| File: | gromacs/gmxlib/nonbonded/nb_kernel_sse4_1_single/nb_kernel_ElecCSTab_VdwCSTab_GeomW4W4_sse4_1_single.c |
| Location: | line 1479, column 22 |
| Description: | Value stored to 'two' during its initialization is never read |
| 1 | /* |
| 2 | * This file is part of the GROMACS molecular simulation package. |
| 3 | * |
| 4 | * Copyright (c) 2012,2013,2014, 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 GROMACS sse4_1_single kernel generator. |
| 37 | */ |
| 38 | #ifdef HAVE_CONFIG_H1 |
| 39 | #include <config.h> |
| 40 | #endif |
| 41 | |
| 42 | #include <math.h> |
| 43 | |
| 44 | #include "../nb_kernel.h" |
| 45 | #include "types/simple.h" |
| 46 | #include "gromacs/math/vec.h" |
| 47 | #include "nrnb.h" |
| 48 | |
| 49 | #include "gromacs/simd/math_x86_sse4_1_single.h" |
| 50 | #include "kernelutil_x86_sse4_1_single.h" |
| 51 | |
| 52 | /* |
| 53 | * Gromacs nonbonded kernel: nb_kernel_ElecCSTab_VdwCSTab_GeomW4W4_VF_sse4_1_single |
| 54 | * Electrostatics interaction: CubicSplineTable |
| 55 | * VdW interaction: CubicSplineTable |
| 56 | * Geometry: Water4-Water4 |
| 57 | * Calculate force/pot: PotentialAndForce |
| 58 | */ |
| 59 | void |
| 60 | nb_kernel_ElecCSTab_VdwCSTab_GeomW4W4_VF_sse4_1_single |
| 61 | (t_nblist * gmx_restrict nlist, |
| 62 | rvec * gmx_restrict xx, |
| 63 | rvec * gmx_restrict ff, |
| 64 | t_forcerec * gmx_restrict fr, |
| 65 | t_mdatoms * gmx_restrict mdatoms, |
| 66 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict kernel_data, |
| 67 | t_nrnb * gmx_restrict nrnb) |
| 68 | { |
| 69 | /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or |
| 70 | * just 0 for non-waters. |
| 71 | * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different |
| 72 | * jnr indices corresponding to data put in the four positions in the SIMD register. |
| 73 | */ |
| 74 | int i_shift_offset,i_coord_offset,outeriter,inneriter; |
| 75 | int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx; |
| 76 | int jnrA,jnrB,jnrC,jnrD; |
| 77 | int jnrlistA,jnrlistB,jnrlistC,jnrlistD; |
| 78 | int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD; |
| 79 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
| 80 | real rcutoff_scalar; |
| 81 | real *shiftvec,*fshift,*x,*f; |
| 82 | real *fjptrA,*fjptrB,*fjptrC,*fjptrD; |
| 83 | real scratch[4*DIM3]; |
| 84 | __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall; |
| 85 | int vdwioffset0; |
| 86 | __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0; |
| 87 | int vdwioffset1; |
| 88 | __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
| 89 | int vdwioffset2; |
| 90 | __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
| 91 | int vdwioffset3; |
| 92 | __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3; |
| 93 | int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D; |
| 94 | __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
| 95 | int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D; |
| 96 | __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
| 97 | int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D; |
| 98 | __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
| 99 | int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D; |
| 100 | __m128 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3; |
| 101 | __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00; |
| 102 | __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11; |
| 103 | __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12; |
| 104 | __m128 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13; |
| 105 | __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21; |
| 106 | __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22; |
| 107 | __m128 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23; |
| 108 | __m128 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31; |
| 109 | __m128 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32; |
| 110 | __m128 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33; |
| 111 | __m128 velec,felec,velecsum,facel,crf,krf,krf2; |
| 112 | real *charge; |
| 113 | int nvdwtype; |
| 114 | __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6; |
| 115 | int *vdwtype; |
| 116 | real *vdwparam; |
| 117 | __m128 one_sixth = _mm_set1_ps(1.0/6.0); |
| 118 | __m128 one_twelfth = _mm_set1_ps(1.0/12.0); |
| 119 | __m128i vfitab; |
| 120 | __m128i ifour = _mm_set1_epi32(4); |
| 121 | __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF; |
| 122 | real *vftab; |
| 123 | __m128 dummy_mask,cutoff_mask; |
| 124 | __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) ); |
| 125 | __m128 one = _mm_set1_ps(1.0); |
| 126 | __m128 two = _mm_set1_ps(2.0); |
| 127 | x = xx[0]; |
| 128 | f = ff[0]; |
| 129 | |
| 130 | nri = nlist->nri; |
| 131 | iinr = nlist->iinr; |
| 132 | jindex = nlist->jindex; |
| 133 | jjnr = nlist->jjnr; |
| 134 | shiftidx = nlist->shift; |
| 135 | gid = nlist->gid; |
| 136 | shiftvec = fr->shift_vec[0]; |
| 137 | fshift = fr->fshift[0]; |
| 138 | facel = _mm_set1_ps(fr->epsfac); |
| 139 | charge = mdatoms->chargeA; |
| 140 | nvdwtype = fr->ntype; |
| 141 | vdwparam = fr->nbfp; |
| 142 | vdwtype = mdatoms->typeA; |
| 143 | |
| 144 | vftab = kernel_data->table_elec_vdw->data; |
| 145 | vftabscale = _mm_set1_ps(kernel_data->table_elec_vdw->scale); |
| 146 | |
| 147 | /* Setup water-specific parameters */ |
| 148 | inr = nlist->iinr[0]; |
| 149 | iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1])); |
| 150 | iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2])); |
| 151 | iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3])); |
| 152 | vdwioffset0 = 2*nvdwtype*vdwtype[inr+0]; |
| 153 | |
| 154 | jq1 = _mm_set1_ps(charge[inr+1]); |
| 155 | jq2 = _mm_set1_ps(charge[inr+2]); |
| 156 | jq3 = _mm_set1_ps(charge[inr+3]); |
| 157 | vdwjidx0A = 2*vdwtype[inr+0]; |
| 158 | c6_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]); |
| 159 | c12_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]); |
| 160 | qq11 = _mm_mul_ps(iq1,jq1); |
| 161 | qq12 = _mm_mul_ps(iq1,jq2); |
| 162 | qq13 = _mm_mul_ps(iq1,jq3); |
| 163 | qq21 = _mm_mul_ps(iq2,jq1); |
| 164 | qq22 = _mm_mul_ps(iq2,jq2); |
| 165 | qq23 = _mm_mul_ps(iq2,jq3); |
| 166 | qq31 = _mm_mul_ps(iq3,jq1); |
| 167 | qq32 = _mm_mul_ps(iq3,jq2); |
| 168 | qq33 = _mm_mul_ps(iq3,jq3); |
| 169 | |
| 170 | /* Avoid stupid compiler warnings */ |
| 171 | jnrA = jnrB = jnrC = jnrD = 0; |
| 172 | j_coord_offsetA = 0; |
| 173 | j_coord_offsetB = 0; |
| 174 | j_coord_offsetC = 0; |
| 175 | j_coord_offsetD = 0; |
| 176 | |
| 177 | outeriter = 0; |
| 178 | inneriter = 0; |
| 179 | |
| 180 | for(iidx=0;iidx<4*DIM3;iidx++) |
| 181 | { |
| 182 | scratch[iidx] = 0.0; |
| 183 | } |
| 184 | |
| 185 | /* Start outer loop over neighborlists */ |
| 186 | for(iidx=0; iidx<nri; iidx++) |
| 187 | { |
| 188 | /* Load shift vector for this list */ |
| 189 | i_shift_offset = DIM3*shiftidx[iidx]; |
| 190 | |
| 191 | /* Load limits for loop over neighbors */ |
| 192 | j_index_start = jindex[iidx]; |
| 193 | j_index_end = jindex[iidx+1]; |
| 194 | |
| 195 | /* Get outer coordinate index */ |
| 196 | inr = iinr[iidx]; |
| 197 | i_coord_offset = DIM3*inr; |
| 198 | |
| 199 | /* Load i particle coords and add shift vector */ |
| 200 | gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset, |
| 201 | &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3); |
| 202 | |
| 203 | fix0 = _mm_setzero_ps(); |
| 204 | fiy0 = _mm_setzero_ps(); |
| 205 | fiz0 = _mm_setzero_ps(); |
| 206 | fix1 = _mm_setzero_ps(); |
| 207 | fiy1 = _mm_setzero_ps(); |
| 208 | fiz1 = _mm_setzero_ps(); |
| 209 | fix2 = _mm_setzero_ps(); |
| 210 | fiy2 = _mm_setzero_ps(); |
| 211 | fiz2 = _mm_setzero_ps(); |
| 212 | fix3 = _mm_setzero_ps(); |
| 213 | fiy3 = _mm_setzero_ps(); |
| 214 | fiz3 = _mm_setzero_ps(); |
| 215 | |
| 216 | /* Reset potential sums */ |
| 217 | velecsum = _mm_setzero_ps(); |
| 218 | vvdwsum = _mm_setzero_ps(); |
| 219 | |
| 220 | /* Start inner kernel loop */ |
| 221 | for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4) |
| 222 | { |
| 223 | |
| 224 | /* Get j neighbor index, and coordinate index */ |
| 225 | jnrA = jjnr[jidx]; |
| 226 | jnrB = jjnr[jidx+1]; |
| 227 | jnrC = jjnr[jidx+2]; |
| 228 | jnrD = jjnr[jidx+3]; |
| 229 | j_coord_offsetA = DIM3*jnrA; |
| 230 | j_coord_offsetB = DIM3*jnrB; |
| 231 | j_coord_offsetC = DIM3*jnrC; |
| 232 | j_coord_offsetD = DIM3*jnrD; |
| 233 | |
| 234 | /* load j atom coordinates */ |
| 235 | gmx_mm_load_4rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
| 236 | x+j_coord_offsetC,x+j_coord_offsetD, |
| 237 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2, |
| 238 | &jy2,&jz2,&jx3,&jy3,&jz3); |
| 239 | |
| 240 | /* Calculate displacement vector */ |
| 241 | dx00 = _mm_sub_ps(ix0,jx0); |
| 242 | dy00 = _mm_sub_ps(iy0,jy0); |
| 243 | dz00 = _mm_sub_ps(iz0,jz0); |
| 244 | dx11 = _mm_sub_ps(ix1,jx1); |
| 245 | dy11 = _mm_sub_ps(iy1,jy1); |
| 246 | dz11 = _mm_sub_ps(iz1,jz1); |
| 247 | dx12 = _mm_sub_ps(ix1,jx2); |
| 248 | dy12 = _mm_sub_ps(iy1,jy2); |
| 249 | dz12 = _mm_sub_ps(iz1,jz2); |
| 250 | dx13 = _mm_sub_ps(ix1,jx3); |
| 251 | dy13 = _mm_sub_ps(iy1,jy3); |
| 252 | dz13 = _mm_sub_ps(iz1,jz3); |
| 253 | dx21 = _mm_sub_ps(ix2,jx1); |
| 254 | dy21 = _mm_sub_ps(iy2,jy1); |
| 255 | dz21 = _mm_sub_ps(iz2,jz1); |
| 256 | dx22 = _mm_sub_ps(ix2,jx2); |
| 257 | dy22 = _mm_sub_ps(iy2,jy2); |
| 258 | dz22 = _mm_sub_ps(iz2,jz2); |
| 259 | dx23 = _mm_sub_ps(ix2,jx3); |
| 260 | dy23 = _mm_sub_ps(iy2,jy3); |
| 261 | dz23 = _mm_sub_ps(iz2,jz3); |
| 262 | dx31 = _mm_sub_ps(ix3,jx1); |
| 263 | dy31 = _mm_sub_ps(iy3,jy1); |
| 264 | dz31 = _mm_sub_ps(iz3,jz1); |
| 265 | dx32 = _mm_sub_ps(ix3,jx2); |
| 266 | dy32 = _mm_sub_ps(iy3,jy2); |
| 267 | dz32 = _mm_sub_ps(iz3,jz2); |
| 268 | dx33 = _mm_sub_ps(ix3,jx3); |
| 269 | dy33 = _mm_sub_ps(iy3,jy3); |
| 270 | dz33 = _mm_sub_ps(iz3,jz3); |
| 271 | |
| 272 | /* Calculate squared distance and things based on it */ |
| 273 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
| 274 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
| 275 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
| 276 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
| 277 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
| 278 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
| 279 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
| 280 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
| 281 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
| 282 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
| 283 | |
| 284 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
| 285 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
| 286 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
| 287 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
| 288 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
| 289 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
| 290 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
| 291 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
| 292 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
| 293 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
| 294 | |
| 295 | fjx0 = _mm_setzero_ps(); |
| 296 | fjy0 = _mm_setzero_ps(); |
| 297 | fjz0 = _mm_setzero_ps(); |
| 298 | fjx1 = _mm_setzero_ps(); |
| 299 | fjy1 = _mm_setzero_ps(); |
| 300 | fjz1 = _mm_setzero_ps(); |
| 301 | fjx2 = _mm_setzero_ps(); |
| 302 | fjy2 = _mm_setzero_ps(); |
| 303 | fjz2 = _mm_setzero_ps(); |
| 304 | fjx3 = _mm_setzero_ps(); |
| 305 | fjy3 = _mm_setzero_ps(); |
| 306 | fjz3 = _mm_setzero_ps(); |
| 307 | |
| 308 | /************************** |
| 309 | * CALCULATE INTERACTIONS * |
| 310 | **************************/ |
| 311 | |
| 312 | r00 = _mm_mul_ps(rsq00,rinv00); |
| 313 | |
| 314 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 315 | rt = _mm_mul_ps(r00,vftabscale); |
| 316 | vfitab = _mm_cvttps_epi32(rt); |
| 317 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 318 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 319 | |
| 320 | /* CUBIC SPLINE TABLE DISPERSION */ |
| 321 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 322 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 323 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 324 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 325 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 326 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 327 | Heps = _mm_mul_ps(vfeps,H); |
| 328 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 329 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 330 | vvdw6 = _mm_mul_ps(c6_00,VV); |
| 331 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 332 | fvdw6 = _mm_mul_ps(c6_00,FF); |
| 333 | |
| 334 | /* CUBIC SPLINE TABLE REPULSION */ |
| 335 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 336 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 337 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 338 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 339 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 340 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 341 | Heps = _mm_mul_ps(vfeps,H); |
| 342 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 343 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 344 | vvdw12 = _mm_mul_ps(c12_00,VV); |
| 345 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 346 | fvdw12 = _mm_mul_ps(c12_00,FF); |
| 347 | vvdw = _mm_add_ps(vvdw12,vvdw6); |
| 348 | fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00))); |
| 349 | |
| 350 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 351 | vvdwsum = _mm_add_ps(vvdwsum,vvdw); |
| 352 | |
| 353 | fscal = fvdw; |
| 354 | |
| 355 | /* Calculate temporary vectorial force */ |
| 356 | tx = _mm_mul_ps(fscal,dx00); |
| 357 | ty = _mm_mul_ps(fscal,dy00); |
| 358 | tz = _mm_mul_ps(fscal,dz00); |
| 359 | |
| 360 | /* Update vectorial force */ |
| 361 | fix0 = _mm_add_ps(fix0,tx); |
| 362 | fiy0 = _mm_add_ps(fiy0,ty); |
| 363 | fiz0 = _mm_add_ps(fiz0,tz); |
| 364 | |
| 365 | fjx0 = _mm_add_ps(fjx0,tx); |
| 366 | fjy0 = _mm_add_ps(fjy0,ty); |
| 367 | fjz0 = _mm_add_ps(fjz0,tz); |
| 368 | |
| 369 | /************************** |
| 370 | * CALCULATE INTERACTIONS * |
| 371 | **************************/ |
| 372 | |
| 373 | r11 = _mm_mul_ps(rsq11,rinv11); |
| 374 | |
| 375 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 376 | rt = _mm_mul_ps(r11,vftabscale); |
| 377 | vfitab = _mm_cvttps_epi32(rt); |
| 378 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 379 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 380 | |
| 381 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 382 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 383 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 384 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 385 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 386 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 387 | Heps = _mm_mul_ps(vfeps,H); |
| 388 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 389 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 390 | velec = _mm_mul_ps(qq11,VV); |
| 391 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 392 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
| 393 | |
| 394 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 395 | velecsum = _mm_add_ps(velecsum,velec); |
| 396 | |
| 397 | fscal = felec; |
| 398 | |
| 399 | /* Calculate temporary vectorial force */ |
| 400 | tx = _mm_mul_ps(fscal,dx11); |
| 401 | ty = _mm_mul_ps(fscal,dy11); |
| 402 | tz = _mm_mul_ps(fscal,dz11); |
| 403 | |
| 404 | /* Update vectorial force */ |
| 405 | fix1 = _mm_add_ps(fix1,tx); |
| 406 | fiy1 = _mm_add_ps(fiy1,ty); |
| 407 | fiz1 = _mm_add_ps(fiz1,tz); |
| 408 | |
| 409 | fjx1 = _mm_add_ps(fjx1,tx); |
| 410 | fjy1 = _mm_add_ps(fjy1,ty); |
| 411 | fjz1 = _mm_add_ps(fjz1,tz); |
| 412 | |
| 413 | /************************** |
| 414 | * CALCULATE INTERACTIONS * |
| 415 | **************************/ |
| 416 | |
| 417 | r12 = _mm_mul_ps(rsq12,rinv12); |
| 418 | |
| 419 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 420 | rt = _mm_mul_ps(r12,vftabscale); |
| 421 | vfitab = _mm_cvttps_epi32(rt); |
| 422 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 423 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 424 | |
| 425 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 426 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 427 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 428 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 429 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 430 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 431 | Heps = _mm_mul_ps(vfeps,H); |
| 432 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 433 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 434 | velec = _mm_mul_ps(qq12,VV); |
| 435 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 436 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
| 437 | |
| 438 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 439 | velecsum = _mm_add_ps(velecsum,velec); |
| 440 | |
| 441 | fscal = felec; |
| 442 | |
| 443 | /* Calculate temporary vectorial force */ |
| 444 | tx = _mm_mul_ps(fscal,dx12); |
| 445 | ty = _mm_mul_ps(fscal,dy12); |
| 446 | tz = _mm_mul_ps(fscal,dz12); |
| 447 | |
| 448 | /* Update vectorial force */ |
| 449 | fix1 = _mm_add_ps(fix1,tx); |
| 450 | fiy1 = _mm_add_ps(fiy1,ty); |
| 451 | fiz1 = _mm_add_ps(fiz1,tz); |
| 452 | |
| 453 | fjx2 = _mm_add_ps(fjx2,tx); |
| 454 | fjy2 = _mm_add_ps(fjy2,ty); |
| 455 | fjz2 = _mm_add_ps(fjz2,tz); |
| 456 | |
| 457 | /************************** |
| 458 | * CALCULATE INTERACTIONS * |
| 459 | **************************/ |
| 460 | |
| 461 | r13 = _mm_mul_ps(rsq13,rinv13); |
| 462 | |
| 463 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 464 | rt = _mm_mul_ps(r13,vftabscale); |
| 465 | vfitab = _mm_cvttps_epi32(rt); |
| 466 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 467 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 468 | |
| 469 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 470 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 471 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 472 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 473 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 474 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 475 | Heps = _mm_mul_ps(vfeps,H); |
| 476 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 477 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 478 | velec = _mm_mul_ps(qq13,VV); |
| 479 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 480 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
| 481 | |
| 482 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 483 | velecsum = _mm_add_ps(velecsum,velec); |
| 484 | |
| 485 | fscal = felec; |
| 486 | |
| 487 | /* Calculate temporary vectorial force */ |
| 488 | tx = _mm_mul_ps(fscal,dx13); |
| 489 | ty = _mm_mul_ps(fscal,dy13); |
| 490 | tz = _mm_mul_ps(fscal,dz13); |
| 491 | |
| 492 | /* Update vectorial force */ |
| 493 | fix1 = _mm_add_ps(fix1,tx); |
| 494 | fiy1 = _mm_add_ps(fiy1,ty); |
| 495 | fiz1 = _mm_add_ps(fiz1,tz); |
| 496 | |
| 497 | fjx3 = _mm_add_ps(fjx3,tx); |
| 498 | fjy3 = _mm_add_ps(fjy3,ty); |
| 499 | fjz3 = _mm_add_ps(fjz3,tz); |
| 500 | |
| 501 | /************************** |
| 502 | * CALCULATE INTERACTIONS * |
| 503 | **************************/ |
| 504 | |
| 505 | r21 = _mm_mul_ps(rsq21,rinv21); |
| 506 | |
| 507 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 508 | rt = _mm_mul_ps(r21,vftabscale); |
| 509 | vfitab = _mm_cvttps_epi32(rt); |
| 510 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 511 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 512 | |
| 513 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 514 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 515 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 516 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 517 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 518 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 519 | Heps = _mm_mul_ps(vfeps,H); |
| 520 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 521 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 522 | velec = _mm_mul_ps(qq21,VV); |
| 523 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 524 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
| 525 | |
| 526 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 527 | velecsum = _mm_add_ps(velecsum,velec); |
| 528 | |
| 529 | fscal = felec; |
| 530 | |
| 531 | /* Calculate temporary vectorial force */ |
| 532 | tx = _mm_mul_ps(fscal,dx21); |
| 533 | ty = _mm_mul_ps(fscal,dy21); |
| 534 | tz = _mm_mul_ps(fscal,dz21); |
| 535 | |
| 536 | /* Update vectorial force */ |
| 537 | fix2 = _mm_add_ps(fix2,tx); |
| 538 | fiy2 = _mm_add_ps(fiy2,ty); |
| 539 | fiz2 = _mm_add_ps(fiz2,tz); |
| 540 | |
| 541 | fjx1 = _mm_add_ps(fjx1,tx); |
| 542 | fjy1 = _mm_add_ps(fjy1,ty); |
| 543 | fjz1 = _mm_add_ps(fjz1,tz); |
| 544 | |
| 545 | /************************** |
| 546 | * CALCULATE INTERACTIONS * |
| 547 | **************************/ |
| 548 | |
| 549 | r22 = _mm_mul_ps(rsq22,rinv22); |
| 550 | |
| 551 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 552 | rt = _mm_mul_ps(r22,vftabscale); |
| 553 | vfitab = _mm_cvttps_epi32(rt); |
| 554 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 555 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 556 | |
| 557 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 558 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 559 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 560 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 561 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 562 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 563 | Heps = _mm_mul_ps(vfeps,H); |
| 564 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 565 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 566 | velec = _mm_mul_ps(qq22,VV); |
| 567 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 568 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
| 569 | |
| 570 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 571 | velecsum = _mm_add_ps(velecsum,velec); |
| 572 | |
| 573 | fscal = felec; |
| 574 | |
| 575 | /* Calculate temporary vectorial force */ |
| 576 | tx = _mm_mul_ps(fscal,dx22); |
| 577 | ty = _mm_mul_ps(fscal,dy22); |
| 578 | tz = _mm_mul_ps(fscal,dz22); |
| 579 | |
| 580 | /* Update vectorial force */ |
| 581 | fix2 = _mm_add_ps(fix2,tx); |
| 582 | fiy2 = _mm_add_ps(fiy2,ty); |
| 583 | fiz2 = _mm_add_ps(fiz2,tz); |
| 584 | |
| 585 | fjx2 = _mm_add_ps(fjx2,tx); |
| 586 | fjy2 = _mm_add_ps(fjy2,ty); |
| 587 | fjz2 = _mm_add_ps(fjz2,tz); |
| 588 | |
| 589 | /************************** |
| 590 | * CALCULATE INTERACTIONS * |
| 591 | **************************/ |
| 592 | |
| 593 | r23 = _mm_mul_ps(rsq23,rinv23); |
| 594 | |
| 595 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 596 | rt = _mm_mul_ps(r23,vftabscale); |
| 597 | vfitab = _mm_cvttps_epi32(rt); |
| 598 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 599 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 600 | |
| 601 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 602 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 603 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 604 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 605 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 606 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 607 | Heps = _mm_mul_ps(vfeps,H); |
| 608 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 609 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 610 | velec = _mm_mul_ps(qq23,VV); |
| 611 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 612 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
| 613 | |
| 614 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 615 | velecsum = _mm_add_ps(velecsum,velec); |
| 616 | |
| 617 | fscal = felec; |
| 618 | |
| 619 | /* Calculate temporary vectorial force */ |
| 620 | tx = _mm_mul_ps(fscal,dx23); |
| 621 | ty = _mm_mul_ps(fscal,dy23); |
| 622 | tz = _mm_mul_ps(fscal,dz23); |
| 623 | |
| 624 | /* Update vectorial force */ |
| 625 | fix2 = _mm_add_ps(fix2,tx); |
| 626 | fiy2 = _mm_add_ps(fiy2,ty); |
| 627 | fiz2 = _mm_add_ps(fiz2,tz); |
| 628 | |
| 629 | fjx3 = _mm_add_ps(fjx3,tx); |
| 630 | fjy3 = _mm_add_ps(fjy3,ty); |
| 631 | fjz3 = _mm_add_ps(fjz3,tz); |
| 632 | |
| 633 | /************************** |
| 634 | * CALCULATE INTERACTIONS * |
| 635 | **************************/ |
| 636 | |
| 637 | r31 = _mm_mul_ps(rsq31,rinv31); |
| 638 | |
| 639 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 640 | rt = _mm_mul_ps(r31,vftabscale); |
| 641 | vfitab = _mm_cvttps_epi32(rt); |
| 642 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 643 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 644 | |
| 645 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 646 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 647 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 648 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 649 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 650 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 651 | Heps = _mm_mul_ps(vfeps,H); |
| 652 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 653 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 654 | velec = _mm_mul_ps(qq31,VV); |
| 655 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 656 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
| 657 | |
| 658 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 659 | velecsum = _mm_add_ps(velecsum,velec); |
| 660 | |
| 661 | fscal = felec; |
| 662 | |
| 663 | /* Calculate temporary vectorial force */ |
| 664 | tx = _mm_mul_ps(fscal,dx31); |
| 665 | ty = _mm_mul_ps(fscal,dy31); |
| 666 | tz = _mm_mul_ps(fscal,dz31); |
| 667 | |
| 668 | /* Update vectorial force */ |
| 669 | fix3 = _mm_add_ps(fix3,tx); |
| 670 | fiy3 = _mm_add_ps(fiy3,ty); |
| 671 | fiz3 = _mm_add_ps(fiz3,tz); |
| 672 | |
| 673 | fjx1 = _mm_add_ps(fjx1,tx); |
| 674 | fjy1 = _mm_add_ps(fjy1,ty); |
| 675 | fjz1 = _mm_add_ps(fjz1,tz); |
| 676 | |
| 677 | /************************** |
| 678 | * CALCULATE INTERACTIONS * |
| 679 | **************************/ |
| 680 | |
| 681 | r32 = _mm_mul_ps(rsq32,rinv32); |
| 682 | |
| 683 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 684 | rt = _mm_mul_ps(r32,vftabscale); |
| 685 | vfitab = _mm_cvttps_epi32(rt); |
| 686 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 687 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 688 | |
| 689 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 690 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 691 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 692 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 693 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 694 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 695 | Heps = _mm_mul_ps(vfeps,H); |
| 696 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 697 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 698 | velec = _mm_mul_ps(qq32,VV); |
| 699 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 700 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
| 701 | |
| 702 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 703 | velecsum = _mm_add_ps(velecsum,velec); |
| 704 | |
| 705 | fscal = felec; |
| 706 | |
| 707 | /* Calculate temporary vectorial force */ |
| 708 | tx = _mm_mul_ps(fscal,dx32); |
| 709 | ty = _mm_mul_ps(fscal,dy32); |
| 710 | tz = _mm_mul_ps(fscal,dz32); |
| 711 | |
| 712 | /* Update vectorial force */ |
| 713 | fix3 = _mm_add_ps(fix3,tx); |
| 714 | fiy3 = _mm_add_ps(fiy3,ty); |
| 715 | fiz3 = _mm_add_ps(fiz3,tz); |
| 716 | |
| 717 | fjx2 = _mm_add_ps(fjx2,tx); |
| 718 | fjy2 = _mm_add_ps(fjy2,ty); |
| 719 | fjz2 = _mm_add_ps(fjz2,tz); |
| 720 | |
| 721 | /************************** |
| 722 | * CALCULATE INTERACTIONS * |
| 723 | **************************/ |
| 724 | |
| 725 | r33 = _mm_mul_ps(rsq33,rinv33); |
| 726 | |
| 727 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 728 | rt = _mm_mul_ps(r33,vftabscale); |
| 729 | vfitab = _mm_cvttps_epi32(rt); |
| 730 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 731 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 732 | |
| 733 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 734 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 735 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 736 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 737 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 738 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 739 | Heps = _mm_mul_ps(vfeps,H); |
| 740 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 741 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 742 | velec = _mm_mul_ps(qq33,VV); |
| 743 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 744 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
| 745 | |
| 746 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 747 | velecsum = _mm_add_ps(velecsum,velec); |
| 748 | |
| 749 | fscal = felec; |
| 750 | |
| 751 | /* Calculate temporary vectorial force */ |
| 752 | tx = _mm_mul_ps(fscal,dx33); |
| 753 | ty = _mm_mul_ps(fscal,dy33); |
| 754 | tz = _mm_mul_ps(fscal,dz33); |
| 755 | |
| 756 | /* Update vectorial force */ |
| 757 | fix3 = _mm_add_ps(fix3,tx); |
| 758 | fiy3 = _mm_add_ps(fiy3,ty); |
| 759 | fiz3 = _mm_add_ps(fiz3,tz); |
| 760 | |
| 761 | fjx3 = _mm_add_ps(fjx3,tx); |
| 762 | fjy3 = _mm_add_ps(fjy3,ty); |
| 763 | fjz3 = _mm_add_ps(fjz3,tz); |
| 764 | |
| 765 | fjptrA = f+j_coord_offsetA; |
| 766 | fjptrB = f+j_coord_offsetB; |
| 767 | fjptrC = f+j_coord_offsetC; |
| 768 | fjptrD = f+j_coord_offsetD; |
| 769 | |
| 770 | gmx_mm_decrement_4rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
| 771 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1, |
| 772 | fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
| 773 | |
| 774 | /* Inner loop uses 446 flops */ |
| 775 | } |
| 776 | |
| 777 | if(jidx<j_index_end) |
| 778 | { |
| 779 | |
| 780 | /* Get j neighbor index, and coordinate index */ |
| 781 | jnrlistA = jjnr[jidx]; |
| 782 | jnrlistB = jjnr[jidx+1]; |
| 783 | jnrlistC = jjnr[jidx+2]; |
| 784 | jnrlistD = jjnr[jidx+3]; |
| 785 | /* Sign of each element will be negative for non-real atoms. |
| 786 | * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones, |
| 787 | * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries. |
| 788 | */ |
| 789 | dummy_mask = gmx_mm_castsi128_ps_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())); |
| 790 | jnrA = (jnrlistA>=0) ? jnrlistA : 0; |
| 791 | jnrB = (jnrlistB>=0) ? jnrlistB : 0; |
| 792 | jnrC = (jnrlistC>=0) ? jnrlistC : 0; |
| 793 | jnrD = (jnrlistD>=0) ? jnrlistD : 0; |
| 794 | j_coord_offsetA = DIM3*jnrA; |
| 795 | j_coord_offsetB = DIM3*jnrB; |
| 796 | j_coord_offsetC = DIM3*jnrC; |
| 797 | j_coord_offsetD = DIM3*jnrD; |
| 798 | |
| 799 | /* load j atom coordinates */ |
| 800 | gmx_mm_load_4rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
| 801 | x+j_coord_offsetC,x+j_coord_offsetD, |
| 802 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2, |
| 803 | &jy2,&jz2,&jx3,&jy3,&jz3); |
| 804 | |
| 805 | /* Calculate displacement vector */ |
| 806 | dx00 = _mm_sub_ps(ix0,jx0); |
| 807 | dy00 = _mm_sub_ps(iy0,jy0); |
| 808 | dz00 = _mm_sub_ps(iz0,jz0); |
| 809 | dx11 = _mm_sub_ps(ix1,jx1); |
| 810 | dy11 = _mm_sub_ps(iy1,jy1); |
| 811 | dz11 = _mm_sub_ps(iz1,jz1); |
| 812 | dx12 = _mm_sub_ps(ix1,jx2); |
| 813 | dy12 = _mm_sub_ps(iy1,jy2); |
| 814 | dz12 = _mm_sub_ps(iz1,jz2); |
| 815 | dx13 = _mm_sub_ps(ix1,jx3); |
| 816 | dy13 = _mm_sub_ps(iy1,jy3); |
| 817 | dz13 = _mm_sub_ps(iz1,jz3); |
| 818 | dx21 = _mm_sub_ps(ix2,jx1); |
| 819 | dy21 = _mm_sub_ps(iy2,jy1); |
| 820 | dz21 = _mm_sub_ps(iz2,jz1); |
| 821 | dx22 = _mm_sub_ps(ix2,jx2); |
| 822 | dy22 = _mm_sub_ps(iy2,jy2); |
| 823 | dz22 = _mm_sub_ps(iz2,jz2); |
| 824 | dx23 = _mm_sub_ps(ix2,jx3); |
| 825 | dy23 = _mm_sub_ps(iy2,jy3); |
| 826 | dz23 = _mm_sub_ps(iz2,jz3); |
| 827 | dx31 = _mm_sub_ps(ix3,jx1); |
| 828 | dy31 = _mm_sub_ps(iy3,jy1); |
| 829 | dz31 = _mm_sub_ps(iz3,jz1); |
| 830 | dx32 = _mm_sub_ps(ix3,jx2); |
| 831 | dy32 = _mm_sub_ps(iy3,jy2); |
| 832 | dz32 = _mm_sub_ps(iz3,jz2); |
| 833 | dx33 = _mm_sub_ps(ix3,jx3); |
| 834 | dy33 = _mm_sub_ps(iy3,jy3); |
| 835 | dz33 = _mm_sub_ps(iz3,jz3); |
| 836 | |
| 837 | /* Calculate squared distance and things based on it */ |
| 838 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
| 839 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
| 840 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
| 841 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
| 842 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
| 843 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
| 844 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
| 845 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
| 846 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
| 847 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
| 848 | |
| 849 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
| 850 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
| 851 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
| 852 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
| 853 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
| 854 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
| 855 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
| 856 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
| 857 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
| 858 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
| 859 | |
| 860 | fjx0 = _mm_setzero_ps(); |
| 861 | fjy0 = _mm_setzero_ps(); |
| 862 | fjz0 = _mm_setzero_ps(); |
| 863 | fjx1 = _mm_setzero_ps(); |
| 864 | fjy1 = _mm_setzero_ps(); |
| 865 | fjz1 = _mm_setzero_ps(); |
| 866 | fjx2 = _mm_setzero_ps(); |
| 867 | fjy2 = _mm_setzero_ps(); |
| 868 | fjz2 = _mm_setzero_ps(); |
| 869 | fjx3 = _mm_setzero_ps(); |
| 870 | fjy3 = _mm_setzero_ps(); |
| 871 | fjz3 = _mm_setzero_ps(); |
| 872 | |
| 873 | /************************** |
| 874 | * CALCULATE INTERACTIONS * |
| 875 | **************************/ |
| 876 | |
| 877 | r00 = _mm_mul_ps(rsq00,rinv00); |
| 878 | r00 = _mm_andnot_ps(dummy_mask,r00); |
| 879 | |
| 880 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 881 | rt = _mm_mul_ps(r00,vftabscale); |
| 882 | vfitab = _mm_cvttps_epi32(rt); |
| 883 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 884 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 885 | |
| 886 | /* CUBIC SPLINE TABLE DISPERSION */ |
| 887 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 888 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 889 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 890 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 891 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 892 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 893 | Heps = _mm_mul_ps(vfeps,H); |
| 894 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 895 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 896 | vvdw6 = _mm_mul_ps(c6_00,VV); |
| 897 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 898 | fvdw6 = _mm_mul_ps(c6_00,FF); |
| 899 | |
| 900 | /* CUBIC SPLINE TABLE REPULSION */ |
| 901 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 902 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 903 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 904 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 905 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 906 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 907 | Heps = _mm_mul_ps(vfeps,H); |
| 908 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 909 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 910 | vvdw12 = _mm_mul_ps(c12_00,VV); |
| 911 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 912 | fvdw12 = _mm_mul_ps(c12_00,FF); |
| 913 | vvdw = _mm_add_ps(vvdw12,vvdw6); |
| 914 | fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00))); |
| 915 | |
| 916 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 917 | vvdw = _mm_andnot_ps(dummy_mask,vvdw); |
| 918 | vvdwsum = _mm_add_ps(vvdwsum,vvdw); |
| 919 | |
| 920 | fscal = fvdw; |
| 921 | |
| 922 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 923 | |
| 924 | /* Calculate temporary vectorial force */ |
| 925 | tx = _mm_mul_ps(fscal,dx00); |
| 926 | ty = _mm_mul_ps(fscal,dy00); |
| 927 | tz = _mm_mul_ps(fscal,dz00); |
| 928 | |
| 929 | /* Update vectorial force */ |
| 930 | fix0 = _mm_add_ps(fix0,tx); |
| 931 | fiy0 = _mm_add_ps(fiy0,ty); |
| 932 | fiz0 = _mm_add_ps(fiz0,tz); |
| 933 | |
| 934 | fjx0 = _mm_add_ps(fjx0,tx); |
| 935 | fjy0 = _mm_add_ps(fjy0,ty); |
| 936 | fjz0 = _mm_add_ps(fjz0,tz); |
| 937 | |
| 938 | /************************** |
| 939 | * CALCULATE INTERACTIONS * |
| 940 | **************************/ |
| 941 | |
| 942 | r11 = _mm_mul_ps(rsq11,rinv11); |
| 943 | r11 = _mm_andnot_ps(dummy_mask,r11); |
| 944 | |
| 945 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 946 | rt = _mm_mul_ps(r11,vftabscale); |
| 947 | vfitab = _mm_cvttps_epi32(rt); |
| 948 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 949 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 950 | |
| 951 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 952 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 953 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 954 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 955 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 956 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 957 | Heps = _mm_mul_ps(vfeps,H); |
| 958 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 959 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 960 | velec = _mm_mul_ps(qq11,VV); |
| 961 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 962 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
| 963 | |
| 964 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 965 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 966 | velecsum = _mm_add_ps(velecsum,velec); |
| 967 | |
| 968 | fscal = felec; |
| 969 | |
| 970 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 971 | |
| 972 | /* Calculate temporary vectorial force */ |
| 973 | tx = _mm_mul_ps(fscal,dx11); |
| 974 | ty = _mm_mul_ps(fscal,dy11); |
| 975 | tz = _mm_mul_ps(fscal,dz11); |
| 976 | |
| 977 | /* Update vectorial force */ |
| 978 | fix1 = _mm_add_ps(fix1,tx); |
| 979 | fiy1 = _mm_add_ps(fiy1,ty); |
| 980 | fiz1 = _mm_add_ps(fiz1,tz); |
| 981 | |
| 982 | fjx1 = _mm_add_ps(fjx1,tx); |
| 983 | fjy1 = _mm_add_ps(fjy1,ty); |
| 984 | fjz1 = _mm_add_ps(fjz1,tz); |
| 985 | |
| 986 | /************************** |
| 987 | * CALCULATE INTERACTIONS * |
| 988 | **************************/ |
| 989 | |
| 990 | r12 = _mm_mul_ps(rsq12,rinv12); |
| 991 | r12 = _mm_andnot_ps(dummy_mask,r12); |
| 992 | |
| 993 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 994 | rt = _mm_mul_ps(r12,vftabscale); |
| 995 | vfitab = _mm_cvttps_epi32(rt); |
| 996 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 997 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 998 | |
| 999 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1000 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1001 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1002 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1003 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1004 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1005 | Heps = _mm_mul_ps(vfeps,H); |
| 1006 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1007 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1008 | velec = _mm_mul_ps(qq12,VV); |
| 1009 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1010 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
| 1011 | |
| 1012 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1013 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1014 | velecsum = _mm_add_ps(velecsum,velec); |
| 1015 | |
| 1016 | fscal = felec; |
| 1017 | |
| 1018 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1019 | |
| 1020 | /* Calculate temporary vectorial force */ |
| 1021 | tx = _mm_mul_ps(fscal,dx12); |
| 1022 | ty = _mm_mul_ps(fscal,dy12); |
| 1023 | tz = _mm_mul_ps(fscal,dz12); |
| 1024 | |
| 1025 | /* Update vectorial force */ |
| 1026 | fix1 = _mm_add_ps(fix1,tx); |
| 1027 | fiy1 = _mm_add_ps(fiy1,ty); |
| 1028 | fiz1 = _mm_add_ps(fiz1,tz); |
| 1029 | |
| 1030 | fjx2 = _mm_add_ps(fjx2,tx); |
| 1031 | fjy2 = _mm_add_ps(fjy2,ty); |
| 1032 | fjz2 = _mm_add_ps(fjz2,tz); |
| 1033 | |
| 1034 | /************************** |
| 1035 | * CALCULATE INTERACTIONS * |
| 1036 | **************************/ |
| 1037 | |
| 1038 | r13 = _mm_mul_ps(rsq13,rinv13); |
| 1039 | r13 = _mm_andnot_ps(dummy_mask,r13); |
| 1040 | |
| 1041 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1042 | rt = _mm_mul_ps(r13,vftabscale); |
| 1043 | vfitab = _mm_cvttps_epi32(rt); |
| 1044 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1045 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1046 | |
| 1047 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1048 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1049 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1050 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1051 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1052 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1053 | Heps = _mm_mul_ps(vfeps,H); |
| 1054 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1055 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1056 | velec = _mm_mul_ps(qq13,VV); |
| 1057 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1058 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
| 1059 | |
| 1060 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1061 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1062 | velecsum = _mm_add_ps(velecsum,velec); |
| 1063 | |
| 1064 | fscal = felec; |
| 1065 | |
| 1066 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1067 | |
| 1068 | /* Calculate temporary vectorial force */ |
| 1069 | tx = _mm_mul_ps(fscal,dx13); |
| 1070 | ty = _mm_mul_ps(fscal,dy13); |
| 1071 | tz = _mm_mul_ps(fscal,dz13); |
| 1072 | |
| 1073 | /* Update vectorial force */ |
| 1074 | fix1 = _mm_add_ps(fix1,tx); |
| 1075 | fiy1 = _mm_add_ps(fiy1,ty); |
| 1076 | fiz1 = _mm_add_ps(fiz1,tz); |
| 1077 | |
| 1078 | fjx3 = _mm_add_ps(fjx3,tx); |
| 1079 | fjy3 = _mm_add_ps(fjy3,ty); |
| 1080 | fjz3 = _mm_add_ps(fjz3,tz); |
| 1081 | |
| 1082 | /************************** |
| 1083 | * CALCULATE INTERACTIONS * |
| 1084 | **************************/ |
| 1085 | |
| 1086 | r21 = _mm_mul_ps(rsq21,rinv21); |
| 1087 | r21 = _mm_andnot_ps(dummy_mask,r21); |
| 1088 | |
| 1089 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1090 | rt = _mm_mul_ps(r21,vftabscale); |
| 1091 | vfitab = _mm_cvttps_epi32(rt); |
| 1092 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1093 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1094 | |
| 1095 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1096 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1097 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1098 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1099 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1100 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1101 | Heps = _mm_mul_ps(vfeps,H); |
| 1102 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1103 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1104 | velec = _mm_mul_ps(qq21,VV); |
| 1105 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1106 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
| 1107 | |
| 1108 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1109 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1110 | velecsum = _mm_add_ps(velecsum,velec); |
| 1111 | |
| 1112 | fscal = felec; |
| 1113 | |
| 1114 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1115 | |
| 1116 | /* Calculate temporary vectorial force */ |
| 1117 | tx = _mm_mul_ps(fscal,dx21); |
| 1118 | ty = _mm_mul_ps(fscal,dy21); |
| 1119 | tz = _mm_mul_ps(fscal,dz21); |
| 1120 | |
| 1121 | /* Update vectorial force */ |
| 1122 | fix2 = _mm_add_ps(fix2,tx); |
| 1123 | fiy2 = _mm_add_ps(fiy2,ty); |
| 1124 | fiz2 = _mm_add_ps(fiz2,tz); |
| 1125 | |
| 1126 | fjx1 = _mm_add_ps(fjx1,tx); |
| 1127 | fjy1 = _mm_add_ps(fjy1,ty); |
| 1128 | fjz1 = _mm_add_ps(fjz1,tz); |
| 1129 | |
| 1130 | /************************** |
| 1131 | * CALCULATE INTERACTIONS * |
| 1132 | **************************/ |
| 1133 | |
| 1134 | r22 = _mm_mul_ps(rsq22,rinv22); |
| 1135 | r22 = _mm_andnot_ps(dummy_mask,r22); |
| 1136 | |
| 1137 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1138 | rt = _mm_mul_ps(r22,vftabscale); |
| 1139 | vfitab = _mm_cvttps_epi32(rt); |
| 1140 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1141 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1142 | |
| 1143 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1144 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1145 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1146 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1147 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1148 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1149 | Heps = _mm_mul_ps(vfeps,H); |
| 1150 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1151 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1152 | velec = _mm_mul_ps(qq22,VV); |
| 1153 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1154 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
| 1155 | |
| 1156 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1157 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1158 | velecsum = _mm_add_ps(velecsum,velec); |
| 1159 | |
| 1160 | fscal = felec; |
| 1161 | |
| 1162 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1163 | |
| 1164 | /* Calculate temporary vectorial force */ |
| 1165 | tx = _mm_mul_ps(fscal,dx22); |
| 1166 | ty = _mm_mul_ps(fscal,dy22); |
| 1167 | tz = _mm_mul_ps(fscal,dz22); |
| 1168 | |
| 1169 | /* Update vectorial force */ |
| 1170 | fix2 = _mm_add_ps(fix2,tx); |
| 1171 | fiy2 = _mm_add_ps(fiy2,ty); |
| 1172 | fiz2 = _mm_add_ps(fiz2,tz); |
| 1173 | |
| 1174 | fjx2 = _mm_add_ps(fjx2,tx); |
| 1175 | fjy2 = _mm_add_ps(fjy2,ty); |
| 1176 | fjz2 = _mm_add_ps(fjz2,tz); |
| 1177 | |
| 1178 | /************************** |
| 1179 | * CALCULATE INTERACTIONS * |
| 1180 | **************************/ |
| 1181 | |
| 1182 | r23 = _mm_mul_ps(rsq23,rinv23); |
| 1183 | r23 = _mm_andnot_ps(dummy_mask,r23); |
| 1184 | |
| 1185 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1186 | rt = _mm_mul_ps(r23,vftabscale); |
| 1187 | vfitab = _mm_cvttps_epi32(rt); |
| 1188 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1189 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1190 | |
| 1191 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1192 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1193 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1194 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1195 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1196 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1197 | Heps = _mm_mul_ps(vfeps,H); |
| 1198 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1199 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1200 | velec = _mm_mul_ps(qq23,VV); |
| 1201 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1202 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
| 1203 | |
| 1204 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1205 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1206 | velecsum = _mm_add_ps(velecsum,velec); |
| 1207 | |
| 1208 | fscal = felec; |
| 1209 | |
| 1210 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1211 | |
| 1212 | /* Calculate temporary vectorial force */ |
| 1213 | tx = _mm_mul_ps(fscal,dx23); |
| 1214 | ty = _mm_mul_ps(fscal,dy23); |
| 1215 | tz = _mm_mul_ps(fscal,dz23); |
| 1216 | |
| 1217 | /* Update vectorial force */ |
| 1218 | fix2 = _mm_add_ps(fix2,tx); |
| 1219 | fiy2 = _mm_add_ps(fiy2,ty); |
| 1220 | fiz2 = _mm_add_ps(fiz2,tz); |
| 1221 | |
| 1222 | fjx3 = _mm_add_ps(fjx3,tx); |
| 1223 | fjy3 = _mm_add_ps(fjy3,ty); |
| 1224 | fjz3 = _mm_add_ps(fjz3,tz); |
| 1225 | |
| 1226 | /************************** |
| 1227 | * CALCULATE INTERACTIONS * |
| 1228 | **************************/ |
| 1229 | |
| 1230 | r31 = _mm_mul_ps(rsq31,rinv31); |
| 1231 | r31 = _mm_andnot_ps(dummy_mask,r31); |
| 1232 | |
| 1233 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1234 | rt = _mm_mul_ps(r31,vftabscale); |
| 1235 | vfitab = _mm_cvttps_epi32(rt); |
| 1236 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1237 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1238 | |
| 1239 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1240 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1241 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1242 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1243 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1244 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1245 | Heps = _mm_mul_ps(vfeps,H); |
| 1246 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1247 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1248 | velec = _mm_mul_ps(qq31,VV); |
| 1249 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1250 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
| 1251 | |
| 1252 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1253 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1254 | velecsum = _mm_add_ps(velecsum,velec); |
| 1255 | |
| 1256 | fscal = felec; |
| 1257 | |
| 1258 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1259 | |
| 1260 | /* Calculate temporary vectorial force */ |
| 1261 | tx = _mm_mul_ps(fscal,dx31); |
| 1262 | ty = _mm_mul_ps(fscal,dy31); |
| 1263 | tz = _mm_mul_ps(fscal,dz31); |
| 1264 | |
| 1265 | /* Update vectorial force */ |
| 1266 | fix3 = _mm_add_ps(fix3,tx); |
| 1267 | fiy3 = _mm_add_ps(fiy3,ty); |
| 1268 | fiz3 = _mm_add_ps(fiz3,tz); |
| 1269 | |
| 1270 | fjx1 = _mm_add_ps(fjx1,tx); |
| 1271 | fjy1 = _mm_add_ps(fjy1,ty); |
| 1272 | fjz1 = _mm_add_ps(fjz1,tz); |
| 1273 | |
| 1274 | /************************** |
| 1275 | * CALCULATE INTERACTIONS * |
| 1276 | **************************/ |
| 1277 | |
| 1278 | r32 = _mm_mul_ps(rsq32,rinv32); |
| 1279 | r32 = _mm_andnot_ps(dummy_mask,r32); |
| 1280 | |
| 1281 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1282 | rt = _mm_mul_ps(r32,vftabscale); |
| 1283 | vfitab = _mm_cvttps_epi32(rt); |
| 1284 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1285 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1286 | |
| 1287 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1288 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1289 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1290 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1291 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1292 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1293 | Heps = _mm_mul_ps(vfeps,H); |
| 1294 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1295 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1296 | velec = _mm_mul_ps(qq32,VV); |
| 1297 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1298 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
| 1299 | |
| 1300 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1301 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1302 | velecsum = _mm_add_ps(velecsum,velec); |
| 1303 | |
| 1304 | fscal = felec; |
| 1305 | |
| 1306 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1307 | |
| 1308 | /* Calculate temporary vectorial force */ |
| 1309 | tx = _mm_mul_ps(fscal,dx32); |
| 1310 | ty = _mm_mul_ps(fscal,dy32); |
| 1311 | tz = _mm_mul_ps(fscal,dz32); |
| 1312 | |
| 1313 | /* Update vectorial force */ |
| 1314 | fix3 = _mm_add_ps(fix3,tx); |
| 1315 | fiy3 = _mm_add_ps(fiy3,ty); |
| 1316 | fiz3 = _mm_add_ps(fiz3,tz); |
| 1317 | |
| 1318 | fjx2 = _mm_add_ps(fjx2,tx); |
| 1319 | fjy2 = _mm_add_ps(fjy2,ty); |
| 1320 | fjz2 = _mm_add_ps(fjz2,tz); |
| 1321 | |
| 1322 | /************************** |
| 1323 | * CALCULATE INTERACTIONS * |
| 1324 | **************************/ |
| 1325 | |
| 1326 | r33 = _mm_mul_ps(rsq33,rinv33); |
| 1327 | r33 = _mm_andnot_ps(dummy_mask,r33); |
| 1328 | |
| 1329 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1330 | rt = _mm_mul_ps(r33,vftabscale); |
| 1331 | vfitab = _mm_cvttps_epi32(rt); |
| 1332 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1333 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1334 | |
| 1335 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1336 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1337 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1338 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1339 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1340 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1341 | Heps = _mm_mul_ps(vfeps,H); |
| 1342 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1343 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
| 1344 | velec = _mm_mul_ps(qq33,VV); |
| 1345 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1346 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
| 1347 | |
| 1348 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
| 1349 | velec = _mm_andnot_ps(dummy_mask,velec); |
| 1350 | velecsum = _mm_add_ps(velecsum,velec); |
| 1351 | |
| 1352 | fscal = felec; |
| 1353 | |
| 1354 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 1355 | |
| 1356 | /* Calculate temporary vectorial force */ |
| 1357 | tx = _mm_mul_ps(fscal,dx33); |
| 1358 | ty = _mm_mul_ps(fscal,dy33); |
| 1359 | tz = _mm_mul_ps(fscal,dz33); |
| 1360 | |
| 1361 | /* Update vectorial force */ |
| 1362 | fix3 = _mm_add_ps(fix3,tx); |
| 1363 | fiy3 = _mm_add_ps(fiy3,ty); |
| 1364 | fiz3 = _mm_add_ps(fiz3,tz); |
| 1365 | |
| 1366 | fjx3 = _mm_add_ps(fjx3,tx); |
| 1367 | fjy3 = _mm_add_ps(fjy3,ty); |
| 1368 | fjz3 = _mm_add_ps(fjz3,tz); |
| 1369 | |
| 1370 | fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch; |
| 1371 | fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch; |
| 1372 | fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch; |
| 1373 | fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch; |
| 1374 | |
| 1375 | gmx_mm_decrement_4rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
| 1376 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1, |
| 1377 | fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
| 1378 | |
| 1379 | /* Inner loop uses 456 flops */ |
| 1380 | } |
| 1381 | |
| 1382 | /* End of innermost loop */ |
| 1383 | |
| 1384 | gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3, |
| 1385 | f+i_coord_offset,fshift+i_shift_offset); |
| 1386 | |
| 1387 | ggid = gid[iidx]; |
| 1388 | /* Update potential energies */ |
| 1389 | gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid); |
| 1390 | gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid); |
| 1391 | |
| 1392 | /* Increment number of inner iterations */ |
| 1393 | inneriter += j_index_end - j_index_start; |
| 1394 | |
| 1395 | /* Outer loop uses 26 flops */ |
| 1396 | } |
| 1397 | |
| 1398 | /* Increment number of outer iterations */ |
| 1399 | outeriter += nri; |
| 1400 | |
| 1401 | /* Update outer/inner flops */ |
| 1402 | |
| 1403 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*26 + inneriter*456)(nrnb)->n[eNR_NBKERNEL_ELEC_VDW_W4W4_VF] += outeriter*26 + inneriter*456; |
| 1404 | } |
| 1405 | /* |
| 1406 | * Gromacs nonbonded kernel: nb_kernel_ElecCSTab_VdwCSTab_GeomW4W4_F_sse4_1_single |
| 1407 | * Electrostatics interaction: CubicSplineTable |
| 1408 | * VdW interaction: CubicSplineTable |
| 1409 | * Geometry: Water4-Water4 |
| 1410 | * Calculate force/pot: Force |
| 1411 | */ |
| 1412 | void |
| 1413 | nb_kernel_ElecCSTab_VdwCSTab_GeomW4W4_F_sse4_1_single |
| 1414 | (t_nblist * gmx_restrict nlist, |
| 1415 | rvec * gmx_restrict xx, |
| 1416 | rvec * gmx_restrict ff, |
| 1417 | t_forcerec * gmx_restrict fr, |
| 1418 | t_mdatoms * gmx_restrict mdatoms, |
| 1419 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict kernel_data, |
| 1420 | t_nrnb * gmx_restrict nrnb) |
| 1421 | { |
| 1422 | /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or |
| 1423 | * just 0 for non-waters. |
| 1424 | * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different |
| 1425 | * jnr indices corresponding to data put in the four positions in the SIMD register. |
| 1426 | */ |
| 1427 | int i_shift_offset,i_coord_offset,outeriter,inneriter; |
| 1428 | int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx; |
| 1429 | int jnrA,jnrB,jnrC,jnrD; |
| 1430 | int jnrlistA,jnrlistB,jnrlistC,jnrlistD; |
| 1431 | int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD; |
| 1432 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
| 1433 | real rcutoff_scalar; |
| 1434 | real *shiftvec,*fshift,*x,*f; |
| 1435 | real *fjptrA,*fjptrB,*fjptrC,*fjptrD; |
| 1436 | real scratch[4*DIM3]; |
| 1437 | __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall; |
| 1438 | int vdwioffset0; |
| 1439 | __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0; |
| 1440 | int vdwioffset1; |
| 1441 | __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
| 1442 | int vdwioffset2; |
| 1443 | __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
| 1444 | int vdwioffset3; |
| 1445 | __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3; |
| 1446 | int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D; |
| 1447 | __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
| 1448 | int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D; |
| 1449 | __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
| 1450 | int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D; |
| 1451 | __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
| 1452 | int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D; |
| 1453 | __m128 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3; |
| 1454 | __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00; |
| 1455 | __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11; |
| 1456 | __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12; |
| 1457 | __m128 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13; |
| 1458 | __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21; |
| 1459 | __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22; |
| 1460 | __m128 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23; |
| 1461 | __m128 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31; |
| 1462 | __m128 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32; |
| 1463 | __m128 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33; |
| 1464 | __m128 velec,felec,velecsum,facel,crf,krf,krf2; |
| 1465 | real *charge; |
| 1466 | int nvdwtype; |
| 1467 | __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6; |
| 1468 | int *vdwtype; |
| 1469 | real *vdwparam; |
| 1470 | __m128 one_sixth = _mm_set1_ps(1.0/6.0); |
| 1471 | __m128 one_twelfth = _mm_set1_ps(1.0/12.0); |
| 1472 | __m128i vfitab; |
| 1473 | __m128i ifour = _mm_set1_epi32(4); |
| 1474 | __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF; |
| 1475 | real *vftab; |
| 1476 | __m128 dummy_mask,cutoff_mask; |
| 1477 | __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) ); |
| 1478 | __m128 one = _mm_set1_ps(1.0); |
| 1479 | __m128 two = _mm_set1_ps(2.0); |
Value stored to 'two' during its initialization is never read | |
| 1480 | x = xx[0]; |
| 1481 | f = ff[0]; |
| 1482 | |
| 1483 | nri = nlist->nri; |
| 1484 | iinr = nlist->iinr; |
| 1485 | jindex = nlist->jindex; |
| 1486 | jjnr = nlist->jjnr; |
| 1487 | shiftidx = nlist->shift; |
| 1488 | gid = nlist->gid; |
| 1489 | shiftvec = fr->shift_vec[0]; |
| 1490 | fshift = fr->fshift[0]; |
| 1491 | facel = _mm_set1_ps(fr->epsfac); |
| 1492 | charge = mdatoms->chargeA; |
| 1493 | nvdwtype = fr->ntype; |
| 1494 | vdwparam = fr->nbfp; |
| 1495 | vdwtype = mdatoms->typeA; |
| 1496 | |
| 1497 | vftab = kernel_data->table_elec_vdw->data; |
| 1498 | vftabscale = _mm_set1_ps(kernel_data->table_elec_vdw->scale); |
| 1499 | |
| 1500 | /* Setup water-specific parameters */ |
| 1501 | inr = nlist->iinr[0]; |
| 1502 | iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1])); |
| 1503 | iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2])); |
| 1504 | iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3])); |
| 1505 | vdwioffset0 = 2*nvdwtype*vdwtype[inr+0]; |
| 1506 | |
| 1507 | jq1 = _mm_set1_ps(charge[inr+1]); |
| 1508 | jq2 = _mm_set1_ps(charge[inr+2]); |
| 1509 | jq3 = _mm_set1_ps(charge[inr+3]); |
| 1510 | vdwjidx0A = 2*vdwtype[inr+0]; |
| 1511 | c6_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]); |
| 1512 | c12_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]); |
| 1513 | qq11 = _mm_mul_ps(iq1,jq1); |
| 1514 | qq12 = _mm_mul_ps(iq1,jq2); |
| 1515 | qq13 = _mm_mul_ps(iq1,jq3); |
| 1516 | qq21 = _mm_mul_ps(iq2,jq1); |
| 1517 | qq22 = _mm_mul_ps(iq2,jq2); |
| 1518 | qq23 = _mm_mul_ps(iq2,jq3); |
| 1519 | qq31 = _mm_mul_ps(iq3,jq1); |
| 1520 | qq32 = _mm_mul_ps(iq3,jq2); |
| 1521 | qq33 = _mm_mul_ps(iq3,jq3); |
| 1522 | |
| 1523 | /* Avoid stupid compiler warnings */ |
| 1524 | jnrA = jnrB = jnrC = jnrD = 0; |
| 1525 | j_coord_offsetA = 0; |
| 1526 | j_coord_offsetB = 0; |
| 1527 | j_coord_offsetC = 0; |
| 1528 | j_coord_offsetD = 0; |
| 1529 | |
| 1530 | outeriter = 0; |
| 1531 | inneriter = 0; |
| 1532 | |
| 1533 | for(iidx=0;iidx<4*DIM3;iidx++) |
| 1534 | { |
| 1535 | scratch[iidx] = 0.0; |
| 1536 | } |
| 1537 | |
| 1538 | /* Start outer loop over neighborlists */ |
| 1539 | for(iidx=0; iidx<nri; iidx++) |
| 1540 | { |
| 1541 | /* Load shift vector for this list */ |
| 1542 | i_shift_offset = DIM3*shiftidx[iidx]; |
| 1543 | |
| 1544 | /* Load limits for loop over neighbors */ |
| 1545 | j_index_start = jindex[iidx]; |
| 1546 | j_index_end = jindex[iidx+1]; |
| 1547 | |
| 1548 | /* Get outer coordinate index */ |
| 1549 | inr = iinr[iidx]; |
| 1550 | i_coord_offset = DIM3*inr; |
| 1551 | |
| 1552 | /* Load i particle coords and add shift vector */ |
| 1553 | gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset, |
| 1554 | &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3); |
| 1555 | |
| 1556 | fix0 = _mm_setzero_ps(); |
| 1557 | fiy0 = _mm_setzero_ps(); |
| 1558 | fiz0 = _mm_setzero_ps(); |
| 1559 | fix1 = _mm_setzero_ps(); |
| 1560 | fiy1 = _mm_setzero_ps(); |
| 1561 | fiz1 = _mm_setzero_ps(); |
| 1562 | fix2 = _mm_setzero_ps(); |
| 1563 | fiy2 = _mm_setzero_ps(); |
| 1564 | fiz2 = _mm_setzero_ps(); |
| 1565 | fix3 = _mm_setzero_ps(); |
| 1566 | fiy3 = _mm_setzero_ps(); |
| 1567 | fiz3 = _mm_setzero_ps(); |
| 1568 | |
| 1569 | /* Start inner kernel loop */ |
| 1570 | for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4) |
| 1571 | { |
| 1572 | |
| 1573 | /* Get j neighbor index, and coordinate index */ |
| 1574 | jnrA = jjnr[jidx]; |
| 1575 | jnrB = jjnr[jidx+1]; |
| 1576 | jnrC = jjnr[jidx+2]; |
| 1577 | jnrD = jjnr[jidx+3]; |
| 1578 | j_coord_offsetA = DIM3*jnrA; |
| 1579 | j_coord_offsetB = DIM3*jnrB; |
| 1580 | j_coord_offsetC = DIM3*jnrC; |
| 1581 | j_coord_offsetD = DIM3*jnrD; |
| 1582 | |
| 1583 | /* load j atom coordinates */ |
| 1584 | gmx_mm_load_4rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
| 1585 | x+j_coord_offsetC,x+j_coord_offsetD, |
| 1586 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2, |
| 1587 | &jy2,&jz2,&jx3,&jy3,&jz3); |
| 1588 | |
| 1589 | /* Calculate displacement vector */ |
| 1590 | dx00 = _mm_sub_ps(ix0,jx0); |
| 1591 | dy00 = _mm_sub_ps(iy0,jy0); |
| 1592 | dz00 = _mm_sub_ps(iz0,jz0); |
| 1593 | dx11 = _mm_sub_ps(ix1,jx1); |
| 1594 | dy11 = _mm_sub_ps(iy1,jy1); |
| 1595 | dz11 = _mm_sub_ps(iz1,jz1); |
| 1596 | dx12 = _mm_sub_ps(ix1,jx2); |
| 1597 | dy12 = _mm_sub_ps(iy1,jy2); |
| 1598 | dz12 = _mm_sub_ps(iz1,jz2); |
| 1599 | dx13 = _mm_sub_ps(ix1,jx3); |
| 1600 | dy13 = _mm_sub_ps(iy1,jy3); |
| 1601 | dz13 = _mm_sub_ps(iz1,jz3); |
| 1602 | dx21 = _mm_sub_ps(ix2,jx1); |
| 1603 | dy21 = _mm_sub_ps(iy2,jy1); |
| 1604 | dz21 = _mm_sub_ps(iz2,jz1); |
| 1605 | dx22 = _mm_sub_ps(ix2,jx2); |
| 1606 | dy22 = _mm_sub_ps(iy2,jy2); |
| 1607 | dz22 = _mm_sub_ps(iz2,jz2); |
| 1608 | dx23 = _mm_sub_ps(ix2,jx3); |
| 1609 | dy23 = _mm_sub_ps(iy2,jy3); |
| 1610 | dz23 = _mm_sub_ps(iz2,jz3); |
| 1611 | dx31 = _mm_sub_ps(ix3,jx1); |
| 1612 | dy31 = _mm_sub_ps(iy3,jy1); |
| 1613 | dz31 = _mm_sub_ps(iz3,jz1); |
| 1614 | dx32 = _mm_sub_ps(ix3,jx2); |
| 1615 | dy32 = _mm_sub_ps(iy3,jy2); |
| 1616 | dz32 = _mm_sub_ps(iz3,jz2); |
| 1617 | dx33 = _mm_sub_ps(ix3,jx3); |
| 1618 | dy33 = _mm_sub_ps(iy3,jy3); |
| 1619 | dz33 = _mm_sub_ps(iz3,jz3); |
| 1620 | |
| 1621 | /* Calculate squared distance and things based on it */ |
| 1622 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
| 1623 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
| 1624 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
| 1625 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
| 1626 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
| 1627 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
| 1628 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
| 1629 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
| 1630 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
| 1631 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
| 1632 | |
| 1633 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
| 1634 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
| 1635 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
| 1636 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
| 1637 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
| 1638 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
| 1639 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
| 1640 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
| 1641 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
| 1642 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
| 1643 | |
| 1644 | fjx0 = _mm_setzero_ps(); |
| 1645 | fjy0 = _mm_setzero_ps(); |
| 1646 | fjz0 = _mm_setzero_ps(); |
| 1647 | fjx1 = _mm_setzero_ps(); |
| 1648 | fjy1 = _mm_setzero_ps(); |
| 1649 | fjz1 = _mm_setzero_ps(); |
| 1650 | fjx2 = _mm_setzero_ps(); |
| 1651 | fjy2 = _mm_setzero_ps(); |
| 1652 | fjz2 = _mm_setzero_ps(); |
| 1653 | fjx3 = _mm_setzero_ps(); |
| 1654 | fjy3 = _mm_setzero_ps(); |
| 1655 | fjz3 = _mm_setzero_ps(); |
| 1656 | |
| 1657 | /************************** |
| 1658 | * CALCULATE INTERACTIONS * |
| 1659 | **************************/ |
| 1660 | |
| 1661 | r00 = _mm_mul_ps(rsq00,rinv00); |
| 1662 | |
| 1663 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1664 | rt = _mm_mul_ps(r00,vftabscale); |
| 1665 | vfitab = _mm_cvttps_epi32(rt); |
| 1666 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1667 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1668 | |
| 1669 | /* CUBIC SPLINE TABLE DISPERSION */ |
| 1670 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 1671 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1672 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1673 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1674 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1675 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1676 | Heps = _mm_mul_ps(vfeps,H); |
| 1677 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1678 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1679 | fvdw6 = _mm_mul_ps(c6_00,FF); |
| 1680 | |
| 1681 | /* CUBIC SPLINE TABLE REPULSION */ |
| 1682 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 1683 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1684 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1685 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1686 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1687 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1688 | Heps = _mm_mul_ps(vfeps,H); |
| 1689 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1690 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1691 | fvdw12 = _mm_mul_ps(c12_00,FF); |
| 1692 | fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00))); |
| 1693 | |
| 1694 | fscal = fvdw; |
| 1695 | |
| 1696 | /* Calculate temporary vectorial force */ |
| 1697 | tx = _mm_mul_ps(fscal,dx00); |
| 1698 | ty = _mm_mul_ps(fscal,dy00); |
| 1699 | tz = _mm_mul_ps(fscal,dz00); |
| 1700 | |
| 1701 | /* Update vectorial force */ |
| 1702 | fix0 = _mm_add_ps(fix0,tx); |
| 1703 | fiy0 = _mm_add_ps(fiy0,ty); |
| 1704 | fiz0 = _mm_add_ps(fiz0,tz); |
| 1705 | |
| 1706 | fjx0 = _mm_add_ps(fjx0,tx); |
| 1707 | fjy0 = _mm_add_ps(fjy0,ty); |
| 1708 | fjz0 = _mm_add_ps(fjz0,tz); |
| 1709 | |
| 1710 | /************************** |
| 1711 | * CALCULATE INTERACTIONS * |
| 1712 | **************************/ |
| 1713 | |
| 1714 | r11 = _mm_mul_ps(rsq11,rinv11); |
| 1715 | |
| 1716 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1717 | rt = _mm_mul_ps(r11,vftabscale); |
| 1718 | vfitab = _mm_cvttps_epi32(rt); |
| 1719 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1720 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1721 | |
| 1722 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1723 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1724 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1725 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1726 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1727 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1728 | Heps = _mm_mul_ps(vfeps,H); |
| 1729 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1730 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1731 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
| 1732 | |
| 1733 | fscal = felec; |
| 1734 | |
| 1735 | /* Calculate temporary vectorial force */ |
| 1736 | tx = _mm_mul_ps(fscal,dx11); |
| 1737 | ty = _mm_mul_ps(fscal,dy11); |
| 1738 | tz = _mm_mul_ps(fscal,dz11); |
| 1739 | |
| 1740 | /* Update vectorial force */ |
| 1741 | fix1 = _mm_add_ps(fix1,tx); |
| 1742 | fiy1 = _mm_add_ps(fiy1,ty); |
| 1743 | fiz1 = _mm_add_ps(fiz1,tz); |
| 1744 | |
| 1745 | fjx1 = _mm_add_ps(fjx1,tx); |
| 1746 | fjy1 = _mm_add_ps(fjy1,ty); |
| 1747 | fjz1 = _mm_add_ps(fjz1,tz); |
| 1748 | |
| 1749 | /************************** |
| 1750 | * CALCULATE INTERACTIONS * |
| 1751 | **************************/ |
| 1752 | |
| 1753 | r12 = _mm_mul_ps(rsq12,rinv12); |
| 1754 | |
| 1755 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1756 | rt = _mm_mul_ps(r12,vftabscale); |
| 1757 | vfitab = _mm_cvttps_epi32(rt); |
| 1758 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1759 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1760 | |
| 1761 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1762 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1763 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1764 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1765 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1766 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1767 | Heps = _mm_mul_ps(vfeps,H); |
| 1768 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1769 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1770 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
| 1771 | |
| 1772 | fscal = felec; |
| 1773 | |
| 1774 | /* Calculate temporary vectorial force */ |
| 1775 | tx = _mm_mul_ps(fscal,dx12); |
| 1776 | ty = _mm_mul_ps(fscal,dy12); |
| 1777 | tz = _mm_mul_ps(fscal,dz12); |
| 1778 | |
| 1779 | /* Update vectorial force */ |
| 1780 | fix1 = _mm_add_ps(fix1,tx); |
| 1781 | fiy1 = _mm_add_ps(fiy1,ty); |
| 1782 | fiz1 = _mm_add_ps(fiz1,tz); |
| 1783 | |
| 1784 | fjx2 = _mm_add_ps(fjx2,tx); |
| 1785 | fjy2 = _mm_add_ps(fjy2,ty); |
| 1786 | fjz2 = _mm_add_ps(fjz2,tz); |
| 1787 | |
| 1788 | /************************** |
| 1789 | * CALCULATE INTERACTIONS * |
| 1790 | **************************/ |
| 1791 | |
| 1792 | r13 = _mm_mul_ps(rsq13,rinv13); |
| 1793 | |
| 1794 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1795 | rt = _mm_mul_ps(r13,vftabscale); |
| 1796 | vfitab = _mm_cvttps_epi32(rt); |
| 1797 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1798 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1799 | |
| 1800 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1801 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1802 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1803 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1804 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1805 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1806 | Heps = _mm_mul_ps(vfeps,H); |
| 1807 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1808 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1809 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
| 1810 | |
| 1811 | fscal = felec; |
| 1812 | |
| 1813 | /* Calculate temporary vectorial force */ |
| 1814 | tx = _mm_mul_ps(fscal,dx13); |
| 1815 | ty = _mm_mul_ps(fscal,dy13); |
| 1816 | tz = _mm_mul_ps(fscal,dz13); |
| 1817 | |
| 1818 | /* Update vectorial force */ |
| 1819 | fix1 = _mm_add_ps(fix1,tx); |
| 1820 | fiy1 = _mm_add_ps(fiy1,ty); |
| 1821 | fiz1 = _mm_add_ps(fiz1,tz); |
| 1822 | |
| 1823 | fjx3 = _mm_add_ps(fjx3,tx); |
| 1824 | fjy3 = _mm_add_ps(fjy3,ty); |
| 1825 | fjz3 = _mm_add_ps(fjz3,tz); |
| 1826 | |
| 1827 | /************************** |
| 1828 | * CALCULATE INTERACTIONS * |
| 1829 | **************************/ |
| 1830 | |
| 1831 | r21 = _mm_mul_ps(rsq21,rinv21); |
| 1832 | |
| 1833 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1834 | rt = _mm_mul_ps(r21,vftabscale); |
| 1835 | vfitab = _mm_cvttps_epi32(rt); |
| 1836 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1837 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1838 | |
| 1839 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1840 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1841 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1842 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1843 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1844 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1845 | Heps = _mm_mul_ps(vfeps,H); |
| 1846 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1847 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1848 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
| 1849 | |
| 1850 | fscal = felec; |
| 1851 | |
| 1852 | /* Calculate temporary vectorial force */ |
| 1853 | tx = _mm_mul_ps(fscal,dx21); |
| 1854 | ty = _mm_mul_ps(fscal,dy21); |
| 1855 | tz = _mm_mul_ps(fscal,dz21); |
| 1856 | |
| 1857 | /* Update vectorial force */ |
| 1858 | fix2 = _mm_add_ps(fix2,tx); |
| 1859 | fiy2 = _mm_add_ps(fiy2,ty); |
| 1860 | fiz2 = _mm_add_ps(fiz2,tz); |
| 1861 | |
| 1862 | fjx1 = _mm_add_ps(fjx1,tx); |
| 1863 | fjy1 = _mm_add_ps(fjy1,ty); |
| 1864 | fjz1 = _mm_add_ps(fjz1,tz); |
| 1865 | |
| 1866 | /************************** |
| 1867 | * CALCULATE INTERACTIONS * |
| 1868 | **************************/ |
| 1869 | |
| 1870 | r22 = _mm_mul_ps(rsq22,rinv22); |
| 1871 | |
| 1872 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1873 | rt = _mm_mul_ps(r22,vftabscale); |
| 1874 | vfitab = _mm_cvttps_epi32(rt); |
| 1875 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1876 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1877 | |
| 1878 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1879 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1880 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1881 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1882 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1883 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1884 | Heps = _mm_mul_ps(vfeps,H); |
| 1885 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1886 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1887 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
| 1888 | |
| 1889 | fscal = felec; |
| 1890 | |
| 1891 | /* Calculate temporary vectorial force */ |
| 1892 | tx = _mm_mul_ps(fscal,dx22); |
| 1893 | ty = _mm_mul_ps(fscal,dy22); |
| 1894 | tz = _mm_mul_ps(fscal,dz22); |
| 1895 | |
| 1896 | /* Update vectorial force */ |
| 1897 | fix2 = _mm_add_ps(fix2,tx); |
| 1898 | fiy2 = _mm_add_ps(fiy2,ty); |
| 1899 | fiz2 = _mm_add_ps(fiz2,tz); |
| 1900 | |
| 1901 | fjx2 = _mm_add_ps(fjx2,tx); |
| 1902 | fjy2 = _mm_add_ps(fjy2,ty); |
| 1903 | fjz2 = _mm_add_ps(fjz2,tz); |
| 1904 | |
| 1905 | /************************** |
| 1906 | * CALCULATE INTERACTIONS * |
| 1907 | **************************/ |
| 1908 | |
| 1909 | r23 = _mm_mul_ps(rsq23,rinv23); |
| 1910 | |
| 1911 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1912 | rt = _mm_mul_ps(r23,vftabscale); |
| 1913 | vfitab = _mm_cvttps_epi32(rt); |
| 1914 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1915 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1916 | |
| 1917 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1918 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1919 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1920 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1921 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1922 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1923 | Heps = _mm_mul_ps(vfeps,H); |
| 1924 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1925 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1926 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
| 1927 | |
| 1928 | fscal = felec; |
| 1929 | |
| 1930 | /* Calculate temporary vectorial force */ |
| 1931 | tx = _mm_mul_ps(fscal,dx23); |
| 1932 | ty = _mm_mul_ps(fscal,dy23); |
| 1933 | tz = _mm_mul_ps(fscal,dz23); |
| 1934 | |
| 1935 | /* Update vectorial force */ |
| 1936 | fix2 = _mm_add_ps(fix2,tx); |
| 1937 | fiy2 = _mm_add_ps(fiy2,ty); |
| 1938 | fiz2 = _mm_add_ps(fiz2,tz); |
| 1939 | |
| 1940 | fjx3 = _mm_add_ps(fjx3,tx); |
| 1941 | fjy3 = _mm_add_ps(fjy3,ty); |
| 1942 | fjz3 = _mm_add_ps(fjz3,tz); |
| 1943 | |
| 1944 | /************************** |
| 1945 | * CALCULATE INTERACTIONS * |
| 1946 | **************************/ |
| 1947 | |
| 1948 | r31 = _mm_mul_ps(rsq31,rinv31); |
| 1949 | |
| 1950 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1951 | rt = _mm_mul_ps(r31,vftabscale); |
| 1952 | vfitab = _mm_cvttps_epi32(rt); |
| 1953 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1954 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1955 | |
| 1956 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1957 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1958 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1959 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1960 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 1961 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 1962 | Heps = _mm_mul_ps(vfeps,H); |
| 1963 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 1964 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 1965 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
| 1966 | |
| 1967 | fscal = felec; |
| 1968 | |
| 1969 | /* Calculate temporary vectorial force */ |
| 1970 | tx = _mm_mul_ps(fscal,dx31); |
| 1971 | ty = _mm_mul_ps(fscal,dy31); |
| 1972 | tz = _mm_mul_ps(fscal,dz31); |
| 1973 | |
| 1974 | /* Update vectorial force */ |
| 1975 | fix3 = _mm_add_ps(fix3,tx); |
| 1976 | fiy3 = _mm_add_ps(fiy3,ty); |
| 1977 | fiz3 = _mm_add_ps(fiz3,tz); |
| 1978 | |
| 1979 | fjx1 = _mm_add_ps(fjx1,tx); |
| 1980 | fjy1 = _mm_add_ps(fjy1,ty); |
| 1981 | fjz1 = _mm_add_ps(fjz1,tz); |
| 1982 | |
| 1983 | /************************** |
| 1984 | * CALCULATE INTERACTIONS * |
| 1985 | **************************/ |
| 1986 | |
| 1987 | r32 = _mm_mul_ps(rsq32,rinv32); |
| 1988 | |
| 1989 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 1990 | rt = _mm_mul_ps(r32,vftabscale); |
| 1991 | vfitab = _mm_cvttps_epi32(rt); |
| 1992 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 1993 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 1994 | |
| 1995 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 1996 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 1997 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 1998 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 1999 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2000 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2001 | Heps = _mm_mul_ps(vfeps,H); |
| 2002 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2003 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2004 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
| 2005 | |
| 2006 | fscal = felec; |
| 2007 | |
| 2008 | /* Calculate temporary vectorial force */ |
| 2009 | tx = _mm_mul_ps(fscal,dx32); |
| 2010 | ty = _mm_mul_ps(fscal,dy32); |
| 2011 | tz = _mm_mul_ps(fscal,dz32); |
| 2012 | |
| 2013 | /* Update vectorial force */ |
| 2014 | fix3 = _mm_add_ps(fix3,tx); |
| 2015 | fiy3 = _mm_add_ps(fiy3,ty); |
| 2016 | fiz3 = _mm_add_ps(fiz3,tz); |
| 2017 | |
| 2018 | fjx2 = _mm_add_ps(fjx2,tx); |
| 2019 | fjy2 = _mm_add_ps(fjy2,ty); |
| 2020 | fjz2 = _mm_add_ps(fjz2,tz); |
| 2021 | |
| 2022 | /************************** |
| 2023 | * CALCULATE INTERACTIONS * |
| 2024 | **************************/ |
| 2025 | |
| 2026 | r33 = _mm_mul_ps(rsq33,rinv33); |
| 2027 | |
| 2028 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2029 | rt = _mm_mul_ps(r33,vftabscale); |
| 2030 | vfitab = _mm_cvttps_epi32(rt); |
| 2031 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2032 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2033 | |
| 2034 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2035 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2036 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2037 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2038 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2039 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2040 | Heps = _mm_mul_ps(vfeps,H); |
| 2041 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2042 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2043 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
| 2044 | |
| 2045 | fscal = felec; |
| 2046 | |
| 2047 | /* Calculate temporary vectorial force */ |
| 2048 | tx = _mm_mul_ps(fscal,dx33); |
| 2049 | ty = _mm_mul_ps(fscal,dy33); |
| 2050 | tz = _mm_mul_ps(fscal,dz33); |
| 2051 | |
| 2052 | /* Update vectorial force */ |
| 2053 | fix3 = _mm_add_ps(fix3,tx); |
| 2054 | fiy3 = _mm_add_ps(fiy3,ty); |
| 2055 | fiz3 = _mm_add_ps(fiz3,tz); |
| 2056 | |
| 2057 | fjx3 = _mm_add_ps(fjx3,tx); |
| 2058 | fjy3 = _mm_add_ps(fjy3,ty); |
| 2059 | fjz3 = _mm_add_ps(fjz3,tz); |
| 2060 | |
| 2061 | fjptrA = f+j_coord_offsetA; |
| 2062 | fjptrB = f+j_coord_offsetB; |
| 2063 | fjptrC = f+j_coord_offsetC; |
| 2064 | fjptrD = f+j_coord_offsetD; |
| 2065 | |
| 2066 | gmx_mm_decrement_4rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
| 2067 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1, |
| 2068 | fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
| 2069 | |
| 2070 | /* Inner loop uses 402 flops */ |
| 2071 | } |
| 2072 | |
| 2073 | if(jidx<j_index_end) |
| 2074 | { |
| 2075 | |
| 2076 | /* Get j neighbor index, and coordinate index */ |
| 2077 | jnrlistA = jjnr[jidx]; |
| 2078 | jnrlistB = jjnr[jidx+1]; |
| 2079 | jnrlistC = jjnr[jidx+2]; |
| 2080 | jnrlistD = jjnr[jidx+3]; |
| 2081 | /* Sign of each element will be negative for non-real atoms. |
| 2082 | * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones, |
| 2083 | * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries. |
| 2084 | */ |
| 2085 | dummy_mask = gmx_mm_castsi128_ps_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())); |
| 2086 | jnrA = (jnrlistA>=0) ? jnrlistA : 0; |
| 2087 | jnrB = (jnrlistB>=0) ? jnrlistB : 0; |
| 2088 | jnrC = (jnrlistC>=0) ? jnrlistC : 0; |
| 2089 | jnrD = (jnrlistD>=0) ? jnrlistD : 0; |
| 2090 | j_coord_offsetA = DIM3*jnrA; |
| 2091 | j_coord_offsetB = DIM3*jnrB; |
| 2092 | j_coord_offsetC = DIM3*jnrC; |
| 2093 | j_coord_offsetD = DIM3*jnrD; |
| 2094 | |
| 2095 | /* load j atom coordinates */ |
| 2096 | gmx_mm_load_4rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
| 2097 | x+j_coord_offsetC,x+j_coord_offsetD, |
| 2098 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2, |
| 2099 | &jy2,&jz2,&jx3,&jy3,&jz3); |
| 2100 | |
| 2101 | /* Calculate displacement vector */ |
| 2102 | dx00 = _mm_sub_ps(ix0,jx0); |
| 2103 | dy00 = _mm_sub_ps(iy0,jy0); |
| 2104 | dz00 = _mm_sub_ps(iz0,jz0); |
| 2105 | dx11 = _mm_sub_ps(ix1,jx1); |
| 2106 | dy11 = _mm_sub_ps(iy1,jy1); |
| 2107 | dz11 = _mm_sub_ps(iz1,jz1); |
| 2108 | dx12 = _mm_sub_ps(ix1,jx2); |
| 2109 | dy12 = _mm_sub_ps(iy1,jy2); |
| 2110 | dz12 = _mm_sub_ps(iz1,jz2); |
| 2111 | dx13 = _mm_sub_ps(ix1,jx3); |
| 2112 | dy13 = _mm_sub_ps(iy1,jy3); |
| 2113 | dz13 = _mm_sub_ps(iz1,jz3); |
| 2114 | dx21 = _mm_sub_ps(ix2,jx1); |
| 2115 | dy21 = _mm_sub_ps(iy2,jy1); |
| 2116 | dz21 = _mm_sub_ps(iz2,jz1); |
| 2117 | dx22 = _mm_sub_ps(ix2,jx2); |
| 2118 | dy22 = _mm_sub_ps(iy2,jy2); |
| 2119 | dz22 = _mm_sub_ps(iz2,jz2); |
| 2120 | dx23 = _mm_sub_ps(ix2,jx3); |
| 2121 | dy23 = _mm_sub_ps(iy2,jy3); |
| 2122 | dz23 = _mm_sub_ps(iz2,jz3); |
| 2123 | dx31 = _mm_sub_ps(ix3,jx1); |
| 2124 | dy31 = _mm_sub_ps(iy3,jy1); |
| 2125 | dz31 = _mm_sub_ps(iz3,jz1); |
| 2126 | dx32 = _mm_sub_ps(ix3,jx2); |
| 2127 | dy32 = _mm_sub_ps(iy3,jy2); |
| 2128 | dz32 = _mm_sub_ps(iz3,jz2); |
| 2129 | dx33 = _mm_sub_ps(ix3,jx3); |
| 2130 | dy33 = _mm_sub_ps(iy3,jy3); |
| 2131 | dz33 = _mm_sub_ps(iz3,jz3); |
| 2132 | |
| 2133 | /* Calculate squared distance and things based on it */ |
| 2134 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
| 2135 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
| 2136 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
| 2137 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
| 2138 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
| 2139 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
| 2140 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
| 2141 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
| 2142 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
| 2143 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
| 2144 | |
| 2145 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
| 2146 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
| 2147 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
| 2148 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
| 2149 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
| 2150 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
| 2151 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
| 2152 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
| 2153 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
| 2154 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
| 2155 | |
| 2156 | fjx0 = _mm_setzero_ps(); |
| 2157 | fjy0 = _mm_setzero_ps(); |
| 2158 | fjz0 = _mm_setzero_ps(); |
| 2159 | fjx1 = _mm_setzero_ps(); |
| 2160 | fjy1 = _mm_setzero_ps(); |
| 2161 | fjz1 = _mm_setzero_ps(); |
| 2162 | fjx2 = _mm_setzero_ps(); |
| 2163 | fjy2 = _mm_setzero_ps(); |
| 2164 | fjz2 = _mm_setzero_ps(); |
| 2165 | fjx3 = _mm_setzero_ps(); |
| 2166 | fjy3 = _mm_setzero_ps(); |
| 2167 | fjz3 = _mm_setzero_ps(); |
| 2168 | |
| 2169 | /************************** |
| 2170 | * CALCULATE INTERACTIONS * |
| 2171 | **************************/ |
| 2172 | |
| 2173 | r00 = _mm_mul_ps(rsq00,rinv00); |
| 2174 | r00 = _mm_andnot_ps(dummy_mask,r00); |
| 2175 | |
| 2176 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2177 | rt = _mm_mul_ps(r00,vftabscale); |
| 2178 | vfitab = _mm_cvttps_epi32(rt); |
| 2179 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2180 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2181 | |
| 2182 | /* CUBIC SPLINE TABLE DISPERSION */ |
| 2183 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 2184 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2185 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2186 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2187 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2188 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2189 | Heps = _mm_mul_ps(vfeps,H); |
| 2190 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2191 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2192 | fvdw6 = _mm_mul_ps(c6_00,FF); |
| 2193 | |
| 2194 | /* CUBIC SPLINE TABLE REPULSION */ |
| 2195 | vfitab = _mm_add_epi32(vfitab,ifour); |
| 2196 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2197 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2198 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2199 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2200 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2201 | Heps = _mm_mul_ps(vfeps,H); |
| 2202 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2203 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2204 | fvdw12 = _mm_mul_ps(c12_00,FF); |
| 2205 | fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00))); |
| 2206 | |
| 2207 | fscal = fvdw; |
| 2208 | |
| 2209 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2210 | |
| 2211 | /* Calculate temporary vectorial force */ |
| 2212 | tx = _mm_mul_ps(fscal,dx00); |
| 2213 | ty = _mm_mul_ps(fscal,dy00); |
| 2214 | tz = _mm_mul_ps(fscal,dz00); |
| 2215 | |
| 2216 | /* Update vectorial force */ |
| 2217 | fix0 = _mm_add_ps(fix0,tx); |
| 2218 | fiy0 = _mm_add_ps(fiy0,ty); |
| 2219 | fiz0 = _mm_add_ps(fiz0,tz); |
| 2220 | |
| 2221 | fjx0 = _mm_add_ps(fjx0,tx); |
| 2222 | fjy0 = _mm_add_ps(fjy0,ty); |
| 2223 | fjz0 = _mm_add_ps(fjz0,tz); |
| 2224 | |
| 2225 | /************************** |
| 2226 | * CALCULATE INTERACTIONS * |
| 2227 | **************************/ |
| 2228 | |
| 2229 | r11 = _mm_mul_ps(rsq11,rinv11); |
| 2230 | r11 = _mm_andnot_ps(dummy_mask,r11); |
| 2231 | |
| 2232 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2233 | rt = _mm_mul_ps(r11,vftabscale); |
| 2234 | vfitab = _mm_cvttps_epi32(rt); |
| 2235 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2236 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2237 | |
| 2238 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2239 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2240 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2241 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2242 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2243 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2244 | Heps = _mm_mul_ps(vfeps,H); |
| 2245 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2246 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2247 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
| 2248 | |
| 2249 | fscal = felec; |
| 2250 | |
| 2251 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2252 | |
| 2253 | /* Calculate temporary vectorial force */ |
| 2254 | tx = _mm_mul_ps(fscal,dx11); |
| 2255 | ty = _mm_mul_ps(fscal,dy11); |
| 2256 | tz = _mm_mul_ps(fscal,dz11); |
| 2257 | |
| 2258 | /* Update vectorial force */ |
| 2259 | fix1 = _mm_add_ps(fix1,tx); |
| 2260 | fiy1 = _mm_add_ps(fiy1,ty); |
| 2261 | fiz1 = _mm_add_ps(fiz1,tz); |
| 2262 | |
| 2263 | fjx1 = _mm_add_ps(fjx1,tx); |
| 2264 | fjy1 = _mm_add_ps(fjy1,ty); |
| 2265 | fjz1 = _mm_add_ps(fjz1,tz); |
| 2266 | |
| 2267 | /************************** |
| 2268 | * CALCULATE INTERACTIONS * |
| 2269 | **************************/ |
| 2270 | |
| 2271 | r12 = _mm_mul_ps(rsq12,rinv12); |
| 2272 | r12 = _mm_andnot_ps(dummy_mask,r12); |
| 2273 | |
| 2274 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2275 | rt = _mm_mul_ps(r12,vftabscale); |
| 2276 | vfitab = _mm_cvttps_epi32(rt); |
| 2277 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2278 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2279 | |
| 2280 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2281 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2282 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2283 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2284 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2285 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2286 | Heps = _mm_mul_ps(vfeps,H); |
| 2287 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2288 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2289 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
| 2290 | |
| 2291 | fscal = felec; |
| 2292 | |
| 2293 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2294 | |
| 2295 | /* Calculate temporary vectorial force */ |
| 2296 | tx = _mm_mul_ps(fscal,dx12); |
| 2297 | ty = _mm_mul_ps(fscal,dy12); |
| 2298 | tz = _mm_mul_ps(fscal,dz12); |
| 2299 | |
| 2300 | /* Update vectorial force */ |
| 2301 | fix1 = _mm_add_ps(fix1,tx); |
| 2302 | fiy1 = _mm_add_ps(fiy1,ty); |
| 2303 | fiz1 = _mm_add_ps(fiz1,tz); |
| 2304 | |
| 2305 | fjx2 = _mm_add_ps(fjx2,tx); |
| 2306 | fjy2 = _mm_add_ps(fjy2,ty); |
| 2307 | fjz2 = _mm_add_ps(fjz2,tz); |
| 2308 | |
| 2309 | /************************** |
| 2310 | * CALCULATE INTERACTIONS * |
| 2311 | **************************/ |
| 2312 | |
| 2313 | r13 = _mm_mul_ps(rsq13,rinv13); |
| 2314 | r13 = _mm_andnot_ps(dummy_mask,r13); |
| 2315 | |
| 2316 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2317 | rt = _mm_mul_ps(r13,vftabscale); |
| 2318 | vfitab = _mm_cvttps_epi32(rt); |
| 2319 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2320 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2321 | |
| 2322 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2323 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2324 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2325 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2326 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2327 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2328 | Heps = _mm_mul_ps(vfeps,H); |
| 2329 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2330 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2331 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
| 2332 | |
| 2333 | fscal = felec; |
| 2334 | |
| 2335 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2336 | |
| 2337 | /* Calculate temporary vectorial force */ |
| 2338 | tx = _mm_mul_ps(fscal,dx13); |
| 2339 | ty = _mm_mul_ps(fscal,dy13); |
| 2340 | tz = _mm_mul_ps(fscal,dz13); |
| 2341 | |
| 2342 | /* Update vectorial force */ |
| 2343 | fix1 = _mm_add_ps(fix1,tx); |
| 2344 | fiy1 = _mm_add_ps(fiy1,ty); |
| 2345 | fiz1 = _mm_add_ps(fiz1,tz); |
| 2346 | |
| 2347 | fjx3 = _mm_add_ps(fjx3,tx); |
| 2348 | fjy3 = _mm_add_ps(fjy3,ty); |
| 2349 | fjz3 = _mm_add_ps(fjz3,tz); |
| 2350 | |
| 2351 | /************************** |
| 2352 | * CALCULATE INTERACTIONS * |
| 2353 | **************************/ |
| 2354 | |
| 2355 | r21 = _mm_mul_ps(rsq21,rinv21); |
| 2356 | r21 = _mm_andnot_ps(dummy_mask,r21); |
| 2357 | |
| 2358 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2359 | rt = _mm_mul_ps(r21,vftabscale); |
| 2360 | vfitab = _mm_cvttps_epi32(rt); |
| 2361 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2362 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2363 | |
| 2364 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2365 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2366 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2367 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2368 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2369 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2370 | Heps = _mm_mul_ps(vfeps,H); |
| 2371 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2372 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2373 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
| 2374 | |
| 2375 | fscal = felec; |
| 2376 | |
| 2377 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2378 | |
| 2379 | /* Calculate temporary vectorial force */ |
| 2380 | tx = _mm_mul_ps(fscal,dx21); |
| 2381 | ty = _mm_mul_ps(fscal,dy21); |
| 2382 | tz = _mm_mul_ps(fscal,dz21); |
| 2383 | |
| 2384 | /* Update vectorial force */ |
| 2385 | fix2 = _mm_add_ps(fix2,tx); |
| 2386 | fiy2 = _mm_add_ps(fiy2,ty); |
| 2387 | fiz2 = _mm_add_ps(fiz2,tz); |
| 2388 | |
| 2389 | fjx1 = _mm_add_ps(fjx1,tx); |
| 2390 | fjy1 = _mm_add_ps(fjy1,ty); |
| 2391 | fjz1 = _mm_add_ps(fjz1,tz); |
| 2392 | |
| 2393 | /************************** |
| 2394 | * CALCULATE INTERACTIONS * |
| 2395 | **************************/ |
| 2396 | |
| 2397 | r22 = _mm_mul_ps(rsq22,rinv22); |
| 2398 | r22 = _mm_andnot_ps(dummy_mask,r22); |
| 2399 | |
| 2400 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2401 | rt = _mm_mul_ps(r22,vftabscale); |
| 2402 | vfitab = _mm_cvttps_epi32(rt); |
| 2403 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2404 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2405 | |
| 2406 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2407 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2408 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2409 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2410 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2411 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2412 | Heps = _mm_mul_ps(vfeps,H); |
| 2413 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2414 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2415 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
| 2416 | |
| 2417 | fscal = felec; |
| 2418 | |
| 2419 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2420 | |
| 2421 | /* Calculate temporary vectorial force */ |
| 2422 | tx = _mm_mul_ps(fscal,dx22); |
| 2423 | ty = _mm_mul_ps(fscal,dy22); |
| 2424 | tz = _mm_mul_ps(fscal,dz22); |
| 2425 | |
| 2426 | /* Update vectorial force */ |
| 2427 | fix2 = _mm_add_ps(fix2,tx); |
| 2428 | fiy2 = _mm_add_ps(fiy2,ty); |
| 2429 | fiz2 = _mm_add_ps(fiz2,tz); |
| 2430 | |
| 2431 | fjx2 = _mm_add_ps(fjx2,tx); |
| 2432 | fjy2 = _mm_add_ps(fjy2,ty); |
| 2433 | fjz2 = _mm_add_ps(fjz2,tz); |
| 2434 | |
| 2435 | /************************** |
| 2436 | * CALCULATE INTERACTIONS * |
| 2437 | **************************/ |
| 2438 | |
| 2439 | r23 = _mm_mul_ps(rsq23,rinv23); |
| 2440 | r23 = _mm_andnot_ps(dummy_mask,r23); |
| 2441 | |
| 2442 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2443 | rt = _mm_mul_ps(r23,vftabscale); |
| 2444 | vfitab = _mm_cvttps_epi32(rt); |
| 2445 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2446 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2447 | |
| 2448 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2449 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2450 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2451 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2452 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2453 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2454 | Heps = _mm_mul_ps(vfeps,H); |
| 2455 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2456 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2457 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
| 2458 | |
| 2459 | fscal = felec; |
| 2460 | |
| 2461 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2462 | |
| 2463 | /* Calculate temporary vectorial force */ |
| 2464 | tx = _mm_mul_ps(fscal,dx23); |
| 2465 | ty = _mm_mul_ps(fscal,dy23); |
| 2466 | tz = _mm_mul_ps(fscal,dz23); |
| 2467 | |
| 2468 | /* Update vectorial force */ |
| 2469 | fix2 = _mm_add_ps(fix2,tx); |
| 2470 | fiy2 = _mm_add_ps(fiy2,ty); |
| 2471 | fiz2 = _mm_add_ps(fiz2,tz); |
| 2472 | |
| 2473 | fjx3 = _mm_add_ps(fjx3,tx); |
| 2474 | fjy3 = _mm_add_ps(fjy3,ty); |
| 2475 | fjz3 = _mm_add_ps(fjz3,tz); |
| 2476 | |
| 2477 | /************************** |
| 2478 | * CALCULATE INTERACTIONS * |
| 2479 | **************************/ |
| 2480 | |
| 2481 | r31 = _mm_mul_ps(rsq31,rinv31); |
| 2482 | r31 = _mm_andnot_ps(dummy_mask,r31); |
| 2483 | |
| 2484 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2485 | rt = _mm_mul_ps(r31,vftabscale); |
| 2486 | vfitab = _mm_cvttps_epi32(rt); |
| 2487 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2488 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2489 | |
| 2490 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2491 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2492 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2493 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2494 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2495 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2496 | Heps = _mm_mul_ps(vfeps,H); |
| 2497 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2498 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2499 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
| 2500 | |
| 2501 | fscal = felec; |
| 2502 | |
| 2503 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2504 | |
| 2505 | /* Calculate temporary vectorial force */ |
| 2506 | tx = _mm_mul_ps(fscal,dx31); |
| 2507 | ty = _mm_mul_ps(fscal,dy31); |
| 2508 | tz = _mm_mul_ps(fscal,dz31); |
| 2509 | |
| 2510 | /* Update vectorial force */ |
| 2511 | fix3 = _mm_add_ps(fix3,tx); |
| 2512 | fiy3 = _mm_add_ps(fiy3,ty); |
| 2513 | fiz3 = _mm_add_ps(fiz3,tz); |
| 2514 | |
| 2515 | fjx1 = _mm_add_ps(fjx1,tx); |
| 2516 | fjy1 = _mm_add_ps(fjy1,ty); |
| 2517 | fjz1 = _mm_add_ps(fjz1,tz); |
| 2518 | |
| 2519 | /************************** |
| 2520 | * CALCULATE INTERACTIONS * |
| 2521 | **************************/ |
| 2522 | |
| 2523 | r32 = _mm_mul_ps(rsq32,rinv32); |
| 2524 | r32 = _mm_andnot_ps(dummy_mask,r32); |
| 2525 | |
| 2526 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2527 | rt = _mm_mul_ps(r32,vftabscale); |
| 2528 | vfitab = _mm_cvttps_epi32(rt); |
| 2529 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2530 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2531 | |
| 2532 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2533 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2534 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2535 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2536 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2537 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2538 | Heps = _mm_mul_ps(vfeps,H); |
| 2539 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2540 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2541 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
| 2542 | |
| 2543 | fscal = felec; |
| 2544 | |
| 2545 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2546 | |
| 2547 | /* Calculate temporary vectorial force */ |
| 2548 | tx = _mm_mul_ps(fscal,dx32); |
| 2549 | ty = _mm_mul_ps(fscal,dy32); |
| 2550 | tz = _mm_mul_ps(fscal,dz32); |
| 2551 | |
| 2552 | /* Update vectorial force */ |
| 2553 | fix3 = _mm_add_ps(fix3,tx); |
| 2554 | fiy3 = _mm_add_ps(fiy3,ty); |
| 2555 | fiz3 = _mm_add_ps(fiz3,tz); |
| 2556 | |
| 2557 | fjx2 = _mm_add_ps(fjx2,tx); |
| 2558 | fjy2 = _mm_add_ps(fjy2,ty); |
| 2559 | fjz2 = _mm_add_ps(fjz2,tz); |
| 2560 | |
| 2561 | /************************** |
| 2562 | * CALCULATE INTERACTIONS * |
| 2563 | **************************/ |
| 2564 | |
| 2565 | r33 = _mm_mul_ps(rsq33,rinv33); |
| 2566 | r33 = _mm_andnot_ps(dummy_mask,r33); |
| 2567 | |
| 2568 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
| 2569 | rt = _mm_mul_ps(r33,vftabscale); |
| 2570 | vfitab = _mm_cvttps_epi32(rt); |
| 2571 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
| 2572 | vfitab = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2); |
| 2573 | |
| 2574 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
| 2575 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
| 2576 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
| 2577 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
| 2578 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
| 2579 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
| 2580 | Heps = _mm_mul_ps(vfeps,H); |
| 2581 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
| 2582 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
| 2583 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
| 2584 | |
| 2585 | fscal = felec; |
| 2586 | |
| 2587 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
| 2588 | |
| 2589 | /* Calculate temporary vectorial force */ |
| 2590 | tx = _mm_mul_ps(fscal,dx33); |
| 2591 | ty = _mm_mul_ps(fscal,dy33); |
| 2592 | tz = _mm_mul_ps(fscal,dz33); |
| 2593 | |
| 2594 | /* Update vectorial force */ |
| 2595 | fix3 = _mm_add_ps(fix3,tx); |
| 2596 | fiy3 = _mm_add_ps(fiy3,ty); |
| 2597 | fiz3 = _mm_add_ps(fiz3,tz); |
| 2598 | |
| 2599 | fjx3 = _mm_add_ps(fjx3,tx); |
| 2600 | fjy3 = _mm_add_ps(fjy3,ty); |
| 2601 | fjz3 = _mm_add_ps(fjz3,tz); |
| 2602 | |
| 2603 | fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch; |
| 2604 | fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch; |
| 2605 | fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch; |
| 2606 | fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch; |
| 2607 | |
| 2608 | gmx_mm_decrement_4rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
| 2609 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1, |
| 2610 | fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
| 2611 | |
| 2612 | /* Inner loop uses 412 flops */ |
| 2613 | } |
| 2614 | |
| 2615 | /* End of innermost loop */ |
| 2616 | |
| 2617 | gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3, |
| 2618 | f+i_coord_offset,fshift+i_shift_offset); |
| 2619 | |
| 2620 | /* Increment number of inner iterations */ |
| 2621 | inneriter += j_index_end - j_index_start; |
| 2622 | |
| 2623 | /* Outer loop uses 24 flops */ |
| 2624 | } |
| 2625 | |
| 2626 | /* Increment number of outer iterations */ |
| 2627 | outeriter += nri; |
| 2628 | |
| 2629 | /* Update outer/inner flops */ |
| 2630 | |
| 2631 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*24 + inneriter*412)(nrnb)->n[eNR_NBKERNEL_ELEC_VDW_W4W4_F] += outeriter*24 + inneriter *412; |
| 2632 | } |