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.
44 #include "../nb_kernel.h"
45 #include "types/simple.h"
46 #include "gromacs/math/vec.h"
49 #include "gromacs/simd/math_x86_avx_256_single.h"
50 #include "kernelutil_x86_avx_256_single.h"
53 * Gromacs nonbonded kernel: nb_kernel_ElecEw_VdwLJ_GeomW3P1_VF_avx_256_single
54 * Electrostatics interaction: Ewald
55 * VdW interaction: LennardJones
56 * Geometry: Water3-Particle
57 * Calculate force/pot: PotentialAndForce
60 nb_kernel_ElecEw_VdwLJ_GeomW3P1_VF_avx_256_single
61 (t_nblist * gmx_restrict nlist,
62 rvec * gmx_restrict xx,
63 rvec * gmx_restrict ff,
64 t_forcerec * gmx_restrict fr,
65 t_mdatoms * gmx_restrict mdatoms,
66 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
67 t_nrnb * gmx_restrict nrnb)
69 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
70 * just 0 for non-waters.
71 * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
72 * jnr indices corresponding to data put in the four positions in the SIMD register.
74 int i_shift_offset,i_coord_offset,outeriter,inneriter;
75 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
76 int jnrA,jnrB,jnrC,jnrD;
77 int jnrE,jnrF,jnrG,jnrH;
78 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
79 int jnrlistE,jnrlistF,jnrlistG,jnrlistH;
80 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
81 int j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
82 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
84 real *shiftvec,*fshift,*x,*f;
85 real *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
87 __m256 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
88 real * vdwioffsetptr0;
89 __m256 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
90 real * vdwioffsetptr1;
91 __m256 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
92 real * vdwioffsetptr2;
93 __m256 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
94 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
95 __m256 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
96 __m256 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
97 __m256 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
98 __m256 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
99 __m256 velec,felec,velecsum,facel,crf,krf,krf2;
102 __m256 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
105 __m256 one_sixth = _mm256_set1_ps(1.0/6.0);
106 __m256 one_twelfth = _mm256_set1_ps(1.0/12.0);
108 __m128i ewitab_lo,ewitab_hi;
109 __m256 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
110 __m256 beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
112 __m256 dummy_mask,cutoff_mask;
113 __m256 signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
114 __m256 one = _mm256_set1_ps(1.0);
115 __m256 two = _mm256_set1_ps(2.0);
121 jindex = nlist->jindex;
123 shiftidx = nlist->shift;
125 shiftvec = fr->shift_vec[0];
126 fshift = fr->fshift[0];
127 facel = _mm256_set1_ps(fr->epsfac);
128 charge = mdatoms->chargeA;
129 nvdwtype = fr->ntype;
131 vdwtype = mdatoms->typeA;
133 sh_ewald = _mm256_set1_ps(fr->ic->sh_ewald);
134 beta = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
135 beta2 = _mm256_mul_ps(beta,beta);
136 beta3 = _mm256_mul_ps(beta,beta2);
138 ewtab = fr->ic->tabq_coul_FDV0;
139 ewtabscale = _mm256_set1_ps(fr->ic->tabq_scale);
140 ewtabhalfspace = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
142 /* Setup water-specific parameters */
143 inr = nlist->iinr[0];
144 iq0 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
145 iq1 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
146 iq2 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
147 vdwioffsetptr0 = vdwparam+2*nvdwtype*vdwtype[inr+0];
149 /* Avoid stupid compiler warnings */
150 jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
163 for(iidx=0;iidx<4*DIM;iidx++)
168 /* Start outer loop over neighborlists */
169 for(iidx=0; iidx<nri; iidx++)
171 /* Load shift vector for this list */
172 i_shift_offset = DIM*shiftidx[iidx];
174 /* Load limits for loop over neighbors */
175 j_index_start = jindex[iidx];
176 j_index_end = jindex[iidx+1];
178 /* Get outer coordinate index */
180 i_coord_offset = DIM*inr;
182 /* Load i particle coords and add shift vector */
183 gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
184 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
186 fix0 = _mm256_setzero_ps();
187 fiy0 = _mm256_setzero_ps();
188 fiz0 = _mm256_setzero_ps();
189 fix1 = _mm256_setzero_ps();
190 fiy1 = _mm256_setzero_ps();
191 fiz1 = _mm256_setzero_ps();
192 fix2 = _mm256_setzero_ps();
193 fiy2 = _mm256_setzero_ps();
194 fiz2 = _mm256_setzero_ps();
196 /* Reset potential sums */
197 velecsum = _mm256_setzero_ps();
198 vvdwsum = _mm256_setzero_ps();
200 /* Start inner kernel loop */
201 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
204 /* Get j neighbor index, and coordinate index */
213 j_coord_offsetA = DIM*jnrA;
214 j_coord_offsetB = DIM*jnrB;
215 j_coord_offsetC = DIM*jnrC;
216 j_coord_offsetD = DIM*jnrD;
217 j_coord_offsetE = DIM*jnrE;
218 j_coord_offsetF = DIM*jnrF;
219 j_coord_offsetG = DIM*jnrG;
220 j_coord_offsetH = DIM*jnrH;
222 /* load j atom coordinates */
223 gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
224 x+j_coord_offsetC,x+j_coord_offsetD,
225 x+j_coord_offsetE,x+j_coord_offsetF,
226 x+j_coord_offsetG,x+j_coord_offsetH,
229 /* Calculate displacement vector */
230 dx00 = _mm256_sub_ps(ix0,jx0);
231 dy00 = _mm256_sub_ps(iy0,jy0);
232 dz00 = _mm256_sub_ps(iz0,jz0);
233 dx10 = _mm256_sub_ps(ix1,jx0);
234 dy10 = _mm256_sub_ps(iy1,jy0);
235 dz10 = _mm256_sub_ps(iz1,jz0);
236 dx20 = _mm256_sub_ps(ix2,jx0);
237 dy20 = _mm256_sub_ps(iy2,jy0);
238 dz20 = _mm256_sub_ps(iz2,jz0);
240 /* Calculate squared distance and things based on it */
241 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
242 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
243 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
245 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
246 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
247 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
249 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
250 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
251 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
253 /* Load parameters for j particles */
254 jq0 = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
255 charge+jnrC+0,charge+jnrD+0,
256 charge+jnrE+0,charge+jnrF+0,
257 charge+jnrG+0,charge+jnrH+0);
258 vdwjidx0A = 2*vdwtype[jnrA+0];
259 vdwjidx0B = 2*vdwtype[jnrB+0];
260 vdwjidx0C = 2*vdwtype[jnrC+0];
261 vdwjidx0D = 2*vdwtype[jnrD+0];
262 vdwjidx0E = 2*vdwtype[jnrE+0];
263 vdwjidx0F = 2*vdwtype[jnrF+0];
264 vdwjidx0G = 2*vdwtype[jnrG+0];
265 vdwjidx0H = 2*vdwtype[jnrH+0];
267 fjx0 = _mm256_setzero_ps();
268 fjy0 = _mm256_setzero_ps();
269 fjz0 = _mm256_setzero_ps();
271 /**************************
272 * CALCULATE INTERACTIONS *
273 **************************/
275 r00 = _mm256_mul_ps(rsq00,rinv00);
277 /* Compute parameters for interactions between i and j atoms */
278 qq00 = _mm256_mul_ps(iq0,jq0);
279 gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
280 vdwioffsetptr0+vdwjidx0B,
281 vdwioffsetptr0+vdwjidx0C,
282 vdwioffsetptr0+vdwjidx0D,
283 vdwioffsetptr0+vdwjidx0E,
284 vdwioffsetptr0+vdwjidx0F,
285 vdwioffsetptr0+vdwjidx0G,
286 vdwioffsetptr0+vdwjidx0H,
289 /* EWALD ELECTROSTATICS */
291 /* Analytical PME correction */
292 zeta2 = _mm256_mul_ps(beta2,rsq00);
293 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
294 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
295 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
296 felec = _mm256_mul_ps(qq00,felec);
297 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
298 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
299 velec = _mm256_sub_ps(rinv00,pmecorrV);
300 velec = _mm256_mul_ps(qq00,velec);
302 /* LENNARD-JONES DISPERSION/REPULSION */
304 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
305 vvdw6 = _mm256_mul_ps(c6_00,rinvsix);
306 vvdw12 = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
307 vvdw = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
308 fvdw = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
310 /* Update potential sum for this i atom from the interaction with this j atom. */
311 velecsum = _mm256_add_ps(velecsum,velec);
312 vvdwsum = _mm256_add_ps(vvdwsum,vvdw);
314 fscal = _mm256_add_ps(felec,fvdw);
316 /* Calculate temporary vectorial force */
317 tx = _mm256_mul_ps(fscal,dx00);
318 ty = _mm256_mul_ps(fscal,dy00);
319 tz = _mm256_mul_ps(fscal,dz00);
321 /* Update vectorial force */
322 fix0 = _mm256_add_ps(fix0,tx);
323 fiy0 = _mm256_add_ps(fiy0,ty);
324 fiz0 = _mm256_add_ps(fiz0,tz);
326 fjx0 = _mm256_add_ps(fjx0,tx);
327 fjy0 = _mm256_add_ps(fjy0,ty);
328 fjz0 = _mm256_add_ps(fjz0,tz);
330 /**************************
331 * CALCULATE INTERACTIONS *
332 **************************/
334 r10 = _mm256_mul_ps(rsq10,rinv10);
336 /* Compute parameters for interactions between i and j atoms */
337 qq10 = _mm256_mul_ps(iq1,jq0);
339 /* EWALD ELECTROSTATICS */
341 /* Analytical PME correction */
342 zeta2 = _mm256_mul_ps(beta2,rsq10);
343 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
344 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
345 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
346 felec = _mm256_mul_ps(qq10,felec);
347 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
348 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
349 velec = _mm256_sub_ps(rinv10,pmecorrV);
350 velec = _mm256_mul_ps(qq10,velec);
352 /* Update potential sum for this i atom from the interaction with this j atom. */
353 velecsum = _mm256_add_ps(velecsum,velec);
357 /* Calculate temporary vectorial force */
358 tx = _mm256_mul_ps(fscal,dx10);
359 ty = _mm256_mul_ps(fscal,dy10);
360 tz = _mm256_mul_ps(fscal,dz10);
362 /* Update vectorial force */
363 fix1 = _mm256_add_ps(fix1,tx);
364 fiy1 = _mm256_add_ps(fiy1,ty);
365 fiz1 = _mm256_add_ps(fiz1,tz);
367 fjx0 = _mm256_add_ps(fjx0,tx);
368 fjy0 = _mm256_add_ps(fjy0,ty);
369 fjz0 = _mm256_add_ps(fjz0,tz);
371 /**************************
372 * CALCULATE INTERACTIONS *
373 **************************/
375 r20 = _mm256_mul_ps(rsq20,rinv20);
377 /* Compute parameters for interactions between i and j atoms */
378 qq20 = _mm256_mul_ps(iq2,jq0);
380 /* EWALD ELECTROSTATICS */
382 /* Analytical PME correction */
383 zeta2 = _mm256_mul_ps(beta2,rsq20);
384 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
385 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
386 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
387 felec = _mm256_mul_ps(qq20,felec);
388 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
389 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
390 velec = _mm256_sub_ps(rinv20,pmecorrV);
391 velec = _mm256_mul_ps(qq20,velec);
393 /* Update potential sum for this i atom from the interaction with this j atom. */
394 velecsum = _mm256_add_ps(velecsum,velec);
398 /* Calculate temporary vectorial force */
399 tx = _mm256_mul_ps(fscal,dx20);
400 ty = _mm256_mul_ps(fscal,dy20);
401 tz = _mm256_mul_ps(fscal,dz20);
403 /* Update vectorial force */
404 fix2 = _mm256_add_ps(fix2,tx);
405 fiy2 = _mm256_add_ps(fiy2,ty);
406 fiz2 = _mm256_add_ps(fiz2,tz);
408 fjx0 = _mm256_add_ps(fjx0,tx);
409 fjy0 = _mm256_add_ps(fjy0,ty);
410 fjz0 = _mm256_add_ps(fjz0,tz);
412 fjptrA = f+j_coord_offsetA;
413 fjptrB = f+j_coord_offsetB;
414 fjptrC = f+j_coord_offsetC;
415 fjptrD = f+j_coord_offsetD;
416 fjptrE = f+j_coord_offsetE;
417 fjptrF = f+j_coord_offsetF;
418 fjptrG = f+j_coord_offsetG;
419 fjptrH = f+j_coord_offsetH;
421 gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
423 /* Inner loop uses 267 flops */
429 /* Get j neighbor index, and coordinate index */
430 jnrlistA = jjnr[jidx];
431 jnrlistB = jjnr[jidx+1];
432 jnrlistC = jjnr[jidx+2];
433 jnrlistD = jjnr[jidx+3];
434 jnrlistE = jjnr[jidx+4];
435 jnrlistF = jjnr[jidx+5];
436 jnrlistG = jjnr[jidx+6];
437 jnrlistH = jjnr[jidx+7];
438 /* Sign of each element will be negative for non-real atoms.
439 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
440 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
442 dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
443 gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
445 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
446 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
447 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
448 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
449 jnrE = (jnrlistE>=0) ? jnrlistE : 0;
450 jnrF = (jnrlistF>=0) ? jnrlistF : 0;
451 jnrG = (jnrlistG>=0) ? jnrlistG : 0;
452 jnrH = (jnrlistH>=0) ? jnrlistH : 0;
453 j_coord_offsetA = DIM*jnrA;
454 j_coord_offsetB = DIM*jnrB;
455 j_coord_offsetC = DIM*jnrC;
456 j_coord_offsetD = DIM*jnrD;
457 j_coord_offsetE = DIM*jnrE;
458 j_coord_offsetF = DIM*jnrF;
459 j_coord_offsetG = DIM*jnrG;
460 j_coord_offsetH = DIM*jnrH;
462 /* load j atom coordinates */
463 gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
464 x+j_coord_offsetC,x+j_coord_offsetD,
465 x+j_coord_offsetE,x+j_coord_offsetF,
466 x+j_coord_offsetG,x+j_coord_offsetH,
469 /* Calculate displacement vector */
470 dx00 = _mm256_sub_ps(ix0,jx0);
471 dy00 = _mm256_sub_ps(iy0,jy0);
472 dz00 = _mm256_sub_ps(iz0,jz0);
473 dx10 = _mm256_sub_ps(ix1,jx0);
474 dy10 = _mm256_sub_ps(iy1,jy0);
475 dz10 = _mm256_sub_ps(iz1,jz0);
476 dx20 = _mm256_sub_ps(ix2,jx0);
477 dy20 = _mm256_sub_ps(iy2,jy0);
478 dz20 = _mm256_sub_ps(iz2,jz0);
480 /* Calculate squared distance and things based on it */
481 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
482 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
483 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
485 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
486 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
487 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
489 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
490 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
491 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
493 /* Load parameters for j particles */
494 jq0 = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
495 charge+jnrC+0,charge+jnrD+0,
496 charge+jnrE+0,charge+jnrF+0,
497 charge+jnrG+0,charge+jnrH+0);
498 vdwjidx0A = 2*vdwtype[jnrA+0];
499 vdwjidx0B = 2*vdwtype[jnrB+0];
500 vdwjidx0C = 2*vdwtype[jnrC+0];
501 vdwjidx0D = 2*vdwtype[jnrD+0];
502 vdwjidx0E = 2*vdwtype[jnrE+0];
503 vdwjidx0F = 2*vdwtype[jnrF+0];
504 vdwjidx0G = 2*vdwtype[jnrG+0];
505 vdwjidx0H = 2*vdwtype[jnrH+0];
507 fjx0 = _mm256_setzero_ps();
508 fjy0 = _mm256_setzero_ps();
509 fjz0 = _mm256_setzero_ps();
511 /**************************
512 * CALCULATE INTERACTIONS *
513 **************************/
515 r00 = _mm256_mul_ps(rsq00,rinv00);
516 r00 = _mm256_andnot_ps(dummy_mask,r00);
518 /* Compute parameters for interactions between i and j atoms */
519 qq00 = _mm256_mul_ps(iq0,jq0);
520 gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
521 vdwioffsetptr0+vdwjidx0B,
522 vdwioffsetptr0+vdwjidx0C,
523 vdwioffsetptr0+vdwjidx0D,
524 vdwioffsetptr0+vdwjidx0E,
525 vdwioffsetptr0+vdwjidx0F,
526 vdwioffsetptr0+vdwjidx0G,
527 vdwioffsetptr0+vdwjidx0H,
530 /* EWALD ELECTROSTATICS */
532 /* Analytical PME correction */
533 zeta2 = _mm256_mul_ps(beta2,rsq00);
534 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
535 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
536 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
537 felec = _mm256_mul_ps(qq00,felec);
538 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
539 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
540 velec = _mm256_sub_ps(rinv00,pmecorrV);
541 velec = _mm256_mul_ps(qq00,velec);
543 /* LENNARD-JONES DISPERSION/REPULSION */
545 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
546 vvdw6 = _mm256_mul_ps(c6_00,rinvsix);
547 vvdw12 = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
548 vvdw = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
549 fvdw = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
551 /* Update potential sum for this i atom from the interaction with this j atom. */
552 velec = _mm256_andnot_ps(dummy_mask,velec);
553 velecsum = _mm256_add_ps(velecsum,velec);
554 vvdw = _mm256_andnot_ps(dummy_mask,vvdw);
555 vvdwsum = _mm256_add_ps(vvdwsum,vvdw);
557 fscal = _mm256_add_ps(felec,fvdw);
559 fscal = _mm256_andnot_ps(dummy_mask,fscal);
561 /* Calculate temporary vectorial force */
562 tx = _mm256_mul_ps(fscal,dx00);
563 ty = _mm256_mul_ps(fscal,dy00);
564 tz = _mm256_mul_ps(fscal,dz00);
566 /* Update vectorial force */
567 fix0 = _mm256_add_ps(fix0,tx);
568 fiy0 = _mm256_add_ps(fiy0,ty);
569 fiz0 = _mm256_add_ps(fiz0,tz);
571 fjx0 = _mm256_add_ps(fjx0,tx);
572 fjy0 = _mm256_add_ps(fjy0,ty);
573 fjz0 = _mm256_add_ps(fjz0,tz);
575 /**************************
576 * CALCULATE INTERACTIONS *
577 **************************/
579 r10 = _mm256_mul_ps(rsq10,rinv10);
580 r10 = _mm256_andnot_ps(dummy_mask,r10);
582 /* Compute parameters for interactions between i and j atoms */
583 qq10 = _mm256_mul_ps(iq1,jq0);
585 /* EWALD ELECTROSTATICS */
587 /* Analytical PME correction */
588 zeta2 = _mm256_mul_ps(beta2,rsq10);
589 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
590 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
591 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
592 felec = _mm256_mul_ps(qq10,felec);
593 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
594 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
595 velec = _mm256_sub_ps(rinv10,pmecorrV);
596 velec = _mm256_mul_ps(qq10,velec);
598 /* Update potential sum for this i atom from the interaction with this j atom. */
599 velec = _mm256_andnot_ps(dummy_mask,velec);
600 velecsum = _mm256_add_ps(velecsum,velec);
604 fscal = _mm256_andnot_ps(dummy_mask,fscal);
606 /* Calculate temporary vectorial force */
607 tx = _mm256_mul_ps(fscal,dx10);
608 ty = _mm256_mul_ps(fscal,dy10);
609 tz = _mm256_mul_ps(fscal,dz10);
611 /* Update vectorial force */
612 fix1 = _mm256_add_ps(fix1,tx);
613 fiy1 = _mm256_add_ps(fiy1,ty);
614 fiz1 = _mm256_add_ps(fiz1,tz);
616 fjx0 = _mm256_add_ps(fjx0,tx);
617 fjy0 = _mm256_add_ps(fjy0,ty);
618 fjz0 = _mm256_add_ps(fjz0,tz);
620 /**************************
621 * CALCULATE INTERACTIONS *
622 **************************/
624 r20 = _mm256_mul_ps(rsq20,rinv20);
625 r20 = _mm256_andnot_ps(dummy_mask,r20);
627 /* Compute parameters for interactions between i and j atoms */
628 qq20 = _mm256_mul_ps(iq2,jq0);
630 /* EWALD ELECTROSTATICS */
632 /* Analytical PME correction */
633 zeta2 = _mm256_mul_ps(beta2,rsq20);
634 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
635 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
636 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
637 felec = _mm256_mul_ps(qq20,felec);
638 pmecorrV = gmx_mm256_pmecorrV_ps(zeta2);
639 pmecorrV = _mm256_mul_ps(pmecorrV,beta);
640 velec = _mm256_sub_ps(rinv20,pmecorrV);
641 velec = _mm256_mul_ps(qq20,velec);
643 /* Update potential sum for this i atom from the interaction with this j atom. */
644 velec = _mm256_andnot_ps(dummy_mask,velec);
645 velecsum = _mm256_add_ps(velecsum,velec);
649 fscal = _mm256_andnot_ps(dummy_mask,fscal);
651 /* Calculate temporary vectorial force */
652 tx = _mm256_mul_ps(fscal,dx20);
653 ty = _mm256_mul_ps(fscal,dy20);
654 tz = _mm256_mul_ps(fscal,dz20);
656 /* Update vectorial force */
657 fix2 = _mm256_add_ps(fix2,tx);
658 fiy2 = _mm256_add_ps(fiy2,ty);
659 fiz2 = _mm256_add_ps(fiz2,tz);
661 fjx0 = _mm256_add_ps(fjx0,tx);
662 fjy0 = _mm256_add_ps(fjy0,ty);
663 fjz0 = _mm256_add_ps(fjz0,tz);
665 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
666 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
667 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
668 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
669 fjptrE = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
670 fjptrF = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
671 fjptrG = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
672 fjptrH = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
674 gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
676 /* Inner loop uses 270 flops */
679 /* End of innermost loop */
681 gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
682 f+i_coord_offset,fshift+i_shift_offset);
685 /* Update potential energies */
686 gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
687 gmx_mm256_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
689 /* Increment number of inner iterations */
690 inneriter += j_index_end - j_index_start;
692 /* Outer loop uses 20 flops */
695 /* Increment number of outer iterations */
698 /* Update outer/inner flops */
700 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*20 + inneriter*270);
703 * Gromacs nonbonded kernel: nb_kernel_ElecEw_VdwLJ_GeomW3P1_F_avx_256_single
704 * Electrostatics interaction: Ewald
705 * VdW interaction: LennardJones
706 * Geometry: Water3-Particle
707 * Calculate force/pot: Force
710 nb_kernel_ElecEw_VdwLJ_GeomW3P1_F_avx_256_single
711 (t_nblist * gmx_restrict nlist,
712 rvec * gmx_restrict xx,
713 rvec * gmx_restrict ff,
714 t_forcerec * gmx_restrict fr,
715 t_mdatoms * gmx_restrict mdatoms,
716 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
717 t_nrnb * gmx_restrict nrnb)
719 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
720 * just 0 for non-waters.
721 * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
722 * jnr indices corresponding to data put in the four positions in the SIMD register.
724 int i_shift_offset,i_coord_offset,outeriter,inneriter;
725 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
726 int jnrA,jnrB,jnrC,jnrD;
727 int jnrE,jnrF,jnrG,jnrH;
728 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
729 int jnrlistE,jnrlistF,jnrlistG,jnrlistH;
730 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
731 int j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
732 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
734 real *shiftvec,*fshift,*x,*f;
735 real *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
737 __m256 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
738 real * vdwioffsetptr0;
739 __m256 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
740 real * vdwioffsetptr1;
741 __m256 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
742 real * vdwioffsetptr2;
743 __m256 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
744 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
745 __m256 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
746 __m256 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
747 __m256 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
748 __m256 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
749 __m256 velec,felec,velecsum,facel,crf,krf,krf2;
752 __m256 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
755 __m256 one_sixth = _mm256_set1_ps(1.0/6.0);
756 __m256 one_twelfth = _mm256_set1_ps(1.0/12.0);
758 __m128i ewitab_lo,ewitab_hi;
759 __m256 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
760 __m256 beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
762 __m256 dummy_mask,cutoff_mask;
763 __m256 signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
764 __m256 one = _mm256_set1_ps(1.0);
765 __m256 two = _mm256_set1_ps(2.0);
771 jindex = nlist->jindex;
773 shiftidx = nlist->shift;
775 shiftvec = fr->shift_vec[0];
776 fshift = fr->fshift[0];
777 facel = _mm256_set1_ps(fr->epsfac);
778 charge = mdatoms->chargeA;
779 nvdwtype = fr->ntype;
781 vdwtype = mdatoms->typeA;
783 sh_ewald = _mm256_set1_ps(fr->ic->sh_ewald);
784 beta = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
785 beta2 = _mm256_mul_ps(beta,beta);
786 beta3 = _mm256_mul_ps(beta,beta2);
788 ewtab = fr->ic->tabq_coul_F;
789 ewtabscale = _mm256_set1_ps(fr->ic->tabq_scale);
790 ewtabhalfspace = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
792 /* Setup water-specific parameters */
793 inr = nlist->iinr[0];
794 iq0 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
795 iq1 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
796 iq2 = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
797 vdwioffsetptr0 = vdwparam+2*nvdwtype*vdwtype[inr+0];
799 /* Avoid stupid compiler warnings */
800 jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
813 for(iidx=0;iidx<4*DIM;iidx++)
818 /* Start outer loop over neighborlists */
819 for(iidx=0; iidx<nri; iidx++)
821 /* Load shift vector for this list */
822 i_shift_offset = DIM*shiftidx[iidx];
824 /* Load limits for loop over neighbors */
825 j_index_start = jindex[iidx];
826 j_index_end = jindex[iidx+1];
828 /* Get outer coordinate index */
830 i_coord_offset = DIM*inr;
832 /* Load i particle coords and add shift vector */
833 gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
834 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
836 fix0 = _mm256_setzero_ps();
837 fiy0 = _mm256_setzero_ps();
838 fiz0 = _mm256_setzero_ps();
839 fix1 = _mm256_setzero_ps();
840 fiy1 = _mm256_setzero_ps();
841 fiz1 = _mm256_setzero_ps();
842 fix2 = _mm256_setzero_ps();
843 fiy2 = _mm256_setzero_ps();
844 fiz2 = _mm256_setzero_ps();
846 /* Start inner kernel loop */
847 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
850 /* Get j neighbor index, and coordinate index */
859 j_coord_offsetA = DIM*jnrA;
860 j_coord_offsetB = DIM*jnrB;
861 j_coord_offsetC = DIM*jnrC;
862 j_coord_offsetD = DIM*jnrD;
863 j_coord_offsetE = DIM*jnrE;
864 j_coord_offsetF = DIM*jnrF;
865 j_coord_offsetG = DIM*jnrG;
866 j_coord_offsetH = DIM*jnrH;
868 /* load j atom coordinates */
869 gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
870 x+j_coord_offsetC,x+j_coord_offsetD,
871 x+j_coord_offsetE,x+j_coord_offsetF,
872 x+j_coord_offsetG,x+j_coord_offsetH,
875 /* Calculate displacement vector */
876 dx00 = _mm256_sub_ps(ix0,jx0);
877 dy00 = _mm256_sub_ps(iy0,jy0);
878 dz00 = _mm256_sub_ps(iz0,jz0);
879 dx10 = _mm256_sub_ps(ix1,jx0);
880 dy10 = _mm256_sub_ps(iy1,jy0);
881 dz10 = _mm256_sub_ps(iz1,jz0);
882 dx20 = _mm256_sub_ps(ix2,jx0);
883 dy20 = _mm256_sub_ps(iy2,jy0);
884 dz20 = _mm256_sub_ps(iz2,jz0);
886 /* Calculate squared distance and things based on it */
887 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
888 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
889 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
891 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
892 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
893 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
895 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
896 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
897 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
899 /* Load parameters for j particles */
900 jq0 = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
901 charge+jnrC+0,charge+jnrD+0,
902 charge+jnrE+0,charge+jnrF+0,
903 charge+jnrG+0,charge+jnrH+0);
904 vdwjidx0A = 2*vdwtype[jnrA+0];
905 vdwjidx0B = 2*vdwtype[jnrB+0];
906 vdwjidx0C = 2*vdwtype[jnrC+0];
907 vdwjidx0D = 2*vdwtype[jnrD+0];
908 vdwjidx0E = 2*vdwtype[jnrE+0];
909 vdwjidx0F = 2*vdwtype[jnrF+0];
910 vdwjidx0G = 2*vdwtype[jnrG+0];
911 vdwjidx0H = 2*vdwtype[jnrH+0];
913 fjx0 = _mm256_setzero_ps();
914 fjy0 = _mm256_setzero_ps();
915 fjz0 = _mm256_setzero_ps();
917 /**************************
918 * CALCULATE INTERACTIONS *
919 **************************/
921 r00 = _mm256_mul_ps(rsq00,rinv00);
923 /* Compute parameters for interactions between i and j atoms */
924 qq00 = _mm256_mul_ps(iq0,jq0);
925 gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
926 vdwioffsetptr0+vdwjidx0B,
927 vdwioffsetptr0+vdwjidx0C,
928 vdwioffsetptr0+vdwjidx0D,
929 vdwioffsetptr0+vdwjidx0E,
930 vdwioffsetptr0+vdwjidx0F,
931 vdwioffsetptr0+vdwjidx0G,
932 vdwioffsetptr0+vdwjidx0H,
935 /* EWALD ELECTROSTATICS */
937 /* Analytical PME correction */
938 zeta2 = _mm256_mul_ps(beta2,rsq00);
939 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
940 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
941 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
942 felec = _mm256_mul_ps(qq00,felec);
944 /* LENNARD-JONES DISPERSION/REPULSION */
946 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
947 fvdw = _mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),c6_00),_mm256_mul_ps(rinvsix,rinvsq00));
949 fscal = _mm256_add_ps(felec,fvdw);
951 /* Calculate temporary vectorial force */
952 tx = _mm256_mul_ps(fscal,dx00);
953 ty = _mm256_mul_ps(fscal,dy00);
954 tz = _mm256_mul_ps(fscal,dz00);
956 /* Update vectorial force */
957 fix0 = _mm256_add_ps(fix0,tx);
958 fiy0 = _mm256_add_ps(fiy0,ty);
959 fiz0 = _mm256_add_ps(fiz0,tz);
961 fjx0 = _mm256_add_ps(fjx0,tx);
962 fjy0 = _mm256_add_ps(fjy0,ty);
963 fjz0 = _mm256_add_ps(fjz0,tz);
965 /**************************
966 * CALCULATE INTERACTIONS *
967 **************************/
969 r10 = _mm256_mul_ps(rsq10,rinv10);
971 /* Compute parameters for interactions between i and j atoms */
972 qq10 = _mm256_mul_ps(iq1,jq0);
974 /* EWALD ELECTROSTATICS */
976 /* Analytical PME correction */
977 zeta2 = _mm256_mul_ps(beta2,rsq10);
978 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
979 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
980 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
981 felec = _mm256_mul_ps(qq10,felec);
985 /* Calculate temporary vectorial force */
986 tx = _mm256_mul_ps(fscal,dx10);
987 ty = _mm256_mul_ps(fscal,dy10);
988 tz = _mm256_mul_ps(fscal,dz10);
990 /* Update vectorial force */
991 fix1 = _mm256_add_ps(fix1,tx);
992 fiy1 = _mm256_add_ps(fiy1,ty);
993 fiz1 = _mm256_add_ps(fiz1,tz);
995 fjx0 = _mm256_add_ps(fjx0,tx);
996 fjy0 = _mm256_add_ps(fjy0,ty);
997 fjz0 = _mm256_add_ps(fjz0,tz);
999 /**************************
1000 * CALCULATE INTERACTIONS *
1001 **************************/
1003 r20 = _mm256_mul_ps(rsq20,rinv20);
1005 /* Compute parameters for interactions between i and j atoms */
1006 qq20 = _mm256_mul_ps(iq2,jq0);
1008 /* EWALD ELECTROSTATICS */
1010 /* Analytical PME correction */
1011 zeta2 = _mm256_mul_ps(beta2,rsq20);
1012 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
1013 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1014 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1015 felec = _mm256_mul_ps(qq20,felec);
1019 /* Calculate temporary vectorial force */
1020 tx = _mm256_mul_ps(fscal,dx20);
1021 ty = _mm256_mul_ps(fscal,dy20);
1022 tz = _mm256_mul_ps(fscal,dz20);
1024 /* Update vectorial force */
1025 fix2 = _mm256_add_ps(fix2,tx);
1026 fiy2 = _mm256_add_ps(fiy2,ty);
1027 fiz2 = _mm256_add_ps(fiz2,tz);
1029 fjx0 = _mm256_add_ps(fjx0,tx);
1030 fjy0 = _mm256_add_ps(fjy0,ty);
1031 fjz0 = _mm256_add_ps(fjz0,tz);
1033 fjptrA = f+j_coord_offsetA;
1034 fjptrB = f+j_coord_offsetB;
1035 fjptrC = f+j_coord_offsetC;
1036 fjptrD = f+j_coord_offsetD;
1037 fjptrE = f+j_coord_offsetE;
1038 fjptrF = f+j_coord_offsetF;
1039 fjptrG = f+j_coord_offsetG;
1040 fjptrH = f+j_coord_offsetH;
1042 gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
1044 /* Inner loop uses 178 flops */
1047 if(jidx<j_index_end)
1050 /* Get j neighbor index, and coordinate index */
1051 jnrlistA = jjnr[jidx];
1052 jnrlistB = jjnr[jidx+1];
1053 jnrlistC = jjnr[jidx+2];
1054 jnrlistD = jjnr[jidx+3];
1055 jnrlistE = jjnr[jidx+4];
1056 jnrlistF = jjnr[jidx+5];
1057 jnrlistG = jjnr[jidx+6];
1058 jnrlistH = jjnr[jidx+7];
1059 /* Sign of each element will be negative for non-real atoms.
1060 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1061 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1063 dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
1064 gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
1066 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
1067 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
1068 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
1069 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
1070 jnrE = (jnrlistE>=0) ? jnrlistE : 0;
1071 jnrF = (jnrlistF>=0) ? jnrlistF : 0;
1072 jnrG = (jnrlistG>=0) ? jnrlistG : 0;
1073 jnrH = (jnrlistH>=0) ? jnrlistH : 0;
1074 j_coord_offsetA = DIM*jnrA;
1075 j_coord_offsetB = DIM*jnrB;
1076 j_coord_offsetC = DIM*jnrC;
1077 j_coord_offsetD = DIM*jnrD;
1078 j_coord_offsetE = DIM*jnrE;
1079 j_coord_offsetF = DIM*jnrF;
1080 j_coord_offsetG = DIM*jnrG;
1081 j_coord_offsetH = DIM*jnrH;
1083 /* load j atom coordinates */
1084 gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1085 x+j_coord_offsetC,x+j_coord_offsetD,
1086 x+j_coord_offsetE,x+j_coord_offsetF,
1087 x+j_coord_offsetG,x+j_coord_offsetH,
1090 /* Calculate displacement vector */
1091 dx00 = _mm256_sub_ps(ix0,jx0);
1092 dy00 = _mm256_sub_ps(iy0,jy0);
1093 dz00 = _mm256_sub_ps(iz0,jz0);
1094 dx10 = _mm256_sub_ps(ix1,jx0);
1095 dy10 = _mm256_sub_ps(iy1,jy0);
1096 dz10 = _mm256_sub_ps(iz1,jz0);
1097 dx20 = _mm256_sub_ps(ix2,jx0);
1098 dy20 = _mm256_sub_ps(iy2,jy0);
1099 dz20 = _mm256_sub_ps(iz2,jz0);
1101 /* Calculate squared distance and things based on it */
1102 rsq00 = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1103 rsq10 = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1104 rsq20 = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1106 rinv00 = gmx_mm256_invsqrt_ps(rsq00);
1107 rinv10 = gmx_mm256_invsqrt_ps(rsq10);
1108 rinv20 = gmx_mm256_invsqrt_ps(rsq20);
1110 rinvsq00 = _mm256_mul_ps(rinv00,rinv00);
1111 rinvsq10 = _mm256_mul_ps(rinv10,rinv10);
1112 rinvsq20 = _mm256_mul_ps(rinv20,rinv20);
1114 /* Load parameters for j particles */
1115 jq0 = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1116 charge+jnrC+0,charge+jnrD+0,
1117 charge+jnrE+0,charge+jnrF+0,
1118 charge+jnrG+0,charge+jnrH+0);
1119 vdwjidx0A = 2*vdwtype[jnrA+0];
1120 vdwjidx0B = 2*vdwtype[jnrB+0];
1121 vdwjidx0C = 2*vdwtype[jnrC+0];
1122 vdwjidx0D = 2*vdwtype[jnrD+0];
1123 vdwjidx0E = 2*vdwtype[jnrE+0];
1124 vdwjidx0F = 2*vdwtype[jnrF+0];
1125 vdwjidx0G = 2*vdwtype[jnrG+0];
1126 vdwjidx0H = 2*vdwtype[jnrH+0];
1128 fjx0 = _mm256_setzero_ps();
1129 fjy0 = _mm256_setzero_ps();
1130 fjz0 = _mm256_setzero_ps();
1132 /**************************
1133 * CALCULATE INTERACTIONS *
1134 **************************/
1136 r00 = _mm256_mul_ps(rsq00,rinv00);
1137 r00 = _mm256_andnot_ps(dummy_mask,r00);
1139 /* Compute parameters for interactions between i and j atoms */
1140 qq00 = _mm256_mul_ps(iq0,jq0);
1141 gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
1142 vdwioffsetptr0+vdwjidx0B,
1143 vdwioffsetptr0+vdwjidx0C,
1144 vdwioffsetptr0+vdwjidx0D,
1145 vdwioffsetptr0+vdwjidx0E,
1146 vdwioffsetptr0+vdwjidx0F,
1147 vdwioffsetptr0+vdwjidx0G,
1148 vdwioffsetptr0+vdwjidx0H,
1151 /* EWALD ELECTROSTATICS */
1153 /* Analytical PME correction */
1154 zeta2 = _mm256_mul_ps(beta2,rsq00);
1155 rinv3 = _mm256_mul_ps(rinvsq00,rinv00);
1156 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1157 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1158 felec = _mm256_mul_ps(qq00,felec);
1160 /* LENNARD-JONES DISPERSION/REPULSION */
1162 rinvsix = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1163 fvdw = _mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),c6_00),_mm256_mul_ps(rinvsix,rinvsq00));
1165 fscal = _mm256_add_ps(felec,fvdw);
1167 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1169 /* Calculate temporary vectorial force */
1170 tx = _mm256_mul_ps(fscal,dx00);
1171 ty = _mm256_mul_ps(fscal,dy00);
1172 tz = _mm256_mul_ps(fscal,dz00);
1174 /* Update vectorial force */
1175 fix0 = _mm256_add_ps(fix0,tx);
1176 fiy0 = _mm256_add_ps(fiy0,ty);
1177 fiz0 = _mm256_add_ps(fiz0,tz);
1179 fjx0 = _mm256_add_ps(fjx0,tx);
1180 fjy0 = _mm256_add_ps(fjy0,ty);
1181 fjz0 = _mm256_add_ps(fjz0,tz);
1183 /**************************
1184 * CALCULATE INTERACTIONS *
1185 **************************/
1187 r10 = _mm256_mul_ps(rsq10,rinv10);
1188 r10 = _mm256_andnot_ps(dummy_mask,r10);
1190 /* Compute parameters for interactions between i and j atoms */
1191 qq10 = _mm256_mul_ps(iq1,jq0);
1193 /* EWALD ELECTROSTATICS */
1195 /* Analytical PME correction */
1196 zeta2 = _mm256_mul_ps(beta2,rsq10);
1197 rinv3 = _mm256_mul_ps(rinvsq10,rinv10);
1198 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1199 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1200 felec = _mm256_mul_ps(qq10,felec);
1204 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1206 /* Calculate temporary vectorial force */
1207 tx = _mm256_mul_ps(fscal,dx10);
1208 ty = _mm256_mul_ps(fscal,dy10);
1209 tz = _mm256_mul_ps(fscal,dz10);
1211 /* Update vectorial force */
1212 fix1 = _mm256_add_ps(fix1,tx);
1213 fiy1 = _mm256_add_ps(fiy1,ty);
1214 fiz1 = _mm256_add_ps(fiz1,tz);
1216 fjx0 = _mm256_add_ps(fjx0,tx);
1217 fjy0 = _mm256_add_ps(fjy0,ty);
1218 fjz0 = _mm256_add_ps(fjz0,tz);
1220 /**************************
1221 * CALCULATE INTERACTIONS *
1222 **************************/
1224 r20 = _mm256_mul_ps(rsq20,rinv20);
1225 r20 = _mm256_andnot_ps(dummy_mask,r20);
1227 /* Compute parameters for interactions between i and j atoms */
1228 qq20 = _mm256_mul_ps(iq2,jq0);
1230 /* EWALD ELECTROSTATICS */
1232 /* Analytical PME correction */
1233 zeta2 = _mm256_mul_ps(beta2,rsq20);
1234 rinv3 = _mm256_mul_ps(rinvsq20,rinv20);
1235 pmecorrF = gmx_mm256_pmecorrF_ps(zeta2);
1236 felec = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1237 felec = _mm256_mul_ps(qq20,felec);
1241 fscal = _mm256_andnot_ps(dummy_mask,fscal);
1243 /* Calculate temporary vectorial force */
1244 tx = _mm256_mul_ps(fscal,dx20);
1245 ty = _mm256_mul_ps(fscal,dy20);
1246 tz = _mm256_mul_ps(fscal,dz20);
1248 /* Update vectorial force */
1249 fix2 = _mm256_add_ps(fix2,tx);
1250 fiy2 = _mm256_add_ps(fiy2,ty);
1251 fiz2 = _mm256_add_ps(fiz2,tz);
1253 fjx0 = _mm256_add_ps(fjx0,tx);
1254 fjy0 = _mm256_add_ps(fjy0,ty);
1255 fjz0 = _mm256_add_ps(fjz0,tz);
1257 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1258 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1259 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1260 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1261 fjptrE = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1262 fjptrF = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1263 fjptrG = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1264 fjptrH = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1266 gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
1268 /* Inner loop uses 181 flops */
1271 /* End of innermost loop */
1273 gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1274 f+i_coord_offset,fshift+i_shift_offset);
1276 /* Increment number of inner iterations */
1277 inneriter += j_index_end - j_index_start;
1279 /* Outer loop uses 18 flops */
1282 /* Increment number of outer iterations */
1285 /* Update outer/inner flops */
1287 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*18 + inneriter*181);