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 sse4_1_double kernel generator.
44 #include "../nb_kernel.h"
45 #include "gromacs/legacyheaders/types/simple.h"
46 #include "gromacs/math/vec.h"
47 #include "gromacs/legacyheaders/nrnb.h"
49 #include "gromacs/simd/math_x86_sse4_1_double.h"
50 #include "kernelutil_x86_sse4_1_double.h"
53 * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3P1_VF_sse4_1_double
54 * Electrostatics interaction: Ewald
55 * VdW interaction: LJEwald
56 * Geometry: Water3-Particle
57 * Calculate force/pot: PotentialAndForce
60 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3P1_VF_sse4_1_double
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 refer to j loop unrolling done with SSE double precision, e.g. for the two 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;
77 int j_coord_offsetA,j_coord_offsetB;
78 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
80 real *shiftvec,*fshift,*x,*f;
81 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
83 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
85 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
87 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
88 int vdwjidx0A,vdwjidx0B;
89 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
90 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
91 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
92 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
93 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
96 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
99 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
100 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
104 __m128d ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
106 __m128d one_half = _mm_set1_pd(0.5);
107 __m128d minus_one = _mm_set1_pd(-1.0);
109 __m128d ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
111 __m128d dummy_mask,cutoff_mask;
112 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
113 __m128d one = _mm_set1_pd(1.0);
114 __m128d two = _mm_set1_pd(2.0);
120 jindex = nlist->jindex;
122 shiftidx = nlist->shift;
124 shiftvec = fr->shift_vec[0];
125 fshift = fr->fshift[0];
126 facel = _mm_set1_pd(fr->epsfac);
127 charge = mdatoms->chargeA;
128 nvdwtype = fr->ntype;
130 vdwtype = mdatoms->typeA;
131 vdwgridparam = fr->ljpme_c6grid;
132 sh_lj_ewald = _mm_set1_pd(fr->ic->sh_lj_ewald);
133 ewclj = _mm_set1_pd(fr->ewaldcoeff_lj);
134 ewclj2 = _mm_mul_pd(minus_one,_mm_mul_pd(ewclj,ewclj));
136 sh_ewald = _mm_set1_pd(fr->ic->sh_ewald);
137 ewtab = fr->ic->tabq_coul_FDV0;
138 ewtabscale = _mm_set1_pd(fr->ic->tabq_scale);
139 ewtabhalfspace = _mm_set1_pd(0.5/fr->ic->tabq_scale);
141 /* Setup water-specific parameters */
142 inr = nlist->iinr[0];
143 iq0 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
144 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
145 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
146 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
148 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
149 rcutoff_scalar = fr->rcoulomb;
150 rcutoff = _mm_set1_pd(rcutoff_scalar);
151 rcutoff2 = _mm_mul_pd(rcutoff,rcutoff);
153 sh_vdw_invrcut6 = _mm_set1_pd(fr->ic->sh_invrc6);
154 rvdw = _mm_set1_pd(fr->rvdw);
156 /* Avoid stupid compiler warnings */
164 /* Start outer loop over neighborlists */
165 for(iidx=0; iidx<nri; iidx++)
167 /* Load shift vector for this list */
168 i_shift_offset = DIM*shiftidx[iidx];
170 /* Load limits for loop over neighbors */
171 j_index_start = jindex[iidx];
172 j_index_end = jindex[iidx+1];
174 /* Get outer coordinate index */
176 i_coord_offset = DIM*inr;
178 /* Load i particle coords and add shift vector */
179 gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
180 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
182 fix0 = _mm_setzero_pd();
183 fiy0 = _mm_setzero_pd();
184 fiz0 = _mm_setzero_pd();
185 fix1 = _mm_setzero_pd();
186 fiy1 = _mm_setzero_pd();
187 fiz1 = _mm_setzero_pd();
188 fix2 = _mm_setzero_pd();
189 fiy2 = _mm_setzero_pd();
190 fiz2 = _mm_setzero_pd();
192 /* Reset potential sums */
193 velecsum = _mm_setzero_pd();
194 vvdwsum = _mm_setzero_pd();
196 /* Start inner kernel loop */
197 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
200 /* Get j neighbor index, and coordinate index */
203 j_coord_offsetA = DIM*jnrA;
204 j_coord_offsetB = DIM*jnrB;
206 /* load j atom coordinates */
207 gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
210 /* Calculate displacement vector */
211 dx00 = _mm_sub_pd(ix0,jx0);
212 dy00 = _mm_sub_pd(iy0,jy0);
213 dz00 = _mm_sub_pd(iz0,jz0);
214 dx10 = _mm_sub_pd(ix1,jx0);
215 dy10 = _mm_sub_pd(iy1,jy0);
216 dz10 = _mm_sub_pd(iz1,jz0);
217 dx20 = _mm_sub_pd(ix2,jx0);
218 dy20 = _mm_sub_pd(iy2,jy0);
219 dz20 = _mm_sub_pd(iz2,jz0);
221 /* Calculate squared distance and things based on it */
222 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
223 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
224 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
226 rinv00 = gmx_mm_invsqrt_pd(rsq00);
227 rinv10 = gmx_mm_invsqrt_pd(rsq10);
228 rinv20 = gmx_mm_invsqrt_pd(rsq20);
230 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
231 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
232 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
234 /* Load parameters for j particles */
235 jq0 = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
236 vdwjidx0A = 2*vdwtype[jnrA+0];
237 vdwjidx0B = 2*vdwtype[jnrB+0];
239 fjx0 = _mm_setzero_pd();
240 fjy0 = _mm_setzero_pd();
241 fjz0 = _mm_setzero_pd();
243 /**************************
244 * CALCULATE INTERACTIONS *
245 **************************/
247 if (gmx_mm_any_lt(rsq00,rcutoff2))
250 r00 = _mm_mul_pd(rsq00,rinv00);
252 /* Compute parameters for interactions between i and j atoms */
253 qq00 = _mm_mul_pd(iq0,jq0);
254 gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
255 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
256 c6grid_00 = gmx_mm_load_2real_swizzle_pd(vdwgridparam+vdwioffset0+vdwjidx0A,
257 vdwgridparam+vdwioffset0+vdwjidx0B);
259 /* EWALD ELECTROSTATICS */
261 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
262 ewrt = _mm_mul_pd(r00,ewtabscale);
263 ewitab = _mm_cvttpd_epi32(ewrt);
264 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
265 ewitab = _mm_slli_epi32(ewitab,2);
266 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
267 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
268 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
269 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
270 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
271 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
272 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
273 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
274 velec = _mm_mul_pd(qq00,_mm_sub_pd(_mm_sub_pd(rinv00,sh_ewald),velec));
275 felec = _mm_mul_pd(_mm_mul_pd(qq00,rinv00),_mm_sub_pd(rinvsq00,felec));
277 /* Analytical LJ-PME */
278 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
279 ewcljrsq = _mm_mul_pd(ewclj2,rsq00);
280 ewclj6 = _mm_mul_pd(ewclj2,_mm_mul_pd(ewclj2,ewclj2));
281 exponent = gmx_simd_exp_d(ewcljrsq);
282 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
283 poly = _mm_mul_pd(exponent,_mm_add_pd(_mm_sub_pd(one,ewcljrsq),_mm_mul_pd(_mm_mul_pd(ewcljrsq,ewcljrsq),one_half)));
284 /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
285 vvdw6 = _mm_mul_pd(_mm_sub_pd(c6_00,_mm_mul_pd(c6grid_00,_mm_sub_pd(one,poly))),rinvsix);
286 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
287 vvdw = _mm_sub_pd(_mm_mul_pd( _mm_sub_pd(vvdw12 , _mm_mul_pd(c12_00,_mm_mul_pd(sh_vdw_invrcut6,sh_vdw_invrcut6))),one_twelfth),
288 _mm_mul_pd( _mm_sub_pd(vvdw6,_mm_add_pd(_mm_mul_pd(c6_00,sh_vdw_invrcut6),_mm_mul_pd(c6grid_00,sh_lj_ewald))),one_sixth));
289 /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
290 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,_mm_sub_pd(vvdw6,_mm_mul_pd(_mm_mul_pd(c6grid_00,one_sixth),_mm_mul_pd(exponent,ewclj6)))),rinvsq00);
292 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
294 /* Update potential sum for this i atom from the interaction with this j atom. */
295 velec = _mm_and_pd(velec,cutoff_mask);
296 velecsum = _mm_add_pd(velecsum,velec);
297 vvdw = _mm_and_pd(vvdw,cutoff_mask);
298 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
300 fscal = _mm_add_pd(felec,fvdw);
302 fscal = _mm_and_pd(fscal,cutoff_mask);
304 /* Calculate temporary vectorial force */
305 tx = _mm_mul_pd(fscal,dx00);
306 ty = _mm_mul_pd(fscal,dy00);
307 tz = _mm_mul_pd(fscal,dz00);
309 /* Update vectorial force */
310 fix0 = _mm_add_pd(fix0,tx);
311 fiy0 = _mm_add_pd(fiy0,ty);
312 fiz0 = _mm_add_pd(fiz0,tz);
314 fjx0 = _mm_add_pd(fjx0,tx);
315 fjy0 = _mm_add_pd(fjy0,ty);
316 fjz0 = _mm_add_pd(fjz0,tz);
320 /**************************
321 * CALCULATE INTERACTIONS *
322 **************************/
324 if (gmx_mm_any_lt(rsq10,rcutoff2))
327 r10 = _mm_mul_pd(rsq10,rinv10);
329 /* Compute parameters for interactions between i and j atoms */
330 qq10 = _mm_mul_pd(iq1,jq0);
332 /* EWALD ELECTROSTATICS */
334 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
335 ewrt = _mm_mul_pd(r10,ewtabscale);
336 ewitab = _mm_cvttpd_epi32(ewrt);
337 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
338 ewitab = _mm_slli_epi32(ewitab,2);
339 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
340 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
341 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
342 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
343 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
344 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
345 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
346 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
347 velec = _mm_mul_pd(qq10,_mm_sub_pd(_mm_sub_pd(rinv10,sh_ewald),velec));
348 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
350 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
352 /* Update potential sum for this i atom from the interaction with this j atom. */
353 velec = _mm_and_pd(velec,cutoff_mask);
354 velecsum = _mm_add_pd(velecsum,velec);
358 fscal = _mm_and_pd(fscal,cutoff_mask);
360 /* Calculate temporary vectorial force */
361 tx = _mm_mul_pd(fscal,dx10);
362 ty = _mm_mul_pd(fscal,dy10);
363 tz = _mm_mul_pd(fscal,dz10);
365 /* Update vectorial force */
366 fix1 = _mm_add_pd(fix1,tx);
367 fiy1 = _mm_add_pd(fiy1,ty);
368 fiz1 = _mm_add_pd(fiz1,tz);
370 fjx0 = _mm_add_pd(fjx0,tx);
371 fjy0 = _mm_add_pd(fjy0,ty);
372 fjz0 = _mm_add_pd(fjz0,tz);
376 /**************************
377 * CALCULATE INTERACTIONS *
378 **************************/
380 if (gmx_mm_any_lt(rsq20,rcutoff2))
383 r20 = _mm_mul_pd(rsq20,rinv20);
385 /* Compute parameters for interactions between i and j atoms */
386 qq20 = _mm_mul_pd(iq2,jq0);
388 /* EWALD ELECTROSTATICS */
390 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
391 ewrt = _mm_mul_pd(r20,ewtabscale);
392 ewitab = _mm_cvttpd_epi32(ewrt);
393 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
394 ewitab = _mm_slli_epi32(ewitab,2);
395 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
396 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
397 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
398 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
399 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
400 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
401 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
402 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
403 velec = _mm_mul_pd(qq20,_mm_sub_pd(_mm_sub_pd(rinv20,sh_ewald),velec));
404 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
406 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
408 /* Update potential sum for this i atom from the interaction with this j atom. */
409 velec = _mm_and_pd(velec,cutoff_mask);
410 velecsum = _mm_add_pd(velecsum,velec);
414 fscal = _mm_and_pd(fscal,cutoff_mask);
416 /* Calculate temporary vectorial force */
417 tx = _mm_mul_pd(fscal,dx20);
418 ty = _mm_mul_pd(fscal,dy20);
419 tz = _mm_mul_pd(fscal,dz20);
421 /* Update vectorial force */
422 fix2 = _mm_add_pd(fix2,tx);
423 fiy2 = _mm_add_pd(fiy2,ty);
424 fiz2 = _mm_add_pd(fiz2,tz);
426 fjx0 = _mm_add_pd(fjx0,tx);
427 fjy0 = _mm_add_pd(fjy0,ty);
428 fjz0 = _mm_add_pd(fjz0,tz);
432 gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
434 /* Inner loop uses 176 flops */
441 j_coord_offsetA = DIM*jnrA;
443 /* load j atom coordinates */
444 gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
447 /* Calculate displacement vector */
448 dx00 = _mm_sub_pd(ix0,jx0);
449 dy00 = _mm_sub_pd(iy0,jy0);
450 dz00 = _mm_sub_pd(iz0,jz0);
451 dx10 = _mm_sub_pd(ix1,jx0);
452 dy10 = _mm_sub_pd(iy1,jy0);
453 dz10 = _mm_sub_pd(iz1,jz0);
454 dx20 = _mm_sub_pd(ix2,jx0);
455 dy20 = _mm_sub_pd(iy2,jy0);
456 dz20 = _mm_sub_pd(iz2,jz0);
458 /* Calculate squared distance and things based on it */
459 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
460 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
461 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
463 rinv00 = gmx_mm_invsqrt_pd(rsq00);
464 rinv10 = gmx_mm_invsqrt_pd(rsq10);
465 rinv20 = gmx_mm_invsqrt_pd(rsq20);
467 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
468 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
469 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
471 /* Load parameters for j particles */
472 jq0 = _mm_load_sd(charge+jnrA+0);
473 vdwjidx0A = 2*vdwtype[jnrA+0];
475 fjx0 = _mm_setzero_pd();
476 fjy0 = _mm_setzero_pd();
477 fjz0 = _mm_setzero_pd();
479 /**************************
480 * CALCULATE INTERACTIONS *
481 **************************/
483 if (gmx_mm_any_lt(rsq00,rcutoff2))
486 r00 = _mm_mul_pd(rsq00,rinv00);
488 /* Compute parameters for interactions between i and j atoms */
489 qq00 = _mm_mul_pd(iq0,jq0);
490 gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
492 c6grid_00 = gmx_mm_load_1real_pd(vdwgridparam+vdwioffset0+vdwjidx0A);
494 /* EWALD ELECTROSTATICS */
496 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
497 ewrt = _mm_mul_pd(r00,ewtabscale);
498 ewitab = _mm_cvttpd_epi32(ewrt);
499 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
500 ewitab = _mm_slli_epi32(ewitab,2);
501 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
502 ewtabD = _mm_setzero_pd();
503 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
504 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
505 ewtabFn = _mm_setzero_pd();
506 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
507 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
508 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
509 velec = _mm_mul_pd(qq00,_mm_sub_pd(_mm_sub_pd(rinv00,sh_ewald),velec));
510 felec = _mm_mul_pd(_mm_mul_pd(qq00,rinv00),_mm_sub_pd(rinvsq00,felec));
512 /* Analytical LJ-PME */
513 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
514 ewcljrsq = _mm_mul_pd(ewclj2,rsq00);
515 ewclj6 = _mm_mul_pd(ewclj2,_mm_mul_pd(ewclj2,ewclj2));
516 exponent = gmx_simd_exp_d(ewcljrsq);
517 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
518 poly = _mm_mul_pd(exponent,_mm_add_pd(_mm_sub_pd(one,ewcljrsq),_mm_mul_pd(_mm_mul_pd(ewcljrsq,ewcljrsq),one_half)));
519 /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
520 vvdw6 = _mm_mul_pd(_mm_sub_pd(c6_00,_mm_mul_pd(c6grid_00,_mm_sub_pd(one,poly))),rinvsix);
521 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
522 vvdw = _mm_sub_pd(_mm_mul_pd( _mm_sub_pd(vvdw12 , _mm_mul_pd(c12_00,_mm_mul_pd(sh_vdw_invrcut6,sh_vdw_invrcut6))),one_twelfth),
523 _mm_mul_pd( _mm_sub_pd(vvdw6,_mm_add_pd(_mm_mul_pd(c6_00,sh_vdw_invrcut6),_mm_mul_pd(c6grid_00,sh_lj_ewald))),one_sixth));
524 /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
525 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,_mm_sub_pd(vvdw6,_mm_mul_pd(_mm_mul_pd(c6grid_00,one_sixth),_mm_mul_pd(exponent,ewclj6)))),rinvsq00);
527 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
529 /* Update potential sum for this i atom from the interaction with this j atom. */
530 velec = _mm_and_pd(velec,cutoff_mask);
531 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
532 velecsum = _mm_add_pd(velecsum,velec);
533 vvdw = _mm_and_pd(vvdw,cutoff_mask);
534 vvdw = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
535 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
537 fscal = _mm_add_pd(felec,fvdw);
539 fscal = _mm_and_pd(fscal,cutoff_mask);
541 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
543 /* Calculate temporary vectorial force */
544 tx = _mm_mul_pd(fscal,dx00);
545 ty = _mm_mul_pd(fscal,dy00);
546 tz = _mm_mul_pd(fscal,dz00);
548 /* Update vectorial force */
549 fix0 = _mm_add_pd(fix0,tx);
550 fiy0 = _mm_add_pd(fiy0,ty);
551 fiz0 = _mm_add_pd(fiz0,tz);
553 fjx0 = _mm_add_pd(fjx0,tx);
554 fjy0 = _mm_add_pd(fjy0,ty);
555 fjz0 = _mm_add_pd(fjz0,tz);
559 /**************************
560 * CALCULATE INTERACTIONS *
561 **************************/
563 if (gmx_mm_any_lt(rsq10,rcutoff2))
566 r10 = _mm_mul_pd(rsq10,rinv10);
568 /* Compute parameters for interactions between i and j atoms */
569 qq10 = _mm_mul_pd(iq1,jq0);
571 /* EWALD ELECTROSTATICS */
573 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
574 ewrt = _mm_mul_pd(r10,ewtabscale);
575 ewitab = _mm_cvttpd_epi32(ewrt);
576 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
577 ewitab = _mm_slli_epi32(ewitab,2);
578 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
579 ewtabD = _mm_setzero_pd();
580 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
581 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
582 ewtabFn = _mm_setzero_pd();
583 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
584 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
585 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
586 velec = _mm_mul_pd(qq10,_mm_sub_pd(_mm_sub_pd(rinv10,sh_ewald),velec));
587 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
589 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
591 /* Update potential sum for this i atom from the interaction with this j atom. */
592 velec = _mm_and_pd(velec,cutoff_mask);
593 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
594 velecsum = _mm_add_pd(velecsum,velec);
598 fscal = _mm_and_pd(fscal,cutoff_mask);
600 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
602 /* Calculate temporary vectorial force */
603 tx = _mm_mul_pd(fscal,dx10);
604 ty = _mm_mul_pd(fscal,dy10);
605 tz = _mm_mul_pd(fscal,dz10);
607 /* Update vectorial force */
608 fix1 = _mm_add_pd(fix1,tx);
609 fiy1 = _mm_add_pd(fiy1,ty);
610 fiz1 = _mm_add_pd(fiz1,tz);
612 fjx0 = _mm_add_pd(fjx0,tx);
613 fjy0 = _mm_add_pd(fjy0,ty);
614 fjz0 = _mm_add_pd(fjz0,tz);
618 /**************************
619 * CALCULATE INTERACTIONS *
620 **************************/
622 if (gmx_mm_any_lt(rsq20,rcutoff2))
625 r20 = _mm_mul_pd(rsq20,rinv20);
627 /* Compute parameters for interactions between i and j atoms */
628 qq20 = _mm_mul_pd(iq2,jq0);
630 /* EWALD ELECTROSTATICS */
632 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
633 ewrt = _mm_mul_pd(r20,ewtabscale);
634 ewitab = _mm_cvttpd_epi32(ewrt);
635 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
636 ewitab = _mm_slli_epi32(ewitab,2);
637 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
638 ewtabD = _mm_setzero_pd();
639 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
640 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
641 ewtabFn = _mm_setzero_pd();
642 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
643 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
644 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
645 velec = _mm_mul_pd(qq20,_mm_sub_pd(_mm_sub_pd(rinv20,sh_ewald),velec));
646 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
648 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
650 /* Update potential sum for this i atom from the interaction with this j atom. */
651 velec = _mm_and_pd(velec,cutoff_mask);
652 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
653 velecsum = _mm_add_pd(velecsum,velec);
657 fscal = _mm_and_pd(fscal,cutoff_mask);
659 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
661 /* Calculate temporary vectorial force */
662 tx = _mm_mul_pd(fscal,dx20);
663 ty = _mm_mul_pd(fscal,dy20);
664 tz = _mm_mul_pd(fscal,dz20);
666 /* Update vectorial force */
667 fix2 = _mm_add_pd(fix2,tx);
668 fiy2 = _mm_add_pd(fiy2,ty);
669 fiz2 = _mm_add_pd(fiz2,tz);
671 fjx0 = _mm_add_pd(fjx0,tx);
672 fjy0 = _mm_add_pd(fjy0,ty);
673 fjz0 = _mm_add_pd(fjz0,tz);
677 gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
679 /* Inner loop uses 176 flops */
682 /* End of innermost loop */
684 gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
685 f+i_coord_offset,fshift+i_shift_offset);
688 /* Update potential energies */
689 gmx_mm_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
690 gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
692 /* Increment number of inner iterations */
693 inneriter += j_index_end - j_index_start;
695 /* Outer loop uses 20 flops */
698 /* Increment number of outer iterations */
701 /* Update outer/inner flops */
703 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*20 + inneriter*176);
706 * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3P1_F_sse4_1_double
707 * Electrostatics interaction: Ewald
708 * VdW interaction: LJEwald
709 * Geometry: Water3-Particle
710 * Calculate force/pot: Force
713 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3P1_F_sse4_1_double
714 (t_nblist * gmx_restrict nlist,
715 rvec * gmx_restrict xx,
716 rvec * gmx_restrict ff,
717 t_forcerec * gmx_restrict fr,
718 t_mdatoms * gmx_restrict mdatoms,
719 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
720 t_nrnb * gmx_restrict nrnb)
722 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
723 * just 0 for non-waters.
724 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
725 * jnr indices corresponding to data put in the four positions in the SIMD register.
727 int i_shift_offset,i_coord_offset,outeriter,inneriter;
728 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
730 int j_coord_offsetA,j_coord_offsetB;
731 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
733 real *shiftvec,*fshift,*x,*f;
734 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
736 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
738 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
740 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
741 int vdwjidx0A,vdwjidx0B;
742 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
743 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
744 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
745 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
746 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
749 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
752 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
753 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
757 __m128d ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
759 __m128d one_half = _mm_set1_pd(0.5);
760 __m128d minus_one = _mm_set1_pd(-1.0);
762 __m128d ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
764 __m128d dummy_mask,cutoff_mask;
765 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
766 __m128d one = _mm_set1_pd(1.0);
767 __m128d two = _mm_set1_pd(2.0);
773 jindex = nlist->jindex;
775 shiftidx = nlist->shift;
777 shiftvec = fr->shift_vec[0];
778 fshift = fr->fshift[0];
779 facel = _mm_set1_pd(fr->epsfac);
780 charge = mdatoms->chargeA;
781 nvdwtype = fr->ntype;
783 vdwtype = mdatoms->typeA;
784 vdwgridparam = fr->ljpme_c6grid;
785 sh_lj_ewald = _mm_set1_pd(fr->ic->sh_lj_ewald);
786 ewclj = _mm_set1_pd(fr->ewaldcoeff_lj);
787 ewclj2 = _mm_mul_pd(minus_one,_mm_mul_pd(ewclj,ewclj));
789 sh_ewald = _mm_set1_pd(fr->ic->sh_ewald);
790 ewtab = fr->ic->tabq_coul_F;
791 ewtabscale = _mm_set1_pd(fr->ic->tabq_scale);
792 ewtabhalfspace = _mm_set1_pd(0.5/fr->ic->tabq_scale);
794 /* Setup water-specific parameters */
795 inr = nlist->iinr[0];
796 iq0 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
797 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
798 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
799 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
801 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
802 rcutoff_scalar = fr->rcoulomb;
803 rcutoff = _mm_set1_pd(rcutoff_scalar);
804 rcutoff2 = _mm_mul_pd(rcutoff,rcutoff);
806 sh_vdw_invrcut6 = _mm_set1_pd(fr->ic->sh_invrc6);
807 rvdw = _mm_set1_pd(fr->rvdw);
809 /* Avoid stupid compiler warnings */
817 /* Start outer loop over neighborlists */
818 for(iidx=0; iidx<nri; iidx++)
820 /* Load shift vector for this list */
821 i_shift_offset = DIM*shiftidx[iidx];
823 /* Load limits for loop over neighbors */
824 j_index_start = jindex[iidx];
825 j_index_end = jindex[iidx+1];
827 /* Get outer coordinate index */
829 i_coord_offset = DIM*inr;
831 /* Load i particle coords and add shift vector */
832 gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
833 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
835 fix0 = _mm_setzero_pd();
836 fiy0 = _mm_setzero_pd();
837 fiz0 = _mm_setzero_pd();
838 fix1 = _mm_setzero_pd();
839 fiy1 = _mm_setzero_pd();
840 fiz1 = _mm_setzero_pd();
841 fix2 = _mm_setzero_pd();
842 fiy2 = _mm_setzero_pd();
843 fiz2 = _mm_setzero_pd();
845 /* Start inner kernel loop */
846 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
849 /* Get j neighbor index, and coordinate index */
852 j_coord_offsetA = DIM*jnrA;
853 j_coord_offsetB = DIM*jnrB;
855 /* load j atom coordinates */
856 gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
859 /* Calculate displacement vector */
860 dx00 = _mm_sub_pd(ix0,jx0);
861 dy00 = _mm_sub_pd(iy0,jy0);
862 dz00 = _mm_sub_pd(iz0,jz0);
863 dx10 = _mm_sub_pd(ix1,jx0);
864 dy10 = _mm_sub_pd(iy1,jy0);
865 dz10 = _mm_sub_pd(iz1,jz0);
866 dx20 = _mm_sub_pd(ix2,jx0);
867 dy20 = _mm_sub_pd(iy2,jy0);
868 dz20 = _mm_sub_pd(iz2,jz0);
870 /* Calculate squared distance and things based on it */
871 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
872 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
873 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
875 rinv00 = gmx_mm_invsqrt_pd(rsq00);
876 rinv10 = gmx_mm_invsqrt_pd(rsq10);
877 rinv20 = gmx_mm_invsqrt_pd(rsq20);
879 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
880 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
881 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
883 /* Load parameters for j particles */
884 jq0 = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
885 vdwjidx0A = 2*vdwtype[jnrA+0];
886 vdwjidx0B = 2*vdwtype[jnrB+0];
888 fjx0 = _mm_setzero_pd();
889 fjy0 = _mm_setzero_pd();
890 fjz0 = _mm_setzero_pd();
892 /**************************
893 * CALCULATE INTERACTIONS *
894 **************************/
896 if (gmx_mm_any_lt(rsq00,rcutoff2))
899 r00 = _mm_mul_pd(rsq00,rinv00);
901 /* Compute parameters for interactions between i and j atoms */
902 qq00 = _mm_mul_pd(iq0,jq0);
903 gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
904 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
905 c6grid_00 = gmx_mm_load_2real_swizzle_pd(vdwgridparam+vdwioffset0+vdwjidx0A,
906 vdwgridparam+vdwioffset0+vdwjidx0B);
908 /* EWALD ELECTROSTATICS */
910 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
911 ewrt = _mm_mul_pd(r00,ewtabscale);
912 ewitab = _mm_cvttpd_epi32(ewrt);
913 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
914 gmx_mm_load_2pair_swizzle_pd(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
916 felec = _mm_add_pd(_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF),_mm_mul_pd(eweps,ewtabFn));
917 felec = _mm_mul_pd(_mm_mul_pd(qq00,rinv00),_mm_sub_pd(rinvsq00,felec));
919 /* Analytical LJ-PME */
920 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
921 ewcljrsq = _mm_mul_pd(ewclj2,rsq00);
922 ewclj6 = _mm_mul_pd(ewclj2,_mm_mul_pd(ewclj2,ewclj2));
923 exponent = gmx_simd_exp_d(ewcljrsq);
924 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
925 poly = _mm_mul_pd(exponent,_mm_add_pd(_mm_sub_pd(one,ewcljrsq),_mm_mul_pd(_mm_mul_pd(ewcljrsq,ewcljrsq),one_half)));
926 /* f6A = 6 * C6grid * (1 - poly) */
927 f6A = _mm_mul_pd(c6grid_00,_mm_sub_pd(one,poly));
928 /* f6B = C6grid * exponent * beta^6 */
929 f6B = _mm_mul_pd(_mm_mul_pd(c6grid_00,one_sixth),_mm_mul_pd(exponent,ewclj6));
930 /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
931 fvdw = _mm_mul_pd(_mm_add_pd(_mm_mul_pd(_mm_sub_pd(_mm_mul_pd(c12_00,rinvsix),_mm_sub_pd(c6_00,f6A)),rinvsix),f6B),rinvsq00);
933 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
935 fscal = _mm_add_pd(felec,fvdw);
937 fscal = _mm_and_pd(fscal,cutoff_mask);
939 /* Calculate temporary vectorial force */
940 tx = _mm_mul_pd(fscal,dx00);
941 ty = _mm_mul_pd(fscal,dy00);
942 tz = _mm_mul_pd(fscal,dz00);
944 /* Update vectorial force */
945 fix0 = _mm_add_pd(fix0,tx);
946 fiy0 = _mm_add_pd(fiy0,ty);
947 fiz0 = _mm_add_pd(fiz0,tz);
949 fjx0 = _mm_add_pd(fjx0,tx);
950 fjy0 = _mm_add_pd(fjy0,ty);
951 fjz0 = _mm_add_pd(fjz0,tz);
955 /**************************
956 * CALCULATE INTERACTIONS *
957 **************************/
959 if (gmx_mm_any_lt(rsq10,rcutoff2))
962 r10 = _mm_mul_pd(rsq10,rinv10);
964 /* Compute parameters for interactions between i and j atoms */
965 qq10 = _mm_mul_pd(iq1,jq0);
967 /* EWALD ELECTROSTATICS */
969 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
970 ewrt = _mm_mul_pd(r10,ewtabscale);
971 ewitab = _mm_cvttpd_epi32(ewrt);
972 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
973 gmx_mm_load_2pair_swizzle_pd(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
975 felec = _mm_add_pd(_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF),_mm_mul_pd(eweps,ewtabFn));
976 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
978 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
982 fscal = _mm_and_pd(fscal,cutoff_mask);
984 /* Calculate temporary vectorial force */
985 tx = _mm_mul_pd(fscal,dx10);
986 ty = _mm_mul_pd(fscal,dy10);
987 tz = _mm_mul_pd(fscal,dz10);
989 /* Update vectorial force */
990 fix1 = _mm_add_pd(fix1,tx);
991 fiy1 = _mm_add_pd(fiy1,ty);
992 fiz1 = _mm_add_pd(fiz1,tz);
994 fjx0 = _mm_add_pd(fjx0,tx);
995 fjy0 = _mm_add_pd(fjy0,ty);
996 fjz0 = _mm_add_pd(fjz0,tz);
1000 /**************************
1001 * CALCULATE INTERACTIONS *
1002 **************************/
1004 if (gmx_mm_any_lt(rsq20,rcutoff2))
1007 r20 = _mm_mul_pd(rsq20,rinv20);
1009 /* Compute parameters for interactions between i and j atoms */
1010 qq20 = _mm_mul_pd(iq2,jq0);
1012 /* EWALD ELECTROSTATICS */
1014 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1015 ewrt = _mm_mul_pd(r20,ewtabscale);
1016 ewitab = _mm_cvttpd_epi32(ewrt);
1017 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1018 gmx_mm_load_2pair_swizzle_pd(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1020 felec = _mm_add_pd(_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF),_mm_mul_pd(eweps,ewtabFn));
1021 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
1023 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
1027 fscal = _mm_and_pd(fscal,cutoff_mask);
1029 /* Calculate temporary vectorial force */
1030 tx = _mm_mul_pd(fscal,dx20);
1031 ty = _mm_mul_pd(fscal,dy20);
1032 tz = _mm_mul_pd(fscal,dz20);
1034 /* Update vectorial force */
1035 fix2 = _mm_add_pd(fix2,tx);
1036 fiy2 = _mm_add_pd(fiy2,ty);
1037 fiz2 = _mm_add_pd(fiz2,tz);
1039 fjx0 = _mm_add_pd(fjx0,tx);
1040 fjy0 = _mm_add_pd(fjy0,ty);
1041 fjz0 = _mm_add_pd(fjz0,tz);
1045 gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
1047 /* Inner loop uses 143 flops */
1050 if(jidx<j_index_end)
1054 j_coord_offsetA = DIM*jnrA;
1056 /* load j atom coordinates */
1057 gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
1060 /* Calculate displacement vector */
1061 dx00 = _mm_sub_pd(ix0,jx0);
1062 dy00 = _mm_sub_pd(iy0,jy0);
1063 dz00 = _mm_sub_pd(iz0,jz0);
1064 dx10 = _mm_sub_pd(ix1,jx0);
1065 dy10 = _mm_sub_pd(iy1,jy0);
1066 dz10 = _mm_sub_pd(iz1,jz0);
1067 dx20 = _mm_sub_pd(ix2,jx0);
1068 dy20 = _mm_sub_pd(iy2,jy0);
1069 dz20 = _mm_sub_pd(iz2,jz0);
1071 /* Calculate squared distance and things based on it */
1072 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
1073 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
1074 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
1076 rinv00 = gmx_mm_invsqrt_pd(rsq00);
1077 rinv10 = gmx_mm_invsqrt_pd(rsq10);
1078 rinv20 = gmx_mm_invsqrt_pd(rsq20);
1080 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
1081 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
1082 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
1084 /* Load parameters for j particles */
1085 jq0 = _mm_load_sd(charge+jnrA+0);
1086 vdwjidx0A = 2*vdwtype[jnrA+0];
1088 fjx0 = _mm_setzero_pd();
1089 fjy0 = _mm_setzero_pd();
1090 fjz0 = _mm_setzero_pd();
1092 /**************************
1093 * CALCULATE INTERACTIONS *
1094 **************************/
1096 if (gmx_mm_any_lt(rsq00,rcutoff2))
1099 r00 = _mm_mul_pd(rsq00,rinv00);
1101 /* Compute parameters for interactions between i and j atoms */
1102 qq00 = _mm_mul_pd(iq0,jq0);
1103 gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
1105 c6grid_00 = gmx_mm_load_1real_pd(vdwgridparam+vdwioffset0+vdwjidx0A);
1107 /* EWALD ELECTROSTATICS */
1109 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1110 ewrt = _mm_mul_pd(r00,ewtabscale);
1111 ewitab = _mm_cvttpd_epi32(ewrt);
1112 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1113 gmx_mm_load_1pair_swizzle_pd(ewtab+gmx_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
1114 felec = _mm_add_pd(_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF),_mm_mul_pd(eweps,ewtabFn));
1115 felec = _mm_mul_pd(_mm_mul_pd(qq00,rinv00),_mm_sub_pd(rinvsq00,felec));
1117 /* Analytical LJ-PME */
1118 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1119 ewcljrsq = _mm_mul_pd(ewclj2,rsq00);
1120 ewclj6 = _mm_mul_pd(ewclj2,_mm_mul_pd(ewclj2,ewclj2));
1121 exponent = gmx_simd_exp_d(ewcljrsq);
1122 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1123 poly = _mm_mul_pd(exponent,_mm_add_pd(_mm_sub_pd(one,ewcljrsq),_mm_mul_pd(_mm_mul_pd(ewcljrsq,ewcljrsq),one_half)));
1124 /* f6A = 6 * C6grid * (1 - poly) */
1125 f6A = _mm_mul_pd(c6grid_00,_mm_sub_pd(one,poly));
1126 /* f6B = C6grid * exponent * beta^6 */
1127 f6B = _mm_mul_pd(_mm_mul_pd(c6grid_00,one_sixth),_mm_mul_pd(exponent,ewclj6));
1128 /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
1129 fvdw = _mm_mul_pd(_mm_add_pd(_mm_mul_pd(_mm_sub_pd(_mm_mul_pd(c12_00,rinvsix),_mm_sub_pd(c6_00,f6A)),rinvsix),f6B),rinvsq00);
1131 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
1133 fscal = _mm_add_pd(felec,fvdw);
1135 fscal = _mm_and_pd(fscal,cutoff_mask);
1137 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1139 /* Calculate temporary vectorial force */
1140 tx = _mm_mul_pd(fscal,dx00);
1141 ty = _mm_mul_pd(fscal,dy00);
1142 tz = _mm_mul_pd(fscal,dz00);
1144 /* Update vectorial force */
1145 fix0 = _mm_add_pd(fix0,tx);
1146 fiy0 = _mm_add_pd(fiy0,ty);
1147 fiz0 = _mm_add_pd(fiz0,tz);
1149 fjx0 = _mm_add_pd(fjx0,tx);
1150 fjy0 = _mm_add_pd(fjy0,ty);
1151 fjz0 = _mm_add_pd(fjz0,tz);
1155 /**************************
1156 * CALCULATE INTERACTIONS *
1157 **************************/
1159 if (gmx_mm_any_lt(rsq10,rcutoff2))
1162 r10 = _mm_mul_pd(rsq10,rinv10);
1164 /* Compute parameters for interactions between i and j atoms */
1165 qq10 = _mm_mul_pd(iq1,jq0);
1167 /* EWALD ELECTROSTATICS */
1169 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1170 ewrt = _mm_mul_pd(r10,ewtabscale);
1171 ewitab = _mm_cvttpd_epi32(ewrt);
1172 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1173 gmx_mm_load_1pair_swizzle_pd(ewtab+gmx_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
1174 felec = _mm_add_pd(_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF),_mm_mul_pd(eweps,ewtabFn));
1175 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
1177 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
1181 fscal = _mm_and_pd(fscal,cutoff_mask);
1183 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1185 /* Calculate temporary vectorial force */
1186 tx = _mm_mul_pd(fscal,dx10);
1187 ty = _mm_mul_pd(fscal,dy10);
1188 tz = _mm_mul_pd(fscal,dz10);
1190 /* Update vectorial force */
1191 fix1 = _mm_add_pd(fix1,tx);
1192 fiy1 = _mm_add_pd(fiy1,ty);
1193 fiz1 = _mm_add_pd(fiz1,tz);
1195 fjx0 = _mm_add_pd(fjx0,tx);
1196 fjy0 = _mm_add_pd(fjy0,ty);
1197 fjz0 = _mm_add_pd(fjz0,tz);
1201 /**************************
1202 * CALCULATE INTERACTIONS *
1203 **************************/
1205 if (gmx_mm_any_lt(rsq20,rcutoff2))
1208 r20 = _mm_mul_pd(rsq20,rinv20);
1210 /* Compute parameters for interactions between i and j atoms */
1211 qq20 = _mm_mul_pd(iq2,jq0);
1213 /* EWALD ELECTROSTATICS */
1215 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1216 ewrt = _mm_mul_pd(r20,ewtabscale);
1217 ewitab = _mm_cvttpd_epi32(ewrt);
1218 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1219 gmx_mm_load_1pair_swizzle_pd(ewtab+gmx_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
1220 felec = _mm_add_pd(_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF),_mm_mul_pd(eweps,ewtabFn));
1221 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
1223 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
1227 fscal = _mm_and_pd(fscal,cutoff_mask);
1229 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1231 /* Calculate temporary vectorial force */
1232 tx = _mm_mul_pd(fscal,dx20);
1233 ty = _mm_mul_pd(fscal,dy20);
1234 tz = _mm_mul_pd(fscal,dz20);
1236 /* Update vectorial force */
1237 fix2 = _mm_add_pd(fix2,tx);
1238 fiy2 = _mm_add_pd(fiy2,ty);
1239 fiz2 = _mm_add_pd(fiz2,tz);
1241 fjx0 = _mm_add_pd(fjx0,tx);
1242 fjy0 = _mm_add_pd(fjy0,ty);
1243 fjz0 = _mm_add_pd(fjz0,tz);
1247 gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
1249 /* Inner loop uses 143 flops */
1252 /* End of innermost loop */
1254 gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1255 f+i_coord_offset,fshift+i_shift_offset);
1257 /* Increment number of inner iterations */
1258 inneriter += j_index_end - j_index_start;
1260 /* Outer loop uses 18 flops */
1263 /* Increment number of outer iterations */
1266 /* Update outer/inner flops */
1268 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*18 + inneriter*143);