Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 1 | // Copyright 2019 Google LLC |
| 2 | // |
| 3 | // This source code is licensed under the BSD-style license found in the |
| 4 | // LICENSE file in the root directory of this source tree. |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 5 | // |
| 6 | // Auto-generated file. Do not edit! |
| 7 | // Specification: test/f32-raddextexp.yaml |
| 8 | // Generator: tools/generate-raddextexp-test.py |
| 9 | |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 10 | |
| 11 | #include <gtest/gtest.h> |
| 12 | |
| 13 | #include <xnnpack/common.h> |
| 14 | #include <xnnpack/isa-checks.h> |
| 15 | |
| 16 | #include <xnnpack/raddextexp.h> |
| 17 | #include "raddextexp-microkernel-tester.h" |
| 18 | |
| 19 | |
| 20 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 21 | TEST(F32_RADDEXTEXP__AVX2_P5_X64, elements_eq_64) { |
| 22 | TEST_REQUIRES_X86_AVX2; |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 23 | RAddExtExpMicrokernelTester() |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 24 | .elements(64) |
| 25 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64); |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 26 | } |
| 27 | |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 28 | TEST(F32_RADDEXTEXP__AVX2_P5_X64, elements_div_64) { |
| 29 | TEST_REQUIRES_X86_AVX2; |
| 30 | for (size_t elements = 128; elements < 640; elements += 64) { |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 31 | RAddExtExpMicrokernelTester() |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 32 | .elements(elements) |
| 33 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64); |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 34 | } |
| 35 | } |
| 36 | |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 37 | TEST(F32_RADDEXTEXP__AVX2_P5_X64, elements_lt_64) { |
| 38 | TEST_REQUIRES_X86_AVX2; |
| 39 | for (size_t elements = 1; elements < 64; elements++) { |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 40 | RAddExtExpMicrokernelTester() |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 41 | .elements(elements) |
| 42 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64); |
| 43 | } |
| 44 | } |
| 45 | |
| 46 | TEST(F32_RADDEXTEXP__AVX2_P5_X64, elements_gt_64) { |
| 47 | TEST_REQUIRES_X86_AVX2; |
| 48 | for (size_t elements = 65; elements < 128; elements++) { |
| 49 | RAddExtExpMicrokernelTester() |
| 50 | .elements(elements) |
| 51 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64); |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 52 | } |
| 53 | } |
| 54 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 55 | |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 56 | |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 57 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 58 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC2, elements_eq_64) { |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 59 | TEST_REQUIRES_X86_AVX2; |
| 60 | RAddExtExpMicrokernelTester() |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 61 | .elements(64) |
| 62 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc2); |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 63 | } |
| 64 | |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 65 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC2, elements_div_64) { |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 66 | TEST_REQUIRES_X86_AVX2; |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 67 | for (size_t elements = 128; elements < 640; elements += 64) { |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 68 | RAddExtExpMicrokernelTester() |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 69 | .elements(elements) |
| 70 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc2); |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 71 | } |
| 72 | } |
| 73 | |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 74 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC2, elements_lt_64) { |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 75 | TEST_REQUIRES_X86_AVX2; |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 76 | for (size_t elements = 1; elements < 64; elements++) { |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 77 | RAddExtExpMicrokernelTester() |
Marat Dukhan | 4c4eb00 | 2019-12-08 21:27:49 -0800 | [diff] [blame] | 78 | .elements(elements) |
| 79 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc2); |
| 80 | } |
| 81 | } |
| 82 | |
| 83 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC2, elements_gt_64) { |
| 84 | TEST_REQUIRES_X86_AVX2; |
| 85 | for (size_t elements = 65; elements < 128; elements++) { |
| 86 | RAddExtExpMicrokernelTester() |
| 87 | .elements(elements) |
| 88 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc2); |
| 89 | } |
| 90 | } |
| 91 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 92 | |
| 93 | |
| 94 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 95 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC4, elements_eq_64) { |
| 96 | TEST_REQUIRES_X86_AVX2; |
| 97 | RAddExtExpMicrokernelTester() |
| 98 | .elements(64) |
| 99 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc4); |
| 100 | } |
| 101 | |
| 102 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC4, elements_div_64) { |
| 103 | TEST_REQUIRES_X86_AVX2; |
| 104 | for (size_t elements = 128; elements < 640; elements += 64) { |
| 105 | RAddExtExpMicrokernelTester() |
| 106 | .elements(elements) |
| 107 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc4); |
| 108 | } |
| 109 | } |
| 110 | |
| 111 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC4, elements_lt_64) { |
| 112 | TEST_REQUIRES_X86_AVX2; |
| 113 | for (size_t elements = 1; elements < 64; elements++) { |
| 114 | RAddExtExpMicrokernelTester() |
| 115 | .elements(elements) |
| 116 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc4); |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | TEST(F32_RADDEXTEXP__AVX2_P5_X64_ACC4, elements_gt_64) { |
| 121 | TEST_REQUIRES_X86_AVX2; |
| 122 | for (size_t elements = 65; elements < 128; elements++) { |
| 123 | RAddExtExpMicrokernelTester() |
| 124 | .elements(elements) |
| 125 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x64_acc4); |
| 126 | } |
| 127 | } |
| 128 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 129 | |
| 130 | |
| 131 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 132 | TEST(F32_RADDEXTEXP__AVX2_P5_X72, elements_eq_72) { |
| 133 | TEST_REQUIRES_X86_AVX2; |
| 134 | RAddExtExpMicrokernelTester() |
| 135 | .elements(72) |
| 136 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72); |
| 137 | } |
| 138 | |
| 139 | TEST(F32_RADDEXTEXP__AVX2_P5_X72, elements_div_72) { |
| 140 | TEST_REQUIRES_X86_AVX2; |
| 141 | for (size_t elements = 144; elements < 720; elements += 72) { |
| 142 | RAddExtExpMicrokernelTester() |
| 143 | .elements(elements) |
| 144 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72); |
| 145 | } |
| 146 | } |
| 147 | |
| 148 | TEST(F32_RADDEXTEXP__AVX2_P5_X72, elements_lt_72) { |
| 149 | TEST_REQUIRES_X86_AVX2; |
| 150 | for (size_t elements = 1; elements < 72; elements++) { |
| 151 | RAddExtExpMicrokernelTester() |
| 152 | .elements(elements) |
| 153 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72); |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | TEST(F32_RADDEXTEXP__AVX2_P5_X72, elements_gt_72) { |
| 158 | TEST_REQUIRES_X86_AVX2; |
| 159 | for (size_t elements = 73; elements < 144; elements++) { |
| 160 | RAddExtExpMicrokernelTester() |
| 161 | .elements(elements) |
| 162 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72); |
| 163 | } |
| 164 | } |
| 165 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 166 | |
| 167 | |
| 168 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 169 | TEST(F32_RADDEXTEXP__AVX2_P5_X72_ACC3, elements_eq_72) { |
| 170 | TEST_REQUIRES_X86_AVX2; |
| 171 | RAddExtExpMicrokernelTester() |
| 172 | .elements(72) |
| 173 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72_acc3); |
| 174 | } |
| 175 | |
| 176 | TEST(F32_RADDEXTEXP__AVX2_P5_X72_ACC3, elements_div_72) { |
| 177 | TEST_REQUIRES_X86_AVX2; |
| 178 | for (size_t elements = 144; elements < 720; elements += 72) { |
| 179 | RAddExtExpMicrokernelTester() |
| 180 | .elements(elements) |
| 181 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72_acc3); |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | TEST(F32_RADDEXTEXP__AVX2_P5_X72_ACC3, elements_lt_72) { |
| 186 | TEST_REQUIRES_X86_AVX2; |
| 187 | for (size_t elements = 1; elements < 72; elements++) { |
| 188 | RAddExtExpMicrokernelTester() |
| 189 | .elements(elements) |
| 190 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72_acc3); |
| 191 | } |
| 192 | } |
| 193 | |
| 194 | TEST(F32_RADDEXTEXP__AVX2_P5_X72_ACC3, elements_gt_72) { |
| 195 | TEST_REQUIRES_X86_AVX2; |
| 196 | for (size_t elements = 73; elements < 144; elements++) { |
| 197 | RAddExtExpMicrokernelTester() |
| 198 | .elements(elements) |
| 199 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x72_acc3); |
| 200 | } |
| 201 | } |
| 202 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 203 | |
| 204 | |
| 205 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 206 | TEST(F32_RADDEXTEXP__AVX2_P5_X80, elements_eq_80) { |
| 207 | TEST_REQUIRES_X86_AVX2; |
| 208 | RAddExtExpMicrokernelTester() |
| 209 | .elements(80) |
| 210 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80); |
| 211 | } |
| 212 | |
| 213 | TEST(F32_RADDEXTEXP__AVX2_P5_X80, elements_div_80) { |
| 214 | TEST_REQUIRES_X86_AVX2; |
| 215 | for (size_t elements = 160; elements < 800; elements += 80) { |
| 216 | RAddExtExpMicrokernelTester() |
| 217 | .elements(elements) |
| 218 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80); |
| 219 | } |
| 220 | } |
| 221 | |
| 222 | TEST(F32_RADDEXTEXP__AVX2_P5_X80, elements_lt_80) { |
| 223 | TEST_REQUIRES_X86_AVX2; |
| 224 | for (size_t elements = 1; elements < 80; elements++) { |
| 225 | RAddExtExpMicrokernelTester() |
| 226 | .elements(elements) |
| 227 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80); |
| 228 | } |
| 229 | } |
| 230 | |
| 231 | TEST(F32_RADDEXTEXP__AVX2_P5_X80, elements_gt_80) { |
| 232 | TEST_REQUIRES_X86_AVX2; |
| 233 | for (size_t elements = 81; elements < 160; elements++) { |
| 234 | RAddExtExpMicrokernelTester() |
| 235 | .elements(elements) |
| 236 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80); |
| 237 | } |
| 238 | } |
| 239 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 240 | |
| 241 | |
| 242 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 243 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC2, elements_eq_80) { |
| 244 | TEST_REQUIRES_X86_AVX2; |
| 245 | RAddExtExpMicrokernelTester() |
| 246 | .elements(80) |
| 247 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc2); |
| 248 | } |
| 249 | |
| 250 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC2, elements_div_80) { |
| 251 | TEST_REQUIRES_X86_AVX2; |
| 252 | for (size_t elements = 160; elements < 800; elements += 80) { |
| 253 | RAddExtExpMicrokernelTester() |
| 254 | .elements(elements) |
| 255 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc2); |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC2, elements_lt_80) { |
| 260 | TEST_REQUIRES_X86_AVX2; |
| 261 | for (size_t elements = 1; elements < 80; elements++) { |
| 262 | RAddExtExpMicrokernelTester() |
| 263 | .elements(elements) |
| 264 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc2); |
| 265 | } |
| 266 | } |
| 267 | |
| 268 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC2, elements_gt_80) { |
| 269 | TEST_REQUIRES_X86_AVX2; |
| 270 | for (size_t elements = 81; elements < 160; elements++) { |
| 271 | RAddExtExpMicrokernelTester() |
| 272 | .elements(elements) |
| 273 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc2); |
| 274 | } |
| 275 | } |
| 276 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 277 | |
| 278 | |
| 279 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 280 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC5, elements_eq_80) { |
| 281 | TEST_REQUIRES_X86_AVX2; |
| 282 | RAddExtExpMicrokernelTester() |
| 283 | .elements(80) |
| 284 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc5); |
| 285 | } |
| 286 | |
| 287 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC5, elements_div_80) { |
| 288 | TEST_REQUIRES_X86_AVX2; |
| 289 | for (size_t elements = 160; elements < 800; elements += 80) { |
| 290 | RAddExtExpMicrokernelTester() |
| 291 | .elements(elements) |
| 292 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc5); |
| 293 | } |
| 294 | } |
| 295 | |
| 296 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC5, elements_lt_80) { |
| 297 | TEST_REQUIRES_X86_AVX2; |
| 298 | for (size_t elements = 1; elements < 80; elements++) { |
| 299 | RAddExtExpMicrokernelTester() |
| 300 | .elements(elements) |
| 301 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc5); |
| 302 | } |
| 303 | } |
| 304 | |
| 305 | TEST(F32_RADDEXTEXP__AVX2_P5_X80_ACC5, elements_gt_80) { |
| 306 | TEST_REQUIRES_X86_AVX2; |
| 307 | for (size_t elements = 81; elements < 160; elements++) { |
| 308 | RAddExtExpMicrokernelTester() |
| 309 | .elements(elements) |
| 310 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x80_acc5); |
| 311 | } |
| 312 | } |
| 313 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 314 | |
| 315 | |
| 316 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 317 | TEST(F32_RADDEXTEXP__AVX2_P5_X96, elements_eq_96) { |
| 318 | TEST_REQUIRES_X86_AVX2; |
| 319 | RAddExtExpMicrokernelTester() |
| 320 | .elements(96) |
| 321 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96); |
| 322 | } |
| 323 | |
| 324 | TEST(F32_RADDEXTEXP__AVX2_P5_X96, elements_div_96) { |
| 325 | TEST_REQUIRES_X86_AVX2; |
| 326 | for (size_t elements = 192; elements < 960; elements += 96) { |
| 327 | RAddExtExpMicrokernelTester() |
| 328 | .elements(elements) |
| 329 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96); |
| 330 | } |
| 331 | } |
| 332 | |
| 333 | TEST(F32_RADDEXTEXP__AVX2_P5_X96, elements_lt_96) { |
| 334 | TEST_REQUIRES_X86_AVX2; |
| 335 | for (size_t elements = 1; elements < 96; elements++) { |
| 336 | RAddExtExpMicrokernelTester() |
| 337 | .elements(elements) |
| 338 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96); |
| 339 | } |
| 340 | } |
| 341 | |
| 342 | TEST(F32_RADDEXTEXP__AVX2_P5_X96, elements_gt_96) { |
| 343 | TEST_REQUIRES_X86_AVX2; |
| 344 | for (size_t elements = 97; elements < 192; elements++) { |
| 345 | RAddExtExpMicrokernelTester() |
| 346 | .elements(elements) |
| 347 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96); |
| 348 | } |
| 349 | } |
| 350 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 351 | |
| 352 | |
| 353 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 354 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC2, elements_eq_96) { |
| 355 | TEST_REQUIRES_X86_AVX2; |
| 356 | RAddExtExpMicrokernelTester() |
| 357 | .elements(96) |
| 358 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc2); |
| 359 | } |
| 360 | |
| 361 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC2, elements_div_96) { |
| 362 | TEST_REQUIRES_X86_AVX2; |
| 363 | for (size_t elements = 192; elements < 960; elements += 96) { |
| 364 | RAddExtExpMicrokernelTester() |
| 365 | .elements(elements) |
| 366 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc2); |
| 367 | } |
| 368 | } |
| 369 | |
| 370 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC2, elements_lt_96) { |
| 371 | TEST_REQUIRES_X86_AVX2; |
| 372 | for (size_t elements = 1; elements < 96; elements++) { |
| 373 | RAddExtExpMicrokernelTester() |
| 374 | .elements(elements) |
| 375 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc2); |
| 376 | } |
| 377 | } |
| 378 | |
| 379 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC2, elements_gt_96) { |
| 380 | TEST_REQUIRES_X86_AVX2; |
| 381 | for (size_t elements = 97; elements < 192; elements++) { |
| 382 | RAddExtExpMicrokernelTester() |
| 383 | .elements(elements) |
| 384 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc2); |
| 385 | } |
| 386 | } |
| 387 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 388 | |
| 389 | |
| 390 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 391 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC3, elements_eq_96) { |
| 392 | TEST_REQUIRES_X86_AVX2; |
| 393 | RAddExtExpMicrokernelTester() |
| 394 | .elements(96) |
| 395 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc3); |
| 396 | } |
| 397 | |
| 398 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC3, elements_div_96) { |
| 399 | TEST_REQUIRES_X86_AVX2; |
| 400 | for (size_t elements = 192; elements < 960; elements += 96) { |
| 401 | RAddExtExpMicrokernelTester() |
| 402 | .elements(elements) |
| 403 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc3); |
| 404 | } |
| 405 | } |
| 406 | |
| 407 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC3, elements_lt_96) { |
| 408 | TEST_REQUIRES_X86_AVX2; |
| 409 | for (size_t elements = 1; elements < 96; elements++) { |
| 410 | RAddExtExpMicrokernelTester() |
| 411 | .elements(elements) |
| 412 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc3); |
| 413 | } |
| 414 | } |
| 415 | |
| 416 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC3, elements_gt_96) { |
| 417 | TEST_REQUIRES_X86_AVX2; |
| 418 | for (size_t elements = 97; elements < 192; elements++) { |
| 419 | RAddExtExpMicrokernelTester() |
| 420 | .elements(elements) |
| 421 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc3); |
| 422 | } |
| 423 | } |
| 424 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 425 | |
| 426 | |
| 427 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 428 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC6, elements_eq_96) { |
| 429 | TEST_REQUIRES_X86_AVX2; |
| 430 | RAddExtExpMicrokernelTester() |
| 431 | .elements(96) |
| 432 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6); |
| 433 | } |
| 434 | |
| 435 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC6, elements_div_96) { |
| 436 | TEST_REQUIRES_X86_AVX2; |
| 437 | for (size_t elements = 192; elements < 960; elements += 96) { |
| 438 | RAddExtExpMicrokernelTester() |
| 439 | .elements(elements) |
| 440 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6); |
| 441 | } |
| 442 | } |
| 443 | |
| 444 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC6, elements_lt_96) { |
| 445 | TEST_REQUIRES_X86_AVX2; |
| 446 | for (size_t elements = 1; elements < 96; elements++) { |
| 447 | RAddExtExpMicrokernelTester() |
| 448 | .elements(elements) |
| 449 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6); |
| 450 | } |
| 451 | } |
| 452 | |
| 453 | TEST(F32_RADDEXTEXP__AVX2_P5_X96_ACC6, elements_gt_96) { |
| 454 | TEST_REQUIRES_X86_AVX2; |
| 455 | for (size_t elements = 97; elements < 192; elements++) { |
| 456 | RAddExtExpMicrokernelTester() |
| 457 | .elements(elements) |
| 458 | .Test(xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6); |
| 459 | } |
| 460 | } |
| 461 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 462 | |
| 463 | |
| 464 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 465 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128, elements_eq_128) { |
| 466 | TEST_REQUIRES_X86_AVX512F; |
| 467 | RAddExtExpMicrokernelTester() |
| 468 | .elements(128) |
| 469 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128); |
| 470 | } |
| 471 | |
| 472 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128, elements_div_128) { |
| 473 | TEST_REQUIRES_X86_AVX512F; |
| 474 | for (size_t elements = 256; elements < 1280; elements += 128) { |
| 475 | RAddExtExpMicrokernelTester() |
| 476 | .elements(elements) |
| 477 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128); |
| 478 | } |
| 479 | } |
| 480 | |
| 481 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128, elements_lt_128) { |
| 482 | TEST_REQUIRES_X86_AVX512F; |
| 483 | for (size_t elements = 1; elements < 128; elements++) { |
| 484 | RAddExtExpMicrokernelTester() |
| 485 | .elements(elements) |
| 486 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128); |
| 487 | } |
| 488 | } |
| 489 | |
| 490 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128, elements_gt_128) { |
| 491 | TEST_REQUIRES_X86_AVX512F; |
| 492 | for (size_t elements = 129; elements < 256; elements++) { |
| 493 | RAddExtExpMicrokernelTester() |
| 494 | .elements(elements) |
| 495 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128); |
| 496 | } |
| 497 | } |
| 498 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 499 | |
| 500 | |
| 501 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 502 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC2, elements_eq_128) { |
| 503 | TEST_REQUIRES_X86_AVX512F; |
| 504 | RAddExtExpMicrokernelTester() |
| 505 | .elements(128) |
| 506 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc2); |
| 507 | } |
| 508 | |
| 509 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC2, elements_div_128) { |
| 510 | TEST_REQUIRES_X86_AVX512F; |
| 511 | for (size_t elements = 256; elements < 1280; elements += 128) { |
| 512 | RAddExtExpMicrokernelTester() |
| 513 | .elements(elements) |
| 514 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc2); |
| 515 | } |
| 516 | } |
| 517 | |
| 518 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC2, elements_lt_128) { |
| 519 | TEST_REQUIRES_X86_AVX512F; |
| 520 | for (size_t elements = 1; elements < 128; elements++) { |
| 521 | RAddExtExpMicrokernelTester() |
| 522 | .elements(elements) |
| 523 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc2); |
| 524 | } |
| 525 | } |
| 526 | |
| 527 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC2, elements_gt_128) { |
| 528 | TEST_REQUIRES_X86_AVX512F; |
| 529 | for (size_t elements = 129; elements < 256; elements++) { |
| 530 | RAddExtExpMicrokernelTester() |
| 531 | .elements(elements) |
| 532 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc2); |
| 533 | } |
| 534 | } |
| 535 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 536 | |
| 537 | |
| 538 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 539 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC4, elements_eq_128) { |
| 540 | TEST_REQUIRES_X86_AVX512F; |
| 541 | RAddExtExpMicrokernelTester() |
| 542 | .elements(128) |
| 543 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc4); |
| 544 | } |
| 545 | |
| 546 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC4, elements_div_128) { |
| 547 | TEST_REQUIRES_X86_AVX512F; |
| 548 | for (size_t elements = 256; elements < 1280; elements += 128) { |
| 549 | RAddExtExpMicrokernelTester() |
| 550 | .elements(elements) |
| 551 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc4); |
| 552 | } |
| 553 | } |
| 554 | |
| 555 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC4, elements_lt_128) { |
| 556 | TEST_REQUIRES_X86_AVX512F; |
| 557 | for (size_t elements = 1; elements < 128; elements++) { |
| 558 | RAddExtExpMicrokernelTester() |
| 559 | .elements(elements) |
| 560 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc4); |
| 561 | } |
| 562 | } |
| 563 | |
| 564 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X128_ACC4, elements_gt_128) { |
| 565 | TEST_REQUIRES_X86_AVX512F; |
| 566 | for (size_t elements = 129; elements < 256; elements++) { |
| 567 | RAddExtExpMicrokernelTester() |
| 568 | .elements(elements) |
| 569 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x128_acc4); |
| 570 | } |
| 571 | } |
| 572 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 573 | |
| 574 | |
| 575 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 576 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144, elements_eq_144) { |
| 577 | TEST_REQUIRES_X86_AVX512F; |
| 578 | RAddExtExpMicrokernelTester() |
| 579 | .elements(144) |
| 580 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144); |
| 581 | } |
| 582 | |
| 583 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144, elements_div_144) { |
| 584 | TEST_REQUIRES_X86_AVX512F; |
| 585 | for (size_t elements = 288; elements < 1440; elements += 144) { |
| 586 | RAddExtExpMicrokernelTester() |
| 587 | .elements(elements) |
| 588 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144); |
| 589 | } |
| 590 | } |
| 591 | |
| 592 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144, elements_lt_144) { |
| 593 | TEST_REQUIRES_X86_AVX512F; |
| 594 | for (size_t elements = 1; elements < 144; elements++) { |
| 595 | RAddExtExpMicrokernelTester() |
| 596 | .elements(elements) |
| 597 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144); |
| 598 | } |
| 599 | } |
| 600 | |
| 601 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144, elements_gt_144) { |
| 602 | TEST_REQUIRES_X86_AVX512F; |
| 603 | for (size_t elements = 145; elements < 288; elements++) { |
| 604 | RAddExtExpMicrokernelTester() |
| 605 | .elements(elements) |
| 606 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144); |
| 607 | } |
| 608 | } |
| 609 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 610 | |
| 611 | |
| 612 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 613 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144_ACC3, elements_eq_144) { |
| 614 | TEST_REQUIRES_X86_AVX512F; |
| 615 | RAddExtExpMicrokernelTester() |
| 616 | .elements(144) |
| 617 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144_acc3); |
| 618 | } |
| 619 | |
| 620 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144_ACC3, elements_div_144) { |
| 621 | TEST_REQUIRES_X86_AVX512F; |
| 622 | for (size_t elements = 288; elements < 1440; elements += 144) { |
| 623 | RAddExtExpMicrokernelTester() |
| 624 | .elements(elements) |
| 625 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144_acc3); |
| 626 | } |
| 627 | } |
| 628 | |
| 629 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144_ACC3, elements_lt_144) { |
| 630 | TEST_REQUIRES_X86_AVX512F; |
| 631 | for (size_t elements = 1; elements < 144; elements++) { |
| 632 | RAddExtExpMicrokernelTester() |
| 633 | .elements(elements) |
| 634 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144_acc3); |
| 635 | } |
| 636 | } |
| 637 | |
| 638 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X144_ACC3, elements_gt_144) { |
| 639 | TEST_REQUIRES_X86_AVX512F; |
| 640 | for (size_t elements = 145; elements < 288; elements++) { |
| 641 | RAddExtExpMicrokernelTester() |
| 642 | .elements(elements) |
| 643 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x144_acc3); |
| 644 | } |
| 645 | } |
| 646 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 647 | |
| 648 | |
| 649 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 650 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160, elements_eq_160) { |
| 651 | TEST_REQUIRES_X86_AVX512F; |
| 652 | RAddExtExpMicrokernelTester() |
| 653 | .elements(160) |
| 654 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160); |
| 655 | } |
| 656 | |
| 657 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160, elements_div_160) { |
| 658 | TEST_REQUIRES_X86_AVX512F; |
| 659 | for (size_t elements = 320; elements < 1600; elements += 160) { |
| 660 | RAddExtExpMicrokernelTester() |
| 661 | .elements(elements) |
| 662 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160); |
| 663 | } |
| 664 | } |
| 665 | |
| 666 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160, elements_lt_160) { |
| 667 | TEST_REQUIRES_X86_AVX512F; |
| 668 | for (size_t elements = 1; elements < 160; elements++) { |
| 669 | RAddExtExpMicrokernelTester() |
| 670 | .elements(elements) |
| 671 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160); |
| 672 | } |
| 673 | } |
| 674 | |
| 675 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160, elements_gt_160) { |
| 676 | TEST_REQUIRES_X86_AVX512F; |
| 677 | for (size_t elements = 161; elements < 320; elements++) { |
| 678 | RAddExtExpMicrokernelTester() |
| 679 | .elements(elements) |
| 680 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160); |
| 681 | } |
| 682 | } |
| 683 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 684 | |
| 685 | |
| 686 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 687 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC2, elements_eq_160) { |
| 688 | TEST_REQUIRES_X86_AVX512F; |
| 689 | RAddExtExpMicrokernelTester() |
| 690 | .elements(160) |
| 691 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc2); |
| 692 | } |
| 693 | |
| 694 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC2, elements_div_160) { |
| 695 | TEST_REQUIRES_X86_AVX512F; |
| 696 | for (size_t elements = 320; elements < 1600; elements += 160) { |
| 697 | RAddExtExpMicrokernelTester() |
| 698 | .elements(elements) |
| 699 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc2); |
| 700 | } |
| 701 | } |
| 702 | |
| 703 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC2, elements_lt_160) { |
| 704 | TEST_REQUIRES_X86_AVX512F; |
| 705 | for (size_t elements = 1; elements < 160; elements++) { |
| 706 | RAddExtExpMicrokernelTester() |
| 707 | .elements(elements) |
| 708 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc2); |
| 709 | } |
| 710 | } |
| 711 | |
| 712 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC2, elements_gt_160) { |
| 713 | TEST_REQUIRES_X86_AVX512F; |
| 714 | for (size_t elements = 161; elements < 320; elements++) { |
| 715 | RAddExtExpMicrokernelTester() |
| 716 | .elements(elements) |
| 717 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc2); |
| 718 | } |
| 719 | } |
| 720 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 721 | |
| 722 | |
| 723 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 724 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC5, elements_eq_160) { |
| 725 | TEST_REQUIRES_X86_AVX512F; |
| 726 | RAddExtExpMicrokernelTester() |
| 727 | .elements(160) |
| 728 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc5); |
| 729 | } |
| 730 | |
| 731 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC5, elements_div_160) { |
| 732 | TEST_REQUIRES_X86_AVX512F; |
| 733 | for (size_t elements = 320; elements < 1600; elements += 160) { |
| 734 | RAddExtExpMicrokernelTester() |
| 735 | .elements(elements) |
| 736 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc5); |
| 737 | } |
| 738 | } |
| 739 | |
| 740 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC5, elements_lt_160) { |
| 741 | TEST_REQUIRES_X86_AVX512F; |
| 742 | for (size_t elements = 1; elements < 160; elements++) { |
| 743 | RAddExtExpMicrokernelTester() |
| 744 | .elements(elements) |
| 745 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc5); |
| 746 | } |
| 747 | } |
| 748 | |
| 749 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X160_ACC5, elements_gt_160) { |
| 750 | TEST_REQUIRES_X86_AVX512F; |
| 751 | for (size_t elements = 161; elements < 320; elements++) { |
| 752 | RAddExtExpMicrokernelTester() |
| 753 | .elements(elements) |
| 754 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x160_acc5); |
| 755 | } |
| 756 | } |
| 757 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 758 | |
| 759 | |
| 760 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 761 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192, elements_eq_192) { |
| 762 | TEST_REQUIRES_X86_AVX512F; |
| 763 | RAddExtExpMicrokernelTester() |
| 764 | .elements(192) |
| 765 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192); |
| 766 | } |
| 767 | |
| 768 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192, elements_div_192) { |
| 769 | TEST_REQUIRES_X86_AVX512F; |
| 770 | for (size_t elements = 384; elements < 1920; elements += 192) { |
| 771 | RAddExtExpMicrokernelTester() |
| 772 | .elements(elements) |
| 773 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192); |
| 774 | } |
| 775 | } |
| 776 | |
| 777 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192, elements_lt_192) { |
| 778 | TEST_REQUIRES_X86_AVX512F; |
| 779 | for (size_t elements = 1; elements < 192; elements++) { |
| 780 | RAddExtExpMicrokernelTester() |
| 781 | .elements(elements) |
| 782 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192); |
| 783 | } |
| 784 | } |
| 785 | |
| 786 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192, elements_gt_192) { |
| 787 | TEST_REQUIRES_X86_AVX512F; |
| 788 | for (size_t elements = 193; elements < 384; elements++) { |
| 789 | RAddExtExpMicrokernelTester() |
| 790 | .elements(elements) |
| 791 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192); |
| 792 | } |
| 793 | } |
| 794 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 795 | |
| 796 | |
| 797 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 798 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC2, elements_eq_192) { |
| 799 | TEST_REQUIRES_X86_AVX512F; |
| 800 | RAddExtExpMicrokernelTester() |
| 801 | .elements(192) |
| 802 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc2); |
| 803 | } |
| 804 | |
| 805 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC2, elements_div_192) { |
| 806 | TEST_REQUIRES_X86_AVX512F; |
| 807 | for (size_t elements = 384; elements < 1920; elements += 192) { |
| 808 | RAddExtExpMicrokernelTester() |
| 809 | .elements(elements) |
| 810 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc2); |
| 811 | } |
| 812 | } |
| 813 | |
| 814 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC2, elements_lt_192) { |
| 815 | TEST_REQUIRES_X86_AVX512F; |
| 816 | for (size_t elements = 1; elements < 192; elements++) { |
| 817 | RAddExtExpMicrokernelTester() |
| 818 | .elements(elements) |
| 819 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc2); |
| 820 | } |
| 821 | } |
| 822 | |
| 823 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC2, elements_gt_192) { |
| 824 | TEST_REQUIRES_X86_AVX512F; |
| 825 | for (size_t elements = 193; elements < 384; elements++) { |
| 826 | RAddExtExpMicrokernelTester() |
| 827 | .elements(elements) |
| 828 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc2); |
| 829 | } |
| 830 | } |
| 831 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 832 | |
| 833 | |
| 834 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 835 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC3, elements_eq_192) { |
| 836 | TEST_REQUIRES_X86_AVX512F; |
| 837 | RAddExtExpMicrokernelTester() |
| 838 | .elements(192) |
| 839 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc3); |
| 840 | } |
| 841 | |
| 842 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC3, elements_div_192) { |
| 843 | TEST_REQUIRES_X86_AVX512F; |
| 844 | for (size_t elements = 384; elements < 1920; elements += 192) { |
| 845 | RAddExtExpMicrokernelTester() |
| 846 | .elements(elements) |
| 847 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc3); |
| 848 | } |
| 849 | } |
| 850 | |
| 851 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC3, elements_lt_192) { |
| 852 | TEST_REQUIRES_X86_AVX512F; |
| 853 | for (size_t elements = 1; elements < 192; elements++) { |
| 854 | RAddExtExpMicrokernelTester() |
| 855 | .elements(elements) |
| 856 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc3); |
| 857 | } |
| 858 | } |
| 859 | |
| 860 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC3, elements_gt_192) { |
| 861 | TEST_REQUIRES_X86_AVX512F; |
| 862 | for (size_t elements = 193; elements < 384; elements++) { |
| 863 | RAddExtExpMicrokernelTester() |
| 864 | .elements(elements) |
| 865 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc3); |
| 866 | } |
| 867 | } |
| 868 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 869 | |
| 870 | |
| 871 | #if XNN_ARCH_X86 || XNN_ARCH_X86_64 |
| 872 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC6, elements_eq_192) { |
| 873 | TEST_REQUIRES_X86_AVX512F; |
| 874 | RAddExtExpMicrokernelTester() |
| 875 | .elements(192) |
| 876 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc6); |
| 877 | } |
| 878 | |
| 879 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC6, elements_div_192) { |
| 880 | TEST_REQUIRES_X86_AVX512F; |
| 881 | for (size_t elements = 384; elements < 1920; elements += 192) { |
| 882 | RAddExtExpMicrokernelTester() |
| 883 | .elements(elements) |
| 884 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc6); |
| 885 | } |
| 886 | } |
| 887 | |
| 888 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC6, elements_lt_192) { |
| 889 | TEST_REQUIRES_X86_AVX512F; |
| 890 | for (size_t elements = 1; elements < 192; elements++) { |
| 891 | RAddExtExpMicrokernelTester() |
| 892 | .elements(elements) |
| 893 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc6); |
| 894 | } |
| 895 | } |
| 896 | |
| 897 | TEST(F32_RADDEXTEXP__AVX512F_P5_SCALEF_X192_ACC6, elements_gt_192) { |
| 898 | TEST_REQUIRES_X86_AVX512F; |
| 899 | for (size_t elements = 193; elements < 384; elements++) { |
| 900 | RAddExtExpMicrokernelTester() |
| 901 | .elements(elements) |
| 902 | .Test(xnn_f32_raddextexp_ukernel__avx512f_p5_scalef_x192_acc6); |
Marat Dukhan | 6f8d4d3 | 2019-10-25 17:07:09 -0700 | [diff] [blame] | 903 | } |
| 904 | } |
| 905 | #endif // XNN_ARCH_X86 || XNN_ARCH_X86_64 |