Dan Gohman | fce288f | 2009-09-09 00:09:15 +0000 | [diff] [blame] | 1 | ; RUN: llc < %s -march=arm -mattr=+neon > %t |
Bob Wilson | 5bafff3 | 2009-06-22 23:27:02 +0000 | [diff] [blame] | 2 | ; RUN: grep {vshl\\.s8} %t | count 2 |
| 3 | ; RUN: grep {vshl\\.s16} %t | count 2 |
| 4 | ; RUN: grep {vshl\\.s32} %t | count 2 |
| 5 | ; RUN: grep {vshl\\.s64} %t | count 2 |
| 6 | ; RUN: grep {vshl\\.u8} %t | count 4 |
| 7 | ; RUN: grep {vshl\\.u16} %t | count 4 |
| 8 | ; RUN: grep {vshl\\.u32} %t | count 4 |
| 9 | ; RUN: grep {vshl\\.u64} %t | count 4 |
| 10 | ; RUN: grep {vshl\\.i8} %t | count 2 |
| 11 | ; RUN: grep {vshl\\.i16} %t | count 2 |
| 12 | ; RUN: grep {vshl\\.i32} %t | count 2 |
| 13 | ; RUN: grep {vshl\\.i64} %t | count 2 |
| 14 | ; RUN: grep {vshr\\.u8} %t | count 2 |
| 15 | ; RUN: grep {vshr\\.u16} %t | count 2 |
| 16 | ; RUN: grep {vshr\\.u32} %t | count 2 |
| 17 | ; RUN: grep {vshr\\.u64} %t | count 2 |
| 18 | ; RUN: grep {vshr\\.s8} %t | count 2 |
| 19 | ; RUN: grep {vshr\\.s16} %t | count 2 |
| 20 | ; RUN: grep {vshr\\.s32} %t | count 2 |
| 21 | ; RUN: grep {vshr\\.s64} %t | count 2 |
| 22 | ; RUN: grep {vneg\\.s8} %t | count 4 |
| 23 | ; RUN: grep {vneg\\.s16} %t | count 4 |
| 24 | ; RUN: grep {vneg\\.s32} %t | count 4 |
| 25 | ; RUN: grep {vsub\\.i64} %t | count 4 |
| 26 | |
| 27 | define <8 x i8> @vshls8(<8 x i8>* %A, <8 x i8>* %B) nounwind { |
| 28 | %tmp1 = load <8 x i8>* %A |
| 29 | %tmp2 = load <8 x i8>* %B |
| 30 | %tmp3 = shl <8 x i8> %tmp1, %tmp2 |
| 31 | ret <8 x i8> %tmp3 |
| 32 | } |
| 33 | |
| 34 | define <4 x i16> @vshls16(<4 x i16>* %A, <4 x i16>* %B) nounwind { |
| 35 | %tmp1 = load <4 x i16>* %A |
| 36 | %tmp2 = load <4 x i16>* %B |
| 37 | %tmp3 = shl <4 x i16> %tmp1, %tmp2 |
| 38 | ret <4 x i16> %tmp3 |
| 39 | } |
| 40 | |
| 41 | define <2 x i32> @vshls32(<2 x i32>* %A, <2 x i32>* %B) nounwind { |
| 42 | %tmp1 = load <2 x i32>* %A |
| 43 | %tmp2 = load <2 x i32>* %B |
| 44 | %tmp3 = shl <2 x i32> %tmp1, %tmp2 |
| 45 | ret <2 x i32> %tmp3 |
| 46 | } |
| 47 | |
| 48 | define <1 x i64> @vshls64(<1 x i64>* %A, <1 x i64>* %B) nounwind { |
| 49 | %tmp1 = load <1 x i64>* %A |
| 50 | %tmp2 = load <1 x i64>* %B |
| 51 | %tmp3 = shl <1 x i64> %tmp1, %tmp2 |
| 52 | ret <1 x i64> %tmp3 |
| 53 | } |
| 54 | |
| 55 | define <8 x i8> @vshli8(<8 x i8>* %A) nounwind { |
| 56 | %tmp1 = load <8 x i8>* %A |
| 57 | %tmp2 = shl <8 x i8> %tmp1, < i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7 > |
| 58 | ret <8 x i8> %tmp2 |
| 59 | } |
| 60 | |
| 61 | define <4 x i16> @vshli16(<4 x i16>* %A) nounwind { |
| 62 | %tmp1 = load <4 x i16>* %A |
| 63 | %tmp2 = shl <4 x i16> %tmp1, < i16 15, i16 15, i16 15, i16 15 > |
| 64 | ret <4 x i16> %tmp2 |
| 65 | } |
| 66 | |
| 67 | define <2 x i32> @vshli32(<2 x i32>* %A) nounwind { |
| 68 | %tmp1 = load <2 x i32>* %A |
| 69 | %tmp2 = shl <2 x i32> %tmp1, < i32 31, i32 31 > |
| 70 | ret <2 x i32> %tmp2 |
| 71 | } |
| 72 | |
| 73 | define <1 x i64> @vshli64(<1 x i64>* %A) nounwind { |
| 74 | %tmp1 = load <1 x i64>* %A |
| 75 | %tmp2 = shl <1 x i64> %tmp1, < i64 63 > |
| 76 | ret <1 x i64> %tmp2 |
| 77 | } |
| 78 | |
| 79 | define <16 x i8> @vshlQs8(<16 x i8>* %A, <16 x i8>* %B) nounwind { |
| 80 | %tmp1 = load <16 x i8>* %A |
| 81 | %tmp2 = load <16 x i8>* %B |
| 82 | %tmp3 = shl <16 x i8> %tmp1, %tmp2 |
| 83 | ret <16 x i8> %tmp3 |
| 84 | } |
| 85 | |
| 86 | define <8 x i16> @vshlQs16(<8 x i16>* %A, <8 x i16>* %B) nounwind { |
| 87 | %tmp1 = load <8 x i16>* %A |
| 88 | %tmp2 = load <8 x i16>* %B |
| 89 | %tmp3 = shl <8 x i16> %tmp1, %tmp2 |
| 90 | ret <8 x i16> %tmp3 |
| 91 | } |
| 92 | |
| 93 | define <4 x i32> @vshlQs32(<4 x i32>* %A, <4 x i32>* %B) nounwind { |
| 94 | %tmp1 = load <4 x i32>* %A |
| 95 | %tmp2 = load <4 x i32>* %B |
| 96 | %tmp3 = shl <4 x i32> %tmp1, %tmp2 |
| 97 | ret <4 x i32> %tmp3 |
| 98 | } |
| 99 | |
| 100 | define <2 x i64> @vshlQs64(<2 x i64>* %A, <2 x i64>* %B) nounwind { |
| 101 | %tmp1 = load <2 x i64>* %A |
| 102 | %tmp2 = load <2 x i64>* %B |
| 103 | %tmp3 = shl <2 x i64> %tmp1, %tmp2 |
| 104 | ret <2 x i64> %tmp3 |
| 105 | } |
| 106 | |
| 107 | define <16 x i8> @vshlQi8(<16 x i8>* %A) nounwind { |
| 108 | %tmp1 = load <16 x i8>* %A |
| 109 | %tmp2 = shl <16 x i8> %tmp1, < i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7 > |
| 110 | ret <16 x i8> %tmp2 |
| 111 | } |
| 112 | |
| 113 | define <8 x i16> @vshlQi16(<8 x i16>* %A) nounwind { |
| 114 | %tmp1 = load <8 x i16>* %A |
| 115 | %tmp2 = shl <8 x i16> %tmp1, < i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15 > |
| 116 | ret <8 x i16> %tmp2 |
| 117 | } |
| 118 | |
| 119 | define <4 x i32> @vshlQi32(<4 x i32>* %A) nounwind { |
| 120 | %tmp1 = load <4 x i32>* %A |
| 121 | %tmp2 = shl <4 x i32> %tmp1, < i32 31, i32 31, i32 31, i32 31 > |
| 122 | ret <4 x i32> %tmp2 |
| 123 | } |
| 124 | |
| 125 | define <2 x i64> @vshlQi64(<2 x i64>* %A) nounwind { |
| 126 | %tmp1 = load <2 x i64>* %A |
| 127 | %tmp2 = shl <2 x i64> %tmp1, < i64 63, i64 63 > |
| 128 | ret <2 x i64> %tmp2 |
| 129 | } |
| 130 | |
| 131 | define <8 x i8> @vlshru8(<8 x i8>* %A, <8 x i8>* %B) nounwind { |
| 132 | %tmp1 = load <8 x i8>* %A |
| 133 | %tmp2 = load <8 x i8>* %B |
| 134 | %tmp3 = lshr <8 x i8> %tmp1, %tmp2 |
| 135 | ret <8 x i8> %tmp3 |
| 136 | } |
| 137 | |
| 138 | define <4 x i16> @vlshru16(<4 x i16>* %A, <4 x i16>* %B) nounwind { |
| 139 | %tmp1 = load <4 x i16>* %A |
| 140 | %tmp2 = load <4 x i16>* %B |
| 141 | %tmp3 = lshr <4 x i16> %tmp1, %tmp2 |
| 142 | ret <4 x i16> %tmp3 |
| 143 | } |
| 144 | |
| 145 | define <2 x i32> @vlshru32(<2 x i32>* %A, <2 x i32>* %B) nounwind { |
| 146 | %tmp1 = load <2 x i32>* %A |
| 147 | %tmp2 = load <2 x i32>* %B |
| 148 | %tmp3 = lshr <2 x i32> %tmp1, %tmp2 |
| 149 | ret <2 x i32> %tmp3 |
| 150 | } |
| 151 | |
| 152 | define <1 x i64> @vlshru64(<1 x i64>* %A, <1 x i64>* %B) nounwind { |
| 153 | %tmp1 = load <1 x i64>* %A |
| 154 | %tmp2 = load <1 x i64>* %B |
| 155 | %tmp3 = lshr <1 x i64> %tmp1, %tmp2 |
| 156 | ret <1 x i64> %tmp3 |
| 157 | } |
| 158 | |
| 159 | define <8 x i8> @vlshri8(<8 x i8>* %A) nounwind { |
| 160 | %tmp1 = load <8 x i8>* %A |
| 161 | %tmp2 = lshr <8 x i8> %tmp1, < i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8 > |
| 162 | ret <8 x i8> %tmp2 |
| 163 | } |
| 164 | |
| 165 | define <4 x i16> @vlshri16(<4 x i16>* %A) nounwind { |
| 166 | %tmp1 = load <4 x i16>* %A |
| 167 | %tmp2 = lshr <4 x i16> %tmp1, < i16 16, i16 16, i16 16, i16 16 > |
| 168 | ret <4 x i16> %tmp2 |
| 169 | } |
| 170 | |
| 171 | define <2 x i32> @vlshri32(<2 x i32>* %A) nounwind { |
| 172 | %tmp1 = load <2 x i32>* %A |
| 173 | %tmp2 = lshr <2 x i32> %tmp1, < i32 32, i32 32 > |
| 174 | ret <2 x i32> %tmp2 |
| 175 | } |
| 176 | |
| 177 | define <1 x i64> @vlshri64(<1 x i64>* %A) nounwind { |
| 178 | %tmp1 = load <1 x i64>* %A |
| 179 | %tmp2 = lshr <1 x i64> %tmp1, < i64 64 > |
| 180 | ret <1 x i64> %tmp2 |
| 181 | } |
| 182 | |
| 183 | define <16 x i8> @vlshrQu8(<16 x i8>* %A, <16 x i8>* %B) nounwind { |
| 184 | %tmp1 = load <16 x i8>* %A |
| 185 | %tmp2 = load <16 x i8>* %B |
| 186 | %tmp3 = lshr <16 x i8> %tmp1, %tmp2 |
| 187 | ret <16 x i8> %tmp3 |
| 188 | } |
| 189 | |
| 190 | define <8 x i16> @vlshrQu16(<8 x i16>* %A, <8 x i16>* %B) nounwind { |
| 191 | %tmp1 = load <8 x i16>* %A |
| 192 | %tmp2 = load <8 x i16>* %B |
| 193 | %tmp3 = lshr <8 x i16> %tmp1, %tmp2 |
| 194 | ret <8 x i16> %tmp3 |
| 195 | } |
| 196 | |
| 197 | define <4 x i32> @vlshrQu32(<4 x i32>* %A, <4 x i32>* %B) nounwind { |
| 198 | %tmp1 = load <4 x i32>* %A |
| 199 | %tmp2 = load <4 x i32>* %B |
| 200 | %tmp3 = lshr <4 x i32> %tmp1, %tmp2 |
| 201 | ret <4 x i32> %tmp3 |
| 202 | } |
| 203 | |
| 204 | define <2 x i64> @vlshrQu64(<2 x i64>* %A, <2 x i64>* %B) nounwind { |
| 205 | %tmp1 = load <2 x i64>* %A |
| 206 | %tmp2 = load <2 x i64>* %B |
| 207 | %tmp3 = lshr <2 x i64> %tmp1, %tmp2 |
| 208 | ret <2 x i64> %tmp3 |
| 209 | } |
| 210 | |
| 211 | define <16 x i8> @vlshrQi8(<16 x i8>* %A) nounwind { |
| 212 | %tmp1 = load <16 x i8>* %A |
| 213 | %tmp2 = lshr <16 x i8> %tmp1, < i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8 > |
| 214 | ret <16 x i8> %tmp2 |
| 215 | } |
| 216 | |
| 217 | define <8 x i16> @vlshrQi16(<8 x i16>* %A) nounwind { |
| 218 | %tmp1 = load <8 x i16>* %A |
| 219 | %tmp2 = lshr <8 x i16> %tmp1, < i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16 > |
| 220 | ret <8 x i16> %tmp2 |
| 221 | } |
| 222 | |
| 223 | define <4 x i32> @vlshrQi32(<4 x i32>* %A) nounwind { |
| 224 | %tmp1 = load <4 x i32>* %A |
| 225 | %tmp2 = lshr <4 x i32> %tmp1, < i32 32, i32 32, i32 32, i32 32 > |
| 226 | ret <4 x i32> %tmp2 |
| 227 | } |
| 228 | |
| 229 | define <2 x i64> @vlshrQi64(<2 x i64>* %A) nounwind { |
| 230 | %tmp1 = load <2 x i64>* %A |
| 231 | %tmp2 = lshr <2 x i64> %tmp1, < i64 64, i64 64 > |
| 232 | ret <2 x i64> %tmp2 |
| 233 | } |
| 234 | |
| 235 | define <8 x i8> @vashrs8(<8 x i8>* %A, <8 x i8>* %B) nounwind { |
| 236 | %tmp1 = load <8 x i8>* %A |
| 237 | %tmp2 = load <8 x i8>* %B |
| 238 | %tmp3 = ashr <8 x i8> %tmp1, %tmp2 |
| 239 | ret <8 x i8> %tmp3 |
| 240 | } |
| 241 | |
| 242 | define <4 x i16> @vashrs16(<4 x i16>* %A, <4 x i16>* %B) nounwind { |
| 243 | %tmp1 = load <4 x i16>* %A |
| 244 | %tmp2 = load <4 x i16>* %B |
| 245 | %tmp3 = ashr <4 x i16> %tmp1, %tmp2 |
| 246 | ret <4 x i16> %tmp3 |
| 247 | } |
| 248 | |
| 249 | define <2 x i32> @vashrs32(<2 x i32>* %A, <2 x i32>* %B) nounwind { |
| 250 | %tmp1 = load <2 x i32>* %A |
| 251 | %tmp2 = load <2 x i32>* %B |
| 252 | %tmp3 = ashr <2 x i32> %tmp1, %tmp2 |
| 253 | ret <2 x i32> %tmp3 |
| 254 | } |
| 255 | |
| 256 | define <1 x i64> @vashrs64(<1 x i64>* %A, <1 x i64>* %B) nounwind { |
| 257 | %tmp1 = load <1 x i64>* %A |
| 258 | %tmp2 = load <1 x i64>* %B |
| 259 | %tmp3 = ashr <1 x i64> %tmp1, %tmp2 |
| 260 | ret <1 x i64> %tmp3 |
| 261 | } |
| 262 | |
| 263 | define <8 x i8> @vashri8(<8 x i8>* %A) nounwind { |
| 264 | %tmp1 = load <8 x i8>* %A |
| 265 | %tmp2 = ashr <8 x i8> %tmp1, < i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8 > |
| 266 | ret <8 x i8> %tmp2 |
| 267 | } |
| 268 | |
| 269 | define <4 x i16> @vashri16(<4 x i16>* %A) nounwind { |
| 270 | %tmp1 = load <4 x i16>* %A |
| 271 | %tmp2 = ashr <4 x i16> %tmp1, < i16 16, i16 16, i16 16, i16 16 > |
| 272 | ret <4 x i16> %tmp2 |
| 273 | } |
| 274 | |
| 275 | define <2 x i32> @vashri32(<2 x i32>* %A) nounwind { |
| 276 | %tmp1 = load <2 x i32>* %A |
| 277 | %tmp2 = ashr <2 x i32> %tmp1, < i32 32, i32 32 > |
| 278 | ret <2 x i32> %tmp2 |
| 279 | } |
| 280 | |
| 281 | define <1 x i64> @vashri64(<1 x i64>* %A) nounwind { |
| 282 | %tmp1 = load <1 x i64>* %A |
| 283 | %tmp2 = ashr <1 x i64> %tmp1, < i64 64 > |
| 284 | ret <1 x i64> %tmp2 |
| 285 | } |
| 286 | |
| 287 | define <16 x i8> @vashrQs8(<16 x i8>* %A, <16 x i8>* %B) nounwind { |
| 288 | %tmp1 = load <16 x i8>* %A |
| 289 | %tmp2 = load <16 x i8>* %B |
| 290 | %tmp3 = ashr <16 x i8> %tmp1, %tmp2 |
| 291 | ret <16 x i8> %tmp3 |
| 292 | } |
| 293 | |
| 294 | define <8 x i16> @vashrQs16(<8 x i16>* %A, <8 x i16>* %B) nounwind { |
| 295 | %tmp1 = load <8 x i16>* %A |
| 296 | %tmp2 = load <8 x i16>* %B |
| 297 | %tmp3 = ashr <8 x i16> %tmp1, %tmp2 |
| 298 | ret <8 x i16> %tmp3 |
| 299 | } |
| 300 | |
| 301 | define <4 x i32> @vashrQs32(<4 x i32>* %A, <4 x i32>* %B) nounwind { |
| 302 | %tmp1 = load <4 x i32>* %A |
| 303 | %tmp2 = load <4 x i32>* %B |
| 304 | %tmp3 = ashr <4 x i32> %tmp1, %tmp2 |
| 305 | ret <4 x i32> %tmp3 |
| 306 | } |
| 307 | |
| 308 | define <2 x i64> @vashrQs64(<2 x i64>* %A, <2 x i64>* %B) nounwind { |
| 309 | %tmp1 = load <2 x i64>* %A |
| 310 | %tmp2 = load <2 x i64>* %B |
| 311 | %tmp3 = ashr <2 x i64> %tmp1, %tmp2 |
| 312 | ret <2 x i64> %tmp3 |
| 313 | } |
| 314 | |
| 315 | define <16 x i8> @vashrQi8(<16 x i8>* %A) nounwind { |
| 316 | %tmp1 = load <16 x i8>* %A |
| 317 | %tmp2 = ashr <16 x i8> %tmp1, < i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8 > |
| 318 | ret <16 x i8> %tmp2 |
| 319 | } |
| 320 | |
| 321 | define <8 x i16> @vashrQi16(<8 x i16>* %A) nounwind { |
| 322 | %tmp1 = load <8 x i16>* %A |
| 323 | %tmp2 = ashr <8 x i16> %tmp1, < i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16 > |
| 324 | ret <8 x i16> %tmp2 |
| 325 | } |
| 326 | |
| 327 | define <4 x i32> @vashrQi32(<4 x i32>* %A) nounwind { |
| 328 | %tmp1 = load <4 x i32>* %A |
| 329 | %tmp2 = ashr <4 x i32> %tmp1, < i32 32, i32 32, i32 32, i32 32 > |
| 330 | ret <4 x i32> %tmp2 |
| 331 | } |
| 332 | |
| 333 | define <2 x i64> @vashrQi64(<2 x i64>* %A) nounwind { |
| 334 | %tmp1 = load <2 x i64>* %A |
| 335 | %tmp2 = ashr <2 x i64> %tmp1, < i64 64, i64 64 > |
| 336 | ret <2 x i64> %tmp2 |
| 337 | } |