| 1 | // -------------------------------------------------------------------------- |
| 2 | // |
| 3 | // Copyright |
| 4 | // Markus Wittmann, 2016-2017 |
| 5 | // RRZE, University of Erlangen-Nuremberg, Germany |
| 6 | // markus.wittmann -at- fau.de or hpc -at- rrze.fau.de |
| 7 | // |
| 8 | // Viktor Haag, 2016 |
| 9 | // LSS, University of Erlangen-Nuremberg, Germany |
| 10 | // |
| 11 | // This file is part of the Lattice Boltzmann Benchmark Kernels (LbmBenchKernels). |
| 12 | // |
| 13 | // LbmBenchKernels is free software: you can redistribute it and/or modify |
| 14 | // it under the terms of the GNU General Public License as published by |
| 15 | // the Free Software Foundation, either version 3 of the License, or |
| 16 | // (at your option) any later version. |
| 17 | // |
| 18 | // LbmBenchKernels is distributed in the hope that it will be useful, |
| 19 | // but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 20 | // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 21 | // GNU General Public License for more details. |
| 22 | // |
| 23 | // You should have received a copy of the GNU General Public License |
| 24 | // along with LbmBenchKernels. If not, see <http://www.gnu.org/licenses/>. |
| 25 | // |
| 26 | // -------------------------------------------------------------------------- |
| 27 | #include "BenchKernelD3Q19ListAaPvCommon.h" |
| 28 | |
| 29 | #include "Memory.h" |
| 30 | #include "Vtk.h" |
| 31 | #include "Vector.h" |
| 32 | #include "LikwidIf.h" |
| 33 | |
| 34 | #include <inttypes.h> |
| 35 | #include <math.h> |
| 36 | |
| 37 | #ifdef _OPENMP |
| 38 | #include <omp.h> |
| 39 | #endif |
| 40 | |
| 41 | |
| 42 | static void KernelEven(LatticeDesc * ld, KernelData * kernelData, CaseData * cd); |
| 43 | static void KernelOdd( LatticeDesc * ld, KernelData * kernelData, CaseData * cd); |
| 44 | |
| 45 | void FNAME(D3Q19ListAaPvKernel)(LatticeDesc * ld, KernelData * kernelData, CaseData * cd) |
| 46 | { |
| 47 | |
| 48 | Assert(ld != NULL); |
| 49 | Assert(kernelData != NULL); |
| 50 | Assert(cd != NULL); |
| 51 | |
| 52 | Assert(cd->Omega > F(0.0)); |
| 53 | Assert(cd->Omega < F(2.0));; |
| 54 | |
| 55 | #if defined(VTK_OUTPUT) || defined(STATISTICS) || defined(VERIFICATION) |
| 56 | KernelData * kd = (KernelData *)kernelData; |
| 57 | KernelDataList * kdl = KDL(kernelData); |
| 58 | #endif |
| 59 | |
| 60 | int maxIterations = cd->MaxIterations; |
| 61 | |
| 62 | int nThreads = 1; |
| 63 | #ifdef _OPENMP |
| 64 | nThreads = omp_get_max_threads(); |
| 65 | #endif |
| 66 | |
| 67 | #ifdef VTK_OUTPUT |
| 68 | if (cd->VtkOutput) { |
| 69 | kd->PdfsActive = kd->Pdfs[0]; |
| 70 | VtkWrite(ld, kd, cd, -1); |
| 71 | } |
| 72 | #endif |
| 73 | |
| 74 | #ifdef STATISTICS |
| 75 | kd->PdfsActive = kd->Pdfs[0]; |
| 76 | KernelStatistics(kd, ld, cd, 0); |
| 77 | #endif |
| 78 | |
| 79 | X_KERNEL_START(kernelData); |
| 80 | |
| 81 | // TODO: outer openmp parallel |
| 82 | |
| 83 | for(int iter = 0; iter < maxIterations; iter += 2) { |
| 84 | |
| 85 | // --------------------------------------------------- |
| 86 | // even time step |
| 87 | // --------------------------------------------------- |
| 88 | |
| 89 | X_LIKWID_START("list-aa-pv-even"); |
| 90 | |
| 91 | #ifdef _OPENMP |
| 92 | #pragma omp parallel default(none) shared(ld, kernelData, cd) |
| 93 | #endif |
| 94 | { |
| 95 | KernelEven(ld, kernelData, cd); |
| 96 | } |
| 97 | |
| 98 | X_LIKWID_STOP("list-aa-pv-even"); |
| 99 | |
| 100 | #ifdef VERIFICATION |
| 101 | kdl->Iteration = iter; |
| 102 | kd->PdfsActive = kd->Pdfs[0]; |
| 103 | KernelAddBodyForce(kd, ld, cd); |
| 104 | #endif |
| 105 | |
| 106 | // --------------------------------------------------- |
| 107 | // odd time step |
| 108 | // --------------------------------------------------- |
| 109 | |
| 110 | X_LIKWID_START("list-aa-pv-odd"); |
| 111 | |
| 112 | #ifdef _OPENMP |
| 113 | #pragma omp parallel default(none) shared(ld, kernelData, cd) |
| 114 | #endif |
| 115 | { |
| 116 | KernelOdd(ld, kernelData, cd); |
| 117 | } |
| 118 | |
| 119 | X_LIKWID_STOP("list-aa-pv-odd"); |
| 120 | |
| 121 | |
| 122 | #ifdef VERIFICATION |
| 123 | kdl->Iteration = iter + 1; |
| 124 | kd->PdfsActive = kd->Pdfs[0]; |
| 125 | KernelAddBodyForce(kd, ld, cd); |
| 126 | #endif |
| 127 | |
| 128 | #ifdef VTK_OUTPUT |
| 129 | if (cd->VtkOutput && (iter % cd->VtkModulus) == 0) { |
| 130 | kdl->Iteration = iter + 1; |
| 131 | kd->PdfsActive = kd->Pdfs[0]; |
| 132 | VtkWrite(ld, kd, cd, iter); |
| 133 | } |
| 134 | #endif |
| 135 | |
| 136 | #ifdef STATISTICS |
| 137 | kdl->Iteration = iter + 1; |
| 138 | kd->PdfsActive = kd->Pdfs[0]; |
| 139 | KernelStatistics(kd, ld, cd, iter); |
| 140 | #endif |
| 141 | |
| 142 | } // for (int iter = 0; ... |
| 143 | |
| 144 | X_KERNEL_END(kernelData); |
| 145 | |
| 146 | #ifdef VTK_OUTPUT |
| 147 | if (cd->VtkOutput) { |
| 148 | kd->PdfsActive = kd->Pdfs[0]; |
| 149 | VtkWrite(ld, kd, cd, maxIterations); |
| 150 | } |
| 151 | #endif |
| 152 | |
| 153 | #ifdef STATISTICS |
| 154 | kd->PdfsActive = kd->Pdfs[0]; |
| 155 | KernelStatistics(kd, ld, cd, maxIterations); |
| 156 | #endif |
| 157 | |
| 158 | return; |
| 159 | } |
| 160 | |
| 161 | static void KernelEven(LatticeDesc * ld, KernelData * kernelData, CaseData * cd) |
| 162 | { |
| 163 | Assert(ld != NULL); |
| 164 | Assert(kernelData != NULL); |
| 165 | Assert(cd != NULL); |
| 166 | |
| 167 | Assert(cd->Omega > F(0.0)); |
| 168 | Assert(cd->Omega < F(2.0)); |
| 169 | |
| 170 | KernelData * kd = (KernelData *)kernelData; |
| 171 | KernelDataList * kdl = KDL(kernelData); |
| 172 | KernelDataListRia * kdlr = KDLR(kernelData); |
| 173 | |
| 174 | PdfT omega = cd->Omega; |
| 175 | PdfT omegaEven = omega; |
| 176 | |
| 177 | PdfT magicParam = F(1.0) / F(12.0); |
| 178 | PdfT omegaOdd = F(1.0) / (F(0.5) + magicParam / (F(1.0) / omega - F(0.5))); |
| 179 | |
| 180 | PdfT evenPart = F(0.0); |
| 181 | PdfT oddPart = F(0.0); |
| 182 | PdfT dir_indep_trm = F(0.0); |
| 183 | |
| 184 | const PdfT w_0 = F(1.0) / F( 3.0); |
| 185 | const PdfT w_1 = F(1.0) / F(18.0); |
| 186 | const PdfT w_2 = F(1.0) / F(36.0); |
| 187 | |
| 188 | const PdfT w_1_x3 = w_1 * F(3.0); const PdfT w_1_nine_half = w_1 * F(9.0) / F(2.0); PdfT w_1_indep = F(0.0); |
| 189 | const PdfT w_2_x3 = w_2 * F(3.0); const PdfT w_2_nine_half = w_2 * F(9.0) / F(2.0); PdfT w_2_indep = F(0.0); |
| 190 | |
| 191 | PdfT ui; |
| 192 | |
| 193 | PdfT ux, uy, uz; |
| 194 | PdfT dens; |
| 195 | |
| 196 | |
| 197 | VPDFT VONE_HALF = VSET(F(0.5)); |
| 198 | VPDFT VTHREE_HALF = VSET(F(3.0) / F(2.0)); |
| 199 | |
| 200 | VPDFT vw_1_indep, vw_2_indep; |
| 201 | VPDFT vw_0 = VSET(w_0); |
| 202 | VPDFT vw_1 = VSET(w_1); |
| 203 | VPDFT vw_2 = VSET(w_2); |
| 204 | |
| 205 | VPDFT vw_1_x3 = VSET(w_1_x3); |
| 206 | VPDFT vw_2_x3 = VSET(w_2_x3); |
| 207 | VPDFT vw_1_nine_half = VSET(w_1_nine_half); |
| 208 | VPDFT vw_2_nine_half = VSET(w_2_nine_half); |
| 209 | |
| 210 | VPDFT vui, vux, vuy, vuz, vdens; |
| 211 | |
| 212 | VPDFT vevenPart, voddPart, vdir_indep_trm; |
| 213 | |
| 214 | VPDFT vomegaEven = VSET(omegaEven); |
| 215 | VPDFT vomegaOdd = VSET(omegaOdd); |
| 216 | |
| 217 | // Declare pdf_N, pdf_E, pdf_S, pdf_W, ... |
| 218 | #define X(name, idx, idxinv, x, y, z) PdfT JOIN(pdf_,name); VPDFT JOIN(vpdf_,name); |
| 219 | D3Q19_LIST |
| 220 | #undef X |
| 221 | |
| 222 | PdfT * src = kd->Pdfs[0]; |
| 223 | |
| 224 | int nCells = kdl->nCells; |
| 225 | |
| 226 | int threadId = 0; |
| 227 | #ifdef _OPENMP |
| 228 | threadId = omp_get_thread_num(); |
| 229 | #endif |
| 230 | |
| 231 | int * threadIndices = kdlr->FluidNodeThreadIndices; |
| 232 | |
| 233 | int nFluidThread = threadIndices[threadId + 1] - threadIndices[threadId]; |
| 234 | int nFluidVec = nFluidThread - (nFluidThread % VSIZE); |
| 235 | |
| 236 | int indexStartVec = threadIndices[threadId]; |
| 237 | int indexStopVec = threadIndices[threadId] + nFluidVec; |
| 238 | int indexStop = threadIndices[threadId] + nFluidThread; |
| 239 | |
| 240 | #define I(index, dir) P_INDEX_3((nCells), (index), (dir)) |
| 241 | |
| 242 | for (int index = indexStartVec; index < indexStopVec; index += VSIZE) { // LOOP list-aa-pv-even-vec |
| 243 | |
| 244 | |
| 245 | #define X(name, idx, idxinv, _x, _y, _z) JOIN(vpdf_,name) = VLDU(&src[I(index, idx)]); |
| 246 | D3Q19_LIST |
| 247 | #undef X |
| 248 | |
| 249 | |
| 250 | //vux = vpdf_E + vpdf_NE + vpdf_SE + vpdf_TE + vpdf_BE - |
| 251 | // vpdf_W - vpdf_NW - vpdf_SW - vpdf_TW - vpdf_BW; |
| 252 | vux = VSUB(VSUB(VSUB(VSUB(VSUB(VADD(VADD(vpdf_E,VADD(vpdf_NE,vpdf_SE)),VADD(vpdf_TE,vpdf_BE)),vpdf_W),vpdf_NW),vpdf_SW),vpdf_TW),vpdf_BW); |
| 253 | //vuy = vpdf_N + vpdf_NE + vpdf_NW + vpdf_TN + vpdf_BN - |
| 254 | // vpdf_S - vpdf_SE - vpdf_SW - vpdf_TS - vpdf_BS; |
| 255 | vuy = VSUB(VSUB(VSUB(VSUB(VSUB(VADD(VADD(vpdf_N,VADD(vpdf_NE,vpdf_NW)),VADD(vpdf_TN,vpdf_BN)),vpdf_S),vpdf_SE),vpdf_SW),vpdf_TS),vpdf_BS); |
| 256 | //vuz = vpdf_T + vpdf_TE + vpdf_TW + vpdf_TN + vpdf_TS - |
| 257 | // vpdf_B - vpdf_BE - vpdf_BW - vpdf_BN - vpdf_BS; |
| 258 | vuz = VSUB(VSUB(VSUB(VSUB(VSUB(VADD(VADD(vpdf_T,VADD(vpdf_TE,vpdf_TW)),VADD(vpdf_TN,vpdf_TS)),vpdf_B),vpdf_BE),vpdf_BW),vpdf_BN),vpdf_BS); |
| 259 | |
| 260 | //vdens = vpdf_C + |
| 261 | // vpdf_N + vpdf_E + vpdf_S + vpdf_W + |
| 262 | // vpdf_NE + vpdf_SE + vpdf_SW + vpdf_NW + |
| 263 | // vpdf_T + vpdf_TN + vpdf_TE + vpdf_TS + vpdf_TW + |
| 264 | // vpdf_B + vpdf_BN + vpdf_BE + vpdf_BS + vpdf_BW; |
| 265 | vdens = VADD(VADD(VADD(VADD(VADD(VADD(VADD(VADD(VADD(vpdf_C,VADD(vpdf_N,vpdf_E)),VADD(vpdf_S,vpdf_W)),VADD(vpdf_NE,vpdf_SE)),VADD(vpdf_SW,vpdf_NW)),VADD(vpdf_T,vpdf_TN)),VADD(vpdf_TE,vpdf_TS)),VADD(vpdf_TW,vpdf_B)),VADD(vpdf_BN,vpdf_BE)),VADD(vpdf_BS,vpdf_BW)); |
| 266 | |
| 267 | //vdir_indep_trm = vdens - (vux * vux + vuy * vuy + vuz * vuz) * VTHREE_HALF; |
| 268 | vdir_indep_trm = VSUB(vdens,VMUL(VADD(VADD(VMUL(vux,vux),VMUL(vuy,vuy)),VMUL(vuz,vuz)),VTHREE_HALF)); |
| 269 | |
| 270 | //src[I(index, D3Q19_C) ] =[UA] vpdf_C - vomegaEven * (vpdf_C - vw_0 * vdir_indep_trm); |
| 271 | VSTU(&src[I(index, D3Q19_C)],VSUB(vpdf_C,VMUL(vomegaEven,VSUB(vpdf_C,VMUL(vw_0,vdir_indep_trm))))); |
| 272 | |
| 273 | //vw_1_indep = vw_1 * vdir_indep_trm; |
| 274 | vw_1_indep = VMUL(vw_1,vdir_indep_trm); |
| 275 | |
| 276 | //vui = vuy; |
| 277 | vui = vuy; |
| 278 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_N + vpdf_S) - vui * vui * vw_1_nine_half - vw_1_indep); |
| 279 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_N,vpdf_S)),VMUL(vui,VMUL(vui,vw_1_nine_half))),vw_1_indep)); |
| 280 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_N - vpdf_S) - vui * vw_1_x3); |
| 281 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_N,vpdf_S)),VMUL(vui,vw_1_x3))); |
| 282 | //src[I(index, D3Q19_S)] =[UA] vpdf_N - vevenPart - voddPart; |
| 283 | VSTU(&src[I(index, D3Q19_S)],VSUB(VSUB(vpdf_N,vevenPart),voddPart)); |
| 284 | //src[I(index, D3Q19_N)] =[UA] vpdf_S - vevenPart + voddPart; |
| 285 | VSTU(&src[I(index, D3Q19_N)],VADD(VSUB(vpdf_S,vevenPart),voddPart)); |
| 286 | |
| 287 | //vui = vux; |
| 288 | vui = vux; |
| 289 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_E + vpdf_W) - vui * vui * vw_1_nine_half - vw_1_indep); |
| 290 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_E,vpdf_W)),VMUL(vui,VMUL(vui,vw_1_nine_half))),vw_1_indep)); |
| 291 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_E - vpdf_W) - vui * vw_1_x3 ); |
| 292 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_E,vpdf_W)),VMUL(vui,vw_1_x3))); |
| 293 | //src[I(index, D3Q19_W)] =[UA] vpdf_E - vevenPart - voddPart; |
| 294 | VSTU(&src[I(index, D3Q19_W)],VSUB(VSUB(vpdf_E,vevenPart),voddPart)); |
| 295 | //src[I(index, D3Q19_E)] =[UA] vpdf_W - vevenPart + voddPart; |
| 296 | VSTU(&src[I(index, D3Q19_E)],VADD(VSUB(vpdf_W,vevenPart),voddPart)); |
| 297 | |
| 298 | //vui = vuz; |
| 299 | vui = vuz; |
| 300 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_T + vpdf_B) - vui * vui * vw_1_nine_half - vw_1_indep); |
| 301 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_T,vpdf_B)),VMUL(vui,VMUL(vui,vw_1_nine_half))),vw_1_indep)); |
| 302 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_T - vpdf_B) - vui * vw_1_x3); |
| 303 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_T,vpdf_B)),VMUL(vui,vw_1_x3))); |
| 304 | //src[I(index, D3Q19_B)] =[UA] vpdf_T - vevenPart - voddPart; |
| 305 | VSTU(&src[I(index, D3Q19_B)],VSUB(VSUB(vpdf_T,vevenPart),voddPart)); |
| 306 | //src[I(index, D3Q19_T)] =[UA] vpdf_B - vevenPart + voddPart; |
| 307 | VSTU(&src[I(index, D3Q19_T)],VADD(VSUB(vpdf_B,vevenPart),voddPart)); |
| 308 | |
| 309 | //vw_2_indep = vw_2 * vdir_indep_trm; |
| 310 | vw_2_indep = VMUL(vw_2,vdir_indep_trm); |
| 311 | |
| 312 | //vui = vuy - vux; |
| 313 | vui = VSUB(vuy,vux); |
| 314 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_NW + vpdf_SE) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 315 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_NW,vpdf_SE)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 316 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_NW - vpdf_SE) - vui * vw_2_x3); |
| 317 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_NW,vpdf_SE)),VMUL(vui,vw_2_x3))); |
| 318 | //src[I(index, D3Q19_SE)] =[UA] vpdf_NW - vevenPart - voddPart; |
| 319 | VSTU(&src[I(index, D3Q19_SE)],VSUB(VSUB(vpdf_NW,vevenPart),voddPart)); |
| 320 | //src[I(index, D3Q19_NW)] =[UA] vpdf_SE - vevenPart + voddPart; |
| 321 | VSTU(&src[I(index, D3Q19_NW)],VADD(VSUB(vpdf_SE,vevenPart),voddPart)); |
| 322 | |
| 323 | //vui = vux + vuy; |
| 324 | vui = VADD(vux,vuy); |
| 325 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_NE + vpdf_SW) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 326 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_NE,vpdf_SW)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 327 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_NE - vpdf_SW) - vui * vw_2_x3); |
| 328 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_NE,vpdf_SW)),VMUL(vui,vw_2_x3))); |
| 329 | //src[I(index, D3Q19_SW)] =[UA] vpdf_NE - vevenPart - voddPart; |
| 330 | VSTU(&src[I(index, D3Q19_SW)],VSUB(VSUB(vpdf_NE,vevenPart),voddPart)); |
| 331 | //src[I(index, D3Q19_NE)] =[UA] vpdf_SW - vevenPart + voddPart; |
| 332 | VSTU(&src[I(index, D3Q19_NE)],VADD(VSUB(vpdf_SW,vevenPart),voddPart)); |
| 333 | |
| 334 | //vui = vuz - vux; |
| 335 | vui = VSUB(vuz,vux); |
| 336 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TW + vpdf_BE) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 337 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TW,vpdf_BE)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 338 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TW - vpdf_BE) - vui * vw_2_x3); |
| 339 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TW,vpdf_BE)),VMUL(vui,vw_2_x3))); |
| 340 | //src[I(index, D3Q19_BE)] =[UA] vpdf_TW - vevenPart - voddPart; |
| 341 | VSTU(&src[I(index, D3Q19_BE)],VSUB(VSUB(vpdf_TW,vevenPart),voddPart)); |
| 342 | //src[I(index, D3Q19_TW)] =[UA] vpdf_BE - vevenPart + voddPart; |
| 343 | VSTU(&src[I(index, D3Q19_TW)],VADD(VSUB(vpdf_BE,vevenPart),voddPart)); |
| 344 | |
| 345 | //vui = vux + vuz; |
| 346 | vui = VADD(vux,vuz); |
| 347 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TE + vpdf_BW) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 348 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TE,vpdf_BW)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 349 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TE - vpdf_BW) - vui * vw_2_x3); |
| 350 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TE,vpdf_BW)),VMUL(vui,vw_2_x3))); |
| 351 | //src[I(index, D3Q19_BW)] =[UA] vpdf_TE - vevenPart - voddPart; |
| 352 | VSTU(&src[I(index, D3Q19_BW)],VSUB(VSUB(vpdf_TE,vevenPart),voddPart)); |
| 353 | //src[I(index, D3Q19_TE)] =[UA] vpdf_BW - vevenPart + voddPart; |
| 354 | VSTU(&src[I(index, D3Q19_TE)],VADD(VSUB(vpdf_BW,vevenPart),voddPart)); |
| 355 | |
| 356 | //vui = vuz - vuy; |
| 357 | vui = VSUB(vuz,vuy); |
| 358 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TS + vpdf_BN) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 359 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TS,vpdf_BN)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 360 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TS - vpdf_BN) - vui * vw_2_x3); |
| 361 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TS,vpdf_BN)),VMUL(vui,vw_2_x3))); |
| 362 | //src[I(index, D3Q19_BN)] =[UA] vpdf_TS - vevenPart - voddPart; |
| 363 | VSTU(&src[I(index, D3Q19_BN)],VSUB(VSUB(vpdf_TS,vevenPart),voddPart)); |
| 364 | //src[I(index, D3Q19_TS)] =[UA] vpdf_BN - vevenPart + voddPart; |
| 365 | VSTU(&src[I(index, D3Q19_TS)],VADD(VSUB(vpdf_BN,vevenPart),voddPart)); |
| 366 | |
| 367 | //vui = vuy + vuz; |
| 368 | vui = VADD(vuy,vuz); |
| 369 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TN + vpdf_BS) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 370 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TN,vpdf_BS)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 371 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TN - vpdf_BS) - vui * vw_2_x3); |
| 372 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TN,vpdf_BS)),VMUL(vui,vw_2_x3))); |
| 373 | //src[I(index, D3Q19_BS)] =[UA] vpdf_TN - vevenPart - voddPart; |
| 374 | VSTU(&src[I(index, D3Q19_BS)],VSUB(VSUB(vpdf_TN,vevenPart),voddPart)); |
| 375 | //src[I(index, D3Q19_TN)] =[UA] vpdf_BS - vevenPart + voddPart; |
| 376 | VSTU(&src[I(index, D3Q19_TN)],VADD(VSUB(vpdf_BS,vevenPart),voddPart)); |
| 377 | |
| 378 | } // loop over fluid nodes |
| 379 | |
| 380 | for (int index = indexStopVec; index < indexStop; ++index) { // LOOP list-aa-pv-even-scalar |
| 381 | |
| 382 | #define X(name, idx, idxinv, _x, _y, _z) JOIN(pdf_,name) = src[I(index, idx)]; |
| 383 | D3Q19_LIST |
| 384 | #undef X |
| 385 | |
| 386 | |
| 387 | ux = pdf_E + pdf_NE + pdf_SE + pdf_TE + pdf_BE - |
| 388 | pdf_W - pdf_NW - pdf_SW - pdf_TW - pdf_BW; |
| 389 | uy = pdf_N + pdf_NE + pdf_NW + pdf_TN + pdf_BN - |
| 390 | pdf_S - pdf_SE - pdf_SW - pdf_TS - pdf_BS; |
| 391 | uz = pdf_T + pdf_TE + pdf_TW + pdf_TN + pdf_TS - |
| 392 | pdf_B - pdf_BE - pdf_BW - pdf_BN - pdf_BS; |
| 393 | |
| 394 | dens = pdf_C + |
| 395 | pdf_N + pdf_E + pdf_S + pdf_W + |
| 396 | pdf_NE + pdf_SE + pdf_SW + pdf_NW + |
| 397 | pdf_T + pdf_TN + pdf_TE + pdf_TS + pdf_TW + |
| 398 | pdf_B + pdf_BN + pdf_BE + pdf_BS + pdf_BW; |
| 399 | |
| 400 | dir_indep_trm = dens - (ux * ux + uy * uy + uz * uz) * F(3.0) / F(2.0); |
| 401 | |
| 402 | // direction: w_0 |
| 403 | src[I(index, D3Q19_C) ] = pdf_C - omegaEven*(pdf_C - w_0*dir_indep_trm); |
| 404 | |
| 405 | // direction: w_1 |
| 406 | w_1_indep = w_1*dir_indep_trm; |
| 407 | |
| 408 | ui = uy; |
| 409 | evenPart = omegaEven*( F(0.5)*(pdf_N + pdf_S) - ui*ui*w_1_nine_half - w_1_indep ); |
| 410 | oddPart = omegaOdd*(F(0.5)*(pdf_N - pdf_S) - ui*w_1_x3 ); |
| 411 | src[I(index, D3Q19_S)] = pdf_N - evenPart - oddPart; |
| 412 | src[I(index, D3Q19_N)] = pdf_S - evenPart + oddPart; |
| 413 | |
| 414 | ui = ux; |
| 415 | evenPart = omegaEven*( F(0.5)*(pdf_E + pdf_W) - ui*ui*w_1_nine_half - w_1_indep ); |
| 416 | oddPart = omegaOdd*(F(0.5)*(pdf_E - pdf_W) - ui*w_1_x3 ); |
| 417 | src[I(index, D3Q19_W)] = pdf_E - evenPart - oddPart; |
| 418 | src[I(index, D3Q19_E)] = pdf_W - evenPart + oddPart; |
| 419 | |
| 420 | ui = uz; |
| 421 | evenPart = omegaEven*( F(0.5)*(pdf_T + pdf_B) - ui*ui*w_1_nine_half - w_1_indep ); |
| 422 | oddPart = omegaOdd*(F(0.5)*(pdf_T - pdf_B) - ui*w_1_x3 ); |
| 423 | src[I(index, D3Q19_B)] = pdf_T - evenPart - oddPart; |
| 424 | src[I(index, D3Q19_T)] = pdf_B - evenPart + oddPart; |
| 425 | |
| 426 | // direction: w_2 |
| 427 | w_2_indep = w_2*dir_indep_trm; |
| 428 | |
| 429 | ui = -ux + uy; |
| 430 | evenPart = omegaEven*( F(0.5)*(pdf_NW + pdf_SE) - ui*ui*w_2_nine_half - w_2_indep ); |
| 431 | oddPart = omegaOdd*(F(0.5)*(pdf_NW - pdf_SE) - ui*w_2_x3 ); |
| 432 | src[I(index, D3Q19_SE)] = pdf_NW - evenPart - oddPart; |
| 433 | src[I(index, D3Q19_NW)] = pdf_SE - evenPart + oddPart; |
| 434 | |
| 435 | ui = ux + uy; |
| 436 | evenPart = omegaEven*( F(0.5)*(pdf_NE + pdf_SW) - ui*ui*w_2_nine_half - w_2_indep ); |
| 437 | oddPart = omegaOdd*(F(0.5)*(pdf_NE - pdf_SW) - ui*w_2_x3 ); |
| 438 | src[I(index, D3Q19_SW)] = pdf_NE - evenPart - oddPart; |
| 439 | src[I(index, D3Q19_NE)] = pdf_SW - evenPart + oddPart; |
| 440 | |
| 441 | ui = -ux + uz; |
| 442 | evenPart = omegaEven*( F(0.5)*(pdf_TW + pdf_BE) - ui*ui*w_2_nine_half - w_2_indep ); |
| 443 | oddPart = omegaOdd*(F(0.5)*(pdf_TW - pdf_BE) - ui*w_2_x3 ); |
| 444 | src[I(index, D3Q19_BE)] = pdf_TW - evenPart - oddPart; |
| 445 | src[I(index, D3Q19_TW)] = pdf_BE - evenPart + oddPart; |
| 446 | |
| 447 | ui = ux + uz; |
| 448 | evenPart = omegaEven*( F(0.5)*(pdf_TE + pdf_BW) - ui*ui*w_2_nine_half - w_2_indep ); |
| 449 | oddPart = omegaOdd*(F(0.5)*(pdf_TE - pdf_BW) - ui*w_2_x3 ); |
| 450 | src[I(index, D3Q19_BW)] = pdf_TE - evenPart - oddPart; |
| 451 | src[I(index, D3Q19_TE)] = pdf_BW - evenPart + oddPart; |
| 452 | |
| 453 | ui = -uy + uz; |
| 454 | evenPart = omegaEven*( F(0.5)*(pdf_TS + pdf_BN) - ui*ui*w_2_nine_half - w_2_indep ); |
| 455 | oddPart = omegaOdd*(F(0.5)*(pdf_TS - pdf_BN) - ui*w_2_x3 ); |
| 456 | src[I(index, D3Q19_BN)] = pdf_TS - evenPart - oddPart; |
| 457 | src[I(index, D3Q19_TS)] = pdf_BN - evenPart + oddPart; |
| 458 | |
| 459 | ui = uy + uz; |
| 460 | evenPart = omegaEven*( F(0.5)*(pdf_TN + pdf_BS) - ui*ui*w_2_nine_half - w_2_indep ); |
| 461 | oddPart = omegaOdd*(F(0.5)*(pdf_TN - pdf_BS) - ui*w_2_x3 ); |
| 462 | src[I(index, D3Q19_BS)] = pdf_TN - evenPart - oddPart; |
| 463 | src[I(index, D3Q19_TN)] = pdf_BS - evenPart + oddPart; |
| 464 | |
| 465 | } // loop over fluid nodes |
| 466 | |
| 467 | #undef I |
| 468 | |
| 469 | return; |
| 470 | } |
| 471 | |
| 472 | static void KernelOdd(LatticeDesc * ld, KernelData * kernelData, CaseData * cd) |
| 473 | { |
| 474 | |
| 475 | Assert(ld != NULL); |
| 476 | Assert(kernelData != NULL); |
| 477 | Assert(cd != NULL); |
| 478 | |
| 479 | Assert(cd->Omega > F(0.0)); |
| 480 | Assert(cd->Omega < F(2.0)); |
| 481 | |
| 482 | KernelData * kd = (KernelData *)kernelData; |
| 483 | KernelDataList * kdl = KDL(kernelData); |
| 484 | KernelDataListRia * kdlr = KDLR(kernelData); |
| 485 | |
| 486 | PdfT omega = cd->Omega; |
| 487 | PdfT omegaEven = omega; |
| 488 | |
| 489 | PdfT magicParam = F(1.0) / F(12.0); |
| 490 | PdfT omegaOdd = F(1.0) / (F(0.5) + magicParam / (F(1.0) / omega - F(0.5))); |
| 491 | |
| 492 | PdfT evenPart = F(0.0); |
| 493 | PdfT oddPart = F(0.0); |
| 494 | PdfT dir_indep_trm = F(0.0); |
| 495 | |
| 496 | const PdfT w_0 = F(1.0) / F( 3.0); |
| 497 | const PdfT w_1 = F(1.0) / F(18.0); |
| 498 | const PdfT w_2 = F(1.0) / F(36.0); |
| 499 | |
| 500 | const PdfT w_1_x3 = w_1 * F(3.0); const PdfT w_1_nine_half = w_1 * F(9.0) / F(2.0); PdfT w_1_indep = F(0.0); |
| 501 | const PdfT w_2_x3 = w_2 * F(3.0); const PdfT w_2_nine_half = w_2 * F(9.0) / F(2.0); PdfT w_2_indep = F(0.0); |
| 502 | |
| 503 | PdfT ui; |
| 504 | |
| 505 | PdfT ux, uy, uz; |
| 506 | PdfT dens; |
| 507 | |
| 508 | |
| 509 | VPDFT VONE_HALF = VSET(F(0.5)); |
| 510 | VPDFT VTHREE_HALF = VSET(F(3.0) / F(2.0)); |
| 511 | |
| 512 | VPDFT vw_1_indep, vw_2_indep; |
| 513 | VPDFT vw_0 = VSET(w_0); |
| 514 | VPDFT vw_1 = VSET(w_1); |
| 515 | VPDFT vw_2 = VSET(w_2); |
| 516 | |
| 517 | VPDFT vw_1_x3 = VSET(w_1_x3); |
| 518 | VPDFT vw_2_x3 = VSET(w_2_x3); |
| 519 | VPDFT vw_1_nine_half = VSET(w_1_nine_half); |
| 520 | VPDFT vw_2_nine_half = VSET(w_2_nine_half); |
| 521 | |
| 522 | VPDFT vui, vux, vuy, vuz, vdens; |
| 523 | |
| 524 | VPDFT vevenPart, voddPart, vdir_indep_trm; |
| 525 | |
| 526 | VPDFT vomegaEven = VSET(omegaEven); |
| 527 | VPDFT vomegaOdd = VSET(omegaOdd); |
| 528 | |
| 529 | |
| 530 | // Declare pdf_N, pdf_E, pdf_S, pdf_W, ... |
| 531 | #define X(name, idx, idxinv, x, y, z) PdfT JOIN(pdf_,name); VPDFT JOIN(vpdf_,name); |
| 532 | D3Q19_LIST |
| 533 | #undef X |
| 534 | |
| 535 | // Declare pointers to pdfs ppdf_N, ppdf_E, ppdf_S, ppdf_W, ... |
| 536 | #define X(name, idx, idxinv, x, y, z) PdfT * JOIN(ppdf_,name) = NULL; |
| 537 | D3Q19_LIST |
| 538 | #undef X |
| 539 | |
| 540 | uint32_t nConsecNodes = kdlr->nConsecNodes; |
| 541 | uint32_t * consecNodes = kdlr->ConsecNodes; |
| 542 | uint32_t consecIndex = 0; |
| 543 | uint32_t consecValue = 0; |
| 544 | |
| 545 | #ifndef DEBUG |
| 546 | UNUSED(nConsecNodes); |
| 547 | #endif |
| 548 | |
| 549 | PdfT * src = kd->Pdfs[0]; |
| 550 | |
| 551 | int nCells = kdl->nCells; |
| 552 | |
| 553 | uint32_t adjListIndex; |
| 554 | uint32_t * adjList = kdl->AdjList; |
| 555 | |
| 556 | int threadId = 0; |
| 557 | |
| 558 | #ifdef _OPENMP |
| 559 | threadId = omp_get_thread_num(); |
| 560 | #endif |
| 561 | |
| 562 | consecIndex = kdlr->ConsecThreadIndices[threadId]; |
| 563 | consecValue = 0; |
| 564 | |
| 565 | int * threadIndices = kdlr->FluidNodeThreadIndices; |
| 566 | |
| 567 | int nFluidThread = threadIndices[threadId + 1] - threadIndices[threadId]; |
| 568 | |
| 569 | int indexStart = threadIndices[threadId]; |
| 570 | int indexStop = threadIndices[threadId] + nFluidThread; |
| 571 | |
| 572 | #define I(index, dir) P_INDEX_3((nCells), (index), (dir)) |
| 573 | |
| 574 | #define ADJ_LIST(dir) adjList[adjListIndex + (dir)] |
| 575 | |
| 576 | int pointerOffset = 1; |
| 577 | |
| 578 | for (int index = indexStart; index < indexStop; index += 1) { |
| 579 | |
| 580 | if (consecValue > 0) { |
| 581 | --consecValue; |
| 582 | // Increment all pdf pointers by an offset. If the previous iteration was |
| 583 | // scalar, increment only by one. If the previous iteration was vectorized, |
| 584 | // increment by the vector width. These offsets are set in the corresponding |
| 585 | // if branches. |
| 586 | #define X(name, idx, idxinv, _x, _y, _z) JOIN(ppdf_,name) += pointerOffset; |
| 587 | D3Q19_LIST |
| 588 | #undef X |
| 589 | } |
| 590 | else { |
| 591 | Assert(consecIndex < nConsecNodes); |
| 592 | |
| 593 | consecValue = consecNodes[consecIndex] - 1; |
| 594 | // Load new pointers to PDFs of local cell: |
| 595 | |
| 596 | adjListIndex = index * N_D3Q19_IDX; |
| 597 | |
| 598 | #define X(name, idx, idxinv, _x, _y, _z) JOIN(ppdf_,name) = &(src[adjList[adjListIndex + idxinv]]); |
| 599 | D3Q19_LIST_WO_C |
| 600 | #undef X |
| 601 | |
| 602 | ppdf_C = &(src[P_INDEX_3(nCells, index, D3Q19_C)]); |
| 603 | ++consecIndex; |
| 604 | } |
| 605 | |
| 606 | #define X(name, idx, idxinv, _x, _y, _z) JOIN(pdf_,name) = *JOIN(ppdf_,name); |
| 607 | D3Q19_LIST |
| 608 | #undef X |
| 609 | |
| 610 | if (consecValue >= (VSIZE - 1)) { |
| 611 | // Vectorized part. |
| 612 | |
| 613 | #define X(name, idx, idxinv, _x, _y, _z) JOIN(vpdf_,name) = VLDU(JOIN(ppdf_,name)); |
| 614 | D3Q19_LIST_WO_C |
| 615 | #undef X |
| 616 | |
| 617 | vpdf_C = VLDU(ppdf_C); |
| 618 | |
| 619 | //vux = vpdf_E + vpdf_NE + vpdf_SE + vpdf_TE + vpdf_BE - |
| 620 | // vpdf_W - vpdf_NW - vpdf_SW - vpdf_TW - vpdf_BW; |
| 621 | vux = VSUB(VSUB(VSUB(VSUB(VSUB(VADD(VADD(vpdf_E,VADD(vpdf_NE,vpdf_SE)),VADD(vpdf_TE,vpdf_BE)),vpdf_W),vpdf_NW),vpdf_SW),vpdf_TW),vpdf_BW); |
| 622 | //vuy = vpdf_N + vpdf_NE + vpdf_NW + vpdf_TN + vpdf_BN - |
| 623 | // vpdf_S - vpdf_SE - vpdf_SW - vpdf_TS - vpdf_BS; |
| 624 | vuy = VSUB(VSUB(VSUB(VSUB(VSUB(VADD(VADD(vpdf_N,VADD(vpdf_NE,vpdf_NW)),VADD(vpdf_TN,vpdf_BN)),vpdf_S),vpdf_SE),vpdf_SW),vpdf_TS),vpdf_BS); |
| 625 | //vuz = vpdf_T + vpdf_TE + vpdf_TW + vpdf_TN + vpdf_TS - |
| 626 | // vpdf_B - vpdf_BE - vpdf_BW - vpdf_BN - vpdf_BS; |
| 627 | vuz = VSUB(VSUB(VSUB(VSUB(VSUB(VADD(VADD(vpdf_T,VADD(vpdf_TE,vpdf_TW)),VADD(vpdf_TN,vpdf_TS)),vpdf_B),vpdf_BE),vpdf_BW),vpdf_BN),vpdf_BS); |
| 628 | |
| 629 | //vdens = vpdf_C + |
| 630 | // vpdf_N + vpdf_E + vpdf_S + vpdf_W + |
| 631 | // vpdf_NE + vpdf_SE + vpdf_SW + vpdf_NW + |
| 632 | // vpdf_T + vpdf_TN + vpdf_TE + vpdf_TS + vpdf_TW + |
| 633 | // vpdf_B + vpdf_BN + vpdf_BE + vpdf_BS + vpdf_BW; |
| 634 | vdens = VADD(VADD(VADD(VADD(VADD(VADD(VADD(VADD(VADD(vpdf_C,VADD(vpdf_N,vpdf_E)),VADD(vpdf_S,vpdf_W)),VADD(vpdf_NE,vpdf_SE)), |
| 635 | VADD(vpdf_SW,vpdf_NW)),VADD(vpdf_T,vpdf_TN)),VADD(vpdf_TE,vpdf_TS)),VADD(vpdf_TW,vpdf_B)),VADD(vpdf_BN,vpdf_BE)),VADD(vpdf_BS,vpdf_BW)); |
| 636 | |
| 637 | //vdir_indep_trm = vdens - (vux * vux + vuy * vuy + vuz * vuz) * VTHREE_HALF; |
| 638 | vdir_indep_trm = VSUB(vdens,VMUL(VADD(VADD(VMUL(vux,vux),VMUL(vuy,vuy)),VMUL(vuz,vuz)),VTHREE_HALF)); |
| 639 | |
| 640 | adjListIndex = index * N_D3Q19_IDX; |
| 641 | |
| 642 | //src[I(index, D3Q19_C)] =[UA] vpdf_C - vomegaEven * (vpdf_C - vw_0 * vdir_indep_trm); |
| 643 | VSTU(&src[I(index, D3Q19_C)],VSUB(vpdf_C,VMUL(vomegaEven,VSUB(vpdf_C,VMUL(vw_0,vdir_indep_trm))))); |
| 644 | |
| 645 | //vw_1_indep = vw_1 * vdir_indep_trm; |
| 646 | vw_1_indep = VMUL(vw_1,vdir_indep_trm); |
| 647 | |
| 648 | //vui = vuy; |
| 649 | vui = vuy; |
| 650 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_N + vpdf_S) - vui * vui * vw_1_nine_half - vw_1_indep); |
| 651 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_N,vpdf_S)),VMUL(vui,VMUL(vui,vw_1_nine_half))),vw_1_indep)); |
| 652 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_N - vpdf_S) - vui * vw_1_x3); |
| 653 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_N,vpdf_S)),VMUL(vui,vw_1_x3))); |
| 654 | //src[ADJ_LIST(D3Q19_N)] =[UA] vpdf_N - vevenPart - voddPart; |
| 655 | VSTU(ppdf_S, VSUB(VSUB(vpdf_N,vevenPart),voddPart)); |
| 656 | //src[ADJ_LIST(D3Q19_S)] =[UA] vpdf_S - vevenPart + voddPart; |
| 657 | VSTU(ppdf_N, VADD(VSUB(vpdf_S,vevenPart),voddPart)); |
| 658 | |
| 659 | //vui = vux; |
| 660 | vui = vux; |
| 661 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_E + vpdf_W) - vui * vui * vw_1_nine_half - vw_1_indep); |
| 662 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_E,vpdf_W)),VMUL(vui,VMUL(vui,vw_1_nine_half))),vw_1_indep)); |
| 663 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_E - vpdf_W) - vui * vw_1_x3); |
| 664 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_E,vpdf_W)),VMUL(vui,vw_1_x3))); |
| 665 | //src[ADJ_LIST(D3Q19_E)] =[UA] vpdf_E - vevenPart - voddPart; |
| 666 | VSTU(ppdf_W, VSUB(VSUB(vpdf_E,vevenPart),voddPart)); |
| 667 | //src[ADJ_LIST(D3Q19_W)] =[UA] vpdf_W - vevenPart + voddPart; |
| 668 | VSTU(ppdf_E, VADD(VSUB(vpdf_W,vevenPart),voddPart)); |
| 669 | |
| 670 | //vui = vuz; |
| 671 | vui = vuz; |
| 672 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_T + vpdf_B) - vui * vui * vw_1_nine_half - vw_1_indep); |
| 673 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_T,vpdf_B)),VMUL(vui,VMUL(vui,vw_1_nine_half))),vw_1_indep)); |
| 674 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_T - vpdf_B) - vui * vw_1_x3); |
| 675 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_T,vpdf_B)),VMUL(vui,vw_1_x3))); |
| 676 | //src[ADJ_LIST(D3Q19_T)] =[UA] vpdf_T - vevenPart - voddPart; |
| 677 | VSTU(ppdf_B, VSUB(VSUB(vpdf_T,vevenPart),voddPart)); |
| 678 | //src[ADJ_LIST(D3Q19_B)] =[UA] vpdf_B - vevenPart + voddPart; |
| 679 | VSTU(ppdf_T, VADD(VSUB(vpdf_B,vevenPart),voddPart)); |
| 680 | |
| 681 | //vw_2_indep = vw_2 * vdir_indep_trm; |
| 682 | vw_2_indep = VMUL(vw_2,vdir_indep_trm); |
| 683 | |
| 684 | //vui = vuy - vux; |
| 685 | vui = VSUB(vuy,vux); |
| 686 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_NW + vpdf_SE) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 687 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_NW,vpdf_SE)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 688 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_NW - vpdf_SE) - vui * vw_2_x3); |
| 689 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_NW,vpdf_SE)),VMUL(vui,vw_2_x3))); |
| 690 | //src[ADJ_LIST(D3Q19_NW)] =[UA] vpdf_NW - vevenPart - voddPart; |
| 691 | VSTU(ppdf_SE, VSUB(VSUB(vpdf_NW,vevenPart),voddPart)); |
| 692 | //src[ADJ_LIST(D3Q19_SE)] =[UA] vpdf_SE - vevenPart + voddPart; |
| 693 | VSTU(ppdf_NW, VADD(VSUB(vpdf_SE,vevenPart),voddPart)); |
| 694 | |
| 695 | //vui = vux + vuy; |
| 696 | vui = VADD(vux,vuy); |
| 697 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_NE + vpdf_SW) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 698 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_NE,vpdf_SW)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 699 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_NE - vpdf_SW) - vui * vw_2_x3); |
| 700 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_NE,vpdf_SW)),VMUL(vui,vw_2_x3))); |
| 701 | //src[ADJ_LIST(D3Q19_NE)] =[UA] vpdf_NE - vevenPart - voddPart; |
| 702 | VSTU(ppdf_SW, VSUB(VSUB(vpdf_NE,vevenPart),voddPart)); |
| 703 | //src[ADJ_LIST(D3Q19_SW)] =[UA] vpdf_SW - vevenPart + voddPart; |
| 704 | VSTU(ppdf_NE, VADD(VSUB(vpdf_SW,vevenPart),voddPart)); |
| 705 | |
| 706 | //vui = vuz - vux; |
| 707 | vui = VSUB(vuz,vux); |
| 708 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TW + vpdf_BE) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 709 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TW,vpdf_BE)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 710 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TW - vpdf_BE) - vui * vw_2_x3); |
| 711 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TW,vpdf_BE)),VMUL(vui,vw_2_x3))); |
| 712 | //src[ADJ_LIST(D3Q19_TW)] =[UA] vpdf_TW - vevenPart - voddPart; |
| 713 | VSTU(ppdf_BE, VSUB(VSUB(vpdf_TW,vevenPart),voddPart)); |
| 714 | //src[ADJ_LIST(D3Q19_BE)] =[UA] vpdf_BE - vevenPart + voddPart; |
| 715 | VSTU(ppdf_TW, VADD(VSUB(vpdf_BE,vevenPart),voddPart)); |
| 716 | |
| 717 | //vui = vux + vuz; |
| 718 | vui = VADD(vux,vuz); |
| 719 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TE + vpdf_BW) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 720 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TE,vpdf_BW)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 721 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TE - vpdf_BW) - vui * vw_2_x3); |
| 722 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TE,vpdf_BW)),VMUL(vui,vw_2_x3))); |
| 723 | //src[ADJ_LIST(D3Q19_TE)] =[UA] vpdf_TE - vevenPart - voddPart; |
| 724 | VSTU(ppdf_BW, VSUB(VSUB(vpdf_TE,vevenPart),voddPart)); |
| 725 | //src[ADJ_LIST(D3Q19_BW)] =[UA] vpdf_BW - vevenPart + voddPart; |
| 726 | VSTU(ppdf_TE, VADD(VSUB(vpdf_BW,vevenPart),voddPart)); |
| 727 | |
| 728 | //vui = vuz - vuy; |
| 729 | vui = VSUB(vuz,vuy); |
| 730 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TS + vpdf_BN) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 731 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TS,vpdf_BN)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 732 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TS - vpdf_BN) - vui * vw_2_x3); |
| 733 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TS,vpdf_BN)),VMUL(vui,vw_2_x3))); |
| 734 | //src[ADJ_LIST(D3Q19_TS)] =[UA] vpdf_TS - vevenPart - voddPart; |
| 735 | VSTU(ppdf_BN, VSUB(VSUB(vpdf_TS,vevenPart),voddPart)); |
| 736 | //src[ADJ_LIST(D3Q19_BN)] =[UA] vpdf_BN - vevenPart + voddPart; |
| 737 | VSTU(ppdf_TS, VADD(VSUB(vpdf_BN,vevenPart),voddPart)); |
| 738 | |
| 739 | //vui = vuy + vuz; |
| 740 | vui = VADD(vuy,vuz); |
| 741 | //vevenPart = vomegaEven * (VONE_HALF * (vpdf_TN + vpdf_BS) - vui * vui * vw_2_nine_half - vw_2_indep); |
| 742 | vevenPart = VMUL(vomegaEven,VSUB(VSUB(VMUL(VONE_HALF,VADD(vpdf_TN,vpdf_BS)),VMUL(vui,VMUL(vui,vw_2_nine_half))),vw_2_indep)); |
| 743 | //voddPart = vomegaOdd * (VONE_HALF * (vpdf_TN - vpdf_BS) - vui * vw_2_x3); |
| 744 | voddPart = VMUL(vomegaOdd,VSUB(VMUL(VONE_HALF,VSUB(vpdf_TN,vpdf_BS)),VMUL(vui,vw_2_x3))); |
| 745 | //src[ADJ_LIST(D3Q19_TN)] =[UA] vpdf_TN - vevenPart - voddPart; |
| 746 | VSTU(ppdf_BS, VSUB(VSUB(vpdf_TN,vevenPart),voddPart)); |
| 747 | //src[ADJ_LIST(D3Q19_BS)] =[UA] vpdf_BS - vevenPart + voddPart; |
| 748 | VSTU(ppdf_TN, VADD(VSUB(vpdf_BS,vevenPart),voddPart)); |
| 749 | |
| 750 | consecValue -= (VSIZE - 1); |
| 751 | index += (VSIZE - 1); |
| 752 | pointerOffset = VSIZE; |
| 753 | |
| 754 | } |
| 755 | else { |
| 756 | // Scalar part. |
| 757 | |
| 758 | #define X(name, idx, idxinv, _x, _y, _z) JOIN(pdf_,name) = *(JOIN(ppdf_,name)); |
| 759 | D3Q19_LIST_WO_C |
| 760 | #undef X |
| 761 | |
| 762 | pdf_C = *ppdf_C; |
| 763 | |
| 764 | ux = pdf_E + pdf_NE + pdf_SE + pdf_TE + pdf_BE - |
| 765 | pdf_W - pdf_NW - pdf_SW - pdf_TW - pdf_BW; |
| 766 | uy = pdf_N + pdf_NE + pdf_NW + pdf_TN + pdf_BN - |
| 767 | pdf_S - pdf_SE - pdf_SW - pdf_TS - pdf_BS; |
| 768 | uz = pdf_T + pdf_TE + pdf_TW + pdf_TN + pdf_TS - |
| 769 | pdf_B - pdf_BE - pdf_BW - pdf_BN - pdf_BS; |
| 770 | |
| 771 | dens = pdf_C + |
| 772 | pdf_N + pdf_E + pdf_S + pdf_W + |
| 773 | pdf_NE + pdf_SE + pdf_SW + pdf_NW + |
| 774 | pdf_T + pdf_TN + pdf_TE + pdf_TS + pdf_TW + |
| 775 | pdf_B + pdf_BN + pdf_BE + pdf_BS + pdf_BW; |
| 776 | |
| 777 | dir_indep_trm = dens - (ux * ux + uy * uy + uz * uz) * F(3.0) / F(2.0); |
| 778 | |
| 779 | adjListIndex = index * N_D3Q19_IDX; |
| 780 | |
| 781 | // direction: w_0 |
| 782 | src[I(index, D3Q19_C) ] = pdf_C - omegaEven * (pdf_C - w_0 * dir_indep_trm); |
| 783 | |
| 784 | // direction: w_1 |
| 785 | w_1_indep = w_1 * dir_indep_trm; |
| 786 | |
| 787 | ui = uy; |
| 788 | evenPart = omegaEven * (F(0.5) * (pdf_N + pdf_S) - ui * ui * w_1_nine_half - w_1_indep); |
| 789 | oddPart = omegaOdd * (F(0.5) * (pdf_N - pdf_S) - ui * w_1_x3); |
| 790 | *ppdf_S = pdf_N - evenPart - oddPart; |
| 791 | *ppdf_N = pdf_S - evenPart + oddPart; |
| 792 | |
| 793 | ui = ux; |
| 794 | evenPart = omegaEven * (F(0.5) * (pdf_E + pdf_W) - ui * ui * w_1_nine_half - w_1_indep); |
| 795 | oddPart = omegaOdd * (F(0.5) * (pdf_E - pdf_W) - ui * w_1_x3); |
| 796 | *ppdf_W = pdf_E - evenPart - oddPart; |
| 797 | *ppdf_E = pdf_W - evenPart + oddPart; |
| 798 | |
| 799 | ui = uz; |
| 800 | evenPart = omegaEven * (F(0.5) * (pdf_T + pdf_B) - ui * ui * w_1_nine_half - w_1_indep); |
| 801 | oddPart = omegaOdd * (F(0.5) * (pdf_T - pdf_B) - ui * w_1_x3); |
| 802 | *ppdf_B = pdf_T - evenPart - oddPart; |
| 803 | *ppdf_T = pdf_B - evenPart + oddPart; |
| 804 | |
| 805 | // direction: w_2 |
| 806 | w_2_indep = w_2 * dir_indep_trm; |
| 807 | |
| 808 | ui = -ux + uy; |
| 809 | evenPart = omegaEven * (F(0.5) * (pdf_NW + pdf_SE) - ui * ui * w_2_nine_half - w_2_indep); |
| 810 | oddPart = omegaOdd * (F(0.5) * (pdf_NW - pdf_SE) - ui * w_2_x3); |
| 811 | *ppdf_SE = pdf_NW - evenPart - oddPart; |
| 812 | *ppdf_NW = pdf_SE - evenPart + oddPart; |
| 813 | |
| 814 | ui = ux + uy; |
| 815 | evenPart = omegaEven * (F(0.5) * (pdf_NE + pdf_SW) - ui * ui * w_2_nine_half - w_2_indep); |
| 816 | oddPart = omegaOdd * (F(0.5) * (pdf_NE - pdf_SW) - ui * w_2_x3); |
| 817 | *ppdf_SW = pdf_NE - evenPart - oddPart; |
| 818 | *ppdf_NE = pdf_SW - evenPart + oddPart; |
| 819 | |
| 820 | ui = -ux + uz; |
| 821 | evenPart = omegaEven * (F(0.5) * (pdf_TW + pdf_BE) - ui * ui * w_2_nine_half - w_2_indep); |
| 822 | oddPart = omegaOdd * (F(0.5) * (pdf_TW - pdf_BE) - ui * w_2_x3); |
| 823 | *ppdf_BE = pdf_TW - evenPart - oddPart; |
| 824 | *ppdf_TW = pdf_BE - evenPart + oddPart; |
| 825 | |
| 826 | ui = ux + uz; |
| 827 | evenPart = omegaEven * (F(0.5) * (pdf_TE + pdf_BW) - ui * ui * w_2_nine_half - w_2_indep); |
| 828 | oddPart = omegaOdd * (F(0.5) * (pdf_TE - pdf_BW) - ui * w_2_x3); |
| 829 | *ppdf_BW = pdf_TE - evenPart - oddPart; |
| 830 | *ppdf_TE = pdf_BW - evenPart + oddPart; |
| 831 | |
| 832 | ui = -uy + uz; |
| 833 | evenPart = omegaEven * (F(0.5) * (pdf_TS + pdf_BN) - ui * ui * w_2_nine_half - w_2_indep); |
| 834 | oddPart = omegaOdd * (F(0.5) * (pdf_TS - pdf_BN) - ui * w_2_x3); |
| 835 | *ppdf_BN = pdf_TS - evenPart - oddPart; |
| 836 | *ppdf_TS = pdf_BN - evenPart + oddPart; |
| 837 | |
| 838 | ui = uy + uz; |
| 839 | evenPart = omegaEven * (F(0.5) * (pdf_TN + pdf_BS) - ui * ui * w_2_nine_half - w_2_indep); |
| 840 | oddPart = omegaOdd * (F(0.5) * (pdf_TN - pdf_BS) - ui * w_2_x3); |
| 841 | *ppdf_BS = pdf_TN - evenPart - oddPart; |
| 842 | *ppdf_TN = pdf_BS - evenPart + oddPart; |
| 843 | |
| 844 | pointerOffset = 1; |
| 845 | } |
| 846 | |
| 847 | } // loop over fluid nodes |
| 848 | |
| 849 | #undef ADJ_LIST |
| 850 | #undef I |
| 851 | } |