src/gromacs/nbnxm/kernels_simd_2xmm/kernel_prune.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2016,2017,2018,2019, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35
  36 #include "gmxpre.h"
  37
  38 #include "kernel_prune.h"
  39
  40 #include "gromacs/nbnxm/atomdata.h"
  41 #include "gromacs/nbnxm/nbnxm_simd.h"
  42 #include "gromacs/nbnxm/pairlist.h"
  43 #include "gromacs/utility/gmxassert.h"
  44
  45 #ifdef GMX_NBNXN_SIMD_2XNN
  46 #define GMX_SIMD_J_UNROLL_SIZE 2
  47 #include "gromacs/nbnxm/kernels_simd_2xmm/kernel_common.h"
  48 #endif
  49
  50 /* Prune a single nbnxn_pairtlist_t entry with distance rlistInner */
  51 void
  52 nbnxn_kernel_prune_2xnn(NbnxnPairlistCpu *         nbl,
  53                         const nbnxn_atomdata_t *   nbat,
  54                         const rvec * gmx_restrict  shift_vec,
  55                         real                       rlistInner)
  56 {
  57 #ifdef GMX_NBNXN_SIMD_2XNN
  58     using namespace gmx;
  59
  60     /* We avoid push_back() for efficiency reasons and resize after filling */
  61     nbl->ci.resize(nbl->ciOuter.size());
  62     nbl->cj.resize(nbl->cjOuter.size());
  63
  64     const nbnxn_ci_t * gmx_restrict ciOuter  = nbl->ciOuter.data();
  65     nbnxn_ci_t       * gmx_restrict ciInner  = nbl->ci.data();
  66
  67     const nbnxn_cj_t * gmx_restrict cjOuter  = nbl->cjOuter.data();
  68     nbnxn_cj_t       * gmx_restrict cjInner  = nbl->cj.data();
  69
  70     const real       * gmx_restrict shiftvec = shift_vec[0];
  71     const real       * gmx_restrict x        = nbat->x().data();
  72
  73     const SimdReal                  rlist2_S(rlistInner*rlistInner);
  74
  75     /* Initialize the new list count as empty and add pairs that are in range */
  76     int       nciInner = 0;
  77     int       ncjInner = 0;
  78     const int nciOuter = nbl->ciOuter.size();
  79     for (int i = 0; i < nciOuter; i++)
  80     {
  81         const nbnxn_ci_t * gmx_restrict ciEntry = &ciOuter[i];
  82
  83         /* Copy the original list entry to the pruned entry */
  84         ciInner[nciInner].ci           = ciEntry->ci;
  85         ciInner[nciInner].shift        = ciEntry->shift;
  86         ciInner[nciInner].cj_ind_start = ncjInner;
  87
  88         /* Extract shift data */
  89         int      ish     = (ciEntry->shift & NBNXN_CI_SHIFT);
  90         int      ish3    = ish*3;
  91         int      ci      = ciEntry->ci;
  92
  93         SimdReal shX_S   = SimdReal(shiftvec[ish3    ]);
  94         SimdReal shY_S   = SimdReal(shiftvec[ish3 + 1]);
  95         SimdReal shZ_S   = SimdReal(shiftvec[ish3 + 2]);
  96
  97 #if UNROLLJ <= 4
  98         int      scix    = ci*STRIDE*DIM;
  99 #else
 100         int      scix    = (ci >> 1)*STRIDE*DIM + (ci & 1)*(STRIDE >> 1);
 101 #endif
 102
 103         /* Load i atom data */
 104         int      sciy    = scix + STRIDE;
 105         int      sciz    = sciy + STRIDE;
 106         SimdReal ix_S0   = loadU1DualHsimd(x + scix    ) + shX_S;
 107         SimdReal ix_S2   = loadU1DualHsimd(x + scix + 2) + shX_S;
 108         SimdReal iy_S0   = loadU1DualHsimd(x + sciy    ) + shY_S;
 109         SimdReal iy_S2   = loadU1DualHsimd(x + sciy + 2) + shY_S;
 110         SimdReal iz_S0   = loadU1DualHsimd(x + sciz    ) + shZ_S;
 111         SimdReal iz_S2   = loadU1DualHsimd(x + sciz + 2) + shZ_S;
 112
 113         for (int cjind = ciEntry->cj_ind_start; cjind < ciEntry->cj_ind_end; cjind++)
 114         {
 115             /* j-cluster index */
 116             int cj      = cjOuter[cjind].cj;
 117
 118             /* Atom indices (of the first atom in the cluster) */
 119 #if UNROLLJ == STRIDE
 120             int aj      = cj*UNROLLJ;
 121             int ajx     = aj*DIM;
 122 #else
 123             int ajx     = (cj >> 1)*DIM*STRIDE + (cj & 1)*UNROLLJ;
 124 #endif
 125             int ajy     = ajx + STRIDE;
 126             int ajz     = ajy + STRIDE;
 127
 128             /* load j atom coordinates */
 129             SimdReal jx_S   = loadDuplicateHsimd(x + ajx);
 130             SimdReal jy_S   = loadDuplicateHsimd(x + ajy);
 131             SimdReal jz_S   = loadDuplicateHsimd(x + ajz);
 132
 133             /* Calculate distance */
 134             SimdReal dx_S0  = ix_S0 - jx_S;
 135             SimdReal dy_S0  = iy_S0 - jy_S;
 136             SimdReal dz_S0  = iz_S0 - jz_S;
 137             SimdReal dx_S2  = ix_S2 - jx_S;
 138             SimdReal dy_S2  = iy_S2 - jy_S;
 139             SimdReal dz_S2  = iz_S2 - jz_S;
 140
 141             /* rsq = dx*dx+dy*dy+dz*dz */
 142             SimdReal rsq_S0 = norm2(dx_S0, dy_S0, dz_S0);
 143             SimdReal rsq_S2 = norm2(dx_S2, dy_S2, dz_S2);
 144
 145             /* Do the cut-off check */
 146             SimdBool wco_S0 = (rsq_S0 < rlist2_S);
 147             SimdBool wco_S2 = (rsq_S2 < rlist2_S);
 148
 149             wco_S0          = wco_S0 || wco_S2;
 150
 151             /* Putting the assignment inside the conditional is slower */
 152             cjInner[ncjInner] = cjOuter[cjind];
 153             if (anyTrue(wco_S0))
 154             {
 155                 ncjInner++;
 156             }
 157         }
 158
 159         if (ncjInner > ciInner[nciInner].cj_ind_start)
 160         {
 161             ciInner[nciInner].cj_ind_end = ncjInner;
 162             nciInner++;
 163         }
 164     }
 165
 166     nbl->ci.resize(nciInner);
 167     nbl->cj.resize(ncjInner);
 168
 169 #else  /* GMX_NBNXN_SIMD_2XNN */
 170
 171     GMX_RELEASE_ASSERT(false, "2xNN kernel called without 2xNN support");
 172
 173     GMX_UNUSED_VALUE(nbl);
 174     GMX_UNUSED_VALUE(nbat);
 175     GMX_UNUSED_VALUE(shift_vec);
 176     GMX_UNUSED_VALUE(rlistInner);
 177
 178 #endif /* GMX_NBNXN_SIMD_2XNN */
 179 }