| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-options "-O2" } */ |
| 3 |
/* { dg-final { check-function-bodies "**" "" "" } } */ |
| 4 |
|
| 5 |
#include "arm_neon.h" |
| 6 |
|
| 7 |
// SIGNED VADD INTRINSICS |
| 8 |
|
| 9 |
/* |
| 10 |
**test_vadd_s8: |
| 11 |
** addp v0\.8b, v0\.8b, v1\.8b |
| 12 |
** ret |
| 13 |
*/ |
| 14 |
int8x8_t test_vadd_s8(int8x8_t v1, int8x8_t v2) { |
| 15 |
int8x8_t v3 = vpadd_s8(v1, v2); |
| 16 |
return v3; |
| 17 |
} |
| 18 |
|
| 19 |
/* |
| 20 |
**test_vadd_s16: |
| 21 |
**addp v0\.4h, v0\.4h, v1\.4h |
| 22 |
**ret |
| 23 |
*/ |
| 24 |
int16x4_t test_vadd_s16(int16x4_t v1, int16x4_t v2) { |
| 25 |
int16x4_t v3 = vpadd_s16(v1, v2); |
| 26 |
return v3; |
| 27 |
} |
| 28 |
|
| 29 |
/* |
| 30 |
**test_vadd_s32: |
| 31 |
** addp v0\.2s, v0\.2s, v1\.2s |
| 32 |
** ret |
| 33 |
*/ |
| 34 |
int32x2_t test_vadd_s32(int32x2_t v1, int32x2_t v2) { |
| 35 |
int32x2_t v3 = vpadd_s32(v1, v2); |
| 36 |
return v3; |
| 37 |
} |
| 38 |
|
| 39 |
/* |
| 40 |
**test_vaddq_s8: |
| 41 |
**... |
| 42 |
** addp v0\.16b, v0\.16b, v1\.16b |
| 43 |
** ret |
| 44 |
*/ |
| 45 |
int8x16_t test_vaddq_s8(int8x16_t v1, int8x16_t v2) { |
| 46 |
int8x16_t v3 = vpaddq_s8(v1, v2); |
| 47 |
return v3; |
| 48 |
} |
| 49 |
|
| 50 |
/* |
| 51 |
**test_vaddq_s16: |
| 52 |
**... |
| 53 |
** addp v0\.8h, v0\.8h, v1\.8h |
| 54 |
** ret |
| 55 |
*/ |
| 56 |
int16x8_t test_vaddq_s16(int16x8_t v1, int16x8_t v2) { |
| 57 |
int16x8_t v3 = vpaddq_s16(v1, v2); |
| 58 |
return v3; |
| 59 |
} |
| 60 |
|
| 61 |
/* |
| 62 |
**test_vaddq_s32: |
| 63 |
**... |
| 64 |
** addp v0\.4s, v0\.4s, v1\.4s |
| 65 |
** ret |
| 66 |
*/ |
| 67 |
int32x4_t test_vaddq_s32(int32x4_t v1, int32x4_t v2) { |
| 68 |
int32x4_t v3 = vpaddq_s32(v1, v2); |
| 69 |
return v3; |
| 70 |
} |
| 71 |
|
| 72 |
/* |
| 73 |
**test_vaddq_s64: |
| 74 |
**... |
| 75 |
** addp v0\.2d, v0\.2d, v1\.2d |
| 76 |
** ret |
| 77 |
*/ |
| 78 |
int64x2_t test_vaddq_s64(int64x2_t v1, int64x2_t v2) { |
| 79 |
int64x2_t v3 = vpaddq_s64(v1, v2); |
| 80 |
return v3; |
| 81 |
} |
| 82 |
|
| 83 |
/* |
| 84 |
**test_vaddd_s64: |
| 85 |
**... |
| 86 |
** addp (d[0-9]+), v0\.2d |
| 87 |
** fmov x0, \1 |
| 88 |
** ret |
| 89 |
*/ |
| 90 |
int64_t test_vaddd_s64(int64x2_t v1) { |
| 91 |
int64_t v2 = vpaddd_s64(v1); |
| 92 |
return v2; |
| 93 |
} |
| 94 |
|
| 95 |
/* |
| 96 |
**test_vaddl_s8: |
| 97 |
**... |
| 98 |
** saddlp v0\.4h, v0\.8b |
| 99 |
** ret |
| 100 |
*/ |
| 101 |
int16x4_t test_vaddl_s8(int8x8_t v1) { |
| 102 |
int16x4_t v2 = vpaddl_s8(v1); |
| 103 |
return v2; |
| 104 |
} |
| 105 |
|
| 106 |
/* |
| 107 |
**test_vaddlq_s8: |
| 108 |
**... |
| 109 |
** saddlp v0\.8h, v0\.16b |
| 110 |
** ret |
| 111 |
*/ |
| 112 |
int16x8_t test_vaddlq_s8(int8x16_t v1) { |
| 113 |
int16x8_t v2 = vpaddlq_s8(v1); |
| 114 |
return v2; |
| 115 |
} |
| 116 |
/* |
| 117 |
**test_vaddl_s16: |
| 118 |
**... |
| 119 |
** saddlp v0\.2s, v0\.4h |
| 120 |
** ret |
| 121 |
*/ |
| 122 |
int32x2_t test_vaddl_s16(int16x4_t v1) { |
| 123 |
int32x2_t v2 = vpaddl_s16(v1); |
| 124 |
return v2; |
| 125 |
} |
| 126 |
|
| 127 |
/* |
| 128 |
**test_vaddlq_s16: |
| 129 |
**... |
| 130 |
** saddlp v0\.4s, v0\.8h |
| 131 |
** ret |
| 132 |
*/ |
| 133 |
int32x4_t test_vaddlq_s16(int16x8_t v1) { |
| 134 |
int32x4_t v2 = vpaddlq_s16(v1); |
| 135 |
return v2; |
| 136 |
} |
| 137 |
|
| 138 |
/* |
| 139 |
**test_vaddl_s32: |
| 140 |
**... |
| 141 |
** saddlp v0\.1d, v0\.2s |
| 142 |
** ret |
| 143 |
*/ |
| 144 |
int64x1_t test_vaddl_s32(int32x2_t v1) { |
| 145 |
int64x1_t v2 = vpaddl_s32(v1); |
| 146 |
return v2; |
| 147 |
} |
| 148 |
|
| 149 |
/* |
| 150 |
**test_vaddlq_s32: |
| 151 |
**... |
| 152 |
** saddlp v0\.2d, v0\.4s |
| 153 |
** ret |
| 154 |
*/ |
| 155 |
int64x2_t test_vaddlq_s32(int32x4_t v1) { |
| 156 |
int64x2_t v2 = vpaddlq_s32(v1); |
| 157 |
return v2; |
| 158 |
} |
| 159 |
|
| 160 |
// UNSIGNED VADD INTRINSICS |
| 161 |
|
| 162 |
/* |
| 163 |
**test_vadd_u8: |
| 164 |
**... |
| 165 |
** addp v0\.8b, v0\.8b, v1\.8b |
| 166 |
** ret |
| 167 |
*/ |
| 168 |
uint8x8_t test_vadd_u8(uint8x8_t v1, uint8x8_t v2) { |
| 169 |
uint8x8_t v3 = vpadd_u8(v1, v2); |
| 170 |
return v3; |
| 171 |
} |
| 172 |
|
| 173 |
/* |
| 174 |
**test_vadd_u16: |
| 175 |
**... |
| 176 |
** addp v0\.4h, v0\.4h, v1\.4h |
| 177 |
** ret |
| 178 |
*/ |
| 179 |
uint16x4_t test_vadd_u16(uint16x4_t v1, uint16x4_t v2) { |
| 180 |
uint16x4_t v3 = vpadd_u16(v1, v2); |
| 181 |
return v3; |
| 182 |
} |
| 183 |
|
| 184 |
/* |
| 185 |
**test_vadd_u32: |
| 186 |
**... |
| 187 |
** addp v0\.2s, v0\.2s, v1\.2s |
| 188 |
** ret |
| 189 |
*/ |
| 190 |
uint32x2_t test_vadd_u32(uint32x2_t v1, uint32x2_t v2) { |
| 191 |
uint32x2_t v3 = vpadd_u32(v1, v2); |
| 192 |
return v3; |
| 193 |
} |
| 194 |
|
| 195 |
/* |
| 196 |
**test_vaddq_u8: |
| 197 |
**... |
| 198 |
** addp v0\.16b, v0\.16b, v1\.16b |
| 199 |
** ret |
| 200 |
*/ |
| 201 |
uint8x16_t test_vaddq_u8(uint8x16_t v1, uint8x16_t v2) { |
| 202 |
uint8x16_t v3 = vpaddq_u8(v1, v2); |
| 203 |
return v3; |
| 204 |
} |
| 205 |
|
| 206 |
/* |
| 207 |
**test_vaddq_u16: |
| 208 |
**... |
| 209 |
** addp v0\.8h, v0\.8h, v1\.8h |
| 210 |
** ret |
| 211 |
*/ |
| 212 |
uint16x8_t test_vaddq_u16(uint16x8_t v1, uint16x8_t v2) { |
| 213 |
uint16x8_t v3 = vpaddq_u16(v1, v2); |
| 214 |
return v3; |
| 215 |
} |
| 216 |
|
| 217 |
/* |
| 218 |
**test_vaddq_u32: |
| 219 |
**... |
| 220 |
** addp v0\.4s, v0\.4s, v1\.4s |
| 221 |
** ret |
| 222 |
*/ |
| 223 |
uint32x4_t test_vaddq_u32(uint32x4_t v1, uint32x4_t v2) { |
| 224 |
uint32x4_t v3 = vpaddq_u32(v1, v2); |
| 225 |
return v3; |
| 226 |
} |
| 227 |
|
| 228 |
/* |
| 229 |
**test_vaddq_u64: |
| 230 |
**... |
| 231 |
** addp v0\.2d, v0\.2d, v1\.2d |
| 232 |
** ret |
| 233 |
*/ |
| 234 |
uint64x2_t test_vaddq_u64(uint64x2_t v1, uint64x2_t v2) { |
| 235 |
uint64x2_t v3 = vpaddq_u64(v1, v2); |
| 236 |
return v3; |
| 237 |
} |
| 238 |
|
| 239 |
/* |
| 240 |
**test_vaddd_u64: |
| 241 |
**... |
| 242 |
** addp (d[0-9]+), v0\.2d |
| 243 |
** fmov x0, \1 |
| 244 |
** ret |
| 245 |
*/ |
| 246 |
uint64_t test_vaddd_u64(uint64x2_t v1) { |
| 247 |
uint64_t v2 = vpaddd_u64(v1); |
| 248 |
return v2; |
| 249 |
} |
| 250 |
|
| 251 |
/* |
| 252 |
**test_vaddl_u8: |
| 253 |
**... |
| 254 |
** uaddlp v0\.4h, v0\.8b |
| 255 |
** ret |
| 256 |
*/ |
| 257 |
uint16x4_t test_vaddl_u8(uint8x8_t v1) { |
| 258 |
uint16x4_t v2 = vpaddl_u8(v1); |
| 259 |
return v2; |
| 260 |
} |
| 261 |
|
| 262 |
/* |
| 263 |
**test_vaddlq_u8: |
| 264 |
**... |
| 265 |
** uaddlp v0\.8h, v0\.16b |
| 266 |
** ret |
| 267 |
*/ |
| 268 |
uint16x8_t test_vaddlq_u8(uint8x16_t v1) { |
| 269 |
uint16x8_t v2 = vpaddlq_u8(v1); |
| 270 |
return v2; |
| 271 |
} |
| 272 |
/* |
| 273 |
**test_vaddl_u16: |
| 274 |
**... |
| 275 |
** uaddlp v0\.2s, v0\.4h |
| 276 |
** ret |
| 277 |
*/ |
| 278 |
uint32x2_t test_vaddl_u16(uint16x4_t v1) { |
| 279 |
uint32x2_t v2 = vpaddl_u16(v1); |
| 280 |
return v2; |
| 281 |
} |
| 282 |
|
| 283 |
/* |
| 284 |
**test_vaddlq_u16: |
| 285 |
**... |
| 286 |
** uaddlp v0\.4s, v0\.8h |
| 287 |
** ret |
| 288 |
*/ |
| 289 |
uint32x4_t test_vaddlq_u16(uint16x8_t v1) { |
| 290 |
uint32x4_t v2 = vpaddlq_u16(v1); |
| 291 |
return v2; |
| 292 |
} |
| 293 |
|
| 294 |
/* |
| 295 |
**test_vaddl_u32: |
| 296 |
**... |
| 297 |
** uaddlp v0\.1d, v0\.2s |
| 298 |
** ret |
| 299 |
*/ |
| 300 |
uint64x1_t test_vaddl_u32(uint32x2_t v1) { |
| 301 |
uint64x1_t v2 = vpaddl_u32(v1); |
| 302 |
return v2; |
| 303 |
} |
| 304 |
|
| 305 |
/* |
| 306 |
**test_vaddlq_u32: |
| 307 |
**... |
| 308 |
** uaddlp v0\.2d, v0\.4s |
| 309 |
** ret |
| 310 |
*/ |
| 311 |
uint64x2_t test_vaddlq_u32(uint32x4_t v1) { |
| 312 |
uint64x2_t v2 = vpaddlq_u32(v1); |
| 313 |
return v2; |
| 314 |
} |
| 315 |
|
| 316 |
// FLOATING POINT VADD INTRINSICS |
| 317 |
|
| 318 |
/* |
| 319 |
**test_vadd_f32: |
| 320 |
**... |
| 321 |
** faddp v0\.2s, v0\.2s, v1\.2s |
| 322 |
** ret |
| 323 |
*/ |
| 324 |
float32x2_t test_vadd_f32(float32x2_t v1, float32x2_t v2) { |
| 325 |
float32x2_t v3 = vpadd_f32(v1, v2); |
| 326 |
return v3; |
| 327 |
} |
| 328 |
|
| 329 |
/* |
| 330 |
**test_vaddq_f32: |
| 331 |
**... |
| 332 |
** faddp v0\.4s, v0\.4s, v1\.4s |
| 333 |
** ret |
| 334 |
*/ |
| 335 |
float32x4_t test_vaddq_f32(float32x4_t v1, float32x4_t v2) { |
| 336 |
float32x4_t v3 = vpaddq_f32(v1, v2); |
| 337 |
return v3; |
| 338 |
} |
| 339 |
|
| 340 |
/* |
| 341 |
**test_vaddq_f64: |
| 342 |
**... |
| 343 |
** faddp v0\.2d, v0\.2d, v1\.2d |
| 344 |
** ret |
| 345 |
*/ |
| 346 |
float64x2_t test_vaddq_f64(float64x2_t v1, float64x2_t v2) { |
| 347 |
float64x2_t v3 = vpaddq_f64(v1, v2); |
| 348 |
return v3; |
| 349 |
} |
| 350 |
|
| 351 |
/* |
| 352 |
**test_vadds_f32: |
| 353 |
**... |
| 354 |
** faddp s0, v0\.2s |
| 355 |
** ret |
| 356 |
*/ |
| 357 |
float32_t test_vadds_f32(float32x2_t v1) { |
| 358 |
float32_t v2 = vpadds_f32(v1); |
| 359 |
return v2; |
| 360 |
} |
| 361 |
|
| 362 |
/* |
| 363 |
**test_vaddd_f64: |
| 364 |
**... |
| 365 |
** faddp d0, v0\.2d |
| 366 |
** ret |
| 367 |
*/ |
| 368 |
float64_t test_vaddd_f64(float64x2_t v1) { |
| 369 |
float64_t v2 = vpaddd_f64(v1); |
| 370 |
return v2; |
| 371 |
} |