Erich Keane | b30c166 | 2020-05-08 13:14:50 -0700 | [diff] [blame] | 1 | // RUN: %clang_cc1 -std=c++14 -Wno-unused-value %s -disable-llvm-passes -triple x86_64-linux-gnu -emit-llvm -o - | FileCheck %s |
| 2 | |
| 3 | // FIXME: Unfortunately there is no good way to validate that our values are |
| 4 | // correct since Vector types don't have operator [] implemented for constexpr. |
| 5 | // Instead, we need to use filecheck to ensure the emitted IR is correct. Once |
| 6 | // someone implements array subscript operator for these types as constexpr, |
| 7 | // this test should modified to jsut use static asserts. |
| 8 | |
| 9 | using FourCharsVecSize __attribute__((vector_size(4))) = char; |
| 10 | using FourIntsVecSize __attribute__((vector_size(16))) = int; |
| 11 | using FourLongLongsVecSize __attribute__((vector_size(32))) = long long; |
| 12 | using FourFloatsVecSize __attribute__((vector_size(16))) = float; |
| 13 | using FourDoublesVecSize __attribute__((vector_size(32))) = double; |
| 14 | |
| 15 | using FourCharsExtVec __attribute__((ext_vector_type(4))) = char; |
| 16 | using FourIntsExtVec __attribute__((ext_vector_type(4))) = int; |
| 17 | using FourLongLongsExtVec __attribute__((ext_vector_type(4))) = long long; |
| 18 | using FourFloatsExtVec __attribute__((ext_vector_type(4))) = float; |
| 19 | using FourDoublesExtVec __attribute__((ext_vector_type(4))) = double; |
| 20 | |
| 21 | // Next a series of tests to make sure these operations are usable in |
| 22 | // constexpr functions. Template instantiations don't emit Winvalid-constexpr, |
| 23 | // so we have to do these as macros. |
| 24 | #define MathShiftOps(Type) \ |
| 25 | constexpr auto MathShiftOps##Type(Type a, Type b) { \ |
| 26 | a = a + b; \ |
| 27 | a = a - b; \ |
| 28 | a = a * b; \ |
| 29 | a = a / b; \ |
| 30 | b = a + 1; \ |
| 31 | b = a - 1; \ |
| 32 | b = a * 1; \ |
| 33 | b = a / 1; \ |
| 34 | a += a; \ |
| 35 | a -= a; \ |
| 36 | a *= a; \ |
| 37 | a /= a; \ |
| 38 | b += a; \ |
| 39 | b -= a; \ |
| 40 | b *= a; \ |
| 41 | b /= a; \ |
| 42 | a < b; \ |
| 43 | a > b; \ |
| 44 | a <= b; \ |
| 45 | a >= b; \ |
| 46 | a == b; \ |
| 47 | a != b; \ |
| 48 | a &&b; \ |
| 49 | a || b; \ |
| 50 | auto c = (a, b); \ |
| 51 | return c; \ |
| 52 | } |
| 53 | |
| 54 | // Ops specific to Integers. |
| 55 | #define MathShiftOpsInts(Type) \ |
| 56 | constexpr auto MathShiftopsInts##Type(Type a, Type b) { \ |
| 57 | a = a << b; \ |
| 58 | a = a >> b; \ |
| 59 | a = a << 3; \ |
| 60 | a = a >> 3; \ |
| 61 | a = 3 << b; \ |
| 62 | a = 3 >> b; \ |
| 63 | a <<= b; \ |
| 64 | a >>= b; \ |
| 65 | a <<= 3; \ |
| 66 | a >>= 3; \ |
| 67 | a = a % b; \ |
| 68 | a &b; \ |
| 69 | a | b; \ |
| 70 | a ^ b; \ |
| 71 | return a; \ |
| 72 | } |
| 73 | |
| 74 | MathShiftOps(FourCharsVecSize); |
| 75 | MathShiftOps(FourIntsVecSize); |
| 76 | MathShiftOps(FourLongLongsVecSize); |
| 77 | MathShiftOps(FourFloatsVecSize); |
| 78 | MathShiftOps(FourDoublesVecSize); |
| 79 | MathShiftOps(FourCharsExtVec); |
| 80 | MathShiftOps(FourIntsExtVec); |
| 81 | MathShiftOps(FourLongLongsExtVec); |
| 82 | MathShiftOps(FourFloatsExtVec); |
| 83 | MathShiftOps(FourDoublesExtVec); |
| 84 | |
| 85 | MathShiftOpsInts(FourCharsVecSize); |
| 86 | MathShiftOpsInts(FourIntsVecSize); |
| 87 | MathShiftOpsInts(FourLongLongsVecSize); |
| 88 | MathShiftOpsInts(FourCharsExtVec); |
| 89 | MathShiftOpsInts(FourIntsExtVec); |
| 90 | MathShiftOpsInts(FourLongLongsExtVec); |
| 91 | |
| 92 | template <typename T, typename U> |
| 93 | constexpr auto CmpMul(T t, U u) { |
| 94 | t *= u; |
| 95 | return t; |
| 96 | } |
| 97 | template <typename T, typename U> |
| 98 | constexpr auto CmpDiv(T t, U u) { |
| 99 | t /= u; |
| 100 | return t; |
| 101 | } |
| 102 | template <typename T, typename U> |
| 103 | constexpr auto CmpRem(T t, U u) { |
| 104 | t %= u; |
| 105 | return t; |
| 106 | } |
| 107 | |
| 108 | template <typename T, typename U> |
| 109 | constexpr auto CmpAdd(T t, U u) { |
| 110 | t += u; |
| 111 | return t; |
| 112 | } |
| 113 | |
| 114 | template <typename T, typename U> |
| 115 | constexpr auto CmpSub(T t, U u) { |
| 116 | t -= u; |
| 117 | return t; |
| 118 | } |
| 119 | |
| 120 | template <typename T, typename U> |
| 121 | constexpr auto CmpLSH(T t, U u) { |
| 122 | t <<= u; |
| 123 | return t; |
| 124 | } |
| 125 | |
| 126 | template <typename T, typename U> |
| 127 | constexpr auto CmpRSH(T t, U u) { |
| 128 | t >>= u; |
| 129 | return t; |
| 130 | } |
| 131 | |
| 132 | template <typename T, typename U> |
| 133 | constexpr auto CmpBinAnd(T t, U u) { |
| 134 | t &= u; |
| 135 | return t; |
| 136 | } |
| 137 | |
| 138 | template <typename T, typename U> |
| 139 | constexpr auto CmpBinXOr(T t, U u) { |
| 140 | t ^= u; |
| 141 | return t; |
| 142 | } |
| 143 | |
| 144 | template <typename T, typename U> |
| 145 | constexpr auto CmpBinOr(T t, U u) { |
| 146 | t |= u; |
| 147 | return t; |
| 148 | } |
| 149 | |
| 150 | // Only int vs float makes a difference here, so we only need to test 1 of each. |
| 151 | // Test Char to make sure the mixed-nature of shifts around char is evident. |
| 152 | void CharUsage() { |
| 153 | constexpr auto a = FourCharsVecSize{6, 3, 2, 1} + |
| 154 | FourCharsVecSize{12, 15, 5, 7}; |
| 155 | // CHECK: store <4 x i8> <i8 18, i8 18, i8 7, i8 8> |
| 156 | constexpr auto b = FourCharsVecSize{19, 15, 13, 12} - |
| 157 | FourCharsVecSize{13, 14, 5, 3}; |
| 158 | // CHECK: store <4 x i8> <i8 6, i8 1, i8 8, i8 9> |
| 159 | constexpr auto c = FourCharsVecSize{8, 4, 2, 1} * |
| 160 | FourCharsVecSize{3, 4, 5, 6}; |
| 161 | // CHECK: store <4 x i8> <i8 24, i8 16, i8 10, i8 6> |
| 162 | constexpr auto d = FourCharsVecSize{12, 12, 10, 10} / |
| 163 | FourCharsVecSize{6, 4, 5, 2}; |
| 164 | // CHECK: store <4 x i8> <i8 2, i8 3, i8 2, i8 5> |
| 165 | constexpr auto e = FourCharsVecSize{12, 12, 10, 10} % |
| 166 | FourCharsVecSize{6, 4, 4, 3}; |
| 167 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 2, i8 1> |
| 168 | |
| 169 | constexpr auto f = FourCharsVecSize{6, 3, 2, 1} + 3; |
| 170 | // CHECK: store <4 x i8> <i8 9, i8 6, i8 5, i8 4> |
| 171 | constexpr auto g = FourCharsVecSize{19, 15, 12, 10} - 3; |
| 172 | // CHECK: store <4 x i8> <i8 16, i8 12, i8 9, i8 7> |
| 173 | constexpr auto h = FourCharsVecSize{8, 4, 2, 1} * 3; |
| 174 | // CHECK: store <4 x i8> <i8 24, i8 12, i8 6, i8 3> |
| 175 | constexpr auto j = FourCharsVecSize{12, 15, 18, 21} / 3; |
| 176 | // CHECK: store <4 x i8> <i8 4, i8 5, i8 6, i8 7> |
| 177 | constexpr auto k = FourCharsVecSize{12, 17, 19, 22} % 3; |
| 178 | // CHECK: store <4 x i8> <i8 0, i8 2, i8 1, i8 1> |
| 179 | |
| 180 | constexpr auto l = 3 + FourCharsVecSize{6, 3, 2, 1}; |
| 181 | // CHECK: store <4 x i8> <i8 9, i8 6, i8 5, i8 4> |
| 182 | constexpr auto m = 20 - FourCharsVecSize{19, 15, 12, 10}; |
| 183 | // CHECK: store <4 x i8> <i8 1, i8 5, i8 8, i8 10> |
| 184 | constexpr auto n = 3 * FourCharsVecSize{8, 4, 2, 1}; |
| 185 | // CHECK: store <4 x i8> <i8 24, i8 12, i8 6, i8 3> |
| 186 | constexpr auto o = 100 / FourCharsVecSize{12, 15, 18, 21}; |
| 187 | // CHECK: store <4 x i8> <i8 8, i8 6, i8 5, i8 4> |
| 188 | constexpr auto p = 100 % FourCharsVecSize{12, 15, 18, 21}; |
| 189 | // CHECK: store <4 x i8> <i8 4, i8 10, i8 10, i8 16> |
| 190 | |
| 191 | constexpr auto q = FourCharsVecSize{6, 3, 2, 1} << FourCharsVecSize{1, 1, 2, 2}; |
| 192 | // CHECK: store <4 x i8> <i8 12, i8 6, i8 8, i8 4> |
| 193 | constexpr auto r = FourCharsVecSize{19, 15, 12, 10} >> |
| 194 | FourCharsVecSize{1, 1, 2, 2}; |
| 195 | // CHECK: store <4 x i8> <i8 9, i8 7, i8 3, i8 2> |
| 196 | constexpr auto s = FourCharsVecSize{6, 3, 5, 10} << 1; |
| 197 | // CHECK: store <4 x i8> <i8 12, i8 6, i8 10, i8 20> |
| 198 | constexpr auto t = FourCharsVecSize{19, 15, 10, 20} >> 1; |
| 199 | // CHECK: store <4 x i8> <i8 9, i8 7, i8 5, i8 10> |
| 200 | constexpr auto u = 12 << FourCharsVecSize{1, 2, 3, 3}; |
| 201 | // CHECK: store <4 x i8> <i8 24, i8 48, i8 96, i8 96> |
| 202 | constexpr auto v = 12 >> FourCharsVecSize{1, 2, 2, 1}; |
| 203 | // CHECK: store <4 x i8> <i8 6, i8 3, i8 3, i8 6> |
| 204 | |
| 205 | constexpr auto w = FourCharsVecSize{1, 2, 3, 4} < |
| 206 | FourCharsVecSize{4, 3, 2, 1}; |
| 207 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 0> |
| 208 | constexpr auto x = FourCharsVecSize{1, 2, 3, 4} > |
| 209 | FourCharsVecSize{4, 3, 2, 1}; |
| 210 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 1> |
| 211 | constexpr auto y = FourCharsVecSize{1, 2, 3, 4} <= |
| 212 | FourCharsVecSize{4, 3, 3, 1}; |
| 213 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 0> |
| 214 | constexpr auto z = FourCharsVecSize{1, 2, 3, 4} >= |
| 215 | FourCharsVecSize{4, 3, 3, 1}; |
| 216 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 1> |
| 217 | constexpr auto A = FourCharsVecSize{1, 2, 3, 4} == |
| 218 | FourCharsVecSize{4, 3, 3, 1}; |
| 219 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 0> |
| 220 | constexpr auto B = FourCharsVecSize{1, 2, 3, 4} != |
| 221 | FourCharsVecSize{4, 3, 3, 1}; |
| 222 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 1> |
| 223 | |
| 224 | constexpr auto C = FourCharsVecSize{1, 2, 3, 4} < 3; |
| 225 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 0> |
| 226 | constexpr auto D = FourCharsVecSize{1, 2, 3, 4} > 3; |
| 227 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 0, i8 1> |
| 228 | constexpr auto E = FourCharsVecSize{1, 2, 3, 4} <= 3; |
| 229 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 0> |
| 230 | constexpr auto F = FourCharsVecSize{1, 2, 3, 4} >= 3; |
| 231 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 1> |
| 232 | constexpr auto G = FourCharsVecSize{1, 2, 3, 4} == 3; |
| 233 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 0> |
| 234 | constexpr auto H = FourCharsVecSize{1, 2, 3, 4} != 3; |
| 235 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 1> |
| 236 | |
| 237 | constexpr auto I = FourCharsVecSize{1, 2, 3, 4} & |
| 238 | FourCharsVecSize{4, 3, 2, 1}; |
| 239 | // CHECK: store <4 x i8> <i8 0, i8 2, i8 2, i8 0> |
| 240 | constexpr auto J = FourCharsVecSize{1, 2, 3, 4} ^ |
| 241 | FourCharsVecSize { 4, 3, 2, 1 }; |
| 242 | // CHECK: store <4 x i8> <i8 5, i8 1, i8 1, i8 5> |
| 243 | constexpr auto K = FourCharsVecSize{1, 2, 3, 4} | |
| 244 | FourCharsVecSize{4, 3, 2, 1}; |
| 245 | // CHECK: store <4 x i8> <i8 5, i8 3, i8 3, i8 5> |
| 246 | constexpr auto L = FourCharsVecSize{1, 2, 3, 4} & 3; |
| 247 | // CHECK: store <4 x i8> <i8 1, i8 2, i8 3, i8 0> |
| 248 | constexpr auto M = FourCharsVecSize{1, 2, 3, 4} ^ 3; |
| 249 | // CHECK: store <4 x i8> <i8 2, i8 1, i8 0, i8 7> |
| 250 | constexpr auto N = FourCharsVecSize{1, 2, 3, 4} | 3; |
| 251 | // CHECK: store <4 x i8> <i8 3, i8 3, i8 3, i8 7> |
| 252 | |
| 253 | constexpr auto O = FourCharsVecSize{5, 0, 6, 0} && |
| 254 | FourCharsVecSize{5, 5, 0, 0}; |
| 255 | // CHECK: store <4 x i8> <i8 1, i8 0, i8 0, i8 0> |
| 256 | constexpr auto P = FourCharsVecSize{5, 0, 6, 0} || |
| 257 | FourCharsVecSize{5, 5, 0, 0}; |
| 258 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 0> |
| 259 | |
| 260 | constexpr auto Q = FourCharsVecSize{5, 0, 6, 0} && 3; |
| 261 | // CHECK: store <4 x i8> <i8 1, i8 0, i8 1, i8 0> |
| 262 | constexpr auto R = FourCharsVecSize{5, 0, 6, 0} || 3; |
| 263 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 1> |
| 264 | |
| 265 | constexpr auto T = CmpMul(a, b); |
| 266 | // CHECK: store <4 x i8> <i8 108, i8 18, i8 56, i8 72> |
| 267 | |
| 268 | constexpr auto U = CmpDiv(a, b); |
| 269 | // CHECK: store <4 x i8> <i8 3, i8 18, i8 0, i8 0> |
| 270 | |
| 271 | constexpr auto V = CmpRem(a, b); |
| 272 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 7, i8 8> |
| 273 | |
| 274 | constexpr auto X = CmpAdd(a, b); |
| 275 | // CHECK: store <4 x i8> <i8 24, i8 19, i8 15, i8 17> |
| 276 | |
| 277 | constexpr auto Y = CmpSub(a, b); |
| 278 | // CHECK: store <4 x i8> <i8 12, i8 17, i8 -1, i8 -1> |
| 279 | |
| 280 | constexpr auto Z = CmpLSH(a, H); |
| 281 | // CHECK: store <4 x i8> <i8 36, i8 36, i8 7, i8 16> |
| 282 | |
| 283 | constexpr auto aa = CmpRSH(a, H); |
| 284 | // CHECK: store <4 x i8> <i8 9, i8 9, i8 7, i8 4> |
| 285 | |
| 286 | constexpr auto ab = CmpBinAnd(a, b); |
| 287 | // CHECK: store <4 x i8> <i8 2, i8 0, i8 0, i8 8> |
| 288 | |
| 289 | constexpr auto ac = CmpBinXOr(a, b); |
| 290 | // CHECK: store <4 x i8> <i8 20, i8 19, i8 15, i8 1> |
| 291 | |
| 292 | constexpr auto ad = CmpBinOr(a, b); |
| 293 | // CHECK: store <4 x i8> <i8 22, i8 19, i8 15, i8 9> |
| 294 | } |
| 295 | |
| 296 | void CharExtVecUsage() { |
| 297 | constexpr auto a = FourCharsExtVec{6, 3, 2, 1} + |
| 298 | FourCharsExtVec{12, 15, 5, 7}; |
| 299 | // CHECK: store <4 x i8> <i8 18, i8 18, i8 7, i8 8> |
| 300 | constexpr auto b = FourCharsExtVec{19, 15, 13, 12} - |
| 301 | FourCharsExtVec{13, 14, 5, 3}; |
| 302 | // CHECK: store <4 x i8> <i8 6, i8 1, i8 8, i8 9> |
| 303 | constexpr auto c = FourCharsExtVec{8, 4, 2, 1} * |
| 304 | FourCharsExtVec{3, 4, 5, 6}; |
| 305 | // CHECK: store <4 x i8> <i8 24, i8 16, i8 10, i8 6> |
| 306 | constexpr auto d = FourCharsExtVec{12, 12, 10, 10} / |
| 307 | FourCharsExtVec{6, 4, 5, 2}; |
| 308 | // CHECK: store <4 x i8> <i8 2, i8 3, i8 2, i8 5> |
| 309 | constexpr auto e = FourCharsExtVec{12, 12, 10, 10} % |
| 310 | FourCharsExtVec{6, 4, 4, 3}; |
| 311 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 2, i8 1> |
| 312 | |
| 313 | constexpr auto f = FourCharsExtVec{6, 3, 2, 1} + 3; |
| 314 | // CHECK: store <4 x i8> <i8 9, i8 6, i8 5, i8 4> |
| 315 | constexpr auto g = FourCharsExtVec{19, 15, 12, 10} - 3; |
| 316 | // CHECK: store <4 x i8> <i8 16, i8 12, i8 9, i8 7> |
| 317 | constexpr auto h = FourCharsExtVec{8, 4, 2, 1} * 3; |
| 318 | // CHECK: store <4 x i8> <i8 24, i8 12, i8 6, i8 3> |
| 319 | constexpr auto j = FourCharsExtVec{12, 15, 18, 21} / 3; |
| 320 | // CHECK: store <4 x i8> <i8 4, i8 5, i8 6, i8 7> |
| 321 | constexpr auto k = FourCharsExtVec{12, 17, 19, 22} % 3; |
| 322 | // CHECK: store <4 x i8> <i8 0, i8 2, i8 1, i8 1> |
| 323 | |
| 324 | constexpr auto l = 3 + FourCharsExtVec{6, 3, 2, 1}; |
| 325 | // CHECK: store <4 x i8> <i8 9, i8 6, i8 5, i8 4> |
| 326 | constexpr auto m = 20 - FourCharsExtVec{19, 15, 12, 10}; |
| 327 | // CHECK: store <4 x i8> <i8 1, i8 5, i8 8, i8 10> |
| 328 | constexpr auto n = 3 * FourCharsExtVec{8, 4, 2, 1}; |
| 329 | // CHECK: store <4 x i8> <i8 24, i8 12, i8 6, i8 3> |
| 330 | constexpr auto o = 100 / FourCharsExtVec{12, 15, 18, 21}; |
| 331 | // CHECK: store <4 x i8> <i8 8, i8 6, i8 5, i8 4> |
| 332 | constexpr auto p = 100 % FourCharsExtVec{12, 15, 18, 21}; |
| 333 | // CHECK: store <4 x i8> <i8 4, i8 10, i8 10, i8 16> |
| 334 | |
| 335 | constexpr auto q = FourCharsExtVec{6, 3, 2, 1} << FourCharsVecSize{1, 1, 2, 2}; |
| 336 | // CHECK: store <4 x i8> <i8 12, i8 6, i8 8, i8 4> |
| 337 | constexpr auto r = FourCharsExtVec{19, 15, 12, 10} >> |
| 338 | FourCharsExtVec{1, 1, 2, 2}; |
| 339 | // CHECK: store <4 x i8> <i8 9, i8 7, i8 3, i8 2> |
| 340 | constexpr auto s = FourCharsExtVec{6, 3, 5, 10} << 1; |
| 341 | // CHECK: store <4 x i8> <i8 12, i8 6, i8 10, i8 20> |
| 342 | constexpr auto t = FourCharsExtVec{19, 15, 10, 20} >> 1; |
| 343 | // CHECK: store <4 x i8> <i8 9, i8 7, i8 5, i8 10> |
| 344 | constexpr auto u = 12 << FourCharsExtVec{1, 2, 3, 3}; |
| 345 | // CHECK: store <4 x i8> <i8 24, i8 48, i8 96, i8 96> |
| 346 | constexpr auto v = 12 >> FourCharsExtVec{1, 2, 2, 1}; |
| 347 | // CHECK: store <4 x i8> <i8 6, i8 3, i8 3, i8 6> |
| 348 | |
| 349 | constexpr auto w = FourCharsExtVec{1, 2, 3, 4} < |
| 350 | FourCharsExtVec{4, 3, 2, 1}; |
| 351 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 0> |
| 352 | constexpr auto x = FourCharsExtVec{1, 2, 3, 4} > |
| 353 | FourCharsExtVec{4, 3, 2, 1}; |
| 354 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 1> |
| 355 | constexpr auto y = FourCharsExtVec{1, 2, 3, 4} <= |
| 356 | FourCharsExtVec{4, 3, 3, 1}; |
| 357 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 0> |
| 358 | constexpr auto z = FourCharsExtVec{1, 2, 3, 4} >= |
| 359 | FourCharsExtVec{4, 3, 3, 1}; |
| 360 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 1> |
| 361 | constexpr auto A = FourCharsExtVec{1, 2, 3, 4} == |
| 362 | FourCharsExtVec{4, 3, 3, 1}; |
| 363 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 0> |
| 364 | constexpr auto B = FourCharsExtVec{1, 2, 3, 4} != |
| 365 | FourCharsExtVec{4, 3, 3, 1}; |
| 366 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 1> |
| 367 | |
| 368 | constexpr auto C = FourCharsExtVec{1, 2, 3, 4} < 3; |
| 369 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 0> |
| 370 | constexpr auto D = FourCharsExtVec{1, 2, 3, 4} > 3; |
| 371 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 0, i8 1> |
| 372 | constexpr auto E = FourCharsExtVec{1, 2, 3, 4} <= 3; |
| 373 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 0> |
| 374 | constexpr auto F = FourCharsExtVec{1, 2, 3, 4} >= 3; |
| 375 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 1> |
| 376 | constexpr auto G = FourCharsExtVec{1, 2, 3, 4} == 3; |
| 377 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 1, i8 0> |
| 378 | constexpr auto H = FourCharsExtVec{1, 2, 3, 4} != 3; |
| 379 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 0, i8 1> |
| 380 | |
| 381 | constexpr auto I = FourCharsExtVec{1, 2, 3, 4} & |
| 382 | FourCharsExtVec{4, 3, 2, 1}; |
| 383 | // CHECK: store <4 x i8> <i8 0, i8 2, i8 2, i8 0> |
| 384 | constexpr auto J = FourCharsExtVec{1, 2, 3, 4} ^ |
| 385 | FourCharsExtVec { 4, 3, 2, 1 }; |
| 386 | // CHECK: store <4 x i8> <i8 5, i8 1, i8 1, i8 5> |
| 387 | constexpr auto K = FourCharsExtVec{1, 2, 3, 4} | |
| 388 | FourCharsExtVec{4, 3, 2, 1}; |
| 389 | // CHECK: store <4 x i8> <i8 5, i8 3, i8 3, i8 5> |
| 390 | constexpr auto L = FourCharsExtVec{1, 2, 3, 4} & 3; |
| 391 | // CHECK: store <4 x i8> <i8 1, i8 2, i8 3, i8 0> |
| 392 | constexpr auto M = FourCharsExtVec{1, 2, 3, 4} ^ 3; |
| 393 | // CHECK: store <4 x i8> <i8 2, i8 1, i8 0, i8 7> |
| 394 | constexpr auto N = FourCharsExtVec{1, 2, 3, 4} | 3; |
| 395 | // CHECK: store <4 x i8> <i8 3, i8 3, i8 3, i8 7> |
| 396 | |
| 397 | constexpr auto O = FourCharsExtVec{5, 0, 6, 0} && |
| 398 | FourCharsExtVec{5, 5, 0, 0}; |
| 399 | // CHECK: store <4 x i8> <i8 1, i8 0, i8 0, i8 0> |
| 400 | constexpr auto P = FourCharsExtVec{5, 0, 6, 0} || |
| 401 | FourCharsExtVec{5, 5, 0, 0}; |
| 402 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 0> |
| 403 | |
| 404 | constexpr auto Q = FourCharsExtVec{5, 0, 6, 0} && 3; |
| 405 | // CHECK: store <4 x i8> <i8 1, i8 0, i8 1, i8 0> |
| 406 | constexpr auto R = FourCharsExtVec{5, 0, 6, 0} || 3; |
| 407 | // CHECK: store <4 x i8> <i8 1, i8 1, i8 1, i8 1> |
| 408 | |
| 409 | constexpr auto T = CmpMul(a, b); |
| 410 | // CHECK: store <4 x i8> <i8 108, i8 18, i8 56, i8 72> |
| 411 | |
| 412 | constexpr auto U = CmpDiv(a, b); |
| 413 | // CHECK: store <4 x i8> <i8 3, i8 18, i8 0, i8 0> |
| 414 | |
| 415 | constexpr auto V = CmpRem(a, b); |
| 416 | // CHECK: store <4 x i8> <i8 0, i8 0, i8 7, i8 8> |
| 417 | |
| 418 | constexpr auto X = CmpAdd(a, b); |
| 419 | // CHECK: store <4 x i8> <i8 24, i8 19, i8 15, i8 17> |
| 420 | |
| 421 | constexpr auto Y = CmpSub(a, b); |
| 422 | // CHECK: store <4 x i8> <i8 12, i8 17, i8 -1, i8 -1> |
| 423 | |
| 424 | constexpr auto Z = CmpLSH(a, H); |
| 425 | // CHECK: store <4 x i8> <i8 36, i8 36, i8 7, i8 16> |
| 426 | |
| 427 | constexpr auto aa = CmpRSH(a, H); |
| 428 | // CHECK: store <4 x i8> <i8 9, i8 9, i8 7, i8 4> |
| 429 | |
| 430 | constexpr auto ab = CmpBinAnd(a, b); |
| 431 | // CHECK: store <4 x i8> <i8 2, i8 0, i8 0, i8 8> |
| 432 | |
| 433 | constexpr auto ac = CmpBinXOr(a, b); |
| 434 | // CHECK: store <4 x i8> <i8 20, i8 19, i8 15, i8 1> |
| 435 | |
| 436 | constexpr auto ad = CmpBinOr(a, b); |
| 437 | // CHECK: store <4 x i8> <i8 22, i8 19, i8 15, i8 9> |
| 438 | } |
| 439 | |
| 440 | void FloatUsage() { |
| 441 | constexpr auto a = FourFloatsVecSize{6, 3, 2, 1} + |
| 442 | FourFloatsVecSize{12, 15, 5, 7}; |
| 443 | // CHECK: <4 x float> <float 1.800000e+01, float 1.800000e+01, float 7.000000e+00, float 8.000000e+00> |
| 444 | constexpr auto b = FourFloatsVecSize{19, 15, 13, 12} - |
| 445 | FourFloatsVecSize{13, 14, 5, 3}; |
| 446 | // CHECK: store <4 x float> <float 6.000000e+00, float 1.000000e+00, float 8.000000e+00, float 9.000000e+00> |
| 447 | constexpr auto c = FourFloatsVecSize{8, 4, 2, 1} * |
| 448 | FourFloatsVecSize{3, 4, 5, 6}; |
| 449 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.600000e+01, float 1.000000e+01, float 6.000000e+00> |
| 450 | constexpr auto d = FourFloatsVecSize{12, 12, 10, 10} / |
| 451 | FourFloatsVecSize{6, 4, 5, 2}; |
| 452 | // CHECK: store <4 x float> <float 2.000000e+00, float 3.000000e+00, float 2.000000e+00, float 5.000000e+00> |
| 453 | |
| 454 | constexpr auto f = FourFloatsVecSize{6, 3, 2, 1} + 3; |
| 455 | // CHECK: store <4 x float> <float 9.000000e+00, float 6.000000e+00, float 5.000000e+00, float 4.000000e+00> |
| 456 | constexpr auto g = FourFloatsVecSize{19, 15, 12, 10} - 3; |
| 457 | // CHECK: store <4 x float> <float 1.600000e+01, float 1.200000e+01, float 9.000000e+00, float 7.000000e+00> |
| 458 | constexpr auto h = FourFloatsVecSize{8, 4, 2, 1} * 3; |
| 459 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.200000e+01, float 6.000000e+00, float 3.000000e+00> |
| 460 | constexpr auto j = FourFloatsVecSize{12, 15, 18, 21} / 3; |
| 461 | // CHECK: store <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00> |
| 462 | |
| 463 | constexpr auto l = 3 + FourFloatsVecSize{6, 3, 2, 1}; |
| 464 | // CHECK: store <4 x float> <float 9.000000e+00, float 6.000000e+00, float 5.000000e+00, float 4.000000e+00> |
| 465 | constexpr auto m = 20 - FourFloatsVecSize{19, 15, 12, 10}; |
| 466 | // CHECK: store <4 x float> <float 1.000000e+00, float 5.000000e+00, float 8.000000e+00, float 1.000000e+01> |
| 467 | constexpr auto n = 3 * FourFloatsVecSize{8, 4, 2, 1}; |
| 468 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.200000e+01, float 6.000000e+00, float 3.000000e+00> |
| 469 | constexpr auto o = 100 / FourFloatsVecSize{12, 15, 18, 21}; |
| 470 | // CHECK: store <4 x float> <float 0x4020AAAAA0000000, float 0x401AAAAAA0000000, float 0x401638E380000000, float 0x40130C30C0000000> |
| 471 | |
| 472 | constexpr auto w = FourFloatsVecSize{1, 2, 3, 4} < |
| 473 | FourFloatsVecSize{4, 3, 2, 1}; |
| 474 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 0> |
| 475 | constexpr auto x = FourFloatsVecSize{1, 2, 3, 4} > |
| 476 | FourFloatsVecSize{4, 3, 2, 1}; |
| 477 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 1> |
| 478 | constexpr auto y = FourFloatsVecSize{1, 2, 3, 4} <= |
| 479 | FourFloatsVecSize{4, 3, 3, 1}; |
| 480 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 0> |
| 481 | constexpr auto z = FourFloatsVecSize{1, 2, 3, 4} >= |
| 482 | FourFloatsVecSize{4, 3, 3, 1}; |
| 483 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 1> |
| 484 | constexpr auto A = FourFloatsVecSize{1, 2, 3, 4} == |
| 485 | FourFloatsVecSize{4, 3, 3, 1}; |
| 486 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 0> |
| 487 | constexpr auto B = FourFloatsVecSize{1, 2, 3, 4} != |
| 488 | FourFloatsVecSize{4, 3, 3, 1}; |
| 489 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 1> |
| 490 | |
| 491 | constexpr auto C = FourFloatsVecSize{1, 2, 3, 4} < 3; |
| 492 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 0> |
| 493 | constexpr auto D = FourFloatsVecSize{1, 2, 3, 4} > 3; |
| 494 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 0, i32 1> |
| 495 | constexpr auto E = FourFloatsVecSize{1, 2, 3, 4} <= 3; |
| 496 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 0> |
| 497 | constexpr auto F = FourFloatsVecSize{1, 2, 3, 4} >= 3; |
| 498 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 1> |
| 499 | constexpr auto G = FourFloatsVecSize{1, 2, 3, 4} == 3; |
| 500 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 0> |
| 501 | constexpr auto H = FourFloatsVecSize{1, 2, 3, 4} != 3; |
| 502 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 1> |
| 503 | |
| 504 | constexpr auto O = FourFloatsVecSize{5, 0, 6, 0} && |
| 505 | FourFloatsVecSize{5, 5, 0, 0}; |
| 506 | // CHECK: store <4 x i32> <i32 1, i32 0, i32 0, i32 0> |
| 507 | constexpr auto P = FourFloatsVecSize{5, 0, 6, 0} || |
| 508 | FourFloatsVecSize{5, 5, 0, 0}; |
| 509 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 0> |
| 510 | |
| 511 | constexpr auto Q = FourFloatsVecSize{5, 0, 6, 0} && 3; |
| 512 | // CHECK: store <4 x i32> <i32 1, i32 0, i32 1, i32 0> |
| 513 | constexpr auto R = FourFloatsVecSize{5, 0, 6, 0} || 3; |
| 514 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 1> |
| 515 | |
| 516 | constexpr auto T = CmpMul(a, b); |
| 517 | // CHECK: store <4 x float> <float 1.080000e+02, float 1.800000e+01, float 5.600000e+01, float 7.200000e+01> |
| 518 | |
| 519 | constexpr auto U = CmpDiv(a, b); |
| 520 | // CHECK: store <4 x float> <float 3.000000e+00, float 1.800000e+01, float 8.750000e-01, float 0x3FEC71C720000000> |
| 521 | |
| 522 | constexpr auto X = CmpAdd(a, b); |
| 523 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.900000e+01, float 1.500000e+01, float 1.700000e+01> |
| 524 | |
| 525 | constexpr auto Y = CmpSub(a, b); |
| 526 | // CHECK: store <4 x float> <float 1.200000e+01, float 1.700000e+01, float -1.000000e+00, float -1.000000e+00> |
| 527 | } |
| 528 | |
| 529 | void FloatVecUsage() { |
| 530 | constexpr auto a = FourFloatsVecSize{6, 3, 2, 1} + |
| 531 | FourFloatsVecSize{12, 15, 5, 7}; |
| 532 | // CHECK: <4 x float> <float 1.800000e+01, float 1.800000e+01, float 7.000000e+00, float 8.000000e+00> |
| 533 | constexpr auto b = FourFloatsVecSize{19, 15, 13, 12} - |
| 534 | FourFloatsVecSize{13, 14, 5, 3}; |
| 535 | // CHECK: store <4 x float> <float 6.000000e+00, float 1.000000e+00, float 8.000000e+00, float 9.000000e+00> |
| 536 | constexpr auto c = FourFloatsVecSize{8, 4, 2, 1} * |
| 537 | FourFloatsVecSize{3, 4, 5, 6}; |
| 538 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.600000e+01, float 1.000000e+01, float 6.000000e+00> |
| 539 | constexpr auto d = FourFloatsVecSize{12, 12, 10, 10} / |
| 540 | FourFloatsVecSize{6, 4, 5, 2}; |
| 541 | // CHECK: store <4 x float> <float 2.000000e+00, float 3.000000e+00, float 2.000000e+00, float 5.000000e+00> |
| 542 | |
| 543 | constexpr auto f = FourFloatsVecSize{6, 3, 2, 1} + 3; |
| 544 | // CHECK: store <4 x float> <float 9.000000e+00, float 6.000000e+00, float 5.000000e+00, float 4.000000e+00> |
| 545 | constexpr auto g = FourFloatsVecSize{19, 15, 12, 10} - 3; |
| 546 | // CHECK: store <4 x float> <float 1.600000e+01, float 1.200000e+01, float 9.000000e+00, float 7.000000e+00> |
| 547 | constexpr auto h = FourFloatsVecSize{8, 4, 2, 1} * 3; |
| 548 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.200000e+01, float 6.000000e+00, float 3.000000e+00> |
| 549 | constexpr auto j = FourFloatsVecSize{12, 15, 18, 21} / 3; |
| 550 | // CHECK: store <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00> |
| 551 | |
| 552 | constexpr auto l = 3 + FourFloatsVecSize{6, 3, 2, 1}; |
| 553 | // CHECK: store <4 x float> <float 9.000000e+00, float 6.000000e+00, float 5.000000e+00, float 4.000000e+00> |
| 554 | constexpr auto m = 20 - FourFloatsVecSize{19, 15, 12, 10}; |
| 555 | // CHECK: store <4 x float> <float 1.000000e+00, float 5.000000e+00, float 8.000000e+00, float 1.000000e+01> |
| 556 | constexpr auto n = 3 * FourFloatsVecSize{8, 4, 2, 1}; |
| 557 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.200000e+01, float 6.000000e+00, float 3.000000e+00> |
| 558 | constexpr auto o = 100 / FourFloatsVecSize{12, 15, 18, 21}; |
| 559 | // CHECK: store <4 x float> <float 0x4020AAAAA0000000, float 0x401AAAAAA0000000, float 0x401638E380000000, float 0x40130C30C0000000> |
| 560 | |
| 561 | constexpr auto w = FourFloatsVecSize{1, 2, 3, 4} < |
| 562 | FourFloatsVecSize{4, 3, 2, 1}; |
| 563 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 0> |
| 564 | constexpr auto x = FourFloatsVecSize{1, 2, 3, 4} > |
| 565 | FourFloatsVecSize{4, 3, 2, 1}; |
| 566 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 1> |
| 567 | constexpr auto y = FourFloatsVecSize{1, 2, 3, 4} <= |
| 568 | FourFloatsVecSize{4, 3, 3, 1}; |
| 569 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 0> |
| 570 | constexpr auto z = FourFloatsVecSize{1, 2, 3, 4} >= |
| 571 | FourFloatsVecSize{4, 3, 3, 1}; |
| 572 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 1> |
| 573 | constexpr auto A = FourFloatsVecSize{1, 2, 3, 4} == |
| 574 | FourFloatsVecSize{4, 3, 3, 1}; |
| 575 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 0> |
| 576 | constexpr auto B = FourFloatsVecSize{1, 2, 3, 4} != |
| 577 | FourFloatsVecSize{4, 3, 3, 1}; |
| 578 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 1> |
| 579 | |
| 580 | constexpr auto C = FourFloatsVecSize{1, 2, 3, 4} < 3; |
| 581 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 0> |
| 582 | constexpr auto D = FourFloatsVecSize{1, 2, 3, 4} > 3; |
| 583 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 0, i32 1> |
| 584 | constexpr auto E = FourFloatsVecSize{1, 2, 3, 4} <= 3; |
| 585 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 0> |
| 586 | constexpr auto F = FourFloatsVecSize{1, 2, 3, 4} >= 3; |
| 587 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 1> |
| 588 | constexpr auto G = FourFloatsVecSize{1, 2, 3, 4} == 3; |
| 589 | // CHECK: store <4 x i32> <i32 0, i32 0, i32 1, i32 0> |
| 590 | constexpr auto H = FourFloatsVecSize{1, 2, 3, 4} != 3; |
| 591 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 0, i32 1> |
| 592 | |
| 593 | constexpr auto O = FourFloatsVecSize{5, 0, 6, 0} && |
| 594 | FourFloatsVecSize{5, 5, 0, 0}; |
| 595 | // CHECK: store <4 x i32> <i32 1, i32 0, i32 0, i32 0> |
| 596 | constexpr auto P = FourFloatsVecSize{5, 0, 6, 0} || |
| 597 | FourFloatsVecSize{5, 5, 0, 0}; |
| 598 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 0> |
| 599 | |
| 600 | constexpr auto Q = FourFloatsVecSize{5, 0, 6, 0} && 3; |
| 601 | // CHECK: store <4 x i32> <i32 1, i32 0, i32 1, i32 0> |
| 602 | constexpr auto R = FourFloatsVecSize{5, 0, 6, 0} || 3; |
| 603 | // CHECK: store <4 x i32> <i32 1, i32 1, i32 1, i32 1> |
| 604 | |
| 605 | constexpr auto T = CmpMul(a, b); |
| 606 | // CHECK: store <4 x float> <float 1.080000e+02, float 1.800000e+01, float 5.600000e+01, float 7.200000e+01> |
| 607 | |
| 608 | constexpr auto U = CmpDiv(a, b); |
| 609 | // CHECK: store <4 x float> <float 3.000000e+00, float 1.800000e+01, float 8.750000e-01, float 0x3FEC71C720000000> |
| 610 | |
| 611 | constexpr auto X = CmpAdd(a, b); |
| 612 | // CHECK: store <4 x float> <float 2.400000e+01, float 1.900000e+01, float 1.500000e+01, float 1.700000e+01> |
| 613 | |
| 614 | constexpr auto Y = CmpSub(a, b); |
| 615 | // CHECK: store <4 x float> <float 1.200000e+01, float 1.700000e+01, float -1.000000e+00, float -1.000000e+00> |
| 616 | } |