softfloat.h revision 1.2
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 #include "softfloat-qp.h" 45 46 #include <machine/ieeefp.h> 47 48 /* 49 ------------------------------------------------------------------------------- 50 Software IEC/IEEE floating-point types. 51 ------------------------------------------------------------------------------- 52 */ 53 typedef unsigned int float32; 54 typedef unsigned long long float64; 55 #ifdef FLOATX80 56 typedef struct { 57 unsigned short high; 58 unsigned long long low; 59 } floatx80; 60 #endif 61 #ifdef FLOAT128 62 typedef struct { 63 unsigned long long high, low; 64 } float128; 65 #endif 66 67 /* 68 ------------------------------------------------------------------------------- 69 Software IEC/IEEE floating-point underflow tininess-detection mode. 70 ------------------------------------------------------------------------------- 71 */ 72 extern int8 float_detect_tininess; 73 enum { 74 float_tininess_after_rounding = 0, 75 float_tininess_before_rounding = 1 76 }; 77 78 /* 79 ------------------------------------------------------------------------------- 80 Software IEC/IEEE floating-point rounding mode. 81 ------------------------------------------------------------------------------- 82 */ 83 extern int8 float_rounding_mode; 84 enum { 85 float_round_nearest_even = FP_RN, 86 float_round_to_zero = FP_RZ, 87 float_round_down = FP_RM, 88 float_round_up = FP_RP 89 }; 90 91 /* 92 ------------------------------------------------------------------------------- 93 Software IEC/IEEE floating-point exception flags. 94 ------------------------------------------------------------------------------- 95 */ 96 extern int8 float_exception_flags; 97 enum { 98 float_flag_inexact = FP_X_IMP, 99 float_flag_underflow = FP_X_UFL, 100 float_flag_overflow = FP_X_OFL, 101 float_flag_divbyzero = FP_X_DZ, 102 float_flag_invalid = FP_X_INV 103 }; 104 105 /* 106 ------------------------------------------------------------------------------- 107 Routine to raise any or all of the software IEC/IEEE floating-point 108 exception flags. 109 ------------------------------------------------------------------------------- 110 */ 111 void float_raise( int8 ); 112 113 /* 114 ------------------------------------------------------------------------------- 115 Software IEC/IEEE integer-to-floating-point conversion routines. 116 ------------------------------------------------------------------------------- 117 */ 118 float32 int32_to_float32( int ); 119 float64 int32_to_float64( int ); 120 #ifdef FLOATX80 121 floatx80 int32_to_floatx80( int ); 122 #endif 123 #ifdef FLOAT128 124 float128 int32_to_float128( int ); 125 #endif 126 float32 int64_to_float32( long long ); 127 float64 int64_to_float64( long long ); 128 #ifdef FLOATX80 129 floatx80 int64_to_floatx80( long long ); 130 #endif 131 #ifdef FLOAT128 132 float128 int64_to_float128( long long ); 133 #endif 134 135 /* 136 ------------------------------------------------------------------------------- 137 Software IEC/IEEE single-precision conversion routines. 138 ------------------------------------------------------------------------------- 139 */ 140 int float32_to_int32( float32 ); 141 int float32_to_int32_round_to_zero( float32 ); 142 unsigned int float32_to_uint32_round_to_zero( float32 ); 143 long long float32_to_int64( float32 ); 144 long long float32_to_int64_round_to_zero( float32 ); 145 float64 float32_to_float64( float32 ); 146 #ifdef FLOATX80 147 floatx80 float32_to_floatx80( float32 ); 148 #endif 149 #ifdef FLOAT128 150 float128 float32_to_float128( float32 ); 151 #endif 152 153 /* 154 ------------------------------------------------------------------------------- 155 Software IEC/IEEE single-precision operations. 156 ------------------------------------------------------------------------------- 157 */ 158 float32 float32_round_to_int( float32 ); 159 float32 float32_add( float32, float32 ); 160 float32 float32_sub( float32, float32 ); 161 float32 float32_mul( float32, float32 ); 162 float32 float32_div( float32, float32 ); 163 float32 float32_rem( float32, float32 ); 164 float32 float32_sqrt( float32 ); 165 flag float32_eq( float32, float32 ); 166 flag float32_le( float32, float32 ); 167 flag float32_lt( float32, float32 ); 168 flag float32_eq_signaling( float32, float32 ); 169 flag float32_le_quiet( float32, float32 ); 170 flag float32_lt_quiet( float32, float32 ); 171 flag float32_is_signaling_nan( float32 ); 172 173 /* 174 ------------------------------------------------------------------------------- 175 Software IEC/IEEE double-precision conversion routines. 176 ------------------------------------------------------------------------------- 177 */ 178 int float64_to_int32( float64 ); 179 int float64_to_int32_round_to_zero( float64 ); 180 unsigned int float64_to_uint32_round_to_zero( float64 ); 181 long long float64_to_int64( float64 ); 182 long long float64_to_int64_round_to_zero( float64 ); 183 float32 float64_to_float32( float64 ); 184 #ifdef FLOATX80 185 floatx80 float64_to_floatx80( float64 ); 186 #endif 187 #ifdef FLOAT128 188 float128 float64_to_float128( float64 ); 189 #endif 190 191 /* 192 ------------------------------------------------------------------------------- 193 Software IEC/IEEE double-precision operations. 194 ------------------------------------------------------------------------------- 195 */ 196 float64 float64_round_to_int( float64 ); 197 float64 float64_add( float64, float64 ); 198 float64 float64_sub( float64, float64 ); 199 float64 float64_mul( float64, float64 ); 200 float64 float64_div( float64, float64 ); 201 float64 float64_rem( float64, float64 ); 202 float64 float64_sqrt( float64 ); 203 flag float64_eq( float64, float64 ); 204 flag float64_le( float64, float64 ); 205 flag float64_lt( float64, float64 ); 206 flag float64_eq_signaling( float64, float64 ); 207 flag float64_le_quiet( float64, float64 ); 208 flag float64_lt_quiet( float64, float64 ); 209 flag float64_is_signaling_nan( float64 ); 210 211 #ifdef FLOATX80 212 213 /* 214 ------------------------------------------------------------------------------- 215 Software IEC/IEEE extended double-precision conversion routines. 216 ------------------------------------------------------------------------------- 217 */ 218 int floatx80_to_int32( floatx80 ); 219 int floatx80_to_int32_round_to_zero( floatx80 ); 220 long long floatx80_to_int64( floatx80 ); 221 long long floatx80_to_int64_round_to_zero( floatx80 ); 222 float32 floatx80_to_float32( floatx80 ); 223 float64 floatx80_to_float64( floatx80 ); 224 #ifdef FLOAT128 225 float128 floatx80_to_float128( floatx80 ); 226 #endif 227 228 /* 229 ------------------------------------------------------------------------------- 230 Software IEC/IEEE extended double-precision rounding precision. Valid 231 values are 32, 64, and 80. 232 ------------------------------------------------------------------------------- 233 */ 234 extern int floatx80_rounding_precision; 235 236 /* 237 ------------------------------------------------------------------------------- 238 Software IEC/IEEE extended double-precision operations. 239 ------------------------------------------------------------------------------- 240 */ 241 floatx80 floatx80_round_to_int( floatx80 ); 242 floatx80 floatx80_add( floatx80, floatx80 ); 243 floatx80 floatx80_sub( floatx80, floatx80 ); 244 floatx80 floatx80_mul( floatx80, floatx80 ); 245 floatx80 floatx80_div( floatx80, floatx80 ); 246 floatx80 floatx80_rem( floatx80, floatx80 ); 247 floatx80 floatx80_sqrt( floatx80 ); 248 flag floatx80_eq( floatx80, floatx80 ); 249 flag floatx80_le( floatx80, floatx80 ); 250 flag floatx80_lt( floatx80, floatx80 ); 251 flag floatx80_eq_signaling( floatx80, floatx80 ); 252 flag floatx80_le_quiet( floatx80, floatx80 ); 253 flag floatx80_lt_quiet( floatx80, floatx80 ); 254 flag floatx80_is_signaling_nan( floatx80 ); 255 256 #endif 257 258 #ifdef FLOAT128 259 260 /* 261 ------------------------------------------------------------------------------- 262 Software IEC/IEEE quadruple-precision conversion routines. 263 ------------------------------------------------------------------------------- 264 */ 265 int float128_to_int32( float128 ); 266 int float128_to_int32_round_to_zero( float128 ); 267 long long float128_to_int64( float128 ); 268 long long float128_to_int64_round_to_zero( float128 ); 269 float32 float128_to_float32( float128 ); 270 float64 float128_to_float64( float128 ); 271 #ifdef FLOATX80 272 floatx80 float128_to_floatx80( float128 ); 273 #endif 274 275 /* 276 ------------------------------------------------------------------------------- 277 Software IEC/IEEE quadruple-precision operations. 278 ------------------------------------------------------------------------------- 279 */ 280 float128 float128_round_to_int( float128 ); 281 float128 float128_add( float128, float128 ); 282 float128 float128_sub( float128, float128 ); 283 float128 float128_mul( float128, float128 ); 284 float128 float128_div( float128, float128 ); 285 float128 float128_rem( float128, float128 ); 286 float128 float128_sqrt( float128 ); 287 flag float128_eq( float128, float128 ); 288 flag float128_le( float128, float128 ); 289 flag float128_lt( float128, float128 ); 290 flag float128_eq_signaling( float128, float128 ); 291 flag float128_le_quiet( float128, float128 ); 292 flag float128_lt_quiet( float128, float128 ); 293 flag float128_is_signaling_nan( float128 ); 294 flag float128_is_nan( float128 ); 295 296 #endif 297
Indexes created Sat Sep 27 14:09:57 GMT 2025