src/gromacs/nbnxm/kernels_simd_4xm/kernels.h

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2019, 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  * Note: this file was generated by the Verlet kernel generator for
  37  * kernel type 4xm.
  38  */
  39
  40
  41 #include "gromacs/nbnxm/kernel_common.h"
  42
  43 /* Declare all the different kernel functions.
  44  */
  45 nbk_func_noener nbnxm_kernel_ElecRF_VdwLJCombGeom_F_4xm;
  46 nbk_func_noener nbnxm_kernel_ElecRF_VdwLJCombLB_F_4xm;
  47 nbk_func_noener nbnxm_kernel_ElecRF_VdwLJ_F_4xm;
  48 nbk_func_noener nbnxm_kernel_ElecRF_VdwLJFSw_F_4xm;
  49 nbk_func_noener nbnxm_kernel_ElecRF_VdwLJPSw_F_4xm;
  50 nbk_func_noener nbnxm_kernel_ElecRF_VdwLJEwCombGeom_F_4xm;
  51 nbk_func_noener nbnxm_kernel_ElecQSTab_VdwLJCombGeom_F_4xm;
  52 nbk_func_noener nbnxm_kernel_ElecQSTab_VdwLJCombLB_F_4xm;
  53 nbk_func_noener nbnxm_kernel_ElecQSTab_VdwLJ_F_4xm;
  54 nbk_func_noener nbnxm_kernel_ElecQSTab_VdwLJFSw_F_4xm;
  55 nbk_func_noener nbnxm_kernel_ElecQSTab_VdwLJPSw_F_4xm;
  56 nbk_func_noener nbnxm_kernel_ElecQSTab_VdwLJEwCombGeom_F_4xm;
  57 nbk_func_noener nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombGeom_F_4xm;
  58 nbk_func_noener nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombLB_F_4xm;
  59 nbk_func_noener nbnxm_kernel_ElecQSTabTwinCut_VdwLJ_F_4xm;
  60 nbk_func_noener nbnxm_kernel_ElecQSTabTwinCut_VdwLJFSw_F_4xm;
  61 nbk_func_noener nbnxm_kernel_ElecQSTabTwinCut_VdwLJPSw_F_4xm;
  62 nbk_func_noener nbnxm_kernel_ElecQSTabTwinCut_VdwLJEwCombGeom_F_4xm;
  63 nbk_func_noener nbnxm_kernel_ElecEw_VdwLJCombGeom_F_4xm;
  64 nbk_func_noener nbnxm_kernel_ElecEw_VdwLJCombLB_F_4xm;
  65 nbk_func_noener nbnxm_kernel_ElecEw_VdwLJ_F_4xm;
  66 nbk_func_noener nbnxm_kernel_ElecEw_VdwLJFSw_F_4xm;
  67 nbk_func_noener nbnxm_kernel_ElecEw_VdwLJPSw_F_4xm;
  68 nbk_func_noener nbnxm_kernel_ElecEw_VdwLJEwCombGeom_F_4xm;
  69 nbk_func_noener nbnxm_kernel_ElecEwTwinCut_VdwLJCombGeom_F_4xm;
  70 nbk_func_noener nbnxm_kernel_ElecEwTwinCut_VdwLJCombLB_F_4xm;
  71 nbk_func_noener nbnxm_kernel_ElecEwTwinCut_VdwLJ_F_4xm;
  72 nbk_func_noener nbnxm_kernel_ElecEwTwinCut_VdwLJFSw_F_4xm;
  73 nbk_func_noener nbnxm_kernel_ElecEwTwinCut_VdwLJPSw_F_4xm;
  74 nbk_func_noener nbnxm_kernel_ElecEwTwinCut_VdwLJEwCombGeom_F_4xm;
  75
  76 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJCombGeom_VF_4xm;
  77 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJCombLB_VF_4xm;
  78 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJ_VF_4xm;
  79 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJFSw_VF_4xm;
  80 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJPSw_VF_4xm;
  81 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJEwCombGeom_VF_4xm;
  82 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJCombGeom_VF_4xm;
  83 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJCombLB_VF_4xm;
  84 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJ_VF_4xm;
  85 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJFSw_VF_4xm;
  86 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJPSw_VF_4xm;
  87 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJEwCombGeom_VF_4xm;
  88 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VF_4xm;
  89 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombLB_VF_4xm;
  90 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJ_VF_4xm;
  91 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJFSw_VF_4xm;
  92 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJPSw_VF_4xm;
  93 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJEwCombGeom_VF_4xm;
  94 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJCombGeom_VF_4xm;
  95 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJCombLB_VF_4xm;
  96 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJ_VF_4xm;
  97 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJFSw_VF_4xm;
  98 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJPSw_VF_4xm;
  99 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJEwCombGeom_VF_4xm;
 100 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJCombGeom_VF_4xm;
 101 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJCombLB_VF_4xm;
 102 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJ_VF_4xm;
 103 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJFSw_VF_4xm;
 104 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJPSw_VF_4xm;
 105 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJEwCombGeom_VF_4xm;
 106
 107 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJCombGeom_VgrpF_4xm;
 108 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJCombLB_VgrpF_4xm;
 109 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJ_VgrpF_4xm;
 110 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJFSw_VgrpF_4xm;
 111 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJPSw_VgrpF_4xm;
 112 nbk_func_ener nbnxm_kernel_ElecRF_VdwLJEwCombGeom_VgrpF_4xm;
 113 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJCombGeom_VgrpF_4xm;
 114 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJCombLB_VgrpF_4xm;
 115 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJ_VgrpF_4xm;
 116 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJFSw_VgrpF_4xm;
 117 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJPSw_VgrpF_4xm;
 118 nbk_func_ener nbnxm_kernel_ElecQSTab_VdwLJEwCombGeom_VgrpF_4xm;
 119 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VgrpF_4xm;
 120 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombLB_VgrpF_4xm;
 121 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJ_VgrpF_4xm;
 122 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJFSw_VgrpF_4xm;
 123 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJPSw_VgrpF_4xm;
 124 nbk_func_ener nbnxm_kernel_ElecQSTabTwinCut_VdwLJEwCombGeom_VgrpF_4xm;
 125 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJCombGeom_VgrpF_4xm;
 126 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJCombLB_VgrpF_4xm;
 127 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJ_VgrpF_4xm;
 128 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJFSw_VgrpF_4xm;
 129 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJPSw_VgrpF_4xm;
 130 nbk_func_ener nbnxm_kernel_ElecEw_VdwLJEwCombGeom_VgrpF_4xm;
 131 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJCombGeom_VgrpF_4xm;
 132 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJCombLB_VgrpF_4xm;
 133 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJ_VgrpF_4xm;
 134 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJFSw_VgrpF_4xm;
 135 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJPSw_VgrpF_4xm;
 136 nbk_func_ener nbnxm_kernel_ElecEwTwinCut_VdwLJEwCombGeom_VgrpF_4xm;
 137
 138
 139 #ifdef INCLUDE_KERNELFUNCTION_TABLES
 140
 141 /* Declare and define the kernel function pointer lookup tables.
 142  * The minor index of the array goes over both the LJ combination rules,
 143  * which is only supported by plain cut-off, and the LJ switch/PME functions.
 144  */
 145 static p_nbk_func_noener nbnxm_kernel_noener_simd_4xm[coulktNR][vdwktNR] = {
 146     {
 147             nbnxm_kernel_ElecRF_VdwLJCombGeom_F_4xm,
 148             nbnxm_kernel_ElecRF_VdwLJCombLB_F_4xm,
 149             nbnxm_kernel_ElecRF_VdwLJ_F_4xm,
 150             nbnxm_kernel_ElecRF_VdwLJFSw_F_4xm,
 151             nbnxm_kernel_ElecRF_VdwLJPSw_F_4xm,
 152             nbnxm_kernel_ElecRF_VdwLJEwCombGeom_F_4xm,
 153     },
 154     {
 155             nbnxm_kernel_ElecQSTab_VdwLJCombGeom_F_4xm,
 156             nbnxm_kernel_ElecQSTab_VdwLJCombLB_F_4xm,
 157             nbnxm_kernel_ElecQSTab_VdwLJ_F_4xm,
 158             nbnxm_kernel_ElecQSTab_VdwLJFSw_F_4xm,
 159             nbnxm_kernel_ElecQSTab_VdwLJPSw_F_4xm,
 160             nbnxm_kernel_ElecQSTab_VdwLJEwCombGeom_F_4xm,
 161     },
 162     {
 163             nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombGeom_F_4xm,
 164             nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombLB_F_4xm,
 165             nbnxm_kernel_ElecQSTabTwinCut_VdwLJ_F_4xm,
 166             nbnxm_kernel_ElecQSTabTwinCut_VdwLJFSw_F_4xm,
 167             nbnxm_kernel_ElecQSTabTwinCut_VdwLJPSw_F_4xm,
 168             nbnxm_kernel_ElecQSTabTwinCut_VdwLJEwCombGeom_F_4xm,
 169     },
 170     {
 171             nbnxm_kernel_ElecEw_VdwLJCombGeom_F_4xm,
 172             nbnxm_kernel_ElecEw_VdwLJCombLB_F_4xm,
 173             nbnxm_kernel_ElecEw_VdwLJ_F_4xm,
 174             nbnxm_kernel_ElecEw_VdwLJFSw_F_4xm,
 175             nbnxm_kernel_ElecEw_VdwLJPSw_F_4xm,
 176             nbnxm_kernel_ElecEw_VdwLJEwCombGeom_F_4xm,
 177     },
 178     {
 179             nbnxm_kernel_ElecEwTwinCut_VdwLJCombGeom_F_4xm,
 180             nbnxm_kernel_ElecEwTwinCut_VdwLJCombLB_F_4xm,
 181             nbnxm_kernel_ElecEwTwinCut_VdwLJ_F_4xm,
 182             nbnxm_kernel_ElecEwTwinCut_VdwLJFSw_F_4xm,
 183             nbnxm_kernel_ElecEwTwinCut_VdwLJPSw_F_4xm,
 184             nbnxm_kernel_ElecEwTwinCut_VdwLJEwCombGeom_F_4xm,
 185     },
 186 };
 187
 188 static p_nbk_func_ener nbnxm_kernel_ener_simd_4xm[coulktNR][vdwktNR] = {
 189     {
 190             nbnxm_kernel_ElecRF_VdwLJCombGeom_VF_4xm,
 191             nbnxm_kernel_ElecRF_VdwLJCombLB_VF_4xm,
 192             nbnxm_kernel_ElecRF_VdwLJ_VF_4xm,
 193             nbnxm_kernel_ElecRF_VdwLJFSw_VF_4xm,
 194             nbnxm_kernel_ElecRF_VdwLJPSw_VF_4xm,
 195             nbnxm_kernel_ElecRF_VdwLJEwCombGeom_VF_4xm,
 196     },
 197     {
 198             nbnxm_kernel_ElecQSTab_VdwLJCombGeom_VF_4xm,
 199             nbnxm_kernel_ElecQSTab_VdwLJCombLB_VF_4xm,
 200             nbnxm_kernel_ElecQSTab_VdwLJ_VF_4xm,
 201             nbnxm_kernel_ElecQSTab_VdwLJFSw_VF_4xm,
 202             nbnxm_kernel_ElecQSTab_VdwLJPSw_VF_4xm,
 203             nbnxm_kernel_ElecQSTab_VdwLJEwCombGeom_VF_4xm,
 204     },
 205     {
 206             nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VF_4xm,
 207             nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombLB_VF_4xm,
 208             nbnxm_kernel_ElecQSTabTwinCut_VdwLJ_VF_4xm,
 209             nbnxm_kernel_ElecQSTabTwinCut_VdwLJFSw_VF_4xm,
 210             nbnxm_kernel_ElecQSTabTwinCut_VdwLJPSw_VF_4xm,
 211             nbnxm_kernel_ElecQSTabTwinCut_VdwLJEwCombGeom_VF_4xm,
 212     },
 213     {
 214             nbnxm_kernel_ElecEw_VdwLJCombGeom_VF_4xm,
 215             nbnxm_kernel_ElecEw_VdwLJCombLB_VF_4xm,
 216             nbnxm_kernel_ElecEw_VdwLJ_VF_4xm,
 217             nbnxm_kernel_ElecEw_VdwLJFSw_VF_4xm,
 218             nbnxm_kernel_ElecEw_VdwLJPSw_VF_4xm,
 219             nbnxm_kernel_ElecEw_VdwLJEwCombGeom_VF_4xm,
 220     },
 221     {
 222             nbnxm_kernel_ElecEwTwinCut_VdwLJCombGeom_VF_4xm,
 223             nbnxm_kernel_ElecEwTwinCut_VdwLJCombLB_VF_4xm,
 224             nbnxm_kernel_ElecEwTwinCut_VdwLJ_VF_4xm,
 225             nbnxm_kernel_ElecEwTwinCut_VdwLJFSw_VF_4xm,
 226             nbnxm_kernel_ElecEwTwinCut_VdwLJPSw_VF_4xm,
 227             nbnxm_kernel_ElecEwTwinCut_VdwLJEwCombGeom_VF_4xm,
 228     },
 229 };
 230
 231 static p_nbk_func_ener nbnxm_kernel_energrp_simd_4xm[coulktNR][vdwktNR] = {
 232     {
 233             nbnxm_kernel_ElecRF_VdwLJCombGeom_VgrpF_4xm,
 234             nbnxm_kernel_ElecRF_VdwLJCombLB_VgrpF_4xm,
 235             nbnxm_kernel_ElecRF_VdwLJ_VgrpF_4xm,
 236             nbnxm_kernel_ElecRF_VdwLJFSw_VgrpF_4xm,
 237             nbnxm_kernel_ElecRF_VdwLJPSw_VgrpF_4xm,
 238             nbnxm_kernel_ElecRF_VdwLJEwCombGeom_VgrpF_4xm,
 239     },
 240     {
 241             nbnxm_kernel_ElecQSTab_VdwLJCombGeom_VgrpF_4xm,
 242             nbnxm_kernel_ElecQSTab_VdwLJCombLB_VgrpF_4xm,
 243             nbnxm_kernel_ElecQSTab_VdwLJ_VgrpF_4xm,
 244             nbnxm_kernel_ElecQSTab_VdwLJFSw_VgrpF_4xm,
 245             nbnxm_kernel_ElecQSTab_VdwLJPSw_VgrpF_4xm,
 246             nbnxm_kernel_ElecQSTab_VdwLJEwCombGeom_VgrpF_4xm,
 247     },
 248     {
 249             nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VgrpF_4xm,
 250             nbnxm_kernel_ElecQSTabTwinCut_VdwLJCombLB_VgrpF_4xm,
 251             nbnxm_kernel_ElecQSTabTwinCut_VdwLJ_VgrpF_4xm,
 252             nbnxm_kernel_ElecQSTabTwinCut_VdwLJFSw_VgrpF_4xm,
 253             nbnxm_kernel_ElecQSTabTwinCut_VdwLJPSw_VgrpF_4xm,
 254             nbnxm_kernel_ElecQSTabTwinCut_VdwLJEwCombGeom_VgrpF_4xm,
 255     },
 256     {
 257             nbnxm_kernel_ElecEw_VdwLJCombGeom_VgrpF_4xm,
 258             nbnxm_kernel_ElecEw_VdwLJCombLB_VgrpF_4xm,
 259             nbnxm_kernel_ElecEw_VdwLJ_VgrpF_4xm,
 260             nbnxm_kernel_ElecEw_VdwLJFSw_VgrpF_4xm,
 261             nbnxm_kernel_ElecEw_VdwLJPSw_VgrpF_4xm,
 262             nbnxm_kernel_ElecEw_VdwLJEwCombGeom_VgrpF_4xm,
 263     },
 264     {
 265             nbnxm_kernel_ElecEwTwinCut_VdwLJCombGeom_VgrpF_4xm,
 266             nbnxm_kernel_ElecEwTwinCut_VdwLJCombLB_VgrpF_4xm,
 267             nbnxm_kernel_ElecEwTwinCut_VdwLJ_VgrpF_4xm,
 268             nbnxm_kernel_ElecEwTwinCut_VdwLJFSw_VgrpF_4xm,
 269             nbnxm_kernel_ElecEwTwinCut_VdwLJPSw_VgrpF_4xm,
 270             nbnxm_kernel_ElecEwTwinCut_VdwLJEwCombGeom_VgrpF_4xm,
 271     },
 272 };
 273
 274
 275 #endif /* INCLUDE_KERNELFUNCTION_TABLES */