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_VdwLJEwSh_GeomW3W3_VF_avx_256_single
52 * Electrostatics interaction: Ewald
53 * VdW interaction: LJEwald
54 * Geometry: Water3-Water3
55 * Calculate force/pot: PotentialAndForce
58 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_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 real * vdwgridioffsetptr0;
88 __m256 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
89 real * vdwioffsetptr1;
90 real * vdwgridioffsetptr1;
91 __m256 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
92 real * vdwioffsetptr2;
93 real * vdwgridioffsetptr2;
94 __m256 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
95 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
96 __m256 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
97 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
98 __m256 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
99 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
100 __m256 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
101 __m256 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
102 __m256 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
103 __m256 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
104 __m256 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
105 __m256 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
106 __m256 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
107 __m256 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
108 __m256 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
109 __m256 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
110 __m256 velec,felec,velecsum,facel,crf,krf,krf2;
113 __m256 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
116 __m256 one_sixth = _mm256_set1_ps(1.0/6.0);
117 __m256 one_twelfth = _mm256_set1_ps(1.0/12.0);
128 __m256 ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
129 __m256 one_half = _mm256_set1_ps(0.5);
130 __m256 minus_one = _mm256_set1_ps(-1.0);
132 __m128i ewitab_lo,ewitab_hi;
133 __m256 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
134 __m256 beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
136 __m256 dummy_mask,cutoff_mask;
137 __m256 signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
138 __m256 one = _mm256_set1_ps(1.0);
139 __m256 two = _mm256_set1_ps(2.0);
145 jindex = nlist->jindex;
147 shiftidx = nlist->shift;
149 shiftvec = fr->shift_vec[0];
150 fshift = fr->fshift[0];
151 facel = _mm256_set1_ps(fr->epsfac);
152 charge = mdatoms->chargeA;
153 nvdwtype = fr->ntype;
155 vdwtype = mdatoms->typeA;
156 vdwgridparam = fr->ljpme_c6grid;
157 sh_lj_ewald = _mm256_set1_ps(fr->ic->sh_lj_ewald);
158 ewclj = _mm256_set1_ps(fr->ewaldcoeff_lj);
159 ewclj2 = _mm256_mul_ps(minus_one,_mm256_mul_ps(ewclj,ewclj));
161 sh_ewald = _mm256_set1_ps(fr->ic->sh_ewald);
162 beta = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
163 beta2 = _mm256_mul_ps(beta,beta);
164 beta3 = _mm256_mul_ps(beta,beta2);
166 ewtab = fr->ic->tabq_coul_FDV0;
167 ewtabscale = _mm256_set1_ps(fr->ic->tabq_scale);
168 ewtabhalfspace = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
170 /* Setup water-specific parameters */
171 inr = nlist->iinr[0];
172 iq0 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
173 iq1 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
174 iq2 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
175 vdwioffsetptr0 = vdwparam+2*nvdwtype*vdwtype[inr+0];
176 vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
178 jq0 = _mm256_set1_ps(charge[inr+0]);
179 jq1 = _mm256_set1_ps(charge[inr+1]);
180 jq2 = _mm256_set1_ps(charge[inr+2]);
181 vdwjidx0A = 2*vdwtype[inr+0];
182 qq00 = _mm256_mul_ps(iq0,jq0);
183 c6_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A]);
184 c12_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A+1]);
185 c6grid_00 = _mm256_set1_ps(vdwgridioffsetptr0[vdwjidx0A]);
186 qq01 = _mm256_mul_ps(iq0,jq1);
187 qq02 = _mm256_mul_ps(iq0,jq2);
188 qq10 = _mm256_mul_ps(iq1,jq0);
189 qq11 = _mm256_mul_ps(iq1,jq1);
190 qq12 = _mm256_mul_ps(iq1,jq2);
191 qq20 = _mm256_mul_ps(iq2,jq0);
192 qq21 = _mm256_mul_ps(iq2,jq1);
193 qq22 = _mm256_mul_ps(iq2,jq2);
195 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
196 rcutoff_scalar = fr->rcoulomb;
197 rcutoff = _mm256_set1_ps(rcutoff_scalar);
198 rcutoff2 = _mm256_mul_ps(rcutoff,rcutoff);
200 sh_vdw_invrcut6 = _mm256_set1_ps(fr->ic->sh_invrc6);
201 rvdw = _mm256_set1_ps(fr->rvdw);
203 /* Avoid stupid compiler warnings */
204 jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
217 for(iidx=0;iidx<4*DIM;iidx++)
222 /* Start outer loop over neighborlists */
223 for(iidx=0; iidx<nri; iidx++)
225 /* Load shift vector for this list */
226 i_shift_offset = DIM*shiftidx[iidx];
228 /* Load limits for loop over neighbors */
229 j_index_start = jindex[iidx];
230 j_index_end = jindex[iidx+1];
232 /* Get outer coordinate index */
234 i_coord_offset = DIM*inr;
236 /* Load i particle coords and add shift vector */
237 gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
238 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
240 fix0 = _mm256_setzero_ps();
241 fiy0 = _mm256_setzero_ps();
242 fiz0 = _mm256_setzero_ps();
243 fix1 = _mm256_setzero_ps();
244 fiy1 = _mm256_setzero_ps();
245 fiz1 = _mm256_setzero_ps();
246 fix2 = _mm256_setzero_ps();
247 fiy2 = _mm256_setzero_ps();
248 fiz2 = _mm256_setzero_ps();
250 /* Reset potential sums */
251 velecsum = _mm256_setzero_ps();
252 vvdwsum = _mm256_setzero_ps();
254 /* Start inner kernel loop */
255 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
258 /* Get j neighbor index, and coordinate index */
267 j_coord_offsetA = DIM*jnrA;
268 j_coord_offsetB = DIM*jnrB;
269 j_coord_offsetC = DIM*jnrC;
270 j_coord_offsetD = DIM*jnrD;
271 j_coord_offsetE = DIM*jnrE;
272 j_coord_offsetF = DIM*jnrF;
273 j_coord_offsetG = DIM*jnrG;
274 j_coord_offsetH = DIM*jnrH;
276 /* load j atom coordinates */
277 gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
278 x+j_coord_offsetC,x+j_coord_offsetD,
279 x+j_coord_offsetE,x+j_coord_offsetF,
280 x+j_coord_offsetG,x+j_coord_offsetH,
281 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
283 /* Calculate displacement vector */
284 dx00 = _mm256_sub_ps(ix0,jx0);
285 dy00 = _mm256_sub_ps(iy0,jy0);
286 dz00 = _mm256_sub_ps(iz0,jz0);
287 dx01 = _mm256_sub_ps(ix0,jx1);
288 dy01 = _mm256_sub_ps(iy0,jy1);
289 dz01 = _mm256_sub_ps(iz0,jz1);
290 dx02 = _mm256_sub_ps(ix0,jx2);
291 dy02 = _mm256_sub_ps(iy0,jy2);
292 dz02 = _mm256_sub_ps(iz0,jz2);
293 dx10 = _mm256_sub_ps(ix1,jx0);
294 dy10 = _mm256_sub_ps(iy1,jy0);
295 dz10 = _mm256_sub_ps(iz1,jz0);
296 dx11 = _mm256_sub_ps(ix1,jx1);
297 dy11 = _mm256_sub_ps(iy1,jy1);
298 dz11 = _mm256_sub_ps(iz1,jz1);
299 dx12 = _mm256_sub_ps(ix1,jx2);
300 dy12 = _mm256_sub_ps(iy1,jy2);
301 dz12 = _mm256_sub_ps(iz1,jz2);
302 dx20 = _mm256_sub_ps(ix2,jx0);
303 dy20 = _mm256_sub_ps(iy2,jy0);
304 dz20 = _mm256_sub_ps(iz2,jz0);
305 dx21 = _mm256_sub_ps(ix2,jx1);
306 dy21 = _mm256_sub_ps(iy2,jy1);
307 dz21 = _mm256_sub_ps(iz2,jz1);
308 dx22 = _mm256_sub_ps(ix2,jx2);
309 dy22 = _mm256_sub_ps(iy2,jy2);
310 dz22 = _mm256_sub_ps(iz2,jz2);
312 /* Calculate squared distance and things based on it */
313 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
314 rsq01 = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
315 rsq02 = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
316 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
317 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
318 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
319 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
320 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
321 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
323 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
324 rinv01 = gmx_mm256_invsqrt_ps(rsq01);
325 rinv02 = gmx_mm256_invsqrt_ps(rsq02);
326 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
327 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
328 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
329 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
330 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
331 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
333 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
334 rinvsq01 = _mm256_mul_ps(rinv01,rinv01);
335 rinvsq02 = _mm256_mul_ps(rinv02,rinv02);
336 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
337 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
338 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
339 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
340 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
341 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
343 fjx0 = _mm256_setzero_ps();
344 fjy0 = _mm256_setzero_ps();
345 fjz0 = _mm256_setzero_ps();
346 fjx1 = _mm256_setzero_ps();
347 fjy1 = _mm256_setzero_ps();
348 fjz1 = _mm256_setzero_ps();
349 fjx2 = _mm256_setzero_ps();
350 fjy2 = _mm256_setzero_ps();
351 fjz2 = _mm256_setzero_ps();
353 /**************************
354 * CALCULATE INTERACTIONS *
355 **************************/
357 if (gmx_mm256_any_lt(rsq00,rcutoff2))
360 r00 = _mm256_mul_ps(rsq00,rinv00);
362 /* EWALD ELECTROSTATICS */
364 /* Analytical PME correction */
365 zeta2 = _mm256_mul_ps(beta2,rsq00);
366 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
367 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
368 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
369 felec = _mm256_mul_ps(qq00,felec);
370 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
371 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
372 velec = _mm256_sub_ps(_mm256_sub_ps(rinv00,sh_ewald),pmecorrV);
373 velec = _mm256_mul_ps(qq00,velec);
375 /* Analytical LJ-PME */
376 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
377 ewcljrsq = _mm256_mul_ps(ewclj2,rsq00);
378 ewclj6 = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
379 exponent = gmx_simd_exp_r(ewcljrsq);
380 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
381 poly = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
382 /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
383 vvdw6 = _mm256_mul_ps(_mm256_sub_ps(c6_00,_mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly))),rinvsix);
384 vvdw12 = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
385 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) ,
386 _mm256_mul_ps( _mm256_sub_ps(vvdw6,_mm256_add_ps(_mm256_mul_ps(c6_00,sh_vdw_invrcut6),_mm256_mul_ps(c6grid_00,sh_lj_ewald))),one_sixth));
387 /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
388 fvdw = _mm256_mul_ps(_mm256_sub_ps(vvdw12,_mm256_sub_ps(vvdw6,_mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6)))),rinvsq00);
390 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
392 /* Update potential sum for this i atom from the interaction with this j atom. */
393 velec = _mm256_and_ps(velec,cutoff_mask);
394 velecsum = _mm256_add_ps(velecsum,velec);
395 vvdw = _mm256_and_ps(vvdw,cutoff_mask);
396 vvdwsum = _mm256_add_ps(vvdwsum,vvdw);
398 fscal = _mm256_add_ps(felec,fvdw);
400 fscal = _mm256_and_ps(fscal,cutoff_mask);
402 /* Calculate temporary vectorial force */
403 tx = _mm256_mul_ps(fscal,dx00);
404 ty = _mm256_mul_ps(fscal,dy00);
405 tz = _mm256_mul_ps(fscal,dz00);
407 /* Update vectorial force */
408 fix0 = _mm256_add_ps(fix0,tx);
409 fiy0 = _mm256_add_ps(fiy0,ty);
410 fiz0 = _mm256_add_ps(fiz0,tz);
412 fjx0 = _mm256_add_ps(fjx0,tx);
413 fjy0 = _mm256_add_ps(fjy0,ty);
414 fjz0 = _mm256_add_ps(fjz0,tz);
418 /**************************
419 * CALCULATE INTERACTIONS *
420 **************************/
422 if (gmx_mm256_any_lt(rsq01,rcutoff2))
425 r01 = _mm256_mul_ps(rsq01,rinv01);
427 /* EWALD ELECTROSTATICS */
429 /* Analytical PME correction */
430 zeta2 = _mm256_mul_ps(beta2,rsq01);
431 rinv3 = _mm256_mul_ps(rinvsq01,rinv01);
432 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
433 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
434 felec = _mm256_mul_ps(qq01,felec);
435 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
436 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
437 velec = _mm256_sub_ps(_mm256_sub_ps(rinv01,sh_ewald),pmecorrV);
438 velec = _mm256_mul_ps(qq01,velec);
440 cutoff_mask = _mm256_cmp_ps(rsq01,rcutoff2,_CMP_LT_OQ);
442 /* Update potential sum for this i atom from the interaction with this j atom. */
443 velec = _mm256_and_ps(velec,cutoff_mask);
444 velecsum = _mm256_add_ps(velecsum,velec);
448 fscal = _mm256_and_ps(fscal,cutoff_mask);
450 /* Calculate temporary vectorial force */
451 tx = _mm256_mul_ps(fscal,dx01);
452 ty = _mm256_mul_ps(fscal,dy01);
453 tz = _mm256_mul_ps(fscal,dz01);
455 /* Update vectorial force */
456 fix0 = _mm256_add_ps(fix0,tx);
457 fiy0 = _mm256_add_ps(fiy0,ty);
458 fiz0 = _mm256_add_ps(fiz0,tz);
460 fjx1 = _mm256_add_ps(fjx1,tx);
461 fjy1 = _mm256_add_ps(fjy1,ty);
462 fjz1 = _mm256_add_ps(fjz1,tz);
466 /**************************
467 * CALCULATE INTERACTIONS *
468 **************************/
470 if (gmx_mm256_any_lt(rsq02,rcutoff2))
473 r02 = _mm256_mul_ps(rsq02,rinv02);
475 /* EWALD ELECTROSTATICS */
477 /* Analytical PME correction */
478 zeta2 = _mm256_mul_ps(beta2,rsq02);
479 rinv3 = _mm256_mul_ps(rinvsq02,rinv02);
480 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
481 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
482 felec = _mm256_mul_ps(qq02,felec);
483 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
484 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
485 velec = _mm256_sub_ps(_mm256_sub_ps(rinv02,sh_ewald),pmecorrV);
486 velec = _mm256_mul_ps(qq02,velec);
488 cutoff_mask = _mm256_cmp_ps(rsq02,rcutoff2,_CMP_LT_OQ);
490 /* Update potential sum for this i atom from the interaction with this j atom. */
491 velec = _mm256_and_ps(velec,cutoff_mask);
492 velecsum = _mm256_add_ps(velecsum,velec);
496 fscal = _mm256_and_ps(fscal,cutoff_mask);
498 /* Calculate temporary vectorial force */
499 tx = _mm256_mul_ps(fscal,dx02);
500 ty = _mm256_mul_ps(fscal,dy02);
501 tz = _mm256_mul_ps(fscal,dz02);
503 /* Update vectorial force */
504 fix0 = _mm256_add_ps(fix0,tx);
505 fiy0 = _mm256_add_ps(fiy0,ty);
506 fiz0 = _mm256_add_ps(fiz0,tz);
508 fjx2 = _mm256_add_ps(fjx2,tx);
509 fjy2 = _mm256_add_ps(fjy2,ty);
510 fjz2 = _mm256_add_ps(fjz2,tz);
514 /**************************
515 * CALCULATE INTERACTIONS *
516 **************************/
518 if (gmx_mm256_any_lt(rsq10,rcutoff2))
521 r10 = _mm256_mul_ps(rsq10,rinv10);
523 /* EWALD ELECTROSTATICS */
525 /* Analytical PME correction */
526 zeta2 = _mm256_mul_ps(beta2,rsq10);
527 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
528 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
529 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
530 felec = _mm256_mul_ps(qq10,felec);
531 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
532 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
533 velec = _mm256_sub_ps(_mm256_sub_ps(rinv10,sh_ewald),pmecorrV);
534 velec = _mm256_mul_ps(qq10,velec);
536 cutoff_mask = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
538 /* Update potential sum for this i atom from the interaction with this j atom. */
539 velec = _mm256_and_ps(velec,cutoff_mask);
540 velecsum = _mm256_add_ps(velecsum,velec);
544 fscal = _mm256_and_ps(fscal,cutoff_mask);
546 /* Calculate temporary vectorial force */
547 tx = _mm256_mul_ps(fscal,dx10);
548 ty = _mm256_mul_ps(fscal,dy10);
549 tz = _mm256_mul_ps(fscal,dz10);
551 /* Update vectorial force */
552 fix1 = _mm256_add_ps(fix1,tx);
553 fiy1 = _mm256_add_ps(fiy1,ty);
554 fiz1 = _mm256_add_ps(fiz1,tz);
556 fjx0 = _mm256_add_ps(fjx0,tx);
557 fjy0 = _mm256_add_ps(fjy0,ty);
558 fjz0 = _mm256_add_ps(fjz0,tz);
562 /**************************
563 * CALCULATE INTERACTIONS *
564 **************************/
566 if (gmx_mm256_any_lt(rsq11,rcutoff2))
569 r11 = _mm256_mul_ps(rsq11,rinv11);
571 /* EWALD ELECTROSTATICS */
573 /* Analytical PME correction */
574 zeta2 = _mm256_mul_ps(beta2,rsq11);
575 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
576 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
577 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
578 felec = _mm256_mul_ps(qq11,felec);
579 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
580 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
581 velec = _mm256_sub_ps(_mm256_sub_ps(rinv11,sh_ewald),pmecorrV);
582 velec = _mm256_mul_ps(qq11,velec);
584 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
586 /* Update potential sum for this i atom from the interaction with this j atom. */
587 velec = _mm256_and_ps(velec,cutoff_mask);
588 velecsum = _mm256_add_ps(velecsum,velec);
592 fscal = _mm256_and_ps(fscal,cutoff_mask);
594 /* Calculate temporary vectorial force */
595 tx = _mm256_mul_ps(fscal,dx11);
596 ty = _mm256_mul_ps(fscal,dy11);
597 tz = _mm256_mul_ps(fscal,dz11);
599 /* Update vectorial force */
600 fix1 = _mm256_add_ps(fix1,tx);
601 fiy1 = _mm256_add_ps(fiy1,ty);
602 fiz1 = _mm256_add_ps(fiz1,tz);
604 fjx1 = _mm256_add_ps(fjx1,tx);
605 fjy1 = _mm256_add_ps(fjy1,ty);
606 fjz1 = _mm256_add_ps(fjz1,tz);
610 /**************************
611 * CALCULATE INTERACTIONS *
612 **************************/
614 if (gmx_mm256_any_lt(rsq12,rcutoff2))
617 r12 = _mm256_mul_ps(rsq12,rinv12);
619 /* EWALD ELECTROSTATICS */
621 /* Analytical PME correction */
622 zeta2 = _mm256_mul_ps(beta2,rsq12);
623 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
624 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
625 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
626 felec = _mm256_mul_ps(qq12,felec);
627 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
628 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
629 velec = _mm256_sub_ps(_mm256_sub_ps(rinv12,sh_ewald),pmecorrV);
630 velec = _mm256_mul_ps(qq12,velec);
632 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
634 /* Update potential sum for this i atom from the interaction with this j atom. */
635 velec = _mm256_and_ps(velec,cutoff_mask);
636 velecsum = _mm256_add_ps(velecsum,velec);
640 fscal = _mm256_and_ps(fscal,cutoff_mask);
642 /* Calculate temporary vectorial force */
643 tx = _mm256_mul_ps(fscal,dx12);
644 ty = _mm256_mul_ps(fscal,dy12);
645 tz = _mm256_mul_ps(fscal,dz12);
647 /* Update vectorial force */
648 fix1 = _mm256_add_ps(fix1,tx);
649 fiy1 = _mm256_add_ps(fiy1,ty);
650 fiz1 = _mm256_add_ps(fiz1,tz);
652 fjx2 = _mm256_add_ps(fjx2,tx);
653 fjy2 = _mm256_add_ps(fjy2,ty);
654 fjz2 = _mm256_add_ps(fjz2,tz);
658 /**************************
659 * CALCULATE INTERACTIONS *
660 **************************/
662 if (gmx_mm256_any_lt(rsq20,rcutoff2))
665 r20 = _mm256_mul_ps(rsq20,rinv20);
667 /* EWALD ELECTROSTATICS */
669 /* Analytical PME correction */
670 zeta2 = _mm256_mul_ps(beta2,rsq20);
671 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
672 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
673 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
674 felec = _mm256_mul_ps(qq20,felec);
675 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
676 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
677 velec = _mm256_sub_ps(_mm256_sub_ps(rinv20,sh_ewald),pmecorrV);
678 velec = _mm256_mul_ps(qq20,velec);
680 cutoff_mask = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
682 /* Update potential sum for this i atom from the interaction with this j atom. */
683 velec = _mm256_and_ps(velec,cutoff_mask);
684 velecsum = _mm256_add_ps(velecsum,velec);
688 fscal = _mm256_and_ps(fscal,cutoff_mask);
690 /* Calculate temporary vectorial force */
691 tx = _mm256_mul_ps(fscal,dx20);
692 ty = _mm256_mul_ps(fscal,dy20);
693 tz = _mm256_mul_ps(fscal,dz20);
695 /* Update vectorial force */
696 fix2 = _mm256_add_ps(fix2,tx);
697 fiy2 = _mm256_add_ps(fiy2,ty);
698 fiz2 = _mm256_add_ps(fiz2,tz);
700 fjx0 = _mm256_add_ps(fjx0,tx);
701 fjy0 = _mm256_add_ps(fjy0,ty);
702 fjz0 = _mm256_add_ps(fjz0,tz);
706 /**************************
707 * CALCULATE INTERACTIONS *
708 **************************/
710 if (gmx_mm256_any_lt(rsq21,rcutoff2))
713 r21 = _mm256_mul_ps(rsq21,rinv21);
715 /* EWALD ELECTROSTATICS */
717 /* Analytical PME correction */
718 zeta2 = _mm256_mul_ps(beta2,rsq21);
719 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
720 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
721 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
722 felec = _mm256_mul_ps(qq21,felec);
723 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
724 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
725 velec = _mm256_sub_ps(_mm256_sub_ps(rinv21,sh_ewald),pmecorrV);
726 velec = _mm256_mul_ps(qq21,velec);
728 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
730 /* Update potential sum for this i atom from the interaction with this j atom. */
731 velec = _mm256_and_ps(velec,cutoff_mask);
732 velecsum = _mm256_add_ps(velecsum,velec);
736 fscal = _mm256_and_ps(fscal,cutoff_mask);
738 /* Calculate temporary vectorial force */
739 tx = _mm256_mul_ps(fscal,dx21);
740 ty = _mm256_mul_ps(fscal,dy21);
741 tz = _mm256_mul_ps(fscal,dz21);
743 /* Update vectorial force */
744 fix2 = _mm256_add_ps(fix2,tx);
745 fiy2 = _mm256_add_ps(fiy2,ty);
746 fiz2 = _mm256_add_ps(fiz2,tz);
748 fjx1 = _mm256_add_ps(fjx1,tx);
749 fjy1 = _mm256_add_ps(fjy1,ty);
750 fjz1 = _mm256_add_ps(fjz1,tz);
754 /**************************
755 * CALCULATE INTERACTIONS *
756 **************************/
758 if (gmx_mm256_any_lt(rsq22,rcutoff2))
761 r22 = _mm256_mul_ps(rsq22,rinv22);
763 /* EWALD ELECTROSTATICS */
765 /* Analytical PME correction */
766 zeta2 = _mm256_mul_ps(beta2,rsq22);
767 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
768 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
769 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
770 felec = _mm256_mul_ps(qq22,felec);
771 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
772 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
773 velec = _mm256_sub_ps(_mm256_sub_ps(rinv22,sh_ewald),pmecorrV);
774 velec = _mm256_mul_ps(qq22,velec);
776 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
778 /* Update potential sum for this i atom from the interaction with this j atom. */
779 velec = _mm256_and_ps(velec,cutoff_mask);
780 velecsum = _mm256_add_ps(velecsum,velec);
784 fscal = _mm256_and_ps(fscal,cutoff_mask);
786 /* Calculate temporary vectorial force */
787 tx = _mm256_mul_ps(fscal,dx22);
788 ty = _mm256_mul_ps(fscal,dy22);
789 tz = _mm256_mul_ps(fscal,dz22);
791 /* Update vectorial force */
792 fix2 = _mm256_add_ps(fix2,tx);
793 fiy2 = _mm256_add_ps(fiy2,ty);
794 fiz2 = _mm256_add_ps(fiz2,tz);
796 fjx2 = _mm256_add_ps(fjx2,tx);
797 fjy2 = _mm256_add_ps(fjy2,ty);
798 fjz2 = _mm256_add_ps(fjz2,tz);
802 fjptrA = f+j_coord_offsetA;
803 fjptrB = f+j_coord_offsetB;
804 fjptrC = f+j_coord_offsetC;
805 fjptrD = f+j_coord_offsetD;
806 fjptrE = f+j_coord_offsetE;
807 fjptrF = f+j_coord_offsetF;
808 fjptrG = f+j_coord_offsetG;
809 fjptrH = f+j_coord_offsetH;
811 gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
812 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
814 /* Inner loop uses 1017 flops */
820 /* Get j neighbor index, and coordinate index */
821 jnrlistA = jjnr[jidx];
822 jnrlistB = jjnr[jidx+1];
823 jnrlistC = jjnr[jidx+2];
824 jnrlistD = jjnr[jidx+3];
825 jnrlistE = jjnr[jidx+4];
826 jnrlistF = jjnr[jidx+5];
827 jnrlistG = jjnr[jidx+6];
828 jnrlistH = jjnr[jidx+7];
829 /* Sign of each element will be negative for non-real atoms.
830 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
831 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
833 dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
834 gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
836 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
837 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
838 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
839 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
840 jnrE = (jnrlistE>=0) ? jnrlistE : 0;
841 jnrF = (jnrlistF>=0) ? jnrlistF : 0;
842 jnrG = (jnrlistG>=0) ? jnrlistG : 0;
843 jnrH = (jnrlistH>=0) ? jnrlistH : 0;
844 j_coord_offsetA = DIM*jnrA;
845 j_coord_offsetB = DIM*jnrB;
846 j_coord_offsetC = DIM*jnrC;
847 j_coord_offsetD = DIM*jnrD;
848 j_coord_offsetE = DIM*jnrE;
849 j_coord_offsetF = DIM*jnrF;
850 j_coord_offsetG = DIM*jnrG;
851 j_coord_offsetH = DIM*jnrH;
853 /* load j atom coordinates */
854 gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
855 x+j_coord_offsetC,x+j_coord_offsetD,
856 x+j_coord_offsetE,x+j_coord_offsetF,
857 x+j_coord_offsetG,x+j_coord_offsetH,
858 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
860 /* Calculate displacement vector */
861 dx00 = _mm256_sub_ps(ix0,jx0);
862 dy00 = _mm256_sub_ps(iy0,jy0);
863 dz00 = _mm256_sub_ps(iz0,jz0);
864 dx01 = _mm256_sub_ps(ix0,jx1);
865 dy01 = _mm256_sub_ps(iy0,jy1);
866 dz01 = _mm256_sub_ps(iz0,jz1);
867 dx02 = _mm256_sub_ps(ix0,jx2);
868 dy02 = _mm256_sub_ps(iy0,jy2);
869 dz02 = _mm256_sub_ps(iz0,jz2);
870 dx10 = _mm256_sub_ps(ix1,jx0);
871 dy10 = _mm256_sub_ps(iy1,jy0);
872 dz10 = _mm256_sub_ps(iz1,jz0);
873 dx11 = _mm256_sub_ps(ix1,jx1);
874 dy11 = _mm256_sub_ps(iy1,jy1);
875 dz11 = _mm256_sub_ps(iz1,jz1);
876 dx12 = _mm256_sub_ps(ix1,jx2);
877 dy12 = _mm256_sub_ps(iy1,jy2);
878 dz12 = _mm256_sub_ps(iz1,jz2);
879 dx20 = _mm256_sub_ps(ix2,jx0);
880 dy20 = _mm256_sub_ps(iy2,jy0);
881 dz20 = _mm256_sub_ps(iz2,jz0);
882 dx21 = _mm256_sub_ps(ix2,jx1);
883 dy21 = _mm256_sub_ps(iy2,jy1);
884 dz21 = _mm256_sub_ps(iz2,jz1);
885 dx22 = _mm256_sub_ps(ix2,jx2);
886 dy22 = _mm256_sub_ps(iy2,jy2);
887 dz22 = _mm256_sub_ps(iz2,jz2);
889 /* Calculate squared distance and things based on it */
890 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
891 rsq01 = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
892 rsq02 = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
893 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
894 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
895 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
896 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
897 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
898 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
900 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
901 rinv01 = gmx_mm256_invsqrt_ps(rsq01);
902 rinv02 = gmx_mm256_invsqrt_ps(rsq02);
903 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
904 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
905 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
906 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
907 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
908 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
910 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
911 rinvsq01 = _mm256_mul_ps(rinv01,rinv01);
912 rinvsq02 = _mm256_mul_ps(rinv02,rinv02);
913 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
914 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
915 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
916 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
917 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
918 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
920 fjx0 = _mm256_setzero_ps();
921 fjy0 = _mm256_setzero_ps();
922 fjz0 = _mm256_setzero_ps();
923 fjx1 = _mm256_setzero_ps();
924 fjy1 = _mm256_setzero_ps();
925 fjz1 = _mm256_setzero_ps();
926 fjx2 = _mm256_setzero_ps();
927 fjy2 = _mm256_setzero_ps();
928 fjz2 = _mm256_setzero_ps();
930 /**************************
931 * CALCULATE INTERACTIONS *
932 **************************/
934 if (gmx_mm256_any_lt(rsq00,rcutoff2))
937 r00 = _mm256_mul_ps(rsq00,rinv00);
938 r00 = _mm256_andnot_ps(dummy_mask,r00);
940 /* EWALD ELECTROSTATICS */
942 /* Analytical PME correction */
943 zeta2 = _mm256_mul_ps(beta2,rsq00);
944 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
945 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
946 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
947 felec = _mm256_mul_ps(qq00,felec);
948 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
949 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
950 velec = _mm256_sub_ps(_mm256_sub_ps(rinv00,sh_ewald),pmecorrV);
951 velec = _mm256_mul_ps(qq00,velec);
953 /* Analytical LJ-PME */
954 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
955 ewcljrsq = _mm256_mul_ps(ewclj2,rsq00);
956 ewclj6 = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
957 exponent = gmx_simd_exp_r(ewcljrsq);
958 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
959 poly = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
960 /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
961 vvdw6 = _mm256_mul_ps(_mm256_sub_ps(c6_00,_mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly))),rinvsix);
962 vvdw12 = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
963 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) ,
964 _mm256_mul_ps( _mm256_sub_ps(vvdw6,_mm256_add_ps(_mm256_mul_ps(c6_00,sh_vdw_invrcut6),_mm256_mul_ps(c6grid_00,sh_lj_ewald))),one_sixth));
965 /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
966 fvdw = _mm256_mul_ps(_mm256_sub_ps(vvdw12,_mm256_sub_ps(vvdw6,_mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6)))),rinvsq00);
968 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
970 /* Update potential sum for this i atom from the interaction with this j atom. */
971 velec = _mm256_and_ps(velec,cutoff_mask);
972 velec = _mm256_andnot_ps(dummy_mask,velec);
973 velecsum = _mm256_add_ps(velecsum,velec);
974 vvdw = _mm256_and_ps(vvdw,cutoff_mask);
975 vvdw = _mm256_andnot_ps(dummy_mask,vvdw);
976 vvdwsum = _mm256_add_ps(vvdwsum,vvdw);
978 fscal = _mm256_add_ps(felec,fvdw);
980 fscal = _mm256_and_ps(fscal,cutoff_mask);
982 fscal = _mm256_andnot_ps(dummy_mask,fscal);
984 /* Calculate temporary vectorial force */
985 tx = _mm256_mul_ps(fscal,dx00);
986 ty = _mm256_mul_ps(fscal,dy00);
987 tz = _mm256_mul_ps(fscal,dz00);
989 /* Update vectorial force */
990 fix0 = _mm256_add_ps(fix0,tx);
991 fiy0 = _mm256_add_ps(fiy0,ty);
992 fiz0 = _mm256_add_ps(fiz0,tz);
994 fjx0 = _mm256_add_ps(fjx0,tx);
995 fjy0 = _mm256_add_ps(fjy0,ty);
996 fjz0 = _mm256_add_ps(fjz0,tz);
1000 /**************************
1001 * CALCULATE INTERACTIONS *
1002 **************************/
1004 if (gmx_mm256_any_lt(rsq01,rcutoff2))
1007 r01 = _mm256_mul_ps(rsq01,rinv01);
1008 r01 = _mm256_andnot_ps(dummy_mask,r01);
1010 /* EWALD ELECTROSTATICS */
1012 /* Analytical PME correction */
1013 zeta2 = _mm256_mul_ps(beta2,rsq01);
1014 rinv3 = _mm256_mul_ps(rinvsq01,rinv01);
1015 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1016 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1017 felec = _mm256_mul_ps(qq01,felec);
1018 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1019 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1020 velec = _mm256_sub_ps(_mm256_sub_ps(rinv01,sh_ewald),pmecorrV);
1021 velec = _mm256_mul_ps(qq01,velec);
1023 cutoff_mask = _mm256_cmp_ps(rsq01,rcutoff2,_CMP_LT_OQ);
1025 /* Update potential sum for this i atom from the interaction with this j atom. */
1026 velec = _mm256_and_ps(velec,cutoff_mask);
1027 velec = _mm256_andnot_ps(dummy_mask,velec);
1028 velecsum = _mm256_add_ps(velecsum,velec);
1032 fscal = _mm256_and_ps(fscal,cutoff_mask);
1034 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1036 /* Calculate temporary vectorial force */
1037 tx = _mm256_mul_ps(fscal,dx01);
1038 ty = _mm256_mul_ps(fscal,dy01);
1039 tz = _mm256_mul_ps(fscal,dz01);
1041 /* Update vectorial force */
1042 fix0 = _mm256_add_ps(fix0,tx);
1043 fiy0 = _mm256_add_ps(fiy0,ty);
1044 fiz0 = _mm256_add_ps(fiz0,tz);
1046 fjx1 = _mm256_add_ps(fjx1,tx);
1047 fjy1 = _mm256_add_ps(fjy1,ty);
1048 fjz1 = _mm256_add_ps(fjz1,tz);
1052 /**************************
1053 * CALCULATE INTERACTIONS *
1054 **************************/
1056 if (gmx_mm256_any_lt(rsq02,rcutoff2))
1059 r02 = _mm256_mul_ps(rsq02,rinv02);
1060 r02 = _mm256_andnot_ps(dummy_mask,r02);
1062 /* EWALD ELECTROSTATICS */
1064 /* Analytical PME correction */
1065 zeta2 = _mm256_mul_ps(beta2,rsq02);
1066 rinv3 = _mm256_mul_ps(rinvsq02,rinv02);
1067 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1068 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1069 felec = _mm256_mul_ps(qq02,felec);
1070 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1071 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1072 velec = _mm256_sub_ps(_mm256_sub_ps(rinv02,sh_ewald),pmecorrV);
1073 velec = _mm256_mul_ps(qq02,velec);
1075 cutoff_mask = _mm256_cmp_ps(rsq02,rcutoff2,_CMP_LT_OQ);
1077 /* Update potential sum for this i atom from the interaction with this j atom. */
1078 velec = _mm256_and_ps(velec,cutoff_mask);
1079 velec = _mm256_andnot_ps(dummy_mask,velec);
1080 velecsum = _mm256_add_ps(velecsum,velec);
1084 fscal = _mm256_and_ps(fscal,cutoff_mask);
1086 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1088 /* Calculate temporary vectorial force */
1089 tx = _mm256_mul_ps(fscal,dx02);
1090 ty = _mm256_mul_ps(fscal,dy02);
1091 tz = _mm256_mul_ps(fscal,dz02);
1093 /* Update vectorial force */
1094 fix0 = _mm256_add_ps(fix0,tx);
1095 fiy0 = _mm256_add_ps(fiy0,ty);
1096 fiz0 = _mm256_add_ps(fiz0,tz);
1098 fjx2 = _mm256_add_ps(fjx2,tx);
1099 fjy2 = _mm256_add_ps(fjy2,ty);
1100 fjz2 = _mm256_add_ps(fjz2,tz);
1104 /**************************
1105 * CALCULATE INTERACTIONS *
1106 **************************/
1108 if (gmx_mm256_any_lt(rsq10,rcutoff2))
1111 r10 = _mm256_mul_ps(rsq10,rinv10);
1112 r10 = _mm256_andnot_ps(dummy_mask,r10);
1114 /* EWALD ELECTROSTATICS */
1116 /* Analytical PME correction */
1117 zeta2 = _mm256_mul_ps(beta2,rsq10);
1118 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
1119 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1120 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1121 felec = _mm256_mul_ps(qq10,felec);
1122 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1123 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1124 velec = _mm256_sub_ps(_mm256_sub_ps(rinv10,sh_ewald),pmecorrV);
1125 velec = _mm256_mul_ps(qq10,velec);
1127 cutoff_mask = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
1129 /* Update potential sum for this i atom from the interaction with this j atom. */
1130 velec = _mm256_and_ps(velec,cutoff_mask);
1131 velec = _mm256_andnot_ps(dummy_mask,velec);
1132 velecsum = _mm256_add_ps(velecsum,velec);
1136 fscal = _mm256_and_ps(fscal,cutoff_mask);
1138 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1140 /* Calculate temporary vectorial force */
1141 tx = _mm256_mul_ps(fscal,dx10);
1142 ty = _mm256_mul_ps(fscal,dy10);
1143 tz = _mm256_mul_ps(fscal,dz10);
1145 /* Update vectorial force */
1146 fix1 = _mm256_add_ps(fix1,tx);
1147 fiy1 = _mm256_add_ps(fiy1,ty);
1148 fiz1 = _mm256_add_ps(fiz1,tz);
1150 fjx0 = _mm256_add_ps(fjx0,tx);
1151 fjy0 = _mm256_add_ps(fjy0,ty);
1152 fjz0 = _mm256_add_ps(fjz0,tz);
1156 /**************************
1157 * CALCULATE INTERACTIONS *
1158 **************************/
1160 if (gmx_mm256_any_lt(rsq11,rcutoff2))
1163 r11 = _mm256_mul_ps(rsq11,rinv11);
1164 r11 = _mm256_andnot_ps(dummy_mask,r11);
1166 /* EWALD ELECTROSTATICS */
1168 /* Analytical PME correction */
1169 zeta2 = _mm256_mul_ps(beta2,rsq11);
1170 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
1171 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1172 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1173 felec = _mm256_mul_ps(qq11,felec);
1174 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1175 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1176 velec = _mm256_sub_ps(_mm256_sub_ps(rinv11,sh_ewald),pmecorrV);
1177 velec = _mm256_mul_ps(qq11,velec);
1179 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
1181 /* Update potential sum for this i atom from the interaction with this j atom. */
1182 velec = _mm256_and_ps(velec,cutoff_mask);
1183 velec = _mm256_andnot_ps(dummy_mask,velec);
1184 velecsum = _mm256_add_ps(velecsum,velec);
1188 fscal = _mm256_and_ps(fscal,cutoff_mask);
1190 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1192 /* Calculate temporary vectorial force */
1193 tx = _mm256_mul_ps(fscal,dx11);
1194 ty = _mm256_mul_ps(fscal,dy11);
1195 tz = _mm256_mul_ps(fscal,dz11);
1197 /* Update vectorial force */
1198 fix1 = _mm256_add_ps(fix1,tx);
1199 fiy1 = _mm256_add_ps(fiy1,ty);
1200 fiz1 = _mm256_add_ps(fiz1,tz);
1202 fjx1 = _mm256_add_ps(fjx1,tx);
1203 fjy1 = _mm256_add_ps(fjy1,ty);
1204 fjz1 = _mm256_add_ps(fjz1,tz);
1208 /**************************
1209 * CALCULATE INTERACTIONS *
1210 **************************/
1212 if (gmx_mm256_any_lt(rsq12,rcutoff2))
1215 r12 = _mm256_mul_ps(rsq12,rinv12);
1216 r12 = _mm256_andnot_ps(dummy_mask,r12);
1218 /* EWALD ELECTROSTATICS */
1220 /* Analytical PME correction */
1221 zeta2 = _mm256_mul_ps(beta2,rsq12);
1222 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
1223 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1224 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1225 felec = _mm256_mul_ps(qq12,felec);
1226 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1227 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1228 velec = _mm256_sub_ps(_mm256_sub_ps(rinv12,sh_ewald),pmecorrV);
1229 velec = _mm256_mul_ps(qq12,velec);
1231 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
1233 /* Update potential sum for this i atom from the interaction with this j atom. */
1234 velec = _mm256_and_ps(velec,cutoff_mask);
1235 velec = _mm256_andnot_ps(dummy_mask,velec);
1236 velecsum = _mm256_add_ps(velecsum,velec);
1240 fscal = _mm256_and_ps(fscal,cutoff_mask);
1242 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1244 /* Calculate temporary vectorial force */
1245 tx = _mm256_mul_ps(fscal,dx12);
1246 ty = _mm256_mul_ps(fscal,dy12);
1247 tz = _mm256_mul_ps(fscal,dz12);
1249 /* Update vectorial force */
1250 fix1 = _mm256_add_ps(fix1,tx);
1251 fiy1 = _mm256_add_ps(fiy1,ty);
1252 fiz1 = _mm256_add_ps(fiz1,tz);
1254 fjx2 = _mm256_add_ps(fjx2,tx);
1255 fjy2 = _mm256_add_ps(fjy2,ty);
1256 fjz2 = _mm256_add_ps(fjz2,tz);
1260 /**************************
1261 * CALCULATE INTERACTIONS *
1262 **************************/
1264 if (gmx_mm256_any_lt(rsq20,rcutoff2))
1267 r20 = _mm256_mul_ps(rsq20,rinv20);
1268 r20 = _mm256_andnot_ps(dummy_mask,r20);
1270 /* EWALD ELECTROSTATICS */
1272 /* Analytical PME correction */
1273 zeta2 = _mm256_mul_ps(beta2,rsq20);
1274 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
1275 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1276 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1277 felec = _mm256_mul_ps(qq20,felec);
1278 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1279 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1280 velec = _mm256_sub_ps(_mm256_sub_ps(rinv20,sh_ewald),pmecorrV);
1281 velec = _mm256_mul_ps(qq20,velec);
1283 cutoff_mask = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
1285 /* Update potential sum for this i atom from the interaction with this j atom. */
1286 velec = _mm256_and_ps(velec,cutoff_mask);
1287 velec = _mm256_andnot_ps(dummy_mask,velec);
1288 velecsum = _mm256_add_ps(velecsum,velec);
1292 fscal = _mm256_and_ps(fscal,cutoff_mask);
1294 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1296 /* Calculate temporary vectorial force */
1297 tx = _mm256_mul_ps(fscal,dx20);
1298 ty = _mm256_mul_ps(fscal,dy20);
1299 tz = _mm256_mul_ps(fscal,dz20);
1301 /* Update vectorial force */
1302 fix2 = _mm256_add_ps(fix2,tx);
1303 fiy2 = _mm256_add_ps(fiy2,ty);
1304 fiz2 = _mm256_add_ps(fiz2,tz);
1306 fjx0 = _mm256_add_ps(fjx0,tx);
1307 fjy0 = _mm256_add_ps(fjy0,ty);
1308 fjz0 = _mm256_add_ps(fjz0,tz);
1312 /**************************
1313 * CALCULATE INTERACTIONS *
1314 **************************/
1316 if (gmx_mm256_any_lt(rsq21,rcutoff2))
1319 r21 = _mm256_mul_ps(rsq21,rinv21);
1320 r21 = _mm256_andnot_ps(dummy_mask,r21);
1322 /* EWALD ELECTROSTATICS */
1324 /* Analytical PME correction */
1325 zeta2 = _mm256_mul_ps(beta2,rsq21);
1326 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
1327 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1328 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1329 felec = _mm256_mul_ps(qq21,felec);
1330 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1331 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1332 velec = _mm256_sub_ps(_mm256_sub_ps(rinv21,sh_ewald),pmecorrV);
1333 velec = _mm256_mul_ps(qq21,velec);
1335 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
1337 /* Update potential sum for this i atom from the interaction with this j atom. */
1338 velec = _mm256_and_ps(velec,cutoff_mask);
1339 velec = _mm256_andnot_ps(dummy_mask,velec);
1340 velecsum = _mm256_add_ps(velecsum,velec);
1344 fscal = _mm256_and_ps(fscal,cutoff_mask);
1346 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1348 /* Calculate temporary vectorial force */
1349 tx = _mm256_mul_ps(fscal,dx21);
1350 ty = _mm256_mul_ps(fscal,dy21);
1351 tz = _mm256_mul_ps(fscal,dz21);
1353 /* Update vectorial force */
1354 fix2 = _mm256_add_ps(fix2,tx);
1355 fiy2 = _mm256_add_ps(fiy2,ty);
1356 fiz2 = _mm256_add_ps(fiz2,tz);
1358 fjx1 = _mm256_add_ps(fjx1,tx);
1359 fjy1 = _mm256_add_ps(fjy1,ty);
1360 fjz1 = _mm256_add_ps(fjz1,tz);
1364 /**************************
1365 * CALCULATE INTERACTIONS *
1366 **************************/
1368 if (gmx_mm256_any_lt(rsq22,rcutoff2))
1371 r22 = _mm256_mul_ps(rsq22,rinv22);
1372 r22 = _mm256_andnot_ps(dummy_mask,r22);
1374 /* EWALD ELECTROSTATICS */
1376 /* Analytical PME correction */
1377 zeta2 = _mm256_mul_ps(beta2,rsq22);
1378 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
1379 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1380 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1381 felec = _mm256_mul_ps(qq22,felec);
1382 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
1383 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
1384 velec = _mm256_sub_ps(_mm256_sub_ps(rinv22,sh_ewald),pmecorrV);
1385 velec = _mm256_mul_ps(qq22,velec);
1387 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
1389 /* Update potential sum for this i atom from the interaction with this j atom. */
1390 velec = _mm256_and_ps(velec,cutoff_mask);
1391 velec = _mm256_andnot_ps(dummy_mask,velec);
1392 velecsum = _mm256_add_ps(velecsum,velec);
1396 fscal = _mm256_and_ps(fscal,cutoff_mask);
1398 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1400 /* Calculate temporary vectorial force */
1401 tx = _mm256_mul_ps(fscal,dx22);
1402 ty = _mm256_mul_ps(fscal,dy22);
1403 tz = _mm256_mul_ps(fscal,dz22);
1405 /* Update vectorial force */
1406 fix2 = _mm256_add_ps(fix2,tx);
1407 fiy2 = _mm256_add_ps(fiy2,ty);
1408 fiz2 = _mm256_add_ps(fiz2,tz);
1410 fjx2 = _mm256_add_ps(fjx2,tx);
1411 fjy2 = _mm256_add_ps(fjy2,ty);
1412 fjz2 = _mm256_add_ps(fjz2,tz);
1416 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1417 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1418 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1419 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1420 fjptrE = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1421 fjptrF = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1422 fjptrG = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1423 fjptrH = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1425 gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
1426 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1428 /* Inner loop uses 1026 flops */
1431 /* End of innermost loop */
1433 gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1434 f+i_coord_offset,fshift+i_shift_offset);
1437 /* Update potential energies */
1438 gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1439 gmx_mm256_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
1441 /* Increment number of inner iterations */
1442 inneriter += j_index_end - j_index_start;
1444 /* Outer loop uses 20 flops */
1447 /* Increment number of outer iterations */
1450 /* Update outer/inner flops */
1452 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*1026);
1455 * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_F_avx_256_single
1456 * Electrostatics interaction: Ewald
1457 * VdW interaction: LJEwald
1458 * Geometry: Water3-Water3
1459 * Calculate force/pot: Force
1462 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_F_avx_256_single
1463 (t_nblist * gmx_restrict nlist,
1464 rvec * gmx_restrict xx,
1465 rvec * gmx_restrict ff,
1466 t_forcerec * gmx_restrict fr,
1467 t_mdatoms * gmx_restrict mdatoms,
1468 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1469 t_nrnb * gmx_restrict nrnb)
1471 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1472 * just 0 for non-waters.
1473 * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
1474 * jnr indices corresponding to data put in the four positions in the SIMD register.
1476 int i_shift_offset,i_coord_offset,outeriter,inneriter;
1477 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1478 int jnrA,jnrB,jnrC,jnrD;
1479 int jnrE,jnrF,jnrG,jnrH;
1480 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1481 int jnrlistE,jnrlistF,jnrlistG,jnrlistH;
1482 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1483 int j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
1484 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
1485 real rcutoff_scalar;
1486 real *shiftvec,*fshift,*x,*f;
1487 real *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
1488 real scratch[4*DIM];
1489 __m256 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1490 real * vdwioffsetptr0;
1491 real * vdwgridioffsetptr0;
1492 __m256 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1493 real * vdwioffsetptr1;
1494 real * vdwgridioffsetptr1;
1495 __m256 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1496 real * vdwioffsetptr2;
1497 real * vdwgridioffsetptr2;
1498 __m256 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1499 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
1500 __m256 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1501 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
1502 __m256 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1503 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
1504 __m256 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1505 __m256 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1506 __m256 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
1507 __m256 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
1508 __m256 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
1509 __m256 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1510 __m256 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1511 __m256 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
1512 __m256 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1513 __m256 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1514 __m256 velec,felec,velecsum,facel,crf,krf,krf2;
1517 __m256 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1520 __m256 one_sixth = _mm256_set1_ps(1.0/6.0);
1521 __m256 one_twelfth = _mm256_set1_ps(1.0/12.0);
1532 __m256 ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
1533 __m256 one_half = _mm256_set1_ps(0.5);
1534 __m256 minus_one = _mm256_set1_ps(-1.0);
1536 __m128i ewitab_lo,ewitab_hi;
1537 __m256 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1538 __m256 beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
1540 __m256 dummy_mask,cutoff_mask;
1541 __m256 signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
1542 __m256 one = _mm256_set1_ps(1.0);
1543 __m256 two = _mm256_set1_ps(2.0);
1549 jindex = nlist->jindex;
1551 shiftidx = nlist->shift;
1553 shiftvec = fr->shift_vec[0];
1554 fshift = fr->fshift[0];
1555 facel = _mm256_set1_ps(fr->epsfac);
1556 charge = mdatoms->chargeA;
1557 nvdwtype = fr->ntype;
1558 vdwparam = fr->nbfp;
1559 vdwtype = mdatoms->typeA;
1560 vdwgridparam = fr->ljpme_c6grid;
1561 sh_lj_ewald = _mm256_set1_ps(fr->ic->sh_lj_ewald);
1562 ewclj = _mm256_set1_ps(fr->ewaldcoeff_lj);
1563 ewclj2 = _mm256_mul_ps(minus_one,_mm256_mul_ps(ewclj,ewclj));
1565 sh_ewald = _mm256_set1_ps(fr->ic->sh_ewald);
1566 beta = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
1567 beta2 = _mm256_mul_ps(beta,beta);
1568 beta3 = _mm256_mul_ps(beta,beta2);
1570 ewtab = fr->ic->tabq_coul_F;
1571 ewtabscale = _mm256_set1_ps(fr->ic->tabq_scale);
1572 ewtabhalfspace = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
1574 /* Setup water-specific parameters */
1575 inr = nlist->iinr[0];
1576 iq0 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
1577 iq1 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
1578 iq2 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
1579 vdwioffsetptr0 = vdwparam+2*nvdwtype*vdwtype[inr+0];
1580 vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
1582 jq0 = _mm256_set1_ps(charge[inr+0]);
1583 jq1 = _mm256_set1_ps(charge[inr+1]);
1584 jq2 = _mm256_set1_ps(charge[inr+2]);
1585 vdwjidx0A = 2*vdwtype[inr+0];
1586 qq00 = _mm256_mul_ps(iq0,jq0);
1587 c6_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A]);
1588 c12_00 = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A+1]);
1589 c6grid_00 = _mm256_set1_ps(vdwgridioffsetptr0[vdwjidx0A]);
1590 qq01 = _mm256_mul_ps(iq0,jq1);
1591 qq02 = _mm256_mul_ps(iq0,jq2);
1592 qq10 = _mm256_mul_ps(iq1,jq0);
1593 qq11 = _mm256_mul_ps(iq1,jq1);
1594 qq12 = _mm256_mul_ps(iq1,jq2);
1595 qq20 = _mm256_mul_ps(iq2,jq0);
1596 qq21 = _mm256_mul_ps(iq2,jq1);
1597 qq22 = _mm256_mul_ps(iq2,jq2);
1599 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
1600 rcutoff_scalar = fr->rcoulomb;
1601 rcutoff = _mm256_set1_ps(rcutoff_scalar);
1602 rcutoff2 = _mm256_mul_ps(rcutoff,rcutoff);
1604 sh_vdw_invrcut6 = _mm256_set1_ps(fr->ic->sh_invrc6);
1605 rvdw = _mm256_set1_ps(fr->rvdw);
1607 /* Avoid stupid compiler warnings */
1608 jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
1609 j_coord_offsetA = 0;
1610 j_coord_offsetB = 0;
1611 j_coord_offsetC = 0;
1612 j_coord_offsetD = 0;
1613 j_coord_offsetE = 0;
1614 j_coord_offsetF = 0;
1615 j_coord_offsetG = 0;
1616 j_coord_offsetH = 0;
1621 for(iidx=0;iidx<4*DIM;iidx++)
1623 scratch[iidx] = 0.0;
1626 /* Start outer loop over neighborlists */
1627 for(iidx=0; iidx<nri; iidx++)
1629 /* Load shift vector for this list */
1630 i_shift_offset = DIM*shiftidx[iidx];
1632 /* Load limits for loop over neighbors */
1633 j_index_start = jindex[iidx];
1634 j_index_end = jindex[iidx+1];
1636 /* Get outer coordinate index */
1638 i_coord_offset = DIM*inr;
1640 /* Load i particle coords and add shift vector */
1641 gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1642 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1644 fix0 = _mm256_setzero_ps();
1645 fiy0 = _mm256_setzero_ps();
1646 fiz0 = _mm256_setzero_ps();
1647 fix1 = _mm256_setzero_ps();
1648 fiy1 = _mm256_setzero_ps();
1649 fiz1 = _mm256_setzero_ps();
1650 fix2 = _mm256_setzero_ps();
1651 fiy2 = _mm256_setzero_ps();
1652 fiz2 = _mm256_setzero_ps();
1654 /* Start inner kernel loop */
1655 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
1658 /* Get j neighbor index, and coordinate index */
1660 jnrB = jjnr[jidx+1];
1661 jnrC = jjnr[jidx+2];
1662 jnrD = jjnr[jidx+3];
1663 jnrE = jjnr[jidx+4];
1664 jnrF = jjnr[jidx+5];
1665 jnrG = jjnr[jidx+6];
1666 jnrH = jjnr[jidx+7];
1667 j_coord_offsetA = DIM*jnrA;
1668 j_coord_offsetB = DIM*jnrB;
1669 j_coord_offsetC = DIM*jnrC;
1670 j_coord_offsetD = DIM*jnrD;
1671 j_coord_offsetE = DIM*jnrE;
1672 j_coord_offsetF = DIM*jnrF;
1673 j_coord_offsetG = DIM*jnrG;
1674 j_coord_offsetH = DIM*jnrH;
1676 /* load j atom coordinates */
1677 gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1678 x+j_coord_offsetC,x+j_coord_offsetD,
1679 x+j_coord_offsetE,x+j_coord_offsetF,
1680 x+j_coord_offsetG,x+j_coord_offsetH,
1681 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1683 /* Calculate displacement vector */
1684 dx00 = _mm256_sub_ps(ix0,jx0);
1685 dy00 = _mm256_sub_ps(iy0,jy0);
1686 dz00 = _mm256_sub_ps(iz0,jz0);
1687 dx01 = _mm256_sub_ps(ix0,jx1);
1688 dy01 = _mm256_sub_ps(iy0,jy1);
1689 dz01 = _mm256_sub_ps(iz0,jz1);
1690 dx02 = _mm256_sub_ps(ix0,jx2);
1691 dy02 = _mm256_sub_ps(iy0,jy2);
1692 dz02 = _mm256_sub_ps(iz0,jz2);
1693 dx10 = _mm256_sub_ps(ix1,jx0);
1694 dy10 = _mm256_sub_ps(iy1,jy0);
1695 dz10 = _mm256_sub_ps(iz1,jz0);
1696 dx11 = _mm256_sub_ps(ix1,jx1);
1697 dy11 = _mm256_sub_ps(iy1,jy1);
1698 dz11 = _mm256_sub_ps(iz1,jz1);
1699 dx12 = _mm256_sub_ps(ix1,jx2);
1700 dy12 = _mm256_sub_ps(iy1,jy2);
1701 dz12 = _mm256_sub_ps(iz1,jz2);
1702 dx20 = _mm256_sub_ps(ix2,jx0);
1703 dy20 = _mm256_sub_ps(iy2,jy0);
1704 dz20 = _mm256_sub_ps(iz2,jz0);
1705 dx21 = _mm256_sub_ps(ix2,jx1);
1706 dy21 = _mm256_sub_ps(iy2,jy1);
1707 dz21 = _mm256_sub_ps(iz2,jz1);
1708 dx22 = _mm256_sub_ps(ix2,jx2);
1709 dy22 = _mm256_sub_ps(iy2,jy2);
1710 dz22 = _mm256_sub_ps(iz2,jz2);
1712 /* Calculate squared distance and things based on it */
1713 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1714 rsq01 = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
1715 rsq02 = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
1716 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1717 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
1718 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
1719 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1720 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
1721 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
1723 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
1724 rinv01 = gmx_mm256_invsqrt_ps(rsq01);
1725 rinv02 = gmx_mm256_invsqrt_ps(rsq02);
1726 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
1727 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
1728 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
1729 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
1730 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
1731 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
1733 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
1734 rinvsq01 = _mm256_mul_ps(rinv01,rinv01);
1735 rinvsq02 = _mm256_mul_ps(rinv02,rinv02);
1736 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
1737 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
1738 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
1739 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
1740 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
1741 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
1743 fjx0 = _mm256_setzero_ps();
1744 fjy0 = _mm256_setzero_ps();
1745 fjz0 = _mm256_setzero_ps();
1746 fjx1 = _mm256_setzero_ps();
1747 fjy1 = _mm256_setzero_ps();
1748 fjz1 = _mm256_setzero_ps();
1749 fjx2 = _mm256_setzero_ps();
1750 fjy2 = _mm256_setzero_ps();
1751 fjz2 = _mm256_setzero_ps();
1753 /**************************
1754 * CALCULATE INTERACTIONS *
1755 **************************/
1757 if (gmx_mm256_any_lt(rsq00,rcutoff2))
1760 r00 = _mm256_mul_ps(rsq00,rinv00);
1762 /* EWALD ELECTROSTATICS */
1764 /* Analytical PME correction */
1765 zeta2 = _mm256_mul_ps(beta2,rsq00);
1766 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
1767 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1768 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1769 felec = _mm256_mul_ps(qq00,felec);
1771 /* Analytical LJ-PME */
1772 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1773 ewcljrsq = _mm256_mul_ps(ewclj2,rsq00);
1774 ewclj6 = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
1775 exponent = gmx_simd_exp_r(ewcljrsq);
1776 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1777 poly = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
1778 /* f6A = 6 * C6grid * (1 - poly) */
1779 f6A = _mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly));
1780 /* f6B = C6grid * exponent * beta^6 */
1781 f6B = _mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6));
1782 /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
1783 fvdw = _mm256_mul_ps(_mm256_add_ps(_mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),_mm256_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
1785 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
1787 fscal = _mm256_add_ps(felec,fvdw);
1789 fscal = _mm256_and_ps(fscal,cutoff_mask);
1791 /* Calculate temporary vectorial force */
1792 tx = _mm256_mul_ps(fscal,dx00);
1793 ty = _mm256_mul_ps(fscal,dy00);
1794 tz = _mm256_mul_ps(fscal,dz00);
1796 /* Update vectorial force */
1797 fix0 = _mm256_add_ps(fix0,tx);
1798 fiy0 = _mm256_add_ps(fiy0,ty);
1799 fiz0 = _mm256_add_ps(fiz0,tz);
1801 fjx0 = _mm256_add_ps(fjx0,tx);
1802 fjy0 = _mm256_add_ps(fjy0,ty);
1803 fjz0 = _mm256_add_ps(fjz0,tz);
1807 /**************************
1808 * CALCULATE INTERACTIONS *
1809 **************************/
1811 if (gmx_mm256_any_lt(rsq01,rcutoff2))
1814 r01 = _mm256_mul_ps(rsq01,rinv01);
1816 /* EWALD ELECTROSTATICS */
1818 /* Analytical PME correction */
1819 zeta2 = _mm256_mul_ps(beta2,rsq01);
1820 rinv3 = _mm256_mul_ps(rinvsq01,rinv01);
1821 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1822 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1823 felec = _mm256_mul_ps(qq01,felec);
1825 cutoff_mask = _mm256_cmp_ps(rsq01,rcutoff2,_CMP_LT_OQ);
1829 fscal = _mm256_and_ps(fscal,cutoff_mask);
1831 /* Calculate temporary vectorial force */
1832 tx = _mm256_mul_ps(fscal,dx01);
1833 ty = _mm256_mul_ps(fscal,dy01);
1834 tz = _mm256_mul_ps(fscal,dz01);
1836 /* Update vectorial force */
1837 fix0 = _mm256_add_ps(fix0,tx);
1838 fiy0 = _mm256_add_ps(fiy0,ty);
1839 fiz0 = _mm256_add_ps(fiz0,tz);
1841 fjx1 = _mm256_add_ps(fjx1,tx);
1842 fjy1 = _mm256_add_ps(fjy1,ty);
1843 fjz1 = _mm256_add_ps(fjz1,tz);
1847 /**************************
1848 * CALCULATE INTERACTIONS *
1849 **************************/
1851 if (gmx_mm256_any_lt(rsq02,rcutoff2))
1854 r02 = _mm256_mul_ps(rsq02,rinv02);
1856 /* EWALD ELECTROSTATICS */
1858 /* Analytical PME correction */
1859 zeta2 = _mm256_mul_ps(beta2,rsq02);
1860 rinv3 = _mm256_mul_ps(rinvsq02,rinv02);
1861 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1862 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1863 felec = _mm256_mul_ps(qq02,felec);
1865 cutoff_mask = _mm256_cmp_ps(rsq02,rcutoff2,_CMP_LT_OQ);
1869 fscal = _mm256_and_ps(fscal,cutoff_mask);
1871 /* Calculate temporary vectorial force */
1872 tx = _mm256_mul_ps(fscal,dx02);
1873 ty = _mm256_mul_ps(fscal,dy02);
1874 tz = _mm256_mul_ps(fscal,dz02);
1876 /* Update vectorial force */
1877 fix0 = _mm256_add_ps(fix0,tx);
1878 fiy0 = _mm256_add_ps(fiy0,ty);
1879 fiz0 = _mm256_add_ps(fiz0,tz);
1881 fjx2 = _mm256_add_ps(fjx2,tx);
1882 fjy2 = _mm256_add_ps(fjy2,ty);
1883 fjz2 = _mm256_add_ps(fjz2,tz);
1887 /**************************
1888 * CALCULATE INTERACTIONS *
1889 **************************/
1891 if (gmx_mm256_any_lt(rsq10,rcutoff2))
1894 r10 = _mm256_mul_ps(rsq10,rinv10);
1896 /* EWALD ELECTROSTATICS */
1898 /* Analytical PME correction */
1899 zeta2 = _mm256_mul_ps(beta2,rsq10);
1900 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
1901 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1902 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1903 felec = _mm256_mul_ps(qq10,felec);
1905 cutoff_mask = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
1909 fscal = _mm256_and_ps(fscal,cutoff_mask);
1911 /* Calculate temporary vectorial force */
1912 tx = _mm256_mul_ps(fscal,dx10);
1913 ty = _mm256_mul_ps(fscal,dy10);
1914 tz = _mm256_mul_ps(fscal,dz10);
1916 /* Update vectorial force */
1917 fix1 = _mm256_add_ps(fix1,tx);
1918 fiy1 = _mm256_add_ps(fiy1,ty);
1919 fiz1 = _mm256_add_ps(fiz1,tz);
1921 fjx0 = _mm256_add_ps(fjx0,tx);
1922 fjy0 = _mm256_add_ps(fjy0,ty);
1923 fjz0 = _mm256_add_ps(fjz0,tz);
1927 /**************************
1928 * CALCULATE INTERACTIONS *
1929 **************************/
1931 if (gmx_mm256_any_lt(rsq11,rcutoff2))
1934 r11 = _mm256_mul_ps(rsq11,rinv11);
1936 /* EWALD ELECTROSTATICS */
1938 /* Analytical PME correction */
1939 zeta2 = _mm256_mul_ps(beta2,rsq11);
1940 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
1941 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1942 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1943 felec = _mm256_mul_ps(qq11,felec);
1945 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
1949 fscal = _mm256_and_ps(fscal,cutoff_mask);
1951 /* Calculate temporary vectorial force */
1952 tx = _mm256_mul_ps(fscal,dx11);
1953 ty = _mm256_mul_ps(fscal,dy11);
1954 tz = _mm256_mul_ps(fscal,dz11);
1956 /* Update vectorial force */
1957 fix1 = _mm256_add_ps(fix1,tx);
1958 fiy1 = _mm256_add_ps(fiy1,ty);
1959 fiz1 = _mm256_add_ps(fiz1,tz);
1961 fjx1 = _mm256_add_ps(fjx1,tx);
1962 fjy1 = _mm256_add_ps(fjy1,ty);
1963 fjz1 = _mm256_add_ps(fjz1,tz);
1967 /**************************
1968 * CALCULATE INTERACTIONS *
1969 **************************/
1971 if (gmx_mm256_any_lt(rsq12,rcutoff2))
1974 r12 = _mm256_mul_ps(rsq12,rinv12);
1976 /* EWALD ELECTROSTATICS */
1978 /* Analytical PME correction */
1979 zeta2 = _mm256_mul_ps(beta2,rsq12);
1980 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
1981 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1982 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1983 felec = _mm256_mul_ps(qq12,felec);
1985 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
1989 fscal = _mm256_and_ps(fscal,cutoff_mask);
1991 /* Calculate temporary vectorial force */
1992 tx = _mm256_mul_ps(fscal,dx12);
1993 ty = _mm256_mul_ps(fscal,dy12);
1994 tz = _mm256_mul_ps(fscal,dz12);
1996 /* Update vectorial force */
1997 fix1 = _mm256_add_ps(fix1,tx);
1998 fiy1 = _mm256_add_ps(fiy1,ty);
1999 fiz1 = _mm256_add_ps(fiz1,tz);
2001 fjx2 = _mm256_add_ps(fjx2,tx);
2002 fjy2 = _mm256_add_ps(fjy2,ty);
2003 fjz2 = _mm256_add_ps(fjz2,tz);
2007 /**************************
2008 * CALCULATE INTERACTIONS *
2009 **************************/
2011 if (gmx_mm256_any_lt(rsq20,rcutoff2))
2014 r20 = _mm256_mul_ps(rsq20,rinv20);
2016 /* EWALD ELECTROSTATICS */
2018 /* Analytical PME correction */
2019 zeta2 = _mm256_mul_ps(beta2,rsq20);
2020 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
2021 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2022 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2023 felec = _mm256_mul_ps(qq20,felec);
2025 cutoff_mask = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
2029 fscal = _mm256_and_ps(fscal,cutoff_mask);
2031 /* Calculate temporary vectorial force */
2032 tx = _mm256_mul_ps(fscal,dx20);
2033 ty = _mm256_mul_ps(fscal,dy20);
2034 tz = _mm256_mul_ps(fscal,dz20);
2036 /* Update vectorial force */
2037 fix2 = _mm256_add_ps(fix2,tx);
2038 fiy2 = _mm256_add_ps(fiy2,ty);
2039 fiz2 = _mm256_add_ps(fiz2,tz);
2041 fjx0 = _mm256_add_ps(fjx0,tx);
2042 fjy0 = _mm256_add_ps(fjy0,ty);
2043 fjz0 = _mm256_add_ps(fjz0,tz);
2047 /**************************
2048 * CALCULATE INTERACTIONS *
2049 **************************/
2051 if (gmx_mm256_any_lt(rsq21,rcutoff2))
2054 r21 = _mm256_mul_ps(rsq21,rinv21);
2056 /* EWALD ELECTROSTATICS */
2058 /* Analytical PME correction */
2059 zeta2 = _mm256_mul_ps(beta2,rsq21);
2060 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
2061 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2062 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2063 felec = _mm256_mul_ps(qq21,felec);
2065 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
2069 fscal = _mm256_and_ps(fscal,cutoff_mask);
2071 /* Calculate temporary vectorial force */
2072 tx = _mm256_mul_ps(fscal,dx21);
2073 ty = _mm256_mul_ps(fscal,dy21);
2074 tz = _mm256_mul_ps(fscal,dz21);
2076 /* Update vectorial force */
2077 fix2 = _mm256_add_ps(fix2,tx);
2078 fiy2 = _mm256_add_ps(fiy2,ty);
2079 fiz2 = _mm256_add_ps(fiz2,tz);
2081 fjx1 = _mm256_add_ps(fjx1,tx);
2082 fjy1 = _mm256_add_ps(fjy1,ty);
2083 fjz1 = _mm256_add_ps(fjz1,tz);
2087 /**************************
2088 * CALCULATE INTERACTIONS *
2089 **************************/
2091 if (gmx_mm256_any_lt(rsq22,rcutoff2))
2094 r22 = _mm256_mul_ps(rsq22,rinv22);
2096 /* EWALD ELECTROSTATICS */
2098 /* Analytical PME correction */
2099 zeta2 = _mm256_mul_ps(beta2,rsq22);
2100 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
2101 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2102 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2103 felec = _mm256_mul_ps(qq22,felec);
2105 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
2109 fscal = _mm256_and_ps(fscal,cutoff_mask);
2111 /* Calculate temporary vectorial force */
2112 tx = _mm256_mul_ps(fscal,dx22);
2113 ty = _mm256_mul_ps(fscal,dy22);
2114 tz = _mm256_mul_ps(fscal,dz22);
2116 /* Update vectorial force */
2117 fix2 = _mm256_add_ps(fix2,tx);
2118 fiy2 = _mm256_add_ps(fiy2,ty);
2119 fiz2 = _mm256_add_ps(fiz2,tz);
2121 fjx2 = _mm256_add_ps(fjx2,tx);
2122 fjy2 = _mm256_add_ps(fjy2,ty);
2123 fjz2 = _mm256_add_ps(fjz2,tz);
2127 fjptrA = f+j_coord_offsetA;
2128 fjptrB = f+j_coord_offsetB;
2129 fjptrC = f+j_coord_offsetC;
2130 fjptrD = f+j_coord_offsetD;
2131 fjptrE = f+j_coord_offsetE;
2132 fjptrF = f+j_coord_offsetF;
2133 fjptrG = f+j_coord_offsetG;
2134 fjptrH = f+j_coord_offsetH;
2136 gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
2137 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2139 /* Inner loop uses 554 flops */
2142 if(jidx<j_index_end)
2145 /* Get j neighbor index, and coordinate index */
2146 jnrlistA = jjnr[jidx];
2147 jnrlistB = jjnr[jidx+1];
2148 jnrlistC = jjnr[jidx+2];
2149 jnrlistD = jjnr[jidx+3];
2150 jnrlistE = jjnr[jidx+4];
2151 jnrlistF = jjnr[jidx+5];
2152 jnrlistG = jjnr[jidx+6];
2153 jnrlistH = jjnr[jidx+7];
2154 /* Sign of each element will be negative for non-real atoms.
2155 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
2156 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
2158 dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
2159 gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
2161 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
2162 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
2163 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
2164 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
2165 jnrE = (jnrlistE>=0) ? jnrlistE : 0;
2166 jnrF = (jnrlistF>=0) ? jnrlistF : 0;
2167 jnrG = (jnrlistG>=0) ? jnrlistG : 0;
2168 jnrH = (jnrlistH>=0) ? jnrlistH : 0;
2169 j_coord_offsetA = DIM*jnrA;
2170 j_coord_offsetB = DIM*jnrB;
2171 j_coord_offsetC = DIM*jnrC;
2172 j_coord_offsetD = DIM*jnrD;
2173 j_coord_offsetE = DIM*jnrE;
2174 j_coord_offsetF = DIM*jnrF;
2175 j_coord_offsetG = DIM*jnrG;
2176 j_coord_offsetH = DIM*jnrH;
2178 /* load j atom coordinates */
2179 gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
2180 x+j_coord_offsetC,x+j_coord_offsetD,
2181 x+j_coord_offsetE,x+j_coord_offsetF,
2182 x+j_coord_offsetG,x+j_coord_offsetH,
2183 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
2185 /* Calculate displacement vector */
2186 dx00 = _mm256_sub_ps(ix0,jx0);
2187 dy00 = _mm256_sub_ps(iy0,jy0);
2188 dz00 = _mm256_sub_ps(iz0,jz0);
2189 dx01 = _mm256_sub_ps(ix0,jx1);
2190 dy01 = _mm256_sub_ps(iy0,jy1);
2191 dz01 = _mm256_sub_ps(iz0,jz1);
2192 dx02 = _mm256_sub_ps(ix0,jx2);
2193 dy02 = _mm256_sub_ps(iy0,jy2);
2194 dz02 = _mm256_sub_ps(iz0,jz2);
2195 dx10 = _mm256_sub_ps(ix1,jx0);
2196 dy10 = _mm256_sub_ps(iy1,jy0);
2197 dz10 = _mm256_sub_ps(iz1,jz0);
2198 dx11 = _mm256_sub_ps(ix1,jx1);
2199 dy11 = _mm256_sub_ps(iy1,jy1);
2200 dz11 = _mm256_sub_ps(iz1,jz1);
2201 dx12 = _mm256_sub_ps(ix1,jx2);
2202 dy12 = _mm256_sub_ps(iy1,jy2);
2203 dz12 = _mm256_sub_ps(iz1,jz2);
2204 dx20 = _mm256_sub_ps(ix2,jx0);
2205 dy20 = _mm256_sub_ps(iy2,jy0);
2206 dz20 = _mm256_sub_ps(iz2,jz0);
2207 dx21 = _mm256_sub_ps(ix2,jx1);
2208 dy21 = _mm256_sub_ps(iy2,jy1);
2209 dz21 = _mm256_sub_ps(iz2,jz1);
2210 dx22 = _mm256_sub_ps(ix2,jx2);
2211 dy22 = _mm256_sub_ps(iy2,jy2);
2212 dz22 = _mm256_sub_ps(iz2,jz2);
2214 /* Calculate squared distance and things based on it */
2215 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
2216 rsq01 = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
2217 rsq02 = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
2218 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
2219 rsq11 = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
2220 rsq12 = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
2221 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
2222 rsq21 = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
2223 rsq22 = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
2225 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
2226 rinv01 = gmx_mm256_invsqrt_ps(rsq01);
2227 rinv02 = gmx_mm256_invsqrt_ps(rsq02);
2228 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
2229 rinv11 = gmx_mm256_invsqrt_ps(rsq11);
2230 rinv12 = gmx_mm256_invsqrt_ps(rsq12);
2231 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
2232 rinv21 = gmx_mm256_invsqrt_ps(rsq21);
2233 rinv22 = gmx_mm256_invsqrt_ps(rsq22);
2235 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
2236 rinvsq01 = _mm256_mul_ps(rinv01,rinv01);
2237 rinvsq02 = _mm256_mul_ps(rinv02,rinv02);
2238 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
2239 rinvsq11 = _mm256_mul_ps(rinv11,rinv11);
2240 rinvsq12 = _mm256_mul_ps(rinv12,rinv12);
2241 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
2242 rinvsq21 = _mm256_mul_ps(rinv21,rinv21);
2243 rinvsq22 = _mm256_mul_ps(rinv22,rinv22);
2245 fjx0 = _mm256_setzero_ps();
2246 fjy0 = _mm256_setzero_ps();
2247 fjz0 = _mm256_setzero_ps();
2248 fjx1 = _mm256_setzero_ps();
2249 fjy1 = _mm256_setzero_ps();
2250 fjz1 = _mm256_setzero_ps();
2251 fjx2 = _mm256_setzero_ps();
2252 fjy2 = _mm256_setzero_ps();
2253 fjz2 = _mm256_setzero_ps();
2255 /**************************
2256 * CALCULATE INTERACTIONS *
2257 **************************/
2259 if (gmx_mm256_any_lt(rsq00,rcutoff2))
2262 r00 = _mm256_mul_ps(rsq00,rinv00);
2263 r00 = _mm256_andnot_ps(dummy_mask,r00);
2265 /* EWALD ELECTROSTATICS */
2267 /* Analytical PME correction */
2268 zeta2 = _mm256_mul_ps(beta2,rsq00);
2269 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
2270 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2271 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2272 felec = _mm256_mul_ps(qq00,felec);
2274 /* Analytical LJ-PME */
2275 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
2276 ewcljrsq = _mm256_mul_ps(ewclj2,rsq00);
2277 ewclj6 = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
2278 exponent = gmx_simd_exp_r(ewcljrsq);
2279 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
2280 poly = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
2281 /* f6A = 6 * C6grid * (1 - poly) */
2282 f6A = _mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly));
2283 /* f6B = C6grid * exponent * beta^6 */
2284 f6B = _mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6));
2285 /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
2286 fvdw = _mm256_mul_ps(_mm256_add_ps(_mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),_mm256_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
2288 cutoff_mask = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
2290 fscal = _mm256_add_ps(felec,fvdw);
2292 fscal = _mm256_and_ps(fscal,cutoff_mask);
2294 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2296 /* Calculate temporary vectorial force */
2297 tx = _mm256_mul_ps(fscal,dx00);
2298 ty = _mm256_mul_ps(fscal,dy00);
2299 tz = _mm256_mul_ps(fscal,dz00);
2301 /* Update vectorial force */
2302 fix0 = _mm256_add_ps(fix0,tx);
2303 fiy0 = _mm256_add_ps(fiy0,ty);
2304 fiz0 = _mm256_add_ps(fiz0,tz);
2306 fjx0 = _mm256_add_ps(fjx0,tx);
2307 fjy0 = _mm256_add_ps(fjy0,ty);
2308 fjz0 = _mm256_add_ps(fjz0,tz);
2312 /**************************
2313 * CALCULATE INTERACTIONS *
2314 **************************/
2316 if (gmx_mm256_any_lt(rsq01,rcutoff2))
2319 r01 = _mm256_mul_ps(rsq01,rinv01);
2320 r01 = _mm256_andnot_ps(dummy_mask,r01);
2322 /* EWALD ELECTROSTATICS */
2324 /* Analytical PME correction */
2325 zeta2 = _mm256_mul_ps(beta2,rsq01);
2326 rinv3 = _mm256_mul_ps(rinvsq01,rinv01);
2327 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2328 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2329 felec = _mm256_mul_ps(qq01,felec);
2331 cutoff_mask = _mm256_cmp_ps(rsq01,rcutoff2,_CMP_LT_OQ);
2335 fscal = _mm256_and_ps(fscal,cutoff_mask);
2337 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2339 /* Calculate temporary vectorial force */
2340 tx = _mm256_mul_ps(fscal,dx01);
2341 ty = _mm256_mul_ps(fscal,dy01);
2342 tz = _mm256_mul_ps(fscal,dz01);
2344 /* Update vectorial force */
2345 fix0 = _mm256_add_ps(fix0,tx);
2346 fiy0 = _mm256_add_ps(fiy0,ty);
2347 fiz0 = _mm256_add_ps(fiz0,tz);
2349 fjx1 = _mm256_add_ps(fjx1,tx);
2350 fjy1 = _mm256_add_ps(fjy1,ty);
2351 fjz1 = _mm256_add_ps(fjz1,tz);
2355 /**************************
2356 * CALCULATE INTERACTIONS *
2357 **************************/
2359 if (gmx_mm256_any_lt(rsq02,rcutoff2))
2362 r02 = _mm256_mul_ps(rsq02,rinv02);
2363 r02 = _mm256_andnot_ps(dummy_mask,r02);
2365 /* EWALD ELECTROSTATICS */
2367 /* Analytical PME correction */
2368 zeta2 = _mm256_mul_ps(beta2,rsq02);
2369 rinv3 = _mm256_mul_ps(rinvsq02,rinv02);
2370 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2371 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2372 felec = _mm256_mul_ps(qq02,felec);
2374 cutoff_mask = _mm256_cmp_ps(rsq02,rcutoff2,_CMP_LT_OQ);
2378 fscal = _mm256_and_ps(fscal,cutoff_mask);
2380 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2382 /* Calculate temporary vectorial force */
2383 tx = _mm256_mul_ps(fscal,dx02);
2384 ty = _mm256_mul_ps(fscal,dy02);
2385 tz = _mm256_mul_ps(fscal,dz02);
2387 /* Update vectorial force */
2388 fix0 = _mm256_add_ps(fix0,tx);
2389 fiy0 = _mm256_add_ps(fiy0,ty);
2390 fiz0 = _mm256_add_ps(fiz0,tz);
2392 fjx2 = _mm256_add_ps(fjx2,tx);
2393 fjy2 = _mm256_add_ps(fjy2,ty);
2394 fjz2 = _mm256_add_ps(fjz2,tz);
2398 /**************************
2399 * CALCULATE INTERACTIONS *
2400 **************************/
2402 if (gmx_mm256_any_lt(rsq10,rcutoff2))
2405 r10 = _mm256_mul_ps(rsq10,rinv10);
2406 r10 = _mm256_andnot_ps(dummy_mask,r10);
2408 /* EWALD ELECTROSTATICS */
2410 /* Analytical PME correction */
2411 zeta2 = _mm256_mul_ps(beta2,rsq10);
2412 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
2413 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2414 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2415 felec = _mm256_mul_ps(qq10,felec);
2417 cutoff_mask = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
2421 fscal = _mm256_and_ps(fscal,cutoff_mask);
2423 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2425 /* Calculate temporary vectorial force */
2426 tx = _mm256_mul_ps(fscal,dx10);
2427 ty = _mm256_mul_ps(fscal,dy10);
2428 tz = _mm256_mul_ps(fscal,dz10);
2430 /* Update vectorial force */
2431 fix1 = _mm256_add_ps(fix1,tx);
2432 fiy1 = _mm256_add_ps(fiy1,ty);
2433 fiz1 = _mm256_add_ps(fiz1,tz);
2435 fjx0 = _mm256_add_ps(fjx0,tx);
2436 fjy0 = _mm256_add_ps(fjy0,ty);
2437 fjz0 = _mm256_add_ps(fjz0,tz);
2441 /**************************
2442 * CALCULATE INTERACTIONS *
2443 **************************/
2445 if (gmx_mm256_any_lt(rsq11,rcutoff2))
2448 r11 = _mm256_mul_ps(rsq11,rinv11);
2449 r11 = _mm256_andnot_ps(dummy_mask,r11);
2451 /* EWALD ELECTROSTATICS */
2453 /* Analytical PME correction */
2454 zeta2 = _mm256_mul_ps(beta2,rsq11);
2455 rinv3 = _mm256_mul_ps(rinvsq11,rinv11);
2456 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2457 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2458 felec = _mm256_mul_ps(qq11,felec);
2460 cutoff_mask = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
2464 fscal = _mm256_and_ps(fscal,cutoff_mask);
2466 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2468 /* Calculate temporary vectorial force */
2469 tx = _mm256_mul_ps(fscal,dx11);
2470 ty = _mm256_mul_ps(fscal,dy11);
2471 tz = _mm256_mul_ps(fscal,dz11);
2473 /* Update vectorial force */
2474 fix1 = _mm256_add_ps(fix1,tx);
2475 fiy1 = _mm256_add_ps(fiy1,ty);
2476 fiz1 = _mm256_add_ps(fiz1,tz);
2478 fjx1 = _mm256_add_ps(fjx1,tx);
2479 fjy1 = _mm256_add_ps(fjy1,ty);
2480 fjz1 = _mm256_add_ps(fjz1,tz);
2484 /**************************
2485 * CALCULATE INTERACTIONS *
2486 **************************/
2488 if (gmx_mm256_any_lt(rsq12,rcutoff2))
2491 r12 = _mm256_mul_ps(rsq12,rinv12);
2492 r12 = _mm256_andnot_ps(dummy_mask,r12);
2494 /* EWALD ELECTROSTATICS */
2496 /* Analytical PME correction */
2497 zeta2 = _mm256_mul_ps(beta2,rsq12);
2498 rinv3 = _mm256_mul_ps(rinvsq12,rinv12);
2499 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2500 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2501 felec = _mm256_mul_ps(qq12,felec);
2503 cutoff_mask = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
2507 fscal = _mm256_and_ps(fscal,cutoff_mask);
2509 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2511 /* Calculate temporary vectorial force */
2512 tx = _mm256_mul_ps(fscal,dx12);
2513 ty = _mm256_mul_ps(fscal,dy12);
2514 tz = _mm256_mul_ps(fscal,dz12);
2516 /* Update vectorial force */
2517 fix1 = _mm256_add_ps(fix1,tx);
2518 fiy1 = _mm256_add_ps(fiy1,ty);
2519 fiz1 = _mm256_add_ps(fiz1,tz);
2521 fjx2 = _mm256_add_ps(fjx2,tx);
2522 fjy2 = _mm256_add_ps(fjy2,ty);
2523 fjz2 = _mm256_add_ps(fjz2,tz);
2527 /**************************
2528 * CALCULATE INTERACTIONS *
2529 **************************/
2531 if (gmx_mm256_any_lt(rsq20,rcutoff2))
2534 r20 = _mm256_mul_ps(rsq20,rinv20);
2535 r20 = _mm256_andnot_ps(dummy_mask,r20);
2537 /* EWALD ELECTROSTATICS */
2539 /* Analytical PME correction */
2540 zeta2 = _mm256_mul_ps(beta2,rsq20);
2541 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
2542 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2543 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2544 felec = _mm256_mul_ps(qq20,felec);
2546 cutoff_mask = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
2550 fscal = _mm256_and_ps(fscal,cutoff_mask);
2552 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2554 /* Calculate temporary vectorial force */
2555 tx = _mm256_mul_ps(fscal,dx20);
2556 ty = _mm256_mul_ps(fscal,dy20);
2557 tz = _mm256_mul_ps(fscal,dz20);
2559 /* Update vectorial force */
2560 fix2 = _mm256_add_ps(fix2,tx);
2561 fiy2 = _mm256_add_ps(fiy2,ty);
2562 fiz2 = _mm256_add_ps(fiz2,tz);
2564 fjx0 = _mm256_add_ps(fjx0,tx);
2565 fjy0 = _mm256_add_ps(fjy0,ty);
2566 fjz0 = _mm256_add_ps(fjz0,tz);
2570 /**************************
2571 * CALCULATE INTERACTIONS *
2572 **************************/
2574 if (gmx_mm256_any_lt(rsq21,rcutoff2))
2577 r21 = _mm256_mul_ps(rsq21,rinv21);
2578 r21 = _mm256_andnot_ps(dummy_mask,r21);
2580 /* EWALD ELECTROSTATICS */
2582 /* Analytical PME correction */
2583 zeta2 = _mm256_mul_ps(beta2,rsq21);
2584 rinv3 = _mm256_mul_ps(rinvsq21,rinv21);
2585 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2586 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2587 felec = _mm256_mul_ps(qq21,felec);
2589 cutoff_mask = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
2593 fscal = _mm256_and_ps(fscal,cutoff_mask);
2595 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2597 /* Calculate temporary vectorial force */
2598 tx = _mm256_mul_ps(fscal,dx21);
2599 ty = _mm256_mul_ps(fscal,dy21);
2600 tz = _mm256_mul_ps(fscal,dz21);
2602 /* Update vectorial force */
2603 fix2 = _mm256_add_ps(fix2,tx);
2604 fiy2 = _mm256_add_ps(fiy2,ty);
2605 fiz2 = _mm256_add_ps(fiz2,tz);
2607 fjx1 = _mm256_add_ps(fjx1,tx);
2608 fjy1 = _mm256_add_ps(fjy1,ty);
2609 fjz1 = _mm256_add_ps(fjz1,tz);
2613 /**************************
2614 * CALCULATE INTERACTIONS *
2615 **************************/
2617 if (gmx_mm256_any_lt(rsq22,rcutoff2))
2620 r22 = _mm256_mul_ps(rsq22,rinv22);
2621 r22 = _mm256_andnot_ps(dummy_mask,r22);
2623 /* EWALD ELECTROSTATICS */
2625 /* Analytical PME correction */
2626 zeta2 = _mm256_mul_ps(beta2,rsq22);
2627 rinv3 = _mm256_mul_ps(rinvsq22,rinv22);
2628 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
2629 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2630 felec = _mm256_mul_ps(qq22,felec);
2632 cutoff_mask = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
2636 fscal = _mm256_and_ps(fscal,cutoff_mask);
2638 fscal = _mm256_andnot_ps(dummy_mask,fscal);
2640 /* Calculate temporary vectorial force */
2641 tx = _mm256_mul_ps(fscal,dx22);
2642 ty = _mm256_mul_ps(fscal,dy22);
2643 tz = _mm256_mul_ps(fscal,dz22);
2645 /* Update vectorial force */
2646 fix2 = _mm256_add_ps(fix2,tx);
2647 fiy2 = _mm256_add_ps(fiy2,ty);
2648 fiz2 = _mm256_add_ps(fiz2,tz);
2650 fjx2 = _mm256_add_ps(fjx2,tx);
2651 fjy2 = _mm256_add_ps(fjy2,ty);
2652 fjz2 = _mm256_add_ps(fjz2,tz);
2656 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2657 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
2658 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
2659 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
2660 fjptrE = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
2661 fjptrF = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
2662 fjptrG = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
2663 fjptrH = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
2665 gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
2666 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2668 /* Inner loop uses 563 flops */
2671 /* End of innermost loop */
2673 gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
2674 f+i_coord_offset,fshift+i_shift_offset);
2676 /* Increment number of inner iterations */
2677 inneriter += j_index_end - j_index_start;
2679 /* Outer loop uses 18 flops */
2682 /* Increment number of outer iterations */
2685 /* Update outer/inner flops */
2687 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*563);