src/BenchKernelD3Q19ListPullSplitNt1SIntrinsics.h

   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
  28 #ifndef INDEX_START
  29         #error INDEX_START must be defined
  30 #endif
  31
  32 #ifndef INDEX_STOP
  33         #error INDEX_STOP must be defined
  34 #endif
  35
  36         #define I(index, dir)   P_INDEX_3((nCells), (index), (dir))
  37
  38         for (int blockedIndex = (INDEX_START); blockedIndex < (INDEX_STOP); blockedIndex += nTmpArray) {
  39
  40                 indexMax = MinI(nTmpArray, (INDEX_STOP) - blockedIndex);
  41 #ifdef DEBUG
  42                 memset(tmpArray, -1, sizeof(PdfT) * nTmpArray * N_TMP);
  43 #endif
  44                 for (int index = 0; index < indexMax; ++index) {
  45
  46
  47                         adjListIndex = (index + blockedIndex) * N_D3Q19_IDX;
  48
  49                         pdf_C = src[I(index + blockedIndex, D3Q19_C)];
  50
  51                         #define X(name, idx, idxinv, _x, _y, _z)        JOIN(pdf_,name) = src[adjList[adjListIndex + idx]]; tmpArray[TMP_INDEX(index, idx)] = JOIN(pdf_,name);
  52                         D3Q19_LIST_WO_C
  53                         #undef X
  54
  55                         ux = pdf_E + pdf_NE + pdf_SE + pdf_TE + pdf_BE -
  56                                  pdf_W - pdf_NW - pdf_SW - pdf_TW - pdf_BW;
  57                         uy = pdf_N + pdf_NE + pdf_NW + pdf_TN + pdf_BN -
  58                                  pdf_S - pdf_SE - pdf_SW - pdf_TS - pdf_BS;
  59                         uz = pdf_T + pdf_TE + pdf_TW + pdf_TN + pdf_TS -
  60                                  pdf_B - pdf_BE - pdf_BW - pdf_BN - pdf_BS;
  61
  62                         tmpArray[TMP_INDEX(index, TMP_UX)] = ux;
  63                         tmpArray[TMP_INDEX(index, TMP_UY)] = uy;
  64                         tmpArray[TMP_INDEX(index, TMP_UZ)] = uz;
  65
  66                         dens = pdf_C +
  67                                    pdf_N  + pdf_E  + pdf_S  + pdf_W  +
  68                                    pdf_NE + pdf_SE + pdf_SW + pdf_NW +
  69                                    pdf_T  + pdf_TN + pdf_TE + pdf_TS + pdf_TW +
  70                                    pdf_B  + pdf_BN + pdf_BE + pdf_BS + pdf_BW;
  71
  72                         dir_indep_trm = dens - (ux * ux + uy * uy + uz * uz) * 3.0 / 2.0;
  73
  74                         w_1_indep = w_1 * dir_indep_trm;
  75                         w_2_indep = w_2 * dir_indep_trm;
  76
  77                         tmpArray[TMP_INDEX(index, TMP_W1)] = w_1_indep;
  78                         tmpArray[TMP_INDEX(index, TMP_W2)] = w_2_indep;
  79
  80                         dst[I(index + blockedIndex, D3Q19_C )]  = pdf_C - omegaEven * (pdf_C - w_0 * dir_indep_trm);
  81                 }
  82
  83                 #define LOOP_1(_dir1, _dir2, _vel, _vel_tmp) \
  84                         for (int index = 0; index < indexMax; index += VSIZE) { \
  85                                 Assert((unsigned long)&(dst[I(index + blockedIndex, JOIN(D3Q19_,_dir1) )]) % VSIZE == 0); \
  86                                 vui         = VLDU(&tmpArray[TMP_INDEX(index, JOIN(TMP_,_vel_tmp))]); \
  87                                 vpdf_a      = VLDU(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir1))]); \
  88                                 vpdf_b      = VLDU(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir2))]); \
  89                                 vw_1_indep  = VLDU(&tmpArray[TMP_INDEX(index, TMP_W1)]); \
  90                                 \
  91                                 vevenPart = VMUL(vomegaEven, VSUB(VSUB(VMUL(voneHalf, VADD(vpdf_a, vpdf_b)), VMUL(vui, VMUL(vui, vw_1_nine_half))), vw_1_indep)); \
  92                                 voddPart  = VMUL(vomegaOdd,  VSUB(     VMUL(voneHalf, VSUB(vpdf_a, vpdf_b)), VMUL(vui, vw_1_x3))); \
  93                                 VSTNT(&dst[I(index + blockedIndex, JOIN(D3Q19_,_dir1))], VSUB(VSUB(vpdf_a, vevenPart), voddPart)); \
  94                                 VST(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir2))], VADD(VSUB(vpdf_b, vevenPart), voddPart)); \
  95                         } \
  96                         for (int index = 0; index < indexMax; index += VSIZE) { \
  97                                 Assert((unsigned long)&(dst[I(index + blockedIndex, JOIN(D3Q19_,_dir2) )]) % VSIZE == 0); \
  98                                 vpdf_b      = VLDU(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir2))]); \
  99                                 VSTNT(&dst[I(index + blockedIndex, JOIN(D3Q19_,_dir2))], vpdf_b); \
 100                         }
 101
 102                 #define LOOP_2(_dir1, _dir2, _v1, _v2, _v1_tmp, _v2_tmp, _expr) \
 103                         for (int index = 0; index < indexMax; index += VSIZE) { \
 104                                 _v1         = VLDU(&tmpArray[TMP_INDEX(index, JOIN(TMP_,_v1_tmp))]); \
 105                                 _v2         = VLDU(&tmpArray[TMP_INDEX(index, JOIN(TMP_,_v2_tmp))]); \
 106                                 vpdf_a          = VLDU(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir1))]); \
 107                                 vpdf_b          = VLDU(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir2))]); \
 108                                 vw_2_indep  = VLDU(&tmpArray[TMP_INDEX(index, TMP_W2)]); \
 109                                 \
 110                                 vui = _expr; \
 111                                 vevenPart = VMUL(vomegaEven, VSUB(VSUB(VMUL(voneHalf, VADD(vpdf_a, vpdf_b)), VMUL(vui, VMUL(vui, vw_2_nine_half))), vw_2_indep)); \
 112                                 voddPart  = VMUL(vomegaOdd,  VSUB(     VMUL(voneHalf, VSUB(vpdf_a, vpdf_b)), VMUL(vui, vw_2_x3))); \
 113                                 VSTNT(&dst[I(index + blockedIndex, JOIN(D3Q19_,_dir1))], VSUB(VSUB(vpdf_a, vevenPart), voddPart)); \
 114                                 VST(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir2))], VADD(VSUB(vpdf_b, vevenPart), voddPart)); \
 115                         } \
 116                         for (int index = 0; index < indexMax; index += VSIZE) { \
 117                                 Assert((unsigned long)&(dst[I(index + blockedIndex, JOIN(D3Q19_,_dir2) )]) % VSIZE == 0); \
 118                                 vpdf_b      = VLDU(&tmpArray[TMP_INDEX(index, JOIN(D3Q19_,_dir2))]); \
 119                                 VSTNT(&dst[I(index + blockedIndex, JOIN(D3Q19_,_dir2))], vpdf_b); \
 120                         }
 121
 122                 LOOP_1(N, S, vuy, UY);
 123                 LOOP_1(E, W, vux, UX);
 124                 LOOP_1(T, B, vuz, UZ);
 125
 126                 LOOP_2(NW, SE, vuy, vux, UY, UX, VSUB(vuy, vux));
 127                 LOOP_2(NE, SW, vuy, vux, UY, UX, VADD(vuy, vux));
 128                 LOOP_2(TW, BE, vux, vuz, UX, UZ, VSUB(vuz, vux));
 129                 LOOP_2(TE, BW, vux, vuz, UX, UZ, VADD(vuz, vux));
 130                 LOOP_2(TS, BN, vuy, vuz, UY, UZ, VSUB(vuz, vuy));
 131                 LOOP_2(TN, BS, vuy, vuz, UY, UZ, VADD(vuz, vuy));
 132
 133                 #undef LOOP_1
 134                 #undef LOOP_2
 135
 136         } // loop over fluid nodes
 137
 138         #undef I
 139
 140         #undef INDEX_START
 141         #undef INDEX_STOP
 142