2 * This file is part of the GROMACS molecular simulation package.
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.
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.
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.
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.
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.
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.
36 * Note: this file was generated by the GROMACS avx_256_single kernel generator.
42 #include "../nb_kernel.h"
43 #include "gromacs/legacyheaders/types/simple.h"
44 #include "gromacs/math/vec.h"
45 #include "gromacs/legacyheaders/nrnb.h"
47 #include "gromacs/simd/math_x86_avx_256_single.h"
48 #include "kernelutil_x86_avx_256_single.h"
51 * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwLJSh_GeomW4W4_VF_avx_256_single
52 * Electrostatics interaction: Ewald
53 * VdW interaction: LennardJones
54 * Geometry: Water4-Water4
55 * Calculate force/pot: PotentialAndForce
58 nb_kernel_ElecEwSh_VdwLJSh_GeomW4W4_VF_avx_256_single
59 (t_nblist * gmx_restrict nlist,
60 rvec * gmx_restrict xx,
61 rvec * gmx_restrict ff,
62 t_forcerec * gmx_restrict fr,
63 t_mdatoms * gmx_restrict mdatoms,
64 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
65 t_nrnb * gmx_restrict nrnb)
67 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
68 * just 0 for non-waters.
69 * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
70 * jnr indices corresponding to data put in the four positions in the SIMD register.
72 int i_shift_offset,i_coord_offset,outeriter,inneriter;
73 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
74 int jnrA,jnrB,jnrC,jnrD;
75 int jnrE,jnrF,jnrG,jnrH;
76 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
77 int jnrlistE,jnrlistF,jnrlistG,jnrlistH;
78 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
79 int j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
80 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
82 real *shiftvec,*fshift,*x,*f;
83 real *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
85 __m256 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
86 real * vdwioffsetptr0;
87 __m256 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
88 real * vdwioffsetptr1;
89 __m256 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
90 real * vdwioffsetptr2;
91 __m256 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
92 real * vdwioffsetptr3;
93 __m256 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
94 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
95 __m256 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
96 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
97 __m256 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
98 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
99 __m256 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
100 int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D,vdwjidx3E,vdwjidx3F,vdwjidx3G,vdwjidx3H;
101 __m256 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
102 __m256 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
103 __m256 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
104 __m256 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
105 __m256 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
106 __m256 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
107 __m256 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
108 __m256 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
109 __m256 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
110 __m256 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
111 __m256 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
112 __m256 velec,felec,velecsum,facel,crf,krf,krf2;
115 __m256 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
118 __m256 one_sixth = _mm256_set1_ps(1.0/6.0);
119 __m256 one_twelfth = _mm256_set1_ps(1.0/12.0);
121 __m128i ewitab_lo,ewitab_hi;
122 __m256 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
123 __m256 beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
125 __m256 dummy_mask,cutoff_mask;
126 __m256 signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
127 __m256 one = _mm256_set1_ps(1.0);
128 __m256 two = _mm256_set1_ps(2.0);
134 jindex = nlist->jindex;
136 shiftidx = nlist->shift;
138 shiftvec = fr->shift_vec[0];
139 fshift = fr->fshift[0];
140 facel = _mm256_set1_ps(fr->epsfac);
141 charge = mdatoms->chargeA;
142 nvdwtype = fr->ntype;
144 vdwtype = mdatoms->typeA;
146 sh_ewald = _mm256_set1_ps(fr->ic->sh_ewald);
147 beta = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
148 beta2 = _mm256_mul_ps(beta,beta);
149 beta3 = _mm256_mul_ps(beta,beta2);
151 ewtab = fr->ic->tabq_coul_FDV0;
152 ewtabscale = _mm256_set1_ps(fr->ic->tabq_scale);
153 ewtabhalfspace = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
155 /* Setup water-specific parameters */
156 inr = nlist->iinr[0];
157 iq1 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
158 iq2 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
159 iq3 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
160 vdwioffsetptr0 = vdwparam+2*nvdwtype*vdwtype[inr+0];
162 jq1 = _mm256_set1_ps(charge[inr+1]);
163 jq2 = _mm256_set1_ps(charge[inr+2]);
164 jq3 = _mm256_set1_ps(charge[inr+3]);
165 vdwjidx0A = 2*vdwtype[inr+0];
166 c6_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A]);
167 c12_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A+1]);
168 qq11 = _mm256_mul_ps(iq1,jq1);
169 qq12 = _mm256_mul_ps(iq1,jq2);
170 qq13 = _mm256_mul_ps(iq1,jq3);
171 qq21 = _mm256_mul_ps(iq2,jq1);
172 qq22 = _mm256_mul_ps(iq2,jq2);
173 qq23 = _mm256_mul_ps(iq2,jq3);
174 qq31 = _mm256_mul_ps(iq3,jq1);
175 qq32 = _mm256_mul_ps(iq3,jq2);
176 qq33 = _mm256_mul_ps(iq3,jq3);
178 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
179 rcutoff_scalar = fr->rcoulomb;
180 rcutoff = _mm256_set1_ps(rcutoff_scalar);
181 rcutoff2 = _mm256_mul_ps(rcutoff,rcutoff);
183 sh_vdw_invrcut6 = _mm256_set1_ps(fr->ic->sh_invrc6);
184 rvdw = _mm256_set1_ps(fr->rvdw);
186 /* Avoid stupid compiler warnings */
187 jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
200 for(iidx=0;iidx<4*DIM;iidx++)
205 /* Start outer loop over neighborlists */
206 for(iidx=0; iidx<nri; iidx++)
208 /* Load shift vector for this list */
209 i_shift_offset = DIM*shiftidx[iidx];
211 /* Load limits for loop over neighbors */
212 j_index_start = jindex[iidx];
213 j_index_end = jindex[iidx+1];
215 /* Get outer coordinate index */
217 i_coord_offset = DIM*inr;
219 /* Load i particle coords and add shift vector */
220 gmx_mm256_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
221 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
223 fix0 = _mm256_setzero_ps();
224 fiy0 = _mm256_setzero_ps();
225 fiz0 = _mm256_setzero_ps();
226 fix1 = _mm256_setzero_ps();
227 fiy1 = _mm256_setzero_ps();
228 fiz1 = _mm256_setzero_ps();
229 fix2 = _mm256_setzero_ps();
230 fiy2 = _mm256_setzero_ps();
231 fiz2 = _mm256_setzero_ps();
232 fix3 = _mm256_setzero_ps();
233 fiy3 = _mm256_setzero_ps();
234 fiz3 = _mm256_setzero_ps();
236 /* Reset potential sums */
237 velecsum = _mm256_setzero_ps();
238 vvdwsum = _mm256_setzero_ps();
240 /* Start inner kernel loop */
241 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
244 /* Get j neighbor index, and coordinate index */
253 j_coord_offsetA = DIM*jnrA;
254 j_coord_offsetB = DIM*jnrB;
255 j_coord_offsetC = DIM*jnrC;
256 j_coord_offsetD = DIM*jnrD;
257 j_coord_offsetE = DIM*jnrE;
258 j_coord_offsetF = DIM*jnrF;
259 j_coord_offsetG = DIM*jnrG;
260 j_coord_offsetH = DIM*jnrH;
262 /* load j atom coordinates */
263 gmx_mm256_load_4rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
264 x+j_coord_offsetC,x+j_coord_offsetD,
265 x+j_coord_offsetE,x+j_coord_offsetF,
266 x+j_coord_offsetG,x+j_coord_offsetH,
267 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
268 &jy2,&jz2,&jx3,&jy3,&jz3);
270 /* Calculate displacement vector */
271 dx00 = _mm256_sub_ps(ix0,jx0);
272 dy00 = _mm256_sub_ps(iy0,jy0);
273 dz00 = _mm256_sub_ps(iz0,jz0);
274 dx11 = _mm256_sub_ps(ix1,jx1);
275 dy11 = _mm256_sub_ps(iy1,jy1);
276 dz11 = _mm256_sub_ps(iz1,jz1);
277 dx12 = _mm256_sub_ps(ix1,jx2);
278 dy12 = _mm256_sub_ps(iy1,jy2);
279 dz12 = _mm256_sub_ps(iz1,jz2);
280 dx13 = _mm256_sub_ps(ix1,jx3);
281 dy13 = _mm256_sub_ps(iy1,jy3);
282 dz13 = _mm256_sub_ps(iz1,jz3);
283 dx21 = _mm256_sub_ps(ix2,jx1);
284 dy21 = _mm256_sub_ps(iy2,jy1);
285 dz21 = _mm256_sub_ps(iz2,jz1);
286 dx22 = _mm256_sub_ps(ix2,jx2);
287 dy22 = _mm256_sub_ps(iy2,jy2);
288 dz22 = _mm256_sub_ps(iz2,jz2);
289 dx23 = _mm256_sub_ps(ix2,jx3);
290 dy23 = _mm256_sub_ps(iy2,jy3);
291 dz23 = _mm256_sub_ps(iz2,jz3);
292 dx31 = _mm256_sub_ps(ix3,jx1);
293 dy31 = _mm256_sub_ps(iy3,jy1);
294 dz31 = _mm256_sub_ps(iz3,jz1);
295 dx32 = _mm256_sub_ps(ix3,jx2);
296 dy32 = _mm256_sub_ps(iy3,jy2);
297 dz32 = _mm256_sub_ps(iz3,jz2);
298 dx33 = _mm256_sub_ps(ix3,jx3);
299 dy33 = _mm256_sub_ps(iy3,jy3);
300 dz33 = _mm256_sub_ps(iz3,jz3);
302 /* Calculate squared distance and things based on it */
303 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
304 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
305 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
306 rsq13 = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
307 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
308 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
309 rsq23 = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
310 rsq31 = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
311 rsq32 = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
312 rsq33 = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
314 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
315 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
316 rinv13 = gmx_mm256_invsqrt_ps(rsq13);
317 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
318 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
319 rinv23 = gmx_mm256_invsqrt_ps(rsq23);
320 rinv31 = gmx_mm256_invsqrt_ps(rsq31);
321 rinv32 = gmx_mm256_invsqrt_ps(rsq32);
322 rinv33 = gmx_mm256_invsqrt_ps(rsq33);
324 rinvsq00 = gmx_mm256_inv_ps(rsq00);
325 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
326 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
327 rinvsq13 = _mm256_mul_ps(rinv13,rinv13);
328 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
329 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
330 rinvsq23 = _mm256_mul_ps(rinv23,rinv23);
331 rinvsq31 = _mm256_mul_ps(rinv31,rinv31);
332 rinvsq32 = _mm256_mul_ps(rinv32,rinv32);
333 rinvsq33 = _mm256_mul_ps(rinv33,rinv33);
335 fjx0 = _mm256_setzero_ps();
336 fjy0 = _mm256_setzero_ps();
337 fjz0 = _mm256_setzero_ps();
338 fjx1 = _mm256_setzero_ps();
339 fjy1 = _mm256_setzero_ps();
340 fjz1 = _mm256_setzero_ps();
341 fjx2 = _mm256_setzero_ps();
342 fjy2 = _mm256_setzero_ps();
343 fjz2 = _mm256_setzero_ps();
344 fjx3 = _mm256_setzero_ps();
345 fjy3 = _mm256_setzero_ps();
346 fjz3 = _mm256_setzero_ps();
348 /**************************
349 * CALCULATE INTERACTIONS *
350 **************************/
352 if (gmx_mm256_any_lt(rsq00,rcutoff2))
355 /* LENNARD-JONES DISPERSION/REPULSION */
357 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
358 vvdw6 = _mm256_mul_ps(c6_00,rinvsix);
359 vvdw12 = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
360 vvdw = _mm256_sub_ps(_mm256_mul_ps( _mm256_sub_ps(vvdw12 , _mm256_mul_ps(c12_00,_mm256_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6))), one_twelfth) ,
361 _mm256_mul_ps( _mm256_sub_ps(vvdw6,_mm256_mul_ps(c6_00,sh_vdw_invrcut6)),one_sixth));
362 fvdw = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
364 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
366 /* Update potential sum for this i atom from the interaction with this j atom. */
367 vvdw = _mm256_and_ps(vvdw,cutoff_mask);
368 vvdwsum = _mm256_add_ps(vvdwsum,vvdw);
372 fscal = _mm256_and_ps(fscal,cutoff_mask);
374 /* Calculate temporary vectorial force */
375 tx = _mm256_mul_ps(fscal,dx00);
376 ty = _mm256_mul_ps(fscal,dy00);
377 tz = _mm256_mul_ps(fscal,dz00);
379 /* Update vectorial force */
380 fix0 = _mm256_add_ps(fix0,tx);
381 fiy0 = _mm256_add_ps(fiy0,ty);
382 fiz0 = _mm256_add_ps(fiz0,tz);
384 fjx0 = _mm256_add_ps(fjx0,tx);
385 fjy0 = _mm256_add_ps(fjy0,ty);
386 fjz0 = _mm256_add_ps(fjz0,tz);
390 /**************************
391 * CALCULATE INTERACTIONS *
392 **************************/
394 if (gmx_mm256_any_lt(rsq11,rcutoff2))
397 r11 = _mm256_mul_ps(rsq11,rinv11);
399 /* EWALD ELECTROSTATICS */
401 /* Analytical PME correction */
402 zeta2 = _mm256_mul_ps(beta2,rsq11);
403 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
404 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
405 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
406 felec = _mm256_mul_ps(qq11,felec);
407 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
408 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
409 velec = _mm256_sub_ps(_mm256_sub_ps(rinv11,sh_ewald),pmecorrV);
410 velec = _mm256_mul_ps(qq11,velec);
412 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
414 /* Update potential sum for this i atom from the interaction with this j atom. */
415 velec = _mm256_and_ps(velec,cutoff_mask);
416 velecsum = _mm256_add_ps(velecsum,velec);
420 fscal = _mm256_and_ps(fscal,cutoff_mask);
422 /* Calculate temporary vectorial force */
423 tx = _mm256_mul_ps(fscal,dx11);
424 ty = _mm256_mul_ps(fscal,dy11);
425 tz = _mm256_mul_ps(fscal,dz11);
427 /* Update vectorial force */
428 fix1 = _mm256_add_ps(fix1,tx);
429 fiy1 = _mm256_add_ps(fiy1,ty);
430 fiz1 = _mm256_add_ps(fiz1,tz);
432 fjx1 = _mm256_add_ps(fjx1,tx);
433 fjy1 = _mm256_add_ps(fjy1,ty);
434 fjz1 = _mm256_add_ps(fjz1,tz);
438 /**************************
439 * CALCULATE INTERACTIONS *
440 **************************/
442 if (gmx_mm256_any_lt(rsq12,rcutoff2))
445 r12 = _mm256_mul_ps(rsq12,rinv12);
447 /* EWALD ELECTROSTATICS */
449 /* Analytical PME correction */
450 zeta2 = _mm256_mul_ps(beta2,rsq12);
451 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
452 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
453 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
454 felec = _mm256_mul_ps(qq12,felec);
455 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
456 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
457 velec = _mm256_sub_ps(_mm256_sub_ps(rinv12,sh_ewald),pmecorrV);
458 velec = _mm256_mul_ps(qq12,velec);
460 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
462 /* Update potential sum for this i atom from the interaction with this j atom. */
463 velec = _mm256_and_ps(velec,cutoff_mask);
464 velecsum = _mm256_add_ps(velecsum,velec);
468 fscal = _mm256_and_ps(fscal,cutoff_mask);
470 /* Calculate temporary vectorial force */
471 tx = _mm256_mul_ps(fscal,dx12);
472 ty = _mm256_mul_ps(fscal,dy12);
473 tz = _mm256_mul_ps(fscal,dz12);
475 /* Update vectorial force */
476 fix1 = _mm256_add_ps(fix1,tx);
477 fiy1 = _mm256_add_ps(fiy1,ty);
478 fiz1 = _mm256_add_ps(fiz1,tz);
480 fjx2 = _mm256_add_ps(fjx2,tx);
481 fjy2 = _mm256_add_ps(fjy2,ty);
482 fjz2 = _mm256_add_ps(fjz2,tz);
486 /**************************
487 * CALCULATE INTERACTIONS *
488 **************************/
490 if (gmx_mm256_any_lt(rsq13,rcutoff2))
493 r13 = _mm256_mul_ps(rsq13,rinv13);
495 /* EWALD ELECTROSTATICS */
497 /* Analytical PME correction */
498 zeta2 = _mm256_mul_ps(beta2,rsq13);
499 rinv3 = _mm256_mul_ps(rinvsq13,rinv13);
500 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
501 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
502 felec = _mm256_mul_ps(qq13,felec);
503 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
504 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
505 velec = _mm256_sub_ps(_mm256_sub_ps(rinv13,sh_ewald),pmecorrV);
506 velec = _mm256_mul_ps(qq13,velec);
508 cutoff_mask = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
510 /* Update potential sum for this i atom from the interaction with this j atom. */
511 velec = _mm256_and_ps(velec,cutoff_mask);
512 velecsum = _mm256_add_ps(velecsum,velec);
516 fscal = _mm256_and_ps(fscal,cutoff_mask);
518 /* Calculate temporary vectorial force */
519 tx = _mm256_mul_ps(fscal,dx13);
520 ty = _mm256_mul_ps(fscal,dy13);
521 tz = _mm256_mul_ps(fscal,dz13);
523 /* Update vectorial force */
524 fix1 = _mm256_add_ps(fix1,tx);
525 fiy1 = _mm256_add_ps(fiy1,ty);
526 fiz1 = _mm256_add_ps(fiz1,tz);
528 fjx3 = _mm256_add_ps(fjx3,tx);
529 fjy3 = _mm256_add_ps(fjy3,ty);
530 fjz3 = _mm256_add_ps(fjz3,tz);
534 /**************************
535 * CALCULATE INTERACTIONS *
536 **************************/
538 if (gmx_mm256_any_lt(rsq21,rcutoff2))
541 r21 = _mm256_mul_ps(rsq21,rinv21);
543 /* EWALD ELECTROSTATICS */
545 /* Analytical PME correction */
546 zeta2 = _mm256_mul_ps(beta2,rsq21);
547 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
548 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
549 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
550 felec = _mm256_mul_ps(qq21,felec);
551 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
552 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
553 velec = _mm256_sub_ps(_mm256_sub_ps(rinv21,sh_ewald),pmecorrV);
554 velec = _mm256_mul_ps(qq21,velec);
556 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
558 /* Update potential sum for this i atom from the interaction with this j atom. */
559 velec = _mm256_and_ps(velec,cutoff_mask);
560 velecsum = _mm256_add_ps(velecsum,velec);
564 fscal = _mm256_and_ps(fscal,cutoff_mask);
566 /* Calculate temporary vectorial force */
567 tx = _mm256_mul_ps(fscal,dx21);
568 ty = _mm256_mul_ps(fscal,dy21);
569 tz = _mm256_mul_ps(fscal,dz21);
571 /* Update vectorial force */
572 fix2 = _mm256_add_ps(fix2,tx);
573 fiy2 = _mm256_add_ps(fiy2,ty);
574 fiz2 = _mm256_add_ps(fiz2,tz);
576 fjx1 = _mm256_add_ps(fjx1,tx);
577 fjy1 = _mm256_add_ps(fjy1,ty);
578 fjz1 = _mm256_add_ps(fjz1,tz);
582 /**************************
583 * CALCULATE INTERACTIONS *
584 **************************/
586 if (gmx_mm256_any_lt(rsq22,rcutoff2))
589 r22 = _mm256_mul_ps(rsq22,rinv22);
591 /* EWALD ELECTROSTATICS */
593 /* Analytical PME correction */
594 zeta2 = _mm256_mul_ps(beta2,rsq22);
595 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
596 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
597 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
598 felec = _mm256_mul_ps(qq22,felec);
599 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
600 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
601 velec = _mm256_sub_ps(_mm256_sub_ps(rinv22,sh_ewald),pmecorrV);
602 velec = _mm256_mul_ps(qq22,velec);
604 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
606 /* Update potential sum for this i atom from the interaction with this j atom. */
607 velec = _mm256_and_ps(velec,cutoff_mask);
608 velecsum = _mm256_add_ps(velecsum,velec);
612 fscal = _mm256_and_ps(fscal,cutoff_mask);
614 /* Calculate temporary vectorial force */
615 tx = _mm256_mul_ps(fscal,dx22);
616 ty = _mm256_mul_ps(fscal,dy22);
617 tz = _mm256_mul_ps(fscal,dz22);
619 /* Update vectorial force */
620 fix2 = _mm256_add_ps(fix2,tx);
621 fiy2 = _mm256_add_ps(fiy2,ty);
622 fiz2 = _mm256_add_ps(fiz2,tz);
624 fjx2 = _mm256_add_ps(fjx2,tx);
625 fjy2 = _mm256_add_ps(fjy2,ty);
626 fjz2 = _mm256_add_ps(fjz2,tz);
630 /**************************
631 * CALCULATE INTERACTIONS *
632 **************************/
634 if (gmx_mm256_any_lt(rsq23,rcutoff2))
637 r23 = _mm256_mul_ps(rsq23,rinv23);
639 /* EWALD ELECTROSTATICS */
641 /* Analytical PME correction */
642 zeta2 = _mm256_mul_ps(beta2,rsq23);
643 rinv3 = _mm256_mul_ps(rinvsq23,rinv23);
644 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
645 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
646 felec = _mm256_mul_ps(qq23,felec);
647 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
648 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
649 velec = _mm256_sub_ps(_mm256_sub_ps(rinv23,sh_ewald),pmecorrV);
650 velec = _mm256_mul_ps(qq23,velec);
652 cutoff_mask = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
654 /* Update potential sum for this i atom from the interaction with this j atom. */
655 velec = _mm256_and_ps(velec,cutoff_mask);
656 velecsum = _mm256_add_ps(velecsum,velec);
660 fscal = _mm256_and_ps(fscal,cutoff_mask);
662 /* Calculate temporary vectorial force */
663 tx = _mm256_mul_ps(fscal,dx23);
664 ty = _mm256_mul_ps(fscal,dy23);
665 tz = _mm256_mul_ps(fscal,dz23);
667 /* Update vectorial force */
668 fix2 = _mm256_add_ps(fix2,tx);
669 fiy2 = _mm256_add_ps(fiy2,ty);
670 fiz2 = _mm256_add_ps(fiz2,tz);
672 fjx3 = _mm256_add_ps(fjx3,tx);
673 fjy3 = _mm256_add_ps(fjy3,ty);
674 fjz3 = _mm256_add_ps(fjz3,tz);
678 /**************************
679 * CALCULATE INTERACTIONS *
680 **************************/
682 if (gmx_mm256_any_lt(rsq31,rcutoff2))
685 r31 = _mm256_mul_ps(rsq31,rinv31);
687 /* EWALD ELECTROSTATICS */
689 /* Analytical PME correction */
690 zeta2 = _mm256_mul_ps(beta2,rsq31);
691 rinv3 = _mm256_mul_ps(rinvsq31,rinv31);
692 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
693 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
694 felec = _mm256_mul_ps(qq31,felec);
695 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
696 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
697 velec = _mm256_sub_ps(_mm256_sub_ps(rinv31,sh_ewald),pmecorrV);
698 velec = _mm256_mul_ps(qq31,velec);
700 cutoff_mask = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
702 /* Update potential sum for this i atom from the interaction with this j atom. */
703 velec = _mm256_and_ps(velec,cutoff_mask);
704 velecsum = _mm256_add_ps(velecsum,velec);
708 fscal = _mm256_and_ps(fscal,cutoff_mask);
710 /* Calculate temporary vectorial force */
711 tx = _mm256_mul_ps(fscal,dx31);
712 ty = _mm256_mul_ps(fscal,dy31);
713 tz = _mm256_mul_ps(fscal,dz31);
715 /* Update vectorial force */
716 fix3 = _mm256_add_ps(fix3,tx);
717 fiy3 = _mm256_add_ps(fiy3,ty);
718 fiz3 = _mm256_add_ps(fiz3,tz);
720 fjx1 = _mm256_add_ps(fjx1,tx);
721 fjy1 = _mm256_add_ps(fjy1,ty);
722 fjz1 = _mm256_add_ps(fjz1,tz);
726 /**************************
727 * CALCULATE INTERACTIONS *
728 **************************/
730 if (gmx_mm256_any_lt(rsq32,rcutoff2))
733 r32 = _mm256_mul_ps(rsq32,rinv32);
735 /* EWALD ELECTROSTATICS */
737 /* Analytical PME correction */
738 zeta2 = _mm256_mul_ps(beta2,rsq32);
739 rinv3 = _mm256_mul_ps(rinvsq32,rinv32);
740 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
741 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
742 felec = _mm256_mul_ps(qq32,felec);
743 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
744 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
745 velec = _mm256_sub_ps(_mm256_sub_ps(rinv32,sh_ewald),pmecorrV);
746 velec = _mm256_mul_ps(qq32,velec);
748 cutoff_mask = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
750 /* Update potential sum for this i atom from the interaction with this j atom. */
751 velec = _mm256_and_ps(velec,cutoff_mask);
752 velecsum = _mm256_add_ps(velecsum,velec);
756 fscal = _mm256_and_ps(fscal,cutoff_mask);
758 /* Calculate temporary vectorial force */
759 tx = _mm256_mul_ps(fscal,dx32);
760 ty = _mm256_mul_ps(fscal,dy32);
761 tz = _mm256_mul_ps(fscal,dz32);
763 /* Update vectorial force */
764 fix3 = _mm256_add_ps(fix3,tx);
765 fiy3 = _mm256_add_ps(fiy3,ty);
766 fiz3 = _mm256_add_ps(fiz3,tz);
768 fjx2 = _mm256_add_ps(fjx2,tx);
769 fjy2 = _mm256_add_ps(fjy2,ty);
770 fjz2 = _mm256_add_ps(fjz2,tz);
774 /**************************
775 * CALCULATE INTERACTIONS *
776 **************************/
778 if (gmx_mm256_any_lt(rsq33,rcutoff2))
781 r33 = _mm256_mul_ps(rsq33,rinv33);
783 /* EWALD ELECTROSTATICS */
785 /* Analytical PME correction */
786 zeta2 = _mm256_mul_ps(beta2,rsq33);
787 rinv3 = _mm256_mul_ps(rinvsq33,rinv33);
788 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
789 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
790 felec = _mm256_mul_ps(qq33,felec);
791 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
792 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
793 velec = _mm256_sub_ps(_mm256_sub_ps(rinv33,sh_ewald),pmecorrV);
794 velec = _mm256_mul_ps(qq33,velec);
796 cutoff_mask = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
798 /* Update potential sum for this i atom from the interaction with this j atom. */
799 velec = _mm256_and_ps(velec,cutoff_mask);
800 velecsum = _mm256_add_ps(velecsum,velec);
804 fscal = _mm256_and_ps(fscal,cutoff_mask);
806 /* Calculate temporary vectorial force */
807 tx = _mm256_mul_ps(fscal,dx33);
808 ty = _mm256_mul_ps(fscal,dy33);
809 tz = _mm256_mul_ps(fscal,dz33);
811 /* Update vectorial force */
812 fix3 = _mm256_add_ps(fix3,tx);
813 fiy3 = _mm256_add_ps(fiy3,ty);
814 fiz3 = _mm256_add_ps(fiz3,tz);
816 fjx3 = _mm256_add_ps(fjx3,tx);
817 fjy3 = _mm256_add_ps(fjy3,ty);
818 fjz3 = _mm256_add_ps(fjz3,tz);
822 fjptrA = f+j_coord_offsetA;
823 fjptrB = f+j_coord_offsetB;
824 fjptrC = f+j_coord_offsetC;
825 fjptrD = f+j_coord_offsetD;
826 fjptrE = f+j_coord_offsetE;
827 fjptrF = f+j_coord_offsetF;
828 fjptrG = f+j_coord_offsetG;
829 fjptrH = f+j_coord_offsetH;
831 gmx_mm256_decrement_4rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
832 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,
833 fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
835 /* Inner loop uses 1025 flops */
841 /* Get j neighbor index, and coordinate index */
842 jnrlistA = jjnr[jidx];
843 jnrlistB = jjnr[jidx+1];
844 jnrlistC = jjnr[jidx+2];
845 jnrlistD = jjnr[jidx+3];
846 jnrlistE = jjnr[jidx+4];
847 jnrlistF = jjnr[jidx+5];
848 jnrlistG = jjnr[jidx+6];
849 jnrlistH = jjnr[jidx+7];
850 /* Sign of each element will be negative for non-real atoms.
851 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
852 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
854 dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
855 gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
857 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
858 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
859 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
860 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
861 jnrE = (jnrlistE>=0) ? jnrlistE : 0;
862 jnrF = (jnrlistF>=0) ? jnrlistF : 0;
863 jnrG = (jnrlistG>=0) ? jnrlistG : 0;
864 jnrH = (jnrlistH>=0) ? jnrlistH : 0;
865 j_coord_offsetA = DIM*jnrA;
866 j_coord_offsetB = DIM*jnrB;
867 j_coord_offsetC = DIM*jnrC;
868 j_coord_offsetD = DIM*jnrD;
869 j_coord_offsetE = DIM*jnrE;
870 j_coord_offsetF = DIM*jnrF;
871 j_coord_offsetG = DIM*jnrG;
872 j_coord_offsetH = DIM*jnrH;
874 /* load j atom coordinates */
875 gmx_mm256_load_4rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
876 x+j_coord_offsetC,x+j_coord_offsetD,
877 x+j_coord_offsetE,x+j_coord_offsetF,
878 x+j_coord_offsetG,x+j_coord_offsetH,
879 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
880 &jy2,&jz2,&jx3,&jy3,&jz3);
882 /* Calculate displacement vector */
883 dx00 = _mm256_sub_ps(ix0,jx0);
884 dy00 = _mm256_sub_ps(iy0,jy0);
885 dz00 = _mm256_sub_ps(iz0,jz0);
886 dx11 = _mm256_sub_ps(ix1,jx1);
887 dy11 = _mm256_sub_ps(iy1,jy1);
888 dz11 = _mm256_sub_ps(iz1,jz1);
889 dx12 = _mm256_sub_ps(ix1,jx2);
890 dy12 = _mm256_sub_ps(iy1,jy2);
891 dz12 = _mm256_sub_ps(iz1,jz2);
892 dx13 = _mm256_sub_ps(ix1,jx3);
893 dy13 = _mm256_sub_ps(iy1,jy3);
894 dz13 = _mm256_sub_ps(iz1,jz3);
895 dx21 = _mm256_sub_ps(ix2,jx1);
896 dy21 = _mm256_sub_ps(iy2,jy1);
897 dz21 = _mm256_sub_ps(iz2,jz1);
898 dx22 = _mm256_sub_ps(ix2,jx2);
899 dy22 = _mm256_sub_ps(iy2,jy2);
900 dz22 = _mm256_sub_ps(iz2,jz2);
901 dx23 = _mm256_sub_ps(ix2,jx3);
902 dy23 = _mm256_sub_ps(iy2,jy3);
903 dz23 = _mm256_sub_ps(iz2,jz3);
904 dx31 = _mm256_sub_ps(ix3,jx1);
905 dy31 = _mm256_sub_ps(iy3,jy1);
906 dz31 = _mm256_sub_ps(iz3,jz1);
907 dx32 = _mm256_sub_ps(ix3,jx2);
908 dy32 = _mm256_sub_ps(iy3,jy2);
909 dz32 = _mm256_sub_ps(iz3,jz2);
910 dx33 = _mm256_sub_ps(ix3,jx3);
911 dy33 = _mm256_sub_ps(iy3,jy3);
912 dz33 = _mm256_sub_ps(iz3,jz3);
914 /* Calculate squared distance and things based on it */
915 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
916 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
917 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
918 rsq13 = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
919 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
920 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
921 rsq23 = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
922 rsq31 = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
923 rsq32 = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
924 rsq33 = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
926 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
927 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
928 rinv13 = gmx_mm256_invsqrt_ps(rsq13);
929 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
930 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
931 rinv23 = gmx_mm256_invsqrt_ps(rsq23);
932 rinv31 = gmx_mm256_invsqrt_ps(rsq31);
933 rinv32 = gmx_mm256_invsqrt_ps(rsq32);
934 rinv33 = gmx_mm256_invsqrt_ps(rsq33);
936 rinvsq00 = gmx_mm256_inv_ps(rsq00);
937 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
938 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
939 rinvsq13 = _mm256_mul_ps(rinv13,rinv13);
940 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
941 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
942 rinvsq23 = _mm256_mul_ps(rinv23,rinv23);
943 rinvsq31 = _mm256_mul_ps(rinv31,rinv31);
944 rinvsq32 = _mm256_mul_ps(rinv32,rinv32);
945 rinvsq33 = _mm256_mul_ps(rinv33,rinv33);
947 fjx0 = _mm256_setzero_ps();
948 fjy0 = _mm256_setzero_ps();
949 fjz0 = _mm256_setzero_ps();
950 fjx1 = _mm256_setzero_ps();
951 fjy1 = _mm256_setzero_ps();
952 fjz1 = _mm256_setzero_ps();
953 fjx2 = _mm256_setzero_ps();
954 fjy2 = _mm256_setzero_ps();
955 fjz2 = _mm256_setzero_ps();
956 fjx3 = _mm256_setzero_ps();
957 fjy3 = _mm256_setzero_ps();
958 fjz3 = _mm256_setzero_ps();
960 /**************************
961 * CALCULATE INTERACTIONS *
962 **************************/
964 if (gmx_mm256_any_lt(rsq00,rcutoff2))
967 /* LENNARD-JONES DISPERSION/REPULSION */
969 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
970 vvdw6 = _mm256_mul_ps(c6_00,rinvsix);
971 vvdw12 = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
972 vvdw = _mm256_sub_ps(_mm256_mul_ps( _mm256_sub_ps(vvdw12 , _mm256_mul_ps(c12_00,_mm256_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6))), one_twelfth) ,
973 _mm256_mul_ps( _mm256_sub_ps(vvdw6,_mm256_mul_ps(c6_00,sh_vdw_invrcut6)),one_sixth));
974 fvdw = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
976 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
978 /* Update potential sum for this i atom from the interaction with this j atom. */
979 vvdw = _mm256_and_ps(vvdw,cutoff_mask);
980 vvdw = _mm256_andnot_ps(dummy_mask,vvdw);
981 vvdwsum = _mm256_add_ps(vvdwsum,vvdw);
985 fscal = _mm256_and_ps(fscal,cutoff_mask);
987 fscal = _mm256_andnot_ps(dummy_mask,fscal);
989 /* Calculate temporary vectorial force */
990 tx = _mm256_mul_ps(fscal,dx00);
991 ty = _mm256_mul_ps(fscal,dy00);
992 tz = _mm256_mul_ps(fscal,dz00);
994 /* Update vectorial force */
995 fix0 = _mm256_add_ps(fix0,tx);
996 fiy0 = _mm256_add_ps(fiy0,ty);
997 fiz0 = _mm256_add_ps(fiz0,tz);
999 fjx0 = _mm256_add_ps(fjx0,tx);
1000 fjy0 = _mm256_add_ps(fjy0,ty);
1001 fjz0 = _mm256_add_ps(fjz0,tz);
1005 /**************************
1006 * CALCULATE INTERACTIONS *
1007 **************************/
1009 if (gmx_mm256_any_lt(rsq11,rcutoff2))
1012 r11 = _mm256_mul_ps(rsq11,rinv11);
1013 r11 = _mm256_andnot_ps(dummy_mask,r11);
1015 /* EWALD ELECTROSTATICS */
1017 /* Analytical PME correction */
1018 zeta2 = _mm256_mul_ps(beta2,rsq11);
1019 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
1020 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1021 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1022 felec = _mm256_mul_ps(qq11,felec);
1023 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1024 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1025 velec = _mm256_sub_ps(_mm256_sub_ps(rinv11,sh_ewald),pmecorrV);
1026 velec = _mm256_mul_ps(qq11,velec);
1028 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
1030 /* Update potential sum for this i atom from the interaction with this j atom. */
1031 velec = _mm256_and_ps(velec,cutoff_mask);
1032 velec = _mm256_andnot_ps(dummy_mask,velec);
1033 velecsum = _mm256_add_ps(velecsum,velec);
1037 fscal = _mm256_and_ps(fscal,cutoff_mask);
1039 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1041 /* Calculate temporary vectorial force */
1042 tx = _mm256_mul_ps(fscal,dx11);
1043 ty = _mm256_mul_ps(fscal,dy11);
1044 tz = _mm256_mul_ps(fscal,dz11);
1046 /* Update vectorial force */
1047 fix1 = _mm256_add_ps(fix1,tx);
1048 fiy1 = _mm256_add_ps(fiy1,ty);
1049 fiz1 = _mm256_add_ps(fiz1,tz);
1051 fjx1 = _mm256_add_ps(fjx1,tx);
1052 fjy1 = _mm256_add_ps(fjy1,ty);
1053 fjz1 = _mm256_add_ps(fjz1,tz);
1057 /**************************
1058 * CALCULATE INTERACTIONS *
1059 **************************/
1061 if (gmx_mm256_any_lt(rsq12,rcutoff2))
1064 r12 = _mm256_mul_ps(rsq12,rinv12);
1065 r12 = _mm256_andnot_ps(dummy_mask,r12);
1067 /* EWALD ELECTROSTATICS */
1069 /* Analytical PME correction */
1070 zeta2 = _mm256_mul_ps(beta2,rsq12);
1071 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
1072 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1073 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1074 felec = _mm256_mul_ps(qq12,felec);
1075 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1076 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1077 velec = _mm256_sub_ps(_mm256_sub_ps(rinv12,sh_ewald),pmecorrV);
1078 velec = _mm256_mul_ps(qq12,velec);
1080 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
1082 /* Update potential sum for this i atom from the interaction with this j atom. */
1083 velec = _mm256_and_ps(velec,cutoff_mask);
1084 velec = _mm256_andnot_ps(dummy_mask,velec);
1085 velecsum = _mm256_add_ps(velecsum,velec);
1089 fscal = _mm256_and_ps(fscal,cutoff_mask);
1091 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1093 /* Calculate temporary vectorial force */
1094 tx = _mm256_mul_ps(fscal,dx12);
1095 ty = _mm256_mul_ps(fscal,dy12);
1096 tz = _mm256_mul_ps(fscal,dz12);
1098 /* Update vectorial force */
1099 fix1 = _mm256_add_ps(fix1,tx);
1100 fiy1 = _mm256_add_ps(fiy1,ty);
1101 fiz1 = _mm256_add_ps(fiz1,tz);
1103 fjx2 = _mm256_add_ps(fjx2,tx);
1104 fjy2 = _mm256_add_ps(fjy2,ty);
1105 fjz2 = _mm256_add_ps(fjz2,tz);
1109 /**************************
1110 * CALCULATE INTERACTIONS *
1111 **************************/
1113 if (gmx_mm256_any_lt(rsq13,rcutoff2))
1116 r13 = _mm256_mul_ps(rsq13,rinv13);
1117 r13 = _mm256_andnot_ps(dummy_mask,r13);
1119 /* EWALD ELECTROSTATICS */
1121 /* Analytical PME correction */
1122 zeta2 = _mm256_mul_ps(beta2,rsq13);
1123 rinv3 = _mm256_mul_ps(rinvsq13,rinv13);
1124 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1125 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1126 felec = _mm256_mul_ps(qq13,felec);
1127 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1128 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1129 velec = _mm256_sub_ps(_mm256_sub_ps(rinv13,sh_ewald),pmecorrV);
1130 velec = _mm256_mul_ps(qq13,velec);
1132 cutoff_mask = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
1134 /* Update potential sum for this i atom from the interaction with this j atom. */
1135 velec = _mm256_and_ps(velec,cutoff_mask);
1136 velec = _mm256_andnot_ps(dummy_mask,velec);
1137 velecsum = _mm256_add_ps(velecsum,velec);
1141 fscal = _mm256_and_ps(fscal,cutoff_mask);
1143 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1145 /* Calculate temporary vectorial force */
1146 tx = _mm256_mul_ps(fscal,dx13);
1147 ty = _mm256_mul_ps(fscal,dy13);
1148 tz = _mm256_mul_ps(fscal,dz13);
1150 /* Update vectorial force */
1151 fix1 = _mm256_add_ps(fix1,tx);
1152 fiy1 = _mm256_add_ps(fiy1,ty);
1153 fiz1 = _mm256_add_ps(fiz1,tz);
1155 fjx3 = _mm256_add_ps(fjx3,tx);
1156 fjy3 = _mm256_add_ps(fjy3,ty);
1157 fjz3 = _mm256_add_ps(fjz3,tz);
1161 /**************************
1162 * CALCULATE INTERACTIONS *
1163 **************************/
1165 if (gmx_mm256_any_lt(rsq21,rcutoff2))
1168 r21 = _mm256_mul_ps(rsq21,rinv21);
1169 r21 = _mm256_andnot_ps(dummy_mask,r21);
1171 /* EWALD ELECTROSTATICS */
1173 /* Analytical PME correction */
1174 zeta2 = _mm256_mul_ps(beta2,rsq21);
1175 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
1176 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1177 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1178 felec = _mm256_mul_ps(qq21,felec);
1179 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1180 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1181 velec = _mm256_sub_ps(_mm256_sub_ps(rinv21,sh_ewald),pmecorrV);
1182 velec = _mm256_mul_ps(qq21,velec);
1184 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
1186 /* Update potential sum for this i atom from the interaction with this j atom. */
1187 velec = _mm256_and_ps(velec,cutoff_mask);
1188 velec = _mm256_andnot_ps(dummy_mask,velec);
1189 velecsum = _mm256_add_ps(velecsum,velec);
1193 fscal = _mm256_and_ps(fscal,cutoff_mask);
1195 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1197 /* Calculate temporary vectorial force */
1198 tx = _mm256_mul_ps(fscal,dx21);
1199 ty = _mm256_mul_ps(fscal,dy21);
1200 tz = _mm256_mul_ps(fscal,dz21);
1202 /* Update vectorial force */
1203 fix2 = _mm256_add_ps(fix2,tx);
1204 fiy2 = _mm256_add_ps(fiy2,ty);
1205 fiz2 = _mm256_add_ps(fiz2,tz);
1207 fjx1 = _mm256_add_ps(fjx1,tx);
1208 fjy1 = _mm256_add_ps(fjy1,ty);
1209 fjz1 = _mm256_add_ps(fjz1,tz);
1213 /**************************
1214 * CALCULATE INTERACTIONS *
1215 **************************/
1217 if (gmx_mm256_any_lt(rsq22,rcutoff2))
1220 r22 = _mm256_mul_ps(rsq22,rinv22);
1221 r22 = _mm256_andnot_ps(dummy_mask,r22);
1223 /* EWALD ELECTROSTATICS */
1225 /* Analytical PME correction */
1226 zeta2 = _mm256_mul_ps(beta2,rsq22);
1227 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
1228 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1229 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1230 felec = _mm256_mul_ps(qq22,felec);
1231 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1232 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1233 velec = _mm256_sub_ps(_mm256_sub_ps(rinv22,sh_ewald),pmecorrV);
1234 velec = _mm256_mul_ps(qq22,velec);
1236 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
1238 /* Update potential sum for this i atom from the interaction with this j atom. */
1239 velec = _mm256_and_ps(velec,cutoff_mask);
1240 velec = _mm256_andnot_ps(dummy_mask,velec);
1241 velecsum = _mm256_add_ps(velecsum,velec);
1245 fscal = _mm256_and_ps(fscal,cutoff_mask);
1247 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1249 /* Calculate temporary vectorial force */
1250 tx = _mm256_mul_ps(fscal,dx22);
1251 ty = _mm256_mul_ps(fscal,dy22);
1252 tz = _mm256_mul_ps(fscal,dz22);
1254 /* Update vectorial force */
1255 fix2 = _mm256_add_ps(fix2,tx);
1256 fiy2 = _mm256_add_ps(fiy2,ty);
1257 fiz2 = _mm256_add_ps(fiz2,tz);
1259 fjx2 = _mm256_add_ps(fjx2,tx);
1260 fjy2 = _mm256_add_ps(fjy2,ty);
1261 fjz2 = _mm256_add_ps(fjz2,tz);
1265 /**************************
1266 * CALCULATE INTERACTIONS *
1267 **************************/
1269 if (gmx_mm256_any_lt(rsq23,rcutoff2))
1272 r23 = _mm256_mul_ps(rsq23,rinv23);
1273 r23 = _mm256_andnot_ps(dummy_mask,r23);
1275 /* EWALD ELECTROSTATICS */
1277 /* Analytical PME correction */
1278 zeta2 = _mm256_mul_ps(beta2,rsq23);
1279 rinv3 = _mm256_mul_ps(rinvsq23,rinv23);
1280 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1281 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1282 felec = _mm256_mul_ps(qq23,felec);
1283 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1284 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1285 velec = _mm256_sub_ps(_mm256_sub_ps(rinv23,sh_ewald),pmecorrV);
1286 velec = _mm256_mul_ps(qq23,velec);
1288 cutoff_mask = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
1290 /* Update potential sum for this i atom from the interaction with this j atom. */
1291 velec = _mm256_and_ps(velec,cutoff_mask);
1292 velec = _mm256_andnot_ps(dummy_mask,velec);
1293 velecsum = _mm256_add_ps(velecsum,velec);
1297 fscal = _mm256_and_ps(fscal,cutoff_mask);
1299 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1301 /* Calculate temporary vectorial force */
1302 tx = _mm256_mul_ps(fscal,dx23);
1303 ty = _mm256_mul_ps(fscal,dy23);
1304 tz = _mm256_mul_ps(fscal,dz23);
1306 /* Update vectorial force */
1307 fix2 = _mm256_add_ps(fix2,tx);
1308 fiy2 = _mm256_add_ps(fiy2,ty);
1309 fiz2 = _mm256_add_ps(fiz2,tz);
1311 fjx3 = _mm256_add_ps(fjx3,tx);
1312 fjy3 = _mm256_add_ps(fjy3,ty);
1313 fjz3 = _mm256_add_ps(fjz3,tz);
1317 /**************************
1318 * CALCULATE INTERACTIONS *
1319 **************************/
1321 if (gmx_mm256_any_lt(rsq31,rcutoff2))
1324 r31 = _mm256_mul_ps(rsq31,rinv31);
1325 r31 = _mm256_andnot_ps(dummy_mask,r31);
1327 /* EWALD ELECTROSTATICS */
1329 /* Analytical PME correction */
1330 zeta2 = _mm256_mul_ps(beta2,rsq31);
1331 rinv3 = _mm256_mul_ps(rinvsq31,rinv31);
1332 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1333 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1334 felec = _mm256_mul_ps(qq31,felec);
1335 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1336 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1337 velec = _mm256_sub_ps(_mm256_sub_ps(rinv31,sh_ewald),pmecorrV);
1338 velec = _mm256_mul_ps(qq31,velec);
1340 cutoff_mask = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
1342 /* Update potential sum for this i atom from the interaction with this j atom. */
1343 velec = _mm256_and_ps(velec,cutoff_mask);
1344 velec = _mm256_andnot_ps(dummy_mask,velec);
1345 velecsum = _mm256_add_ps(velecsum,velec);
1349 fscal = _mm256_and_ps(fscal,cutoff_mask);
1351 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1353 /* Calculate temporary vectorial force */
1354 tx = _mm256_mul_ps(fscal,dx31);
1355 ty = _mm256_mul_ps(fscal,dy31);
1356 tz = _mm256_mul_ps(fscal,dz31);
1358 /* Update vectorial force */
1359 fix3 = _mm256_add_ps(fix3,tx);
1360 fiy3 = _mm256_add_ps(fiy3,ty);
1361 fiz3 = _mm256_add_ps(fiz3,tz);
1363 fjx1 = _mm256_add_ps(fjx1,tx);
1364 fjy1 = _mm256_add_ps(fjy1,ty);
1365 fjz1 = _mm256_add_ps(fjz1,tz);
1369 /**************************
1370 * CALCULATE INTERACTIONS *
1371 **************************/
1373 if (gmx_mm256_any_lt(rsq32,rcutoff2))
1376 r32 = _mm256_mul_ps(rsq32,rinv32);
1377 r32 = _mm256_andnot_ps(dummy_mask,r32);
1379 /* EWALD ELECTROSTATICS */
1381 /* Analytical PME correction */
1382 zeta2 = _mm256_mul_ps(beta2,rsq32);
1383 rinv3 = _mm256_mul_ps(rinvsq32,rinv32);
1384 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1385 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1386 felec = _mm256_mul_ps(qq32,felec);
1387 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1388 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1389 velec = _mm256_sub_ps(_mm256_sub_ps(rinv32,sh_ewald),pmecorrV);
1390 velec = _mm256_mul_ps(qq32,velec);
1392 cutoff_mask = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
1394 /* Update potential sum for this i atom from the interaction with this j atom. */
1395 velec = _mm256_and_ps(velec,cutoff_mask);
1396 velec = _mm256_andnot_ps(dummy_mask,velec);
1397 velecsum = _mm256_add_ps(velecsum,velec);
1401 fscal = _mm256_and_ps(fscal,cutoff_mask);
1403 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1405 /* Calculate temporary vectorial force */
1406 tx = _mm256_mul_ps(fscal,dx32);
1407 ty = _mm256_mul_ps(fscal,dy32);
1408 tz = _mm256_mul_ps(fscal,dz32);
1410 /* Update vectorial force */
1411 fix3 = _mm256_add_ps(fix3,tx);
1412 fiy3 = _mm256_add_ps(fiy3,ty);
1413 fiz3 = _mm256_add_ps(fiz3,tz);
1415 fjx2 = _mm256_add_ps(fjx2,tx);
1416 fjy2 = _mm256_add_ps(fjy2,ty);
1417 fjz2 = _mm256_add_ps(fjz2,tz);
1421 /**************************
1422 * CALCULATE INTERACTIONS *
1423 **************************/
1425 if (gmx_mm256_any_lt(rsq33,rcutoff2))
1428 r33 = _mm256_mul_ps(rsq33,rinv33);
1429 r33 = _mm256_andnot_ps(dummy_mask,r33);
1431 /* EWALD ELECTROSTATICS */
1433 /* Analytical PME correction */
1434 zeta2 = _mm256_mul_ps(beta2,rsq33);
1435 rinv3 = _mm256_mul_ps(rinvsq33,rinv33);
1436 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1437 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1438 felec = _mm256_mul_ps(qq33,felec);
1439 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1440 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1441 velec = _mm256_sub_ps(_mm256_sub_ps(rinv33,sh_ewald),pmecorrV);
1442 velec = _mm256_mul_ps(qq33,velec);
1444 cutoff_mask = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
1446 /* Update potential sum for this i atom from the interaction with this j atom. */
1447 velec = _mm256_and_ps(velec,cutoff_mask);
1448 velec = _mm256_andnot_ps(dummy_mask,velec);
1449 velecsum = _mm256_add_ps(velecsum,velec);
1453 fscal = _mm256_and_ps(fscal,cutoff_mask);
1455 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1457 /* Calculate temporary vectorial force */
1458 tx = _mm256_mul_ps(fscal,dx33);
1459 ty = _mm256_mul_ps(fscal,dy33);
1460 tz = _mm256_mul_ps(fscal,dz33);
1462 /* Update vectorial force */
1463 fix3 = _mm256_add_ps(fix3,tx);
1464 fiy3 = _mm256_add_ps(fiy3,ty);
1465 fiz3 = _mm256_add_ps(fiz3,tz);
1467 fjx3 = _mm256_add_ps(fjx3,tx);
1468 fjy3 = _mm256_add_ps(fjy3,ty);
1469 fjz3 = _mm256_add_ps(fjz3,tz);
1473 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1474 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1475 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1476 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1477 fjptrE = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1478 fjptrF = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1479 fjptrG = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1480 fjptrH = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1482 gmx_mm256_decrement_4rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
1483 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,
1484 fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1486 /* Inner loop uses 1034 flops */
1489 /* End of innermost loop */
1491 gmx_mm256_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1492 f+i_coord_offset,fshift+i_shift_offset);
1495 /* Update potential energies */
1496 gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1497 gmx_mm256_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
1499 /* Increment number of inner iterations */
1500 inneriter += j_index_end - j_index_start;
1502 /* Outer loop uses 26 flops */
1505 /* Increment number of outer iterations */
1508 /* Update outer/inner flops */
1510 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*26 + inneriter*1034);
1513 * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwLJSh_GeomW4W4_F_avx_256_single
1514 * Electrostatics interaction: Ewald
1515 * VdW interaction: LennardJones
1516 * Geometry: Water4-Water4
1517 * Calculate force/pot: Force
1520 nb_kernel_ElecEwSh_VdwLJSh_GeomW4W4_F_avx_256_single
1521 (t_nblist * gmx_restrict nlist,
1522 rvec * gmx_restrict xx,
1523 rvec * gmx_restrict ff,
1524 t_forcerec * gmx_restrict fr,
1525 t_mdatoms * gmx_restrict mdatoms,
1526 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1527 t_nrnb * gmx_restrict nrnb)
1529 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1530 * just 0 for non-waters.
1531 * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
1532 * jnr indices corresponding to data put in the four positions in the SIMD register.
1534 int i_shift_offset,i_coord_offset,outeriter,inneriter;
1535 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1536 int jnrA,jnrB,jnrC,jnrD;
1537 int jnrE,jnrF,jnrG,jnrH;
1538 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1539 int jnrlistE,jnrlistF,jnrlistG,jnrlistH;
1540 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1541 int j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
1542 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
1543 real rcutoff_scalar;
1544 real *shiftvec,*fshift,*x,*f;
1545 real *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
1546 real scratch[4*DIM];
1547 __m256 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1548 real * vdwioffsetptr0;
1549 __m256 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1550 real * vdwioffsetptr1;
1551 __m256 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1552 real * vdwioffsetptr2;
1553 __m256 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1554 real * vdwioffsetptr3;
1555 __m256 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
1556 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
1557 __m256 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1558 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
1559 __m256 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1560 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
1561 __m256 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1562 int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D,vdwjidx3E,vdwjidx3F,vdwjidx3G,vdwjidx3H;
1563 __m256 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
1564 __m256 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1565 __m256 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1566 __m256 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1567 __m256 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
1568 __m256 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1569 __m256 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1570 __m256 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
1571 __m256 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
1572 __m256 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
1573 __m256 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
1574 __m256 velec,felec,velecsum,facel,crf,krf,krf2;
1577 __m256 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1580 __m256 one_sixth = _mm256_set1_ps(1.0/6.0);
1581 __m256 one_twelfth = _mm256_set1_ps(1.0/12.0);
1583 __m128i ewitab_lo,ewitab_hi;
1584 __m256 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1585 __m256 beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
1587 __m256 dummy_mask,cutoff_mask;
1588 __m256 signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
1589 __m256 one = _mm256_set1_ps(1.0);
1590 __m256 two = _mm256_set1_ps(2.0);
1596 jindex = nlist->jindex;
1598 shiftidx = nlist->shift;
1600 shiftvec = fr->shift_vec[0];
1601 fshift = fr->fshift[0];
1602 facel = _mm256_set1_ps(fr->epsfac);
1603 charge = mdatoms->chargeA;
1604 nvdwtype = fr->ntype;
1605 vdwparam = fr->nbfp;
1606 vdwtype = mdatoms->typeA;
1608 sh_ewald = _mm256_set1_ps(fr->ic->sh_ewald);
1609 beta = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
1610 beta2 = _mm256_mul_ps(beta,beta);
1611 beta3 = _mm256_mul_ps(beta,beta2);
1613 ewtab = fr->ic->tabq_coul_F;
1614 ewtabscale = _mm256_set1_ps(fr->ic->tabq_scale);
1615 ewtabhalfspace = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
1617 /* Setup water-specific parameters */
1618 inr = nlist->iinr[0];
1619 iq1 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
1620 iq2 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
1621 iq3 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
1622 vdwioffsetptr0 = vdwparam+2*nvdwtype*vdwtype[inr+0];
1624 jq1 = _mm256_set1_ps(charge[inr+1]);
1625 jq2 = _mm256_set1_ps(charge[inr+2]);
1626 jq3 = _mm256_set1_ps(charge[inr+3]);
1627 vdwjidx0A = 2*vdwtype[inr+0];
1628 c6_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A]);
1629 c12_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A+1]);
1630 qq11 = _mm256_mul_ps(iq1,jq1);
1631 qq12 = _mm256_mul_ps(iq1,jq2);
1632 qq13 = _mm256_mul_ps(iq1,jq3);
1633 qq21 = _mm256_mul_ps(iq2,jq1);
1634 qq22 = _mm256_mul_ps(iq2,jq2);
1635 qq23 = _mm256_mul_ps(iq2,jq3);
1636 qq31 = _mm256_mul_ps(iq3,jq1);
1637 qq32 = _mm256_mul_ps(iq3,jq2);
1638 qq33 = _mm256_mul_ps(iq3,jq3);
1640 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
1641 rcutoff_scalar = fr->rcoulomb;
1642 rcutoff = _mm256_set1_ps(rcutoff_scalar);
1643 rcutoff2 = _mm256_mul_ps(rcutoff,rcutoff);
1645 sh_vdw_invrcut6 = _mm256_set1_ps(fr->ic->sh_invrc6);
1646 rvdw = _mm256_set1_ps(fr->rvdw);
1648 /* Avoid stupid compiler warnings */
1649 jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
1650 j_coord_offsetA = 0;
1651 j_coord_offsetB = 0;
1652 j_coord_offsetC = 0;
1653 j_coord_offsetD = 0;
1654 j_coord_offsetE = 0;
1655 j_coord_offsetF = 0;
1656 j_coord_offsetG = 0;
1657 j_coord_offsetH = 0;
1662 for(iidx=0;iidx<4*DIM;iidx++)
1664 scratch[iidx] = 0.0;
1667 /* Start outer loop over neighborlists */
1668 for(iidx=0; iidx<nri; iidx++)
1670 /* Load shift vector for this list */
1671 i_shift_offset = DIM*shiftidx[iidx];
1673 /* Load limits for loop over neighbors */
1674 j_index_start = jindex[iidx];
1675 j_index_end = jindex[iidx+1];
1677 /* Get outer coordinate index */
1679 i_coord_offset = DIM*inr;
1681 /* Load i particle coords and add shift vector */
1682 gmx_mm256_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1683 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1685 fix0 = _mm256_setzero_ps();
1686 fiy0 = _mm256_setzero_ps();
1687 fiz0 = _mm256_setzero_ps();
1688 fix1 = _mm256_setzero_ps();
1689 fiy1 = _mm256_setzero_ps();
1690 fiz1 = _mm256_setzero_ps();
1691 fix2 = _mm256_setzero_ps();
1692 fiy2 = _mm256_setzero_ps();
1693 fiz2 = _mm256_setzero_ps();
1694 fix3 = _mm256_setzero_ps();
1695 fiy3 = _mm256_setzero_ps();
1696 fiz3 = _mm256_setzero_ps();
1698 /* Start inner kernel loop */
1699 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
1702 /* Get j neighbor index, and coordinate index */
1704 jnrB = jjnr[jidx+1];
1705 jnrC = jjnr[jidx+2];
1706 jnrD = jjnr[jidx+3];
1707 jnrE = jjnr[jidx+4];
1708 jnrF = jjnr[jidx+5];
1709 jnrG = jjnr[jidx+6];
1710 jnrH = jjnr[jidx+7];
1711 j_coord_offsetA = DIM*jnrA;
1712 j_coord_offsetB = DIM*jnrB;
1713 j_coord_offsetC = DIM*jnrC;
1714 j_coord_offsetD = DIM*jnrD;
1715 j_coord_offsetE = DIM*jnrE;
1716 j_coord_offsetF = DIM*jnrF;
1717 j_coord_offsetG = DIM*jnrG;
1718 j_coord_offsetH = DIM*jnrH;
1720 /* load j atom coordinates */
1721 gmx_mm256_load_4rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1722 x+j_coord_offsetC,x+j_coord_offsetD,
1723 x+j_coord_offsetE,x+j_coord_offsetF,
1724 x+j_coord_offsetG,x+j_coord_offsetH,
1725 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
1726 &jy2,&jz2,&jx3,&jy3,&jz3);
1728 /* Calculate displacement vector */
1729 dx00 = _mm256_sub_ps(ix0,jx0);
1730 dy00 = _mm256_sub_ps(iy0,jy0);
1731 dz00 = _mm256_sub_ps(iz0,jz0);
1732 dx11 = _mm256_sub_ps(ix1,jx1);
1733 dy11 = _mm256_sub_ps(iy1,jy1);
1734 dz11 = _mm256_sub_ps(iz1,jz1);
1735 dx12 = _mm256_sub_ps(ix1,jx2);
1736 dy12 = _mm256_sub_ps(iy1,jy2);
1737 dz12 = _mm256_sub_ps(iz1,jz2);
1738 dx13 = _mm256_sub_ps(ix1,jx3);
1739 dy13 = _mm256_sub_ps(iy1,jy3);
1740 dz13 = _mm256_sub_ps(iz1,jz3);
1741 dx21 = _mm256_sub_ps(ix2,jx1);
1742 dy21 = _mm256_sub_ps(iy2,jy1);
1743 dz21 = _mm256_sub_ps(iz2,jz1);
1744 dx22 = _mm256_sub_ps(ix2,jx2);
1745 dy22 = _mm256_sub_ps(iy2,jy2);
1746 dz22 = _mm256_sub_ps(iz2,jz2);
1747 dx23 = _mm256_sub_ps(ix2,jx3);
1748 dy23 = _mm256_sub_ps(iy2,jy3);
1749 dz23 = _mm256_sub_ps(iz2,jz3);
1750 dx31 = _mm256_sub_ps(ix3,jx1);
1751 dy31 = _mm256_sub_ps(iy3,jy1);
1752 dz31 = _mm256_sub_ps(iz3,jz1);
1753 dx32 = _mm256_sub_ps(ix3,jx2);
1754 dy32 = _mm256_sub_ps(iy3,jy2);
1755 dz32 = _mm256_sub_ps(iz3,jz2);
1756 dx33 = _mm256_sub_ps(ix3,jx3);
1757 dy33 = _mm256_sub_ps(iy3,jy3);
1758 dz33 = _mm256_sub_ps(iz3,jz3);
1760 /* Calculate squared distance and things based on it */
1761 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1762 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
1763 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
1764 rsq13 = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
1765 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
1766 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
1767 rsq23 = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
1768 rsq31 = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
1769 rsq32 = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
1770 rsq33 = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
1772 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
1773 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
1774 rinv13 = gmx_mm256_invsqrt_ps(rsq13);
1775 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
1776 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
1777 rinv23 = gmx_mm256_invsqrt_ps(rsq23);
1778 rinv31 = gmx_mm256_invsqrt_ps(rsq31);
1779 rinv32 = gmx_mm256_invsqrt_ps(rsq32);
1780 rinv33 = gmx_mm256_invsqrt_ps(rsq33);
1782 rinvsq00 = gmx_mm256_inv_ps(rsq00);
1783 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
1784 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
1785 rinvsq13 = _mm256_mul_ps(rinv13,rinv13);
1786 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
1787 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
1788 rinvsq23 = _mm256_mul_ps(rinv23,rinv23);
1789 rinvsq31 = _mm256_mul_ps(rinv31,rinv31);
1790 rinvsq32 = _mm256_mul_ps(rinv32,rinv32);
1791 rinvsq33 = _mm256_mul_ps(rinv33,rinv33);
1793 fjx0 = _mm256_setzero_ps();
1794 fjy0 = _mm256_setzero_ps();
1795 fjz0 = _mm256_setzero_ps();
1796 fjx1 = _mm256_setzero_ps();
1797 fjy1 = _mm256_setzero_ps();
1798 fjz1 = _mm256_setzero_ps();
1799 fjx2 = _mm256_setzero_ps();
1800 fjy2 = _mm256_setzero_ps();
1801 fjz2 = _mm256_setzero_ps();
1802 fjx3 = _mm256_setzero_ps();
1803 fjy3 = _mm256_setzero_ps();
1804 fjz3 = _mm256_setzero_ps();
1806 /**************************
1807 * CALCULATE INTERACTIONS *
1808 **************************/
1810 if (gmx_mm256_any_lt(rsq00,rcutoff2))
1813 /* LENNARD-JONES DISPERSION/REPULSION */
1815 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1816 fvdw = _mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),c6_00),_mm256_mul_ps(rinvsix,rinvsq00));
1818 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
1822 fscal = _mm256_and_ps(fscal,cutoff_mask);
1824 /* Calculate temporary vectorial force */
1825 tx = _mm256_mul_ps(fscal,dx00);
1826 ty = _mm256_mul_ps(fscal,dy00);
1827 tz = _mm256_mul_ps(fscal,dz00);
1829 /* Update vectorial force */
1830 fix0 = _mm256_add_ps(fix0,tx);
1831 fiy0 = _mm256_add_ps(fiy0,ty);
1832 fiz0 = _mm256_add_ps(fiz0,tz);
1834 fjx0 = _mm256_add_ps(fjx0,tx);
1835 fjy0 = _mm256_add_ps(fjy0,ty);
1836 fjz0 = _mm256_add_ps(fjz0,tz);
1840 /**************************
1841 * CALCULATE INTERACTIONS *
1842 **************************/
1844 if (gmx_mm256_any_lt(rsq11,rcutoff2))
1847 r11 = _mm256_mul_ps(rsq11,rinv11);
1849 /* EWALD ELECTROSTATICS */
1851 /* Analytical PME correction */
1852 zeta2 = _mm256_mul_ps(beta2,rsq11);
1853 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
1854 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1855 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1856 felec = _mm256_mul_ps(qq11,felec);
1858 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
1862 fscal = _mm256_and_ps(fscal,cutoff_mask);
1864 /* Calculate temporary vectorial force */
1865 tx = _mm256_mul_ps(fscal,dx11);
1866 ty = _mm256_mul_ps(fscal,dy11);
1867 tz = _mm256_mul_ps(fscal,dz11);
1869 /* Update vectorial force */
1870 fix1 = _mm256_add_ps(fix1,tx);
1871 fiy1 = _mm256_add_ps(fiy1,ty);
1872 fiz1 = _mm256_add_ps(fiz1,tz);
1874 fjx1 = _mm256_add_ps(fjx1,tx);
1875 fjy1 = _mm256_add_ps(fjy1,ty);
1876 fjz1 = _mm256_add_ps(fjz1,tz);
1880 /**************************
1881 * CALCULATE INTERACTIONS *
1882 **************************/
1884 if (gmx_mm256_any_lt(rsq12,rcutoff2))
1887 r12 = _mm256_mul_ps(rsq12,rinv12);
1889 /* EWALD ELECTROSTATICS */
1891 /* Analytical PME correction */
1892 zeta2 = _mm256_mul_ps(beta2,rsq12);
1893 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
1894 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1895 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1896 felec = _mm256_mul_ps(qq12,felec);
1898 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
1902 fscal = _mm256_and_ps(fscal,cutoff_mask);
1904 /* Calculate temporary vectorial force */
1905 tx = _mm256_mul_ps(fscal,dx12);
1906 ty = _mm256_mul_ps(fscal,dy12);
1907 tz = _mm256_mul_ps(fscal,dz12);
1909 /* Update vectorial force */
1910 fix1 = _mm256_add_ps(fix1,tx);
1911 fiy1 = _mm256_add_ps(fiy1,ty);
1912 fiz1 = _mm256_add_ps(fiz1,tz);
1914 fjx2 = _mm256_add_ps(fjx2,tx);
1915 fjy2 = _mm256_add_ps(fjy2,ty);
1916 fjz2 = _mm256_add_ps(fjz2,tz);
1920 /**************************
1921 * CALCULATE INTERACTIONS *
1922 **************************/
1924 if (gmx_mm256_any_lt(rsq13,rcutoff2))
1927 r13 = _mm256_mul_ps(rsq13,rinv13);
1929 /* EWALD ELECTROSTATICS */
1931 /* Analytical PME correction */
1932 zeta2 = _mm256_mul_ps(beta2,rsq13);
1933 rinv3 = _mm256_mul_ps(rinvsq13,rinv13);
1934 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1935 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1936 felec = _mm256_mul_ps(qq13,felec);
1938 cutoff_mask = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
1942 fscal = _mm256_and_ps(fscal,cutoff_mask);
1944 /* Calculate temporary vectorial force */
1945 tx = _mm256_mul_ps(fscal,dx13);
1946 ty = _mm256_mul_ps(fscal,dy13);
1947 tz = _mm256_mul_ps(fscal,dz13);
1949 /* Update vectorial force */
1950 fix1 = _mm256_add_ps(fix1,tx);
1951 fiy1 = _mm256_add_ps(fiy1,ty);
1952 fiz1 = _mm256_add_ps(fiz1,tz);
1954 fjx3 = _mm256_add_ps(fjx3,tx);
1955 fjy3 = _mm256_add_ps(fjy3,ty);
1956 fjz3 = _mm256_add_ps(fjz3,tz);
1960 /**************************
1961 * CALCULATE INTERACTIONS *
1962 **************************/
1964 if (gmx_mm256_any_lt(rsq21,rcutoff2))
1967 r21 = _mm256_mul_ps(rsq21,rinv21);
1969 /* EWALD ELECTROSTATICS */
1971 /* Analytical PME correction */
1972 zeta2 = _mm256_mul_ps(beta2,rsq21);
1973 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
1974 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1975 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1976 felec = _mm256_mul_ps(qq21,felec);
1978 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
1982 fscal = _mm256_and_ps(fscal,cutoff_mask);
1984 /* Calculate temporary vectorial force */
1985 tx = _mm256_mul_ps(fscal,dx21);
1986 ty = _mm256_mul_ps(fscal,dy21);
1987 tz = _mm256_mul_ps(fscal,dz21);
1989 /* Update vectorial force */
1990 fix2 = _mm256_add_ps(fix2,tx);
1991 fiy2 = _mm256_add_ps(fiy2,ty);
1992 fiz2 = _mm256_add_ps(fiz2,tz);
1994 fjx1 = _mm256_add_ps(fjx1,tx);
1995 fjy1 = _mm256_add_ps(fjy1,ty);
1996 fjz1 = _mm256_add_ps(fjz1,tz);
2000 /**************************
2001 * CALCULATE INTERACTIONS *
2002 **************************/
2004 if (gmx_mm256_any_lt(rsq22,rcutoff2))
2007 r22 = _mm256_mul_ps(rsq22,rinv22);
2009 /* EWALD ELECTROSTATICS */
2011 /* Analytical PME correction */
2012 zeta2 = _mm256_mul_ps(beta2,rsq22);
2013 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
2014 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2015 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2016 felec = _mm256_mul_ps(qq22,felec);
2018 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
2022 fscal = _mm256_and_ps(fscal,cutoff_mask);
2024 /* Calculate temporary vectorial force */
2025 tx = _mm256_mul_ps(fscal,dx22);
2026 ty = _mm256_mul_ps(fscal,dy22);
2027 tz = _mm256_mul_ps(fscal,dz22);
2029 /* Update vectorial force */
2030 fix2 = _mm256_add_ps(fix2,tx);
2031 fiy2 = _mm256_add_ps(fiy2,ty);
2032 fiz2 = _mm256_add_ps(fiz2,tz);
2034 fjx2 = _mm256_add_ps(fjx2,tx);
2035 fjy2 = _mm256_add_ps(fjy2,ty);
2036 fjz2 = _mm256_add_ps(fjz2,tz);
2040 /**************************
2041 * CALCULATE INTERACTIONS *
2042 **************************/
2044 if (gmx_mm256_any_lt(rsq23,rcutoff2))
2047 r23 = _mm256_mul_ps(rsq23,rinv23);
2049 /* EWALD ELECTROSTATICS */
2051 /* Analytical PME correction */
2052 zeta2 = _mm256_mul_ps(beta2,rsq23);
2053 rinv3 = _mm256_mul_ps(rinvsq23,rinv23);
2054 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2055 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2056 felec = _mm256_mul_ps(qq23,felec);
2058 cutoff_mask = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
2062 fscal = _mm256_and_ps(fscal,cutoff_mask);
2064 /* Calculate temporary vectorial force */
2065 tx = _mm256_mul_ps(fscal,dx23);
2066 ty = _mm256_mul_ps(fscal,dy23);
2067 tz = _mm256_mul_ps(fscal,dz23);
2069 /* Update vectorial force */
2070 fix2 = _mm256_add_ps(fix2,tx);
2071 fiy2 = _mm256_add_ps(fiy2,ty);
2072 fiz2 = _mm256_add_ps(fiz2,tz);
2074 fjx3 = _mm256_add_ps(fjx3,tx);
2075 fjy3 = _mm256_add_ps(fjy3,ty);
2076 fjz3 = _mm256_add_ps(fjz3,tz);
2080 /**************************
2081 * CALCULATE INTERACTIONS *
2082 **************************/
2084 if (gmx_mm256_any_lt(rsq31,rcutoff2))
2087 r31 = _mm256_mul_ps(rsq31,rinv31);
2089 /* EWALD ELECTROSTATICS */
2091 /* Analytical PME correction */
2092 zeta2 = _mm256_mul_ps(beta2,rsq31);
2093 rinv3 = _mm256_mul_ps(rinvsq31,rinv31);
2094 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2095 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2096 felec = _mm256_mul_ps(qq31,felec);
2098 cutoff_mask = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
2102 fscal = _mm256_and_ps(fscal,cutoff_mask);
2104 /* Calculate temporary vectorial force */
2105 tx = _mm256_mul_ps(fscal,dx31);
2106 ty = _mm256_mul_ps(fscal,dy31);
2107 tz = _mm256_mul_ps(fscal,dz31);
2109 /* Update vectorial force */
2110 fix3 = _mm256_add_ps(fix3,tx);
2111 fiy3 = _mm256_add_ps(fiy3,ty);
2112 fiz3 = _mm256_add_ps(fiz3,tz);
2114 fjx1 = _mm256_add_ps(fjx1,tx);
2115 fjy1 = _mm256_add_ps(fjy1,ty);
2116 fjz1 = _mm256_add_ps(fjz1,tz);
2120 /**************************
2121 * CALCULATE INTERACTIONS *
2122 **************************/
2124 if (gmx_mm256_any_lt(rsq32,rcutoff2))
2127 r32 = _mm256_mul_ps(rsq32,rinv32);
2129 /* EWALD ELECTROSTATICS */
2131 /* Analytical PME correction */
2132 zeta2 = _mm256_mul_ps(beta2,rsq32);
2133 rinv3 = _mm256_mul_ps(rinvsq32,rinv32);
2134 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2135 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2136 felec = _mm256_mul_ps(qq32,felec);
2138 cutoff_mask = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
2142 fscal = _mm256_and_ps(fscal,cutoff_mask);
2144 /* Calculate temporary vectorial force */
2145 tx = _mm256_mul_ps(fscal,dx32);
2146 ty = _mm256_mul_ps(fscal,dy32);
2147 tz = _mm256_mul_ps(fscal,dz32);
2149 /* Update vectorial force */
2150 fix3 = _mm256_add_ps(fix3,tx);
2151 fiy3 = _mm256_add_ps(fiy3,ty);
2152 fiz3 = _mm256_add_ps(fiz3,tz);
2154 fjx2 = _mm256_add_ps(fjx2,tx);
2155 fjy2 = _mm256_add_ps(fjy2,ty);
2156 fjz2 = _mm256_add_ps(fjz2,tz);
2160 /**************************
2161 * CALCULATE INTERACTIONS *
2162 **************************/
2164 if (gmx_mm256_any_lt(rsq33,rcutoff2))
2167 r33 = _mm256_mul_ps(rsq33,rinv33);
2169 /* EWALD ELECTROSTATICS */
2171 /* Analytical PME correction */
2172 zeta2 = _mm256_mul_ps(beta2,rsq33);
2173 rinv3 = _mm256_mul_ps(rinvsq33,rinv33);
2174 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2175 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2176 felec = _mm256_mul_ps(qq33,felec);
2178 cutoff_mask = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
2182 fscal = _mm256_and_ps(fscal,cutoff_mask);
2184 /* Calculate temporary vectorial force */
2185 tx = _mm256_mul_ps(fscal,dx33);
2186 ty = _mm256_mul_ps(fscal,dy33);
2187 tz = _mm256_mul_ps(fscal,dz33);
2189 /* Update vectorial force */
2190 fix3 = _mm256_add_ps(fix3,tx);
2191 fiy3 = _mm256_add_ps(fiy3,ty);
2192 fiz3 = _mm256_add_ps(fiz3,tz);
2194 fjx3 = _mm256_add_ps(fjx3,tx);
2195 fjy3 = _mm256_add_ps(fjy3,ty);
2196 fjz3 = _mm256_add_ps(fjz3,tz);
2200 fjptrA = f+j_coord_offsetA;
2201 fjptrB = f+j_coord_offsetB;
2202 fjptrC = f+j_coord_offsetC;
2203 fjptrD = f+j_coord_offsetD;
2204 fjptrE = f+j_coord_offsetE;
2205 fjptrF = f+j_coord_offsetF;
2206 fjptrG = f+j_coord_offsetG;
2207 fjptrH = f+j_coord_offsetH;
2209 gmx_mm256_decrement_4rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
2210 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,
2211 fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
2213 /* Inner loop uses 564 flops */
2216 if(jidx<j_index_end)
2219 /* Get j neighbor index, and coordinate index */
2220 jnrlistA = jjnr[jidx];
2221 jnrlistB = jjnr[jidx+1];
2222 jnrlistC = jjnr[jidx+2];
2223 jnrlistD = jjnr[jidx+3];
2224 jnrlistE = jjnr[jidx+4];
2225 jnrlistF = jjnr[jidx+5];
2226 jnrlistG = jjnr[jidx+6];
2227 jnrlistH = jjnr[jidx+7];
2228 /* Sign of each element will be negative for non-real atoms.
2229 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
2230 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
2232 dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
2233 gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
2235 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
2236 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
2237 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
2238 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
2239 jnrE = (jnrlistE>=0) ? jnrlistE : 0;
2240 jnrF = (jnrlistF>=0) ? jnrlistF : 0;
2241 jnrG = (jnrlistG>=0) ? jnrlistG : 0;
2242 jnrH = (jnrlistH>=0) ? jnrlistH : 0;
2243 j_coord_offsetA = DIM*jnrA;
2244 j_coord_offsetB = DIM*jnrB;
2245 j_coord_offsetC = DIM*jnrC;
2246 j_coord_offsetD = DIM*jnrD;
2247 j_coord_offsetE = DIM*jnrE;
2248 j_coord_offsetF = DIM*jnrF;
2249 j_coord_offsetG = DIM*jnrG;
2250 j_coord_offsetH = DIM*jnrH;
2252 /* load j atom coordinates */
2253 gmx_mm256_load_4rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
2254 x+j_coord_offsetC,x+j_coord_offsetD,
2255 x+j_coord_offsetE,x+j_coord_offsetF,
2256 x+j_coord_offsetG,x+j_coord_offsetH,
2257 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
2258 &jy2,&jz2,&jx3,&jy3,&jz3);
2260 /* Calculate displacement vector */
2261 dx00 = _mm256_sub_ps(ix0,jx0);
2262 dy00 = _mm256_sub_ps(iy0,jy0);
2263 dz00 = _mm256_sub_ps(iz0,jz0);
2264 dx11 = _mm256_sub_ps(ix1,jx1);
2265 dy11 = _mm256_sub_ps(iy1,jy1);
2266 dz11 = _mm256_sub_ps(iz1,jz1);
2267 dx12 = _mm256_sub_ps(ix1,jx2);
2268 dy12 = _mm256_sub_ps(iy1,jy2);
2269 dz12 = _mm256_sub_ps(iz1,jz2);
2270 dx13 = _mm256_sub_ps(ix1,jx3);
2271 dy13 = _mm256_sub_ps(iy1,jy3);
2272 dz13 = _mm256_sub_ps(iz1,jz3);
2273 dx21 = _mm256_sub_ps(ix2,jx1);
2274 dy21 = _mm256_sub_ps(iy2,jy1);
2275 dz21 = _mm256_sub_ps(iz2,jz1);
2276 dx22 = _mm256_sub_ps(ix2,jx2);
2277 dy22 = _mm256_sub_ps(iy2,jy2);
2278 dz22 = _mm256_sub_ps(iz2,jz2);
2279 dx23 = _mm256_sub_ps(ix2,jx3);
2280 dy23 = _mm256_sub_ps(iy2,jy3);
2281 dz23 = _mm256_sub_ps(iz2,jz3);
2282 dx31 = _mm256_sub_ps(ix3,jx1);
2283 dy31 = _mm256_sub_ps(iy3,jy1);
2284 dz31 = _mm256_sub_ps(iz3,jz1);
2285 dx32 = _mm256_sub_ps(ix3,jx2);
2286 dy32 = _mm256_sub_ps(iy3,jy2);
2287 dz32 = _mm256_sub_ps(iz3,jz2);
2288 dx33 = _mm256_sub_ps(ix3,jx3);
2289 dy33 = _mm256_sub_ps(iy3,jy3);
2290 dz33 = _mm256_sub_ps(iz3,jz3);
2292 /* Calculate squared distance and things based on it */
2293 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
2294 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
2295 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
2296 rsq13 = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
2297 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
2298 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
2299 rsq23 = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
2300 rsq31 = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
2301 rsq32 = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
2302 rsq33 = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
2304 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
2305 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
2306 rinv13 = gmx_mm256_invsqrt_ps(rsq13);
2307 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
2308 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
2309 rinv23 = gmx_mm256_invsqrt_ps(rsq23);
2310 rinv31 = gmx_mm256_invsqrt_ps(rsq31);
2311 rinv32 = gmx_mm256_invsqrt_ps(rsq32);
2312 rinv33 = gmx_mm256_invsqrt_ps(rsq33);
2314 rinvsq00 = gmx_mm256_inv_ps(rsq00);
2315 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
2316 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
2317 rinvsq13 = _mm256_mul_ps(rinv13,rinv13);
2318 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
2319 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
2320 rinvsq23 = _mm256_mul_ps(rinv23,rinv23);
2321 rinvsq31 = _mm256_mul_ps(rinv31,rinv31);
2322 rinvsq32 = _mm256_mul_ps(rinv32,rinv32);
2323 rinvsq33 = _mm256_mul_ps(rinv33,rinv33);
2325 fjx0 = _mm256_setzero_ps();
2326 fjy0 = _mm256_setzero_ps();
2327 fjz0 = _mm256_setzero_ps();
2328 fjx1 = _mm256_setzero_ps();
2329 fjy1 = _mm256_setzero_ps();
2330 fjz1 = _mm256_setzero_ps();
2331 fjx2 = _mm256_setzero_ps();
2332 fjy2 = _mm256_setzero_ps();
2333 fjz2 = _mm256_setzero_ps();
2334 fjx3 = _mm256_setzero_ps();
2335 fjy3 = _mm256_setzero_ps();
2336 fjz3 = _mm256_setzero_ps();
2338 /**************************
2339 * CALCULATE INTERACTIONS *
2340 **************************/
2342 if (gmx_mm256_any_lt(rsq00,rcutoff2))
2345 /* LENNARD-JONES DISPERSION/REPULSION */
2347 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
2348 fvdw = _mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),c6_00),_mm256_mul_ps(rinvsix,rinvsq00));
2350 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
2354 fscal = _mm256_and_ps(fscal,cutoff_mask);
2356 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2358 /* Calculate temporary vectorial force */
2359 tx = _mm256_mul_ps(fscal,dx00);
2360 ty = _mm256_mul_ps(fscal,dy00);
2361 tz = _mm256_mul_ps(fscal,dz00);
2363 /* Update vectorial force */
2364 fix0 = _mm256_add_ps(fix0,tx);
2365 fiy0 = _mm256_add_ps(fiy0,ty);
2366 fiz0 = _mm256_add_ps(fiz0,tz);
2368 fjx0 = _mm256_add_ps(fjx0,tx);
2369 fjy0 = _mm256_add_ps(fjy0,ty);
2370 fjz0 = _mm256_add_ps(fjz0,tz);
2374 /**************************
2375 * CALCULATE INTERACTIONS *
2376 **************************/
2378 if (gmx_mm256_any_lt(rsq11,rcutoff2))
2381 r11 = _mm256_mul_ps(rsq11,rinv11);
2382 r11 = _mm256_andnot_ps(dummy_mask,r11);
2384 /* EWALD ELECTROSTATICS */
2386 /* Analytical PME correction */
2387 zeta2 = _mm256_mul_ps(beta2,rsq11);
2388 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
2389 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2390 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2391 felec = _mm256_mul_ps(qq11,felec);
2393 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
2397 fscal = _mm256_and_ps(fscal,cutoff_mask);
2399 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2401 /* Calculate temporary vectorial force */
2402 tx = _mm256_mul_ps(fscal,dx11);
2403 ty = _mm256_mul_ps(fscal,dy11);
2404 tz = _mm256_mul_ps(fscal,dz11);
2406 /* Update vectorial force */
2407 fix1 = _mm256_add_ps(fix1,tx);
2408 fiy1 = _mm256_add_ps(fiy1,ty);
2409 fiz1 = _mm256_add_ps(fiz1,tz);
2411 fjx1 = _mm256_add_ps(fjx1,tx);
2412 fjy1 = _mm256_add_ps(fjy1,ty);
2413 fjz1 = _mm256_add_ps(fjz1,tz);
2417 /**************************
2418 * CALCULATE INTERACTIONS *
2419 **************************/
2421 if (gmx_mm256_any_lt(rsq12,rcutoff2))
2424 r12 = _mm256_mul_ps(rsq12,rinv12);
2425 r12 = _mm256_andnot_ps(dummy_mask,r12);
2427 /* EWALD ELECTROSTATICS */
2429 /* Analytical PME correction */
2430 zeta2 = _mm256_mul_ps(beta2,rsq12);
2431 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
2432 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2433 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2434 felec = _mm256_mul_ps(qq12,felec);
2436 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
2440 fscal = _mm256_and_ps(fscal,cutoff_mask);
2442 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2444 /* Calculate temporary vectorial force */
2445 tx = _mm256_mul_ps(fscal,dx12);
2446 ty = _mm256_mul_ps(fscal,dy12);
2447 tz = _mm256_mul_ps(fscal,dz12);
2449 /* Update vectorial force */
2450 fix1 = _mm256_add_ps(fix1,tx);
2451 fiy1 = _mm256_add_ps(fiy1,ty);
2452 fiz1 = _mm256_add_ps(fiz1,tz);
2454 fjx2 = _mm256_add_ps(fjx2,tx);
2455 fjy2 = _mm256_add_ps(fjy2,ty);
2456 fjz2 = _mm256_add_ps(fjz2,tz);
2460 /**************************
2461 * CALCULATE INTERACTIONS *
2462 **************************/
2464 if (gmx_mm256_any_lt(rsq13,rcutoff2))
2467 r13 = _mm256_mul_ps(rsq13,rinv13);
2468 r13 = _mm256_andnot_ps(dummy_mask,r13);
2470 /* EWALD ELECTROSTATICS */
2472 /* Analytical PME correction */
2473 zeta2 = _mm256_mul_ps(beta2,rsq13);
2474 rinv3 = _mm256_mul_ps(rinvsq13,rinv13);
2475 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2476 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2477 felec = _mm256_mul_ps(qq13,felec);
2479 cutoff_mask = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
2483 fscal = _mm256_and_ps(fscal,cutoff_mask);
2485 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2487 /* Calculate temporary vectorial force */
2488 tx = _mm256_mul_ps(fscal,dx13);
2489 ty = _mm256_mul_ps(fscal,dy13);
2490 tz = _mm256_mul_ps(fscal,dz13);
2492 /* Update vectorial force */
2493 fix1 = _mm256_add_ps(fix1,tx);
2494 fiy1 = _mm256_add_ps(fiy1,ty);
2495 fiz1 = _mm256_add_ps(fiz1,tz);
2497 fjx3 = _mm256_add_ps(fjx3,tx);
2498 fjy3 = _mm256_add_ps(fjy3,ty);
2499 fjz3 = _mm256_add_ps(fjz3,tz);
2503 /**************************
2504 * CALCULATE INTERACTIONS *
2505 **************************/
2507 if (gmx_mm256_any_lt(rsq21,rcutoff2))
2510 r21 = _mm256_mul_ps(rsq21,rinv21);
2511 r21 = _mm256_andnot_ps(dummy_mask,r21);
2513 /* EWALD ELECTROSTATICS */
2515 /* Analytical PME correction */
2516 zeta2 = _mm256_mul_ps(beta2,rsq21);
2517 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
2518 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2519 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2520 felec = _mm256_mul_ps(qq21,felec);
2522 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
2526 fscal = _mm256_and_ps(fscal,cutoff_mask);
2528 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2530 /* Calculate temporary vectorial force */
2531 tx = _mm256_mul_ps(fscal,dx21);
2532 ty = _mm256_mul_ps(fscal,dy21);
2533 tz = _mm256_mul_ps(fscal,dz21);
2535 /* Update vectorial force */
2536 fix2 = _mm256_add_ps(fix2,tx);
2537 fiy2 = _mm256_add_ps(fiy2,ty);
2538 fiz2 = _mm256_add_ps(fiz2,tz);
2540 fjx1 = _mm256_add_ps(fjx1,tx);
2541 fjy1 = _mm256_add_ps(fjy1,ty);
2542 fjz1 = _mm256_add_ps(fjz1,tz);
2546 /**************************
2547 * CALCULATE INTERACTIONS *
2548 **************************/
2550 if (gmx_mm256_any_lt(rsq22,rcutoff2))
2553 r22 = _mm256_mul_ps(rsq22,rinv22);
2554 r22 = _mm256_andnot_ps(dummy_mask,r22);
2556 /* EWALD ELECTROSTATICS */
2558 /* Analytical PME correction */
2559 zeta2 = _mm256_mul_ps(beta2,rsq22);
2560 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
2561 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2562 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2563 felec = _mm256_mul_ps(qq22,felec);
2565 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
2569 fscal = _mm256_and_ps(fscal,cutoff_mask);
2571 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2573 /* Calculate temporary vectorial force */
2574 tx = _mm256_mul_ps(fscal,dx22);
2575 ty = _mm256_mul_ps(fscal,dy22);
2576 tz = _mm256_mul_ps(fscal,dz22);
2578 /* Update vectorial force */
2579 fix2 = _mm256_add_ps(fix2,tx);
2580 fiy2 = _mm256_add_ps(fiy2,ty);
2581 fiz2 = _mm256_add_ps(fiz2,tz);
2583 fjx2 = _mm256_add_ps(fjx2,tx);
2584 fjy2 = _mm256_add_ps(fjy2,ty);
2585 fjz2 = _mm256_add_ps(fjz2,tz);
2589 /**************************
2590 * CALCULATE INTERACTIONS *
2591 **************************/
2593 if (gmx_mm256_any_lt(rsq23,rcutoff2))
2596 r23 = _mm256_mul_ps(rsq23,rinv23);
2597 r23 = _mm256_andnot_ps(dummy_mask,r23);
2599 /* EWALD ELECTROSTATICS */
2601 /* Analytical PME correction */
2602 zeta2 = _mm256_mul_ps(beta2,rsq23);
2603 rinv3 = _mm256_mul_ps(rinvsq23,rinv23);
2604 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2605 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2606 felec = _mm256_mul_ps(qq23,felec);
2608 cutoff_mask = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
2612 fscal = _mm256_and_ps(fscal,cutoff_mask);
2614 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2616 /* Calculate temporary vectorial force */
2617 tx = _mm256_mul_ps(fscal,dx23);
2618 ty = _mm256_mul_ps(fscal,dy23);
2619 tz = _mm256_mul_ps(fscal,dz23);
2621 /* Update vectorial force */
2622 fix2 = _mm256_add_ps(fix2,tx);
2623 fiy2 = _mm256_add_ps(fiy2,ty);
2624 fiz2 = _mm256_add_ps(fiz2,tz);
2626 fjx3 = _mm256_add_ps(fjx3,tx);
2627 fjy3 = _mm256_add_ps(fjy3,ty);
2628 fjz3 = _mm256_add_ps(fjz3,tz);
2632 /**************************
2633 * CALCULATE INTERACTIONS *
2634 **************************/
2636 if (gmx_mm256_any_lt(rsq31,rcutoff2))
2639 r31 = _mm256_mul_ps(rsq31,rinv31);
2640 r31 = _mm256_andnot_ps(dummy_mask,r31);
2642 /* EWALD ELECTROSTATICS */
2644 /* Analytical PME correction */
2645 zeta2 = _mm256_mul_ps(beta2,rsq31);
2646 rinv3 = _mm256_mul_ps(rinvsq31,rinv31);
2647 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2648 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2649 felec = _mm256_mul_ps(qq31,felec);
2651 cutoff_mask = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
2655 fscal = _mm256_and_ps(fscal,cutoff_mask);
2657 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2659 /* Calculate temporary vectorial force */
2660 tx = _mm256_mul_ps(fscal,dx31);
2661 ty = _mm256_mul_ps(fscal,dy31);
2662 tz = _mm256_mul_ps(fscal,dz31);
2664 /* Update vectorial force */
2665 fix3 = _mm256_add_ps(fix3,tx);
2666 fiy3 = _mm256_add_ps(fiy3,ty);
2667 fiz3 = _mm256_add_ps(fiz3,tz);
2669 fjx1 = _mm256_add_ps(fjx1,tx);
2670 fjy1 = _mm256_add_ps(fjy1,ty);
2671 fjz1 = _mm256_add_ps(fjz1,tz);
2675 /**************************
2676 * CALCULATE INTERACTIONS *
2677 **************************/
2679 if (gmx_mm256_any_lt(rsq32,rcutoff2))
2682 r32 = _mm256_mul_ps(rsq32,rinv32);
2683 r32 = _mm256_andnot_ps(dummy_mask,r32);
2685 /* EWALD ELECTROSTATICS */
2687 /* Analytical PME correction */
2688 zeta2 = _mm256_mul_ps(beta2,rsq32);
2689 rinv3 = _mm256_mul_ps(rinvsq32,rinv32);
2690 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2691 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2692 felec = _mm256_mul_ps(qq32,felec);
2694 cutoff_mask = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
2698 fscal = _mm256_and_ps(fscal,cutoff_mask);
2700 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2702 /* Calculate temporary vectorial force */
2703 tx = _mm256_mul_ps(fscal,dx32);
2704 ty = _mm256_mul_ps(fscal,dy32);
2705 tz = _mm256_mul_ps(fscal,dz32);
2707 /* Update vectorial force */
2708 fix3 = _mm256_add_ps(fix3,tx);
2709 fiy3 = _mm256_add_ps(fiy3,ty);
2710 fiz3 = _mm256_add_ps(fiz3,tz);
2712 fjx2 = _mm256_add_ps(fjx2,tx);
2713 fjy2 = _mm256_add_ps(fjy2,ty);
2714 fjz2 = _mm256_add_ps(fjz2,tz);
2718 /**************************
2719 * CALCULATE INTERACTIONS *
2720 **************************/
2722 if (gmx_mm256_any_lt(rsq33,rcutoff2))
2725 r33 = _mm256_mul_ps(rsq33,rinv33);
2726 r33 = _mm256_andnot_ps(dummy_mask,r33);
2728 /* EWALD ELECTROSTATICS */
2730 /* Analytical PME correction */
2731 zeta2 = _mm256_mul_ps(beta2,rsq33);
2732 rinv3 = _mm256_mul_ps(rinvsq33,rinv33);
2733 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2734 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2735 felec = _mm256_mul_ps(qq33,felec);
2737 cutoff_mask = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
2741 fscal = _mm256_and_ps(fscal,cutoff_mask);
2743 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2745 /* Calculate temporary vectorial force */
2746 tx = _mm256_mul_ps(fscal,dx33);
2747 ty = _mm256_mul_ps(fscal,dy33);
2748 tz = _mm256_mul_ps(fscal,dz33);
2750 /* Update vectorial force */
2751 fix3 = _mm256_add_ps(fix3,tx);
2752 fiy3 = _mm256_add_ps(fiy3,ty);
2753 fiz3 = _mm256_add_ps(fiz3,tz);
2755 fjx3 = _mm256_add_ps(fjx3,tx);
2756 fjy3 = _mm256_add_ps(fjy3,ty);
2757 fjz3 = _mm256_add_ps(fjz3,tz);
2761 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2762 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
2763 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
2764 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
2765 fjptrE = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
2766 fjptrF = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
2767 fjptrG = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
2768 fjptrH = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
2770 gmx_mm256_decrement_4rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
2771 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,
2772 fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
2774 /* Inner loop uses 573 flops */
2777 /* End of innermost loop */
2779 gmx_mm256_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
2780 f+i_coord_offset,fshift+i_shift_offset);
2782 /* Increment number of inner iterations */
2783 inneriter += j_index_end - j_index_start;
2785 /* Outer loop uses 24 flops */
2788 /* Increment number of outer iterations */
2791 /* Update outer/inner flops */
2793 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*24 + inneriter*573);