1 // RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \ 2 // RUN: -disable-O0-optnone -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s 3 4 // Test new aarch64 intrinsics and types 5 6 #include <arm_neon.h> 7 8 // CHECK-LABEL: define i16 @test_vaddlv_s8(<8 x i8> %a) #0 { 9 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.saddlv.i32.v8i8(<8 x i8> %a) #3 10 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDLV_I]] to i16 11 // CHECK: ret i16 [[TMP0]] test_vaddlv_s8(int8x8_t a)12int16_t test_vaddlv_s8(int8x8_t a) { 13 return vaddlv_s8(a); 14 } 15 16 // CHECK-LABEL: define i32 @test_vaddlv_s16(<4 x i16> %a) #0 { 17 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.saddlv.i32.v4i16(<4 x i16> %a) #3 18 // CHECK: ret i32 [[VADDLV_I]] test_vaddlv_s16(int16x4_t a)19int32_t test_vaddlv_s16(int16x4_t a) { 20 return vaddlv_s16(a); 21 } 22 23 // CHECK-LABEL: define i16 @test_vaddlv_u8(<8 x i8> %a) #0 { 24 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddlv.i32.v8i8(<8 x i8> %a) #3 25 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDLV_I]] to i16 26 // CHECK: ret i16 [[TMP0]] test_vaddlv_u8(uint8x8_t a)27uint16_t test_vaddlv_u8(uint8x8_t a) { 28 return vaddlv_u8(a); 29 } 30 31 // CHECK-LABEL: define i32 @test_vaddlv_u16(<4 x i16> %a) #0 { 32 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddlv.i32.v4i16(<4 x i16> %a) #3 33 // CHECK: ret i32 [[VADDLV_I]] test_vaddlv_u16(uint16x4_t a)34uint32_t test_vaddlv_u16(uint16x4_t a) { 35 return vaddlv_u16(a); 36 } 37 38 // CHECK-LABEL: define i16 @test_vaddlvq_s8(<16 x i8> %a) #1 { 39 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.saddlv.i32.v16i8(<16 x i8> %a) #3 40 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDLV_I]] to i16 41 // CHECK: ret i16 [[TMP0]] test_vaddlvq_s8(int8x16_t a)42int16_t test_vaddlvq_s8(int8x16_t a) { 43 return vaddlvq_s8(a); 44 } 45 46 // CHECK-LABEL: define i32 @test_vaddlvq_s16(<8 x i16> %a) #1 { 47 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.saddlv.i32.v8i16(<8 x i16> %a) #3 48 // CHECK: ret i32 [[VADDLV_I]] test_vaddlvq_s16(int16x8_t a)49int32_t test_vaddlvq_s16(int16x8_t a) { 50 return vaddlvq_s16(a); 51 } 52 53 // CHECK-LABEL: define i64 @test_vaddlvq_s32(<4 x i32> %a) #1 { 54 // CHECK: [[VADDLVQ_S32_I:%.*]] = call i64 @llvm.aarch64.neon.saddlv.i64.v4i32(<4 x i32> %a) #3 55 // CHECK: ret i64 [[VADDLVQ_S32_I]] test_vaddlvq_s32(int32x4_t a)56int64_t test_vaddlvq_s32(int32x4_t a) { 57 return vaddlvq_s32(a); 58 } 59 60 // CHECK-LABEL: define i16 @test_vaddlvq_u8(<16 x i8> %a) #1 { 61 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddlv.i32.v16i8(<16 x i8> %a) #3 62 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDLV_I]] to i16 63 // CHECK: ret i16 [[TMP0]] test_vaddlvq_u8(uint8x16_t a)64uint16_t test_vaddlvq_u8(uint8x16_t a) { 65 return vaddlvq_u8(a); 66 } 67 68 // CHECK-LABEL: define i32 @test_vaddlvq_u16(<8 x i16> %a) #1 { 69 // CHECK: [[VADDLV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddlv.i32.v8i16(<8 x i16> %a) #3 70 // CHECK: ret i32 [[VADDLV_I]] test_vaddlvq_u16(uint16x8_t a)71uint32_t test_vaddlvq_u16(uint16x8_t a) { 72 return vaddlvq_u16(a); 73 } 74 75 // CHECK-LABEL: define i64 @test_vaddlvq_u32(<4 x i32> %a) #1 { 76 // CHECK: [[VADDLVQ_U32_I:%.*]] = call i64 @llvm.aarch64.neon.uaddlv.i64.v4i32(<4 x i32> %a) #3 77 // CHECK: ret i64 [[VADDLVQ_U32_I]] test_vaddlvq_u32(uint32x4_t a)78uint64_t test_vaddlvq_u32(uint32x4_t a) { 79 return vaddlvq_u32(a); 80 } 81 82 // CHECK-LABEL: define i8 @test_vmaxv_s8(<8 x i8> %a) #0 { 83 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.smaxv.i32.v8i8(<8 x i8> %a) #3 84 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMAXV_I]] to i8 85 // CHECK: ret i8 [[TMP0]] test_vmaxv_s8(int8x8_t a)86int8_t test_vmaxv_s8(int8x8_t a) { 87 return vmaxv_s8(a); 88 } 89 90 // CHECK-LABEL: define i16 @test_vmaxv_s16(<4 x i16> %a) #0 { 91 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.smaxv.i32.v4i16(<4 x i16> %a) #3 92 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMAXV_I]] to i16 93 // CHECK: ret i16 [[TMP2]] test_vmaxv_s16(int16x4_t a)94int16_t test_vmaxv_s16(int16x4_t a) { 95 return vmaxv_s16(a); 96 } 97 98 // CHECK-LABEL: define i8 @test_vmaxv_u8(<8 x i8> %a) #0 { 99 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.umaxv.i32.v8i8(<8 x i8> %a) #3 100 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMAXV_I]] to i8 101 // CHECK: ret i8 [[TMP0]] test_vmaxv_u8(uint8x8_t a)102uint8_t test_vmaxv_u8(uint8x8_t a) { 103 return vmaxv_u8(a); 104 } 105 106 // CHECK-LABEL: define i16 @test_vmaxv_u16(<4 x i16> %a) #0 { 107 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.umaxv.i32.v4i16(<4 x i16> %a) #3 108 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMAXV_I]] to i16 109 // CHECK: ret i16 [[TMP2]] test_vmaxv_u16(uint16x4_t a)110uint16_t test_vmaxv_u16(uint16x4_t a) { 111 return vmaxv_u16(a); 112 } 113 114 // CHECK-LABEL: define i8 @test_vmaxvq_s8(<16 x i8> %a) #1 { 115 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.smaxv.i32.v16i8(<16 x i8> %a) #3 116 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMAXV_I]] to i8 117 // CHECK: ret i8 [[TMP0]] test_vmaxvq_s8(int8x16_t a)118int8_t test_vmaxvq_s8(int8x16_t a) { 119 return vmaxvq_s8(a); 120 } 121 122 // CHECK-LABEL: define i16 @test_vmaxvq_s16(<8 x i16> %a) #1 { 123 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.smaxv.i32.v8i16(<8 x i16> %a) #3 124 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMAXV_I]] to i16 125 // CHECK: ret i16 [[TMP2]] test_vmaxvq_s16(int16x8_t a)126int16_t test_vmaxvq_s16(int16x8_t a) { 127 return vmaxvq_s16(a); 128 } 129 130 // CHECK-LABEL: define i32 @test_vmaxvq_s32(<4 x i32> %a) #1 { 131 // CHECK: [[VMAXVQ_S32_I:%.*]] = call i32 @llvm.aarch64.neon.smaxv.i32.v4i32(<4 x i32> %a) #3 132 // CHECK: ret i32 [[VMAXVQ_S32_I]] test_vmaxvq_s32(int32x4_t a)133int32_t test_vmaxvq_s32(int32x4_t a) { 134 return vmaxvq_s32(a); 135 } 136 137 // CHECK-LABEL: define i8 @test_vmaxvq_u8(<16 x i8> %a) #1 { 138 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.umaxv.i32.v16i8(<16 x i8> %a) #3 139 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMAXV_I]] to i8 140 // CHECK: ret i8 [[TMP0]] test_vmaxvq_u8(uint8x16_t a)141uint8_t test_vmaxvq_u8(uint8x16_t a) { 142 return vmaxvq_u8(a); 143 } 144 145 // CHECK-LABEL: define i16 @test_vmaxvq_u16(<8 x i16> %a) #1 { 146 // CHECK: [[VMAXV_I:%.*]] = call i32 @llvm.aarch64.neon.umaxv.i32.v8i16(<8 x i16> %a) #3 147 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMAXV_I]] to i16 148 // CHECK: ret i16 [[TMP2]] test_vmaxvq_u16(uint16x8_t a)149uint16_t test_vmaxvq_u16(uint16x8_t a) { 150 return vmaxvq_u16(a); 151 } 152 153 // CHECK-LABEL: define i32 @test_vmaxvq_u32(<4 x i32> %a) #1 { 154 // CHECK: [[VMAXVQ_U32_I:%.*]] = call i32 @llvm.aarch64.neon.umaxv.i32.v4i32(<4 x i32> %a) #3 155 // CHECK: ret i32 [[VMAXVQ_U32_I]] test_vmaxvq_u32(uint32x4_t a)156uint32_t test_vmaxvq_u32(uint32x4_t a) { 157 return vmaxvq_u32(a); 158 } 159 160 // CHECK-LABEL: define i8 @test_vminv_s8(<8 x i8> %a) #0 { 161 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.sminv.i32.v8i8(<8 x i8> %a) #3 162 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMINV_I]] to i8 163 // CHECK: ret i8 [[TMP0]] test_vminv_s8(int8x8_t a)164int8_t test_vminv_s8(int8x8_t a) { 165 return vminv_s8(a); 166 } 167 168 // CHECK-LABEL: define i16 @test_vminv_s16(<4 x i16> %a) #0 { 169 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.sminv.i32.v4i16(<4 x i16> %a) #3 170 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMINV_I]] to i16 171 // CHECK: ret i16 [[TMP2]] test_vminv_s16(int16x4_t a)172int16_t test_vminv_s16(int16x4_t a) { 173 return vminv_s16(a); 174 } 175 176 // CHECK-LABEL: define i8 @test_vminv_u8(<8 x i8> %a) #0 { 177 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.uminv.i32.v8i8(<8 x i8> %a) #3 178 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMINV_I]] to i8 179 // CHECK: ret i8 [[TMP0]] test_vminv_u8(uint8x8_t a)180uint8_t test_vminv_u8(uint8x8_t a) { 181 return vminv_u8(a); 182 } 183 184 // CHECK-LABEL: define i16 @test_vminv_u16(<4 x i16> %a) #0 { 185 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.uminv.i32.v4i16(<4 x i16> %a) #3 186 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMINV_I]] to i16 187 // CHECK: ret i16 [[TMP2]] test_vminv_u16(uint16x4_t a)188uint16_t test_vminv_u16(uint16x4_t a) { 189 return vminv_u16(a); 190 } 191 192 // CHECK-LABEL: define i8 @test_vminvq_s8(<16 x i8> %a) #1 { 193 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.sminv.i32.v16i8(<16 x i8> %a) #3 194 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMINV_I]] to i8 195 // CHECK: ret i8 [[TMP0]] test_vminvq_s8(int8x16_t a)196int8_t test_vminvq_s8(int8x16_t a) { 197 return vminvq_s8(a); 198 } 199 200 // CHECK-LABEL: define i16 @test_vminvq_s16(<8 x i16> %a) #1 { 201 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.sminv.i32.v8i16(<8 x i16> %a) #3 202 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMINV_I]] to i16 203 // CHECK: ret i16 [[TMP2]] test_vminvq_s16(int16x8_t a)204int16_t test_vminvq_s16(int16x8_t a) { 205 return vminvq_s16(a); 206 } 207 208 // CHECK-LABEL: define i32 @test_vminvq_s32(<4 x i32> %a) #1 { 209 // CHECK: [[VMINVQ_S32_I:%.*]] = call i32 @llvm.aarch64.neon.sminv.i32.v4i32(<4 x i32> %a) #3 210 // CHECK: ret i32 [[VMINVQ_S32_I]] test_vminvq_s32(int32x4_t a)211int32_t test_vminvq_s32(int32x4_t a) { 212 return vminvq_s32(a); 213 } 214 215 // CHECK-LABEL: define i8 @test_vminvq_u8(<16 x i8> %a) #1 { 216 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.uminv.i32.v16i8(<16 x i8> %a) #3 217 // CHECK: [[TMP0:%.*]] = trunc i32 [[VMINV_I]] to i8 218 // CHECK: ret i8 [[TMP0]] test_vminvq_u8(uint8x16_t a)219uint8_t test_vminvq_u8(uint8x16_t a) { 220 return vminvq_u8(a); 221 } 222 223 // CHECK-LABEL: define i16 @test_vminvq_u16(<8 x i16> %a) #1 { 224 // CHECK: [[VMINV_I:%.*]] = call i32 @llvm.aarch64.neon.uminv.i32.v8i16(<8 x i16> %a) #3 225 // CHECK: [[TMP2:%.*]] = trunc i32 [[VMINV_I]] to i16 226 // CHECK: ret i16 [[TMP2]] test_vminvq_u16(uint16x8_t a)227uint16_t test_vminvq_u16(uint16x8_t a) { 228 return vminvq_u16(a); 229 } 230 231 // CHECK-LABEL: define i32 @test_vminvq_u32(<4 x i32> %a) #1 { 232 // CHECK: [[VMINVQ_U32_I:%.*]] = call i32 @llvm.aarch64.neon.uminv.i32.v4i32(<4 x i32> %a) #3 233 // CHECK: ret i32 [[VMINVQ_U32_I]] test_vminvq_u32(uint32x4_t a)234uint32_t test_vminvq_u32(uint32x4_t a) { 235 return vminvq_u32(a); 236 } 237 238 // CHECK-LABEL: define i8 @test_vaddv_s8(<8 x i8> %a) #0 { 239 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.saddv.i32.v8i8(<8 x i8> %a) #3 240 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDV_I]] to i8 241 // CHECK: ret i8 [[TMP0]] test_vaddv_s8(int8x8_t a)242int8_t test_vaddv_s8(int8x8_t a) { 243 return vaddv_s8(a); 244 } 245 246 // CHECK-LABEL: define i16 @test_vaddv_s16(<4 x i16> %a) #0 { 247 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.saddv.i32.v4i16(<4 x i16> %a) #3 248 // CHECK: [[TMP2:%.*]] = trunc i32 [[VADDV_I]] to i16 249 // CHECK: ret i16 [[TMP2]] test_vaddv_s16(int16x4_t a)250int16_t test_vaddv_s16(int16x4_t a) { 251 return vaddv_s16(a); 252 } 253 254 // CHECK-LABEL: define i8 @test_vaddv_u8(<8 x i8> %a) #0 { 255 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddv.i32.v8i8(<8 x i8> %a) #3 256 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDV_I]] to i8 257 // CHECK: ret i8 [[TMP0]] test_vaddv_u8(uint8x8_t a)258uint8_t test_vaddv_u8(uint8x8_t a) { 259 return vaddv_u8(a); 260 } 261 262 // CHECK-LABEL: define i16 @test_vaddv_u16(<4 x i16> %a) #0 { 263 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddv.i32.v4i16(<4 x i16> %a) #3 264 // CHECK: [[TMP2:%.*]] = trunc i32 [[VADDV_I]] to i16 265 // CHECK: ret i16 [[TMP2]] test_vaddv_u16(uint16x4_t a)266uint16_t test_vaddv_u16(uint16x4_t a) { 267 return vaddv_u16(a); 268 } 269 270 // CHECK-LABEL: define i8 @test_vaddvq_s8(<16 x i8> %a) #1 { 271 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.saddv.i32.v16i8(<16 x i8> %a) #3 272 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDV_I]] to i8 273 // CHECK: ret i8 [[TMP0]] test_vaddvq_s8(int8x16_t a)274int8_t test_vaddvq_s8(int8x16_t a) { 275 return vaddvq_s8(a); 276 } 277 278 // CHECK-LABEL: define i16 @test_vaddvq_s16(<8 x i16> %a) #1 { 279 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.saddv.i32.v8i16(<8 x i16> %a) #3 280 // CHECK: [[TMP2:%.*]] = trunc i32 [[VADDV_I]] to i16 281 // CHECK: ret i16 [[TMP2]] test_vaddvq_s16(int16x8_t a)282int16_t test_vaddvq_s16(int16x8_t a) { 283 return vaddvq_s16(a); 284 } 285 286 // CHECK-LABEL: define i32 @test_vaddvq_s32(<4 x i32> %a) #1 { 287 // CHECK: [[VADDVQ_S32_I:%.*]] = call i32 @llvm.aarch64.neon.saddv.i32.v4i32(<4 x i32> %a) #3 288 // CHECK: ret i32 [[VADDVQ_S32_I]] test_vaddvq_s32(int32x4_t a)289int32_t test_vaddvq_s32(int32x4_t a) { 290 return vaddvq_s32(a); 291 } 292 293 // CHECK-LABEL: define i8 @test_vaddvq_u8(<16 x i8> %a) #1 { 294 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddv.i32.v16i8(<16 x i8> %a) #3 295 // CHECK: [[TMP0:%.*]] = trunc i32 [[VADDV_I]] to i8 296 // CHECK: ret i8 [[TMP0]] test_vaddvq_u8(uint8x16_t a)297uint8_t test_vaddvq_u8(uint8x16_t a) { 298 return vaddvq_u8(a); 299 } 300 301 // CHECK-LABEL: define i16 @test_vaddvq_u16(<8 x i16> %a) #1 { 302 // CHECK: [[VADDV_I:%.*]] = call i32 @llvm.aarch64.neon.uaddv.i32.v8i16(<8 x i16> %a) #3 303 // CHECK: [[TMP2:%.*]] = trunc i32 [[VADDV_I]] to i16 304 // CHECK: ret i16 [[TMP2]] test_vaddvq_u16(uint16x8_t a)305uint16_t test_vaddvq_u16(uint16x8_t a) { 306 return vaddvq_u16(a); 307 } 308 309 // CHECK-LABEL: define i32 @test_vaddvq_u32(<4 x i32> %a) #1 { 310 // CHECK: [[VADDVQ_U32_I:%.*]] = call i32 @llvm.aarch64.neon.uaddv.i32.v4i32(<4 x i32> %a) #3 311 // CHECK: ret i32 [[VADDVQ_U32_I]] test_vaddvq_u32(uint32x4_t a)312uint32_t test_vaddvq_u32(uint32x4_t a) { 313 return vaddvq_u32(a); 314 } 315 316 // CHECK-LABEL: define float @test_vmaxvq_f32(<4 x float> %a) #1 { 317 // CHECK: [[VMAXVQ_F32_I:%.*]] = call float @llvm.aarch64.neon.fmaxv.f32.v4f32(<4 x float> %a) #3 318 // CHECK: ret float [[VMAXVQ_F32_I]] test_vmaxvq_f32(float32x4_t a)319float32_t test_vmaxvq_f32(float32x4_t a) { 320 return vmaxvq_f32(a); 321 } 322 323 // CHECK-LABEL: define float @test_vminvq_f32(<4 x float> %a) #1 { 324 // CHECK: [[VMINVQ_F32_I:%.*]] = call float @llvm.aarch64.neon.fminv.f32.v4f32(<4 x float> %a) #3 325 // CHECK: ret float [[VMINVQ_F32_I]] test_vminvq_f32(float32x4_t a)326float32_t test_vminvq_f32(float32x4_t a) { 327 return vminvq_f32(a); 328 } 329 330 // CHECK-LABEL: define float @test_vmaxnmvq_f32(<4 x float> %a) #1 { 331 // CHECK: [[VMAXNMVQ_F32_I:%.*]] = call float @llvm.aarch64.neon.fmaxnmv.f32.v4f32(<4 x float> %a) #3 332 // CHECK: ret float [[VMAXNMVQ_F32_I]] test_vmaxnmvq_f32(float32x4_t a)333float32_t test_vmaxnmvq_f32(float32x4_t a) { 334 return vmaxnmvq_f32(a); 335 } 336 337 // CHECK-LABEL: define float @test_vminnmvq_f32(<4 x float> %a) #1 { 338 // CHECK: [[VMINNMVQ_F32_I:%.*]] = call float @llvm.aarch64.neon.fminnmv.f32.v4f32(<4 x float> %a) #3 339 // CHECK: ret float [[VMINNMVQ_F32_I]] test_vminnmvq_f32(float32x4_t a)340float32_t test_vminnmvq_f32(float32x4_t a) { 341 return vminnmvq_f32(a); 342 } 343 344 // CHECK: attributes #0 ={{.*}}"min-legal-vector-width"="64" 345 // CHECK: attributes #1 ={{.*}}"min-legal-vector-width"="128" 346