 121 #endif 122 float32 int64_to_float32( long long ); 123 float64 int64_to_float64( long long ); 124 #ifdef FLOATX80 125 floatx80 int64_to_floatx80( long long ); 126 #endif 127 #ifdef FLOAT128 128 float128 int64_to_float128( long long ); 129 #endif 130 131 /* 132 ------------------------------------------------------------------------------- 133 Software IEC/IEEE single-precision conversion routines. 134 ------------------------------------------------------------------------------- 135 */ 136 int float32_to_int32( float32 ); 137 int float32_to_int32_round_to_zero( float32 ); 138 long long float32_to_int64( float32 ); 139 long long float32_to_int64_round_to_zero( float32 ); 140 float64 float32_to_float64( float32 ); 141 #ifdef FLOATX80 142 floatx80 float32_to_floatx80( float32 ); 143 #endif 144 #ifdef FLOAT128 145 float128 float32_to_float128( float32 ); 146 #endif 147 148 /* 149 ------------------------------------------------------------------------------- 150 Software IEC/IEEE single-precision operations. 151 ------------------------------------------------------------------------------- 152 */ 153 float32 float32_round_to_int( float32 ); 154 float32 float32_add( float32, float32 ); 155 float32 float32_sub( float32, float32 ); 156 float32 float32_mul( float32, float32 ); 157 float32 float32_div( float32, float32 ); 158 float32 float32_rem( float32, float32 ); 159 float32 float32_sqrt( float32 ); 160 int float32_eq( float32, float32 ); 161 int float32_le( float32, float32 ); 162 int float32_lt( float32, float32 ); 163 int float32_eq_signaling( float32, float32 ); 164 int float32_le_quiet( float32, float32 ); 165 int float32_lt_quiet( float32, float32 ); 166 #ifndef SOFTFLOAT_FOR_GCC 167 int float32_is_signaling_nan( float32 ); 168 #endif 169 170 /* 171 ------------------------------------------------------------------------------- 172 Software IEC/IEEE double-precision conversion routines. 173 ------------------------------------------------------------------------------- 174 */ 175 int float64_to_int32( float64 ); 176 int float64_to_int32_round_to_zero( float64 ); 177 long long float64_to_int64( float64 ); 178 long long float64_to_int64_round_to_zero( float64 ); 179 float32 float64_to_float32( float64 ); 180 #ifdef FLOATX80 181 floatx80 float64_to_floatx80( float64 ); 182 #endif 183 #ifdef FLOAT128 184 float128 float64_to_float128( float64 ); 185 #endif 186 187 /* 188 ------------------------------------------------------------------------------- 189 Software IEC/IEEE double-precision operations. 190 ------------------------------------------------------------------------------- 191 */ 192 float64 float64_round_to_int( float64 ); 193 float64 float64_add( float64, float64 ); 194 float64 float64_sub( float64, float64 ); 195 float64 float64_mul( float64, float64 ); 196 float64 float64_div( float64, float64 ); 197 float64 float64_rem( float64, float64 ); 198 float64 float64_sqrt( float64 ); 199 int float64_eq( float64, float64 ); 200 int float64_le( float64, float64 ); 201 int float64_lt( float64, float64 ); 202 int float64_eq_signaling( float64, float64 ); 203 int float64_le_quiet( float64, float64 ); 204 int float64_lt_quiet( float64, float64 ); 205 #ifndef SOFTFLOAT_FOR_GCC 206 int float64_is_signaling_nan( float64 ); 207 #endif 208 209 #ifdef FLOATX80 210 211 /* 212 ------------------------------------------------------------------------------- 213 Software IEC/IEEE extended double-precision conversion routines. 214 ------------------------------------------------------------------------------- 215 */ 216 int floatx80_to_int32( floatx80 ); 217 int floatx80_to_int32_round_to_zero( floatx80 ); 218 long long floatx80_to_int64( floatx80 ); 219 long long floatx80_to_int64_round_to_zero( floatx80 ); 220 float32 floatx80_to_float32( floatx80 ); 221 float64 floatx80_to_float64( floatx80 ); 222 #ifdef FLOAT128 223 float128 floatx80_to_float128( floatx80 ); 224 #endif 225 226 /* 227 ------------------------------------------------------------------------------- 228 Software IEC/IEEE extended double-precision rounding precision. Valid 229 values are 32, 64, and 80. 230 ------------------------------------------------------------------------------- 231 */ 232 extern int floatx80_rounding_precision; 233 234 /* 235 ------------------------------------------------------------------------------- 236 Software IEC/IEEE extended double-precision operations. 237 ------------------------------------------------------------------------------- 238 */ 239 floatx80 floatx80_round_to_int( floatx80 ); 240 floatx80 floatx80_add( floatx80, floatx80 ); 241 floatx80 floatx80_sub( floatx80, floatx80 ); 242 floatx80 floatx80_mul( floatx80, floatx80 ); 243 floatx80 floatx80_div( floatx80, floatx80 ); 244 floatx80 floatx80_rem( floatx80, floatx80 ); 245 floatx80 floatx80_sqrt( floatx80 ); 246 int floatx80_eq( floatx80, floatx80 ); 247 int floatx80_le( floatx80, floatx80 ); 248 int floatx80_lt( floatx80, floatx80 ); 249 int floatx80_eq_signaling( floatx80, floatx80 ); 250 int floatx80_le_quiet( floatx80, floatx80 ); 251 int floatx80_lt_quiet( floatx80, floatx80 ); 252 int floatx80_is_signaling_nan( floatx80 ); 253 254 #endif 255 256 #ifdef FLOAT128 257 258 /* 259 ------------------------------------------------------------------------------- 260 Software IEC/IEEE quadruple-precision conversion routines. 261 ------------------------------------------------------------------------------- 262 */ 263 int float128_to_int32( float128 ); 264 int float128_to_int32_round_to_zero( float128 ); 265 long long float128_to_int64( float128 ); 266 long long float128_to_int64_round_to_zero( float128 ); 267 float32 float128_to_float32( float128 ); 268 float64 float128_to_float64( float128 ); 269 #ifdef FLOATX80 270 floatx80 float128_to_floatx80( float128 ); 271 #endif 272 273 /* 274 ------------------------------------------------------------------------------- 275 Software IEC/IEEE quadruple-precision operations. 276 ------------------------------------------------------------------------------- 277 */ 278 float128 float128_round_to_int( float128 ); 279 float128 float128_add( float128, float128 ); 280 float128 float128_sub( float128, float128 ); 281 float128 float128_mul( float128, float128 ); 282 float128 float128_div( float128, float128 ); 283 float128 float128_rem( float128, float128 ); 284 float128 float128_sqrt( float128 ); 285 int float128_eq( float128, float128 ); 286 int float128_le( float128, float128 ); 287 int float128_lt( float128, float128 ); 288 int float128_eq_signaling( float128, float128 ); 289 int float128_le_quiet( float128, float128 ); 290 int float128_lt_quiet( float128, float128 ); 291 int float128_is_signaling_nan( float128 ); 292 293 #endif 294 295
Indexes created Mon Sep 22 13:09:51 GMT 2025