2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014,2015,2017,2018, 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 sse2_double kernel generator.
44 #include "../nb_kernel.h"
45 #include "gromacs/gmxlib/nrnb.h"
47 #include "kernelutil_x86_sse2_double.h"
50 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_sse2_double
51 * Electrostatics interaction: Ewald
52 * VdW interaction: LennardJones
53 * Geometry: Water4-Particle
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_sse2_double
58 (t_nblist * gmx_restrict nlist,
59 rvec * gmx_restrict xx,
60 rvec * gmx_restrict ff,
61 struct t_forcerec * gmx_restrict fr,
62 t_mdatoms * gmx_restrict mdatoms,
63 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
64 t_nrnb * gmx_restrict nrnb)
66 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
67 * just 0 for non-waters.
68 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
69 * jnr indices corresponding to data put in the four positions in the SIMD register.
71 int i_shift_offset,i_coord_offset,outeriter,inneriter;
72 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
74 int j_coord_offsetA,j_coord_offsetB;
75 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
77 real *shiftvec,*fshift,*x,*f;
78 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
80 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
82 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
84 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
86 __m128d ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
87 int vdwjidx0A,vdwjidx0B;
88 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
89 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
90 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
91 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
92 __m128d dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
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);
102 __m128d ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
104 __m128d rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
105 real rswitch_scalar,d_scalar;
106 __m128d dummy_mask,cutoff_mask;
107 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
108 __m128d one = _mm_set1_pd(1.0);
109 __m128d two = _mm_set1_pd(2.0);
115 jindex = nlist->jindex;
117 shiftidx = nlist->shift;
119 shiftvec = fr->shift_vec[0];
120 fshift = fr->fshift[0];
121 facel = _mm_set1_pd(fr->ic->epsfac);
122 charge = mdatoms->chargeA;
123 nvdwtype = fr->ntype;
125 vdwtype = mdatoms->typeA;
127 sh_ewald = _mm_set1_pd(fr->ic->sh_ewald);
128 ewtab = fr->ic->tabq_coul_FDV0;
129 ewtabscale = _mm_set1_pd(fr->ic->tabq_scale);
130 ewtabhalfspace = _mm_set1_pd(0.5/fr->ic->tabq_scale);
132 /* Setup water-specific parameters */
133 inr = nlist->iinr[0];
134 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
135 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
136 iq3 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+3]));
137 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
139 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
140 rcutoff_scalar = fr->ic->rcoulomb;
141 rcutoff = _mm_set1_pd(rcutoff_scalar);
142 rcutoff2 = _mm_mul_pd(rcutoff,rcutoff);
144 rswitch_scalar = fr->ic->rcoulomb_switch;
145 rswitch = _mm_set1_pd(rswitch_scalar);
146 /* Setup switch parameters */
147 d_scalar = rcutoff_scalar-rswitch_scalar;
148 d = _mm_set1_pd(d_scalar);
149 swV3 = _mm_set1_pd(-10.0/(d_scalar*d_scalar*d_scalar));
150 swV4 = _mm_set1_pd( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
151 swV5 = _mm_set1_pd( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
152 swF2 = _mm_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar));
153 swF3 = _mm_set1_pd( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
154 swF4 = _mm_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
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_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
180 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
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();
191 fix3 = _mm_setzero_pd();
192 fiy3 = _mm_setzero_pd();
193 fiz3 = _mm_setzero_pd();
195 /* Reset potential sums */
196 velecsum = _mm_setzero_pd();
197 vvdwsum = _mm_setzero_pd();
199 /* Start inner kernel loop */
200 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
203 /* Get j neighbor index, and coordinate index */
206 j_coord_offsetA = DIM*jnrA;
207 j_coord_offsetB = DIM*jnrB;
209 /* load j atom coordinates */
210 gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
213 /* Calculate displacement vector */
214 dx00 = _mm_sub_pd(ix0,jx0);
215 dy00 = _mm_sub_pd(iy0,jy0);
216 dz00 = _mm_sub_pd(iz0,jz0);
217 dx10 = _mm_sub_pd(ix1,jx0);
218 dy10 = _mm_sub_pd(iy1,jy0);
219 dz10 = _mm_sub_pd(iz1,jz0);
220 dx20 = _mm_sub_pd(ix2,jx0);
221 dy20 = _mm_sub_pd(iy2,jy0);
222 dz20 = _mm_sub_pd(iz2,jz0);
223 dx30 = _mm_sub_pd(ix3,jx0);
224 dy30 = _mm_sub_pd(iy3,jy0);
225 dz30 = _mm_sub_pd(iz3,jz0);
227 /* Calculate squared distance and things based on it */
228 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
229 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
230 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
231 rsq30 = gmx_mm_calc_rsq_pd(dx30,dy30,dz30);
233 rinv00 = sse2_invsqrt_d(rsq00);
234 rinv10 = sse2_invsqrt_d(rsq10);
235 rinv20 = sse2_invsqrt_d(rsq20);
236 rinv30 = sse2_invsqrt_d(rsq30);
238 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
239 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
240 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
241 rinvsq30 = _mm_mul_pd(rinv30,rinv30);
243 /* Load parameters for j particles */
244 jq0 = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
245 vdwjidx0A = 2*vdwtype[jnrA+0];
246 vdwjidx0B = 2*vdwtype[jnrB+0];
248 fjx0 = _mm_setzero_pd();
249 fjy0 = _mm_setzero_pd();
250 fjz0 = _mm_setzero_pd();
252 /**************************
253 * CALCULATE INTERACTIONS *
254 **************************/
256 if (gmx_mm_any_lt(rsq00,rcutoff2))
259 r00 = _mm_mul_pd(rsq00,rinv00);
261 /* Compute parameters for interactions between i and j atoms */
262 gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
263 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
265 /* LENNARD-JONES DISPERSION/REPULSION */
267 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
268 vvdw6 = _mm_mul_pd(c6_00,rinvsix);
269 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
270 vvdw = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
271 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
273 d = _mm_sub_pd(r00,rswitch);
274 d = _mm_max_pd(d,_mm_setzero_pd());
275 d2 = _mm_mul_pd(d,d);
276 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
278 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
280 /* Evaluate switch function */
281 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
282 fvdw = _mm_sub_pd( _mm_mul_pd(fvdw,sw) , _mm_mul_pd(rinv00,_mm_mul_pd(vvdw,dsw)) );
283 vvdw = _mm_mul_pd(vvdw,sw);
284 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
286 /* Update potential sum for this i atom from the interaction with this j atom. */
287 vvdw = _mm_and_pd(vvdw,cutoff_mask);
288 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
292 fscal = _mm_and_pd(fscal,cutoff_mask);
294 /* Calculate temporary vectorial force */
295 tx = _mm_mul_pd(fscal,dx00);
296 ty = _mm_mul_pd(fscal,dy00);
297 tz = _mm_mul_pd(fscal,dz00);
299 /* Update vectorial force */
300 fix0 = _mm_add_pd(fix0,tx);
301 fiy0 = _mm_add_pd(fiy0,ty);
302 fiz0 = _mm_add_pd(fiz0,tz);
304 fjx0 = _mm_add_pd(fjx0,tx);
305 fjy0 = _mm_add_pd(fjy0,ty);
306 fjz0 = _mm_add_pd(fjz0,tz);
310 /**************************
311 * CALCULATE INTERACTIONS *
312 **************************/
314 if (gmx_mm_any_lt(rsq10,rcutoff2))
317 r10 = _mm_mul_pd(rsq10,rinv10);
319 /* Compute parameters for interactions between i and j atoms */
320 qq10 = _mm_mul_pd(iq1,jq0);
322 /* EWALD ELECTROSTATICS */
324 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
325 ewrt = _mm_mul_pd(r10,ewtabscale);
326 ewitab = _mm_cvttpd_epi32(ewrt);
327 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
328 ewitab = _mm_slli_epi32(ewitab,2);
329 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
330 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
331 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
332 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
333 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
334 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
335 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
336 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
337 velec = _mm_mul_pd(qq10,_mm_sub_pd(rinv10,velec));
338 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
340 d = _mm_sub_pd(r10,rswitch);
341 d = _mm_max_pd(d,_mm_setzero_pd());
342 d2 = _mm_mul_pd(d,d);
343 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
345 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
347 /* Evaluate switch function */
348 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
349 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv10,_mm_mul_pd(velec,dsw)) );
350 velec = _mm_mul_pd(velec,sw);
351 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
353 /* Update potential sum for this i atom from the interaction with this j atom. */
354 velec = _mm_and_pd(velec,cutoff_mask);
355 velecsum = _mm_add_pd(velecsum,velec);
359 fscal = _mm_and_pd(fscal,cutoff_mask);
361 /* Calculate temporary vectorial force */
362 tx = _mm_mul_pd(fscal,dx10);
363 ty = _mm_mul_pd(fscal,dy10);
364 tz = _mm_mul_pd(fscal,dz10);
366 /* Update vectorial force */
367 fix1 = _mm_add_pd(fix1,tx);
368 fiy1 = _mm_add_pd(fiy1,ty);
369 fiz1 = _mm_add_pd(fiz1,tz);
371 fjx0 = _mm_add_pd(fjx0,tx);
372 fjy0 = _mm_add_pd(fjy0,ty);
373 fjz0 = _mm_add_pd(fjz0,tz);
377 /**************************
378 * CALCULATE INTERACTIONS *
379 **************************/
381 if (gmx_mm_any_lt(rsq20,rcutoff2))
384 r20 = _mm_mul_pd(rsq20,rinv20);
386 /* Compute parameters for interactions between i and j atoms */
387 qq20 = _mm_mul_pd(iq2,jq0);
389 /* EWALD ELECTROSTATICS */
391 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
392 ewrt = _mm_mul_pd(r20,ewtabscale);
393 ewitab = _mm_cvttpd_epi32(ewrt);
394 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
395 ewitab = _mm_slli_epi32(ewitab,2);
396 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
397 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
398 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
399 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
400 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
401 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
402 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
403 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
404 velec = _mm_mul_pd(qq20,_mm_sub_pd(rinv20,velec));
405 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
407 d = _mm_sub_pd(r20,rswitch);
408 d = _mm_max_pd(d,_mm_setzero_pd());
409 d2 = _mm_mul_pd(d,d);
410 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
412 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
414 /* Evaluate switch function */
415 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
416 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv20,_mm_mul_pd(velec,dsw)) );
417 velec = _mm_mul_pd(velec,sw);
418 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
420 /* Update potential sum for this i atom from the interaction with this j atom. */
421 velec = _mm_and_pd(velec,cutoff_mask);
422 velecsum = _mm_add_pd(velecsum,velec);
426 fscal = _mm_and_pd(fscal,cutoff_mask);
428 /* Calculate temporary vectorial force */
429 tx = _mm_mul_pd(fscal,dx20);
430 ty = _mm_mul_pd(fscal,dy20);
431 tz = _mm_mul_pd(fscal,dz20);
433 /* Update vectorial force */
434 fix2 = _mm_add_pd(fix2,tx);
435 fiy2 = _mm_add_pd(fiy2,ty);
436 fiz2 = _mm_add_pd(fiz2,tz);
438 fjx0 = _mm_add_pd(fjx0,tx);
439 fjy0 = _mm_add_pd(fjy0,ty);
440 fjz0 = _mm_add_pd(fjz0,tz);
444 /**************************
445 * CALCULATE INTERACTIONS *
446 **************************/
448 if (gmx_mm_any_lt(rsq30,rcutoff2))
451 r30 = _mm_mul_pd(rsq30,rinv30);
453 /* Compute parameters for interactions between i and j atoms */
454 qq30 = _mm_mul_pd(iq3,jq0);
456 /* EWALD ELECTROSTATICS */
458 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
459 ewrt = _mm_mul_pd(r30,ewtabscale);
460 ewitab = _mm_cvttpd_epi32(ewrt);
461 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
462 ewitab = _mm_slli_epi32(ewitab,2);
463 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
464 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
465 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
466 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
467 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
468 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
469 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
470 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
471 velec = _mm_mul_pd(qq30,_mm_sub_pd(rinv30,velec));
472 felec = _mm_mul_pd(_mm_mul_pd(qq30,rinv30),_mm_sub_pd(rinvsq30,felec));
474 d = _mm_sub_pd(r30,rswitch);
475 d = _mm_max_pd(d,_mm_setzero_pd());
476 d2 = _mm_mul_pd(d,d);
477 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
479 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
481 /* Evaluate switch function */
482 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
483 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv30,_mm_mul_pd(velec,dsw)) );
484 velec = _mm_mul_pd(velec,sw);
485 cutoff_mask = _mm_cmplt_pd(rsq30,rcutoff2);
487 /* Update potential sum for this i atom from the interaction with this j atom. */
488 velec = _mm_and_pd(velec,cutoff_mask);
489 velecsum = _mm_add_pd(velecsum,velec);
493 fscal = _mm_and_pd(fscal,cutoff_mask);
495 /* Calculate temporary vectorial force */
496 tx = _mm_mul_pd(fscal,dx30);
497 ty = _mm_mul_pd(fscal,dy30);
498 tz = _mm_mul_pd(fscal,dz30);
500 /* Update vectorial force */
501 fix3 = _mm_add_pd(fix3,tx);
502 fiy3 = _mm_add_pd(fiy3,ty);
503 fiz3 = _mm_add_pd(fiz3,tz);
505 fjx0 = _mm_add_pd(fjx0,tx);
506 fjy0 = _mm_add_pd(fjy0,ty);
507 fjz0 = _mm_add_pd(fjz0,tz);
511 gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
513 /* Inner loop uses 257 flops */
520 j_coord_offsetA = DIM*jnrA;
522 /* load j atom coordinates */
523 gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
526 /* Calculate displacement vector */
527 dx00 = _mm_sub_pd(ix0,jx0);
528 dy00 = _mm_sub_pd(iy0,jy0);
529 dz00 = _mm_sub_pd(iz0,jz0);
530 dx10 = _mm_sub_pd(ix1,jx0);
531 dy10 = _mm_sub_pd(iy1,jy0);
532 dz10 = _mm_sub_pd(iz1,jz0);
533 dx20 = _mm_sub_pd(ix2,jx0);
534 dy20 = _mm_sub_pd(iy2,jy0);
535 dz20 = _mm_sub_pd(iz2,jz0);
536 dx30 = _mm_sub_pd(ix3,jx0);
537 dy30 = _mm_sub_pd(iy3,jy0);
538 dz30 = _mm_sub_pd(iz3,jz0);
540 /* Calculate squared distance and things based on it */
541 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
542 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
543 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
544 rsq30 = gmx_mm_calc_rsq_pd(dx30,dy30,dz30);
546 rinv00 = sse2_invsqrt_d(rsq00);
547 rinv10 = sse2_invsqrt_d(rsq10);
548 rinv20 = sse2_invsqrt_d(rsq20);
549 rinv30 = sse2_invsqrt_d(rsq30);
551 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
552 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
553 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
554 rinvsq30 = _mm_mul_pd(rinv30,rinv30);
556 /* Load parameters for j particles */
557 jq0 = _mm_load_sd(charge+jnrA+0);
558 vdwjidx0A = 2*vdwtype[jnrA+0];
560 fjx0 = _mm_setzero_pd();
561 fjy0 = _mm_setzero_pd();
562 fjz0 = _mm_setzero_pd();
564 /**************************
565 * CALCULATE INTERACTIONS *
566 **************************/
568 if (gmx_mm_any_lt(rsq00,rcutoff2))
571 r00 = _mm_mul_pd(rsq00,rinv00);
573 /* Compute parameters for interactions between i and j atoms */
574 gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
576 /* LENNARD-JONES DISPERSION/REPULSION */
578 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
579 vvdw6 = _mm_mul_pd(c6_00,rinvsix);
580 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
581 vvdw = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
582 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
584 d = _mm_sub_pd(r00,rswitch);
585 d = _mm_max_pd(d,_mm_setzero_pd());
586 d2 = _mm_mul_pd(d,d);
587 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
589 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
591 /* Evaluate switch function */
592 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
593 fvdw = _mm_sub_pd( _mm_mul_pd(fvdw,sw) , _mm_mul_pd(rinv00,_mm_mul_pd(vvdw,dsw)) );
594 vvdw = _mm_mul_pd(vvdw,sw);
595 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
597 /* Update potential sum for this i atom from the interaction with this j atom. */
598 vvdw = _mm_and_pd(vvdw,cutoff_mask);
599 vvdw = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
600 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
604 fscal = _mm_and_pd(fscal,cutoff_mask);
606 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
608 /* Calculate temporary vectorial force */
609 tx = _mm_mul_pd(fscal,dx00);
610 ty = _mm_mul_pd(fscal,dy00);
611 tz = _mm_mul_pd(fscal,dz00);
613 /* Update vectorial force */
614 fix0 = _mm_add_pd(fix0,tx);
615 fiy0 = _mm_add_pd(fiy0,ty);
616 fiz0 = _mm_add_pd(fiz0,tz);
618 fjx0 = _mm_add_pd(fjx0,tx);
619 fjy0 = _mm_add_pd(fjy0,ty);
620 fjz0 = _mm_add_pd(fjz0,tz);
624 /**************************
625 * CALCULATE INTERACTIONS *
626 **************************/
628 if (gmx_mm_any_lt(rsq10,rcutoff2))
631 r10 = _mm_mul_pd(rsq10,rinv10);
633 /* Compute parameters for interactions between i and j atoms */
634 qq10 = _mm_mul_pd(iq1,jq0);
636 /* EWALD ELECTROSTATICS */
638 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
639 ewrt = _mm_mul_pd(r10,ewtabscale);
640 ewitab = _mm_cvttpd_epi32(ewrt);
641 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
642 ewitab = _mm_slli_epi32(ewitab,2);
643 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
644 ewtabD = _mm_setzero_pd();
645 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
646 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
647 ewtabFn = _mm_setzero_pd();
648 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
649 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
650 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
651 velec = _mm_mul_pd(qq10,_mm_sub_pd(rinv10,velec));
652 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
654 d = _mm_sub_pd(r10,rswitch);
655 d = _mm_max_pd(d,_mm_setzero_pd());
656 d2 = _mm_mul_pd(d,d);
657 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
659 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
661 /* Evaluate switch function */
662 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
663 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv10,_mm_mul_pd(velec,dsw)) );
664 velec = _mm_mul_pd(velec,sw);
665 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
667 /* Update potential sum for this i atom from the interaction with this j atom. */
668 velec = _mm_and_pd(velec,cutoff_mask);
669 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
670 velecsum = _mm_add_pd(velecsum,velec);
674 fscal = _mm_and_pd(fscal,cutoff_mask);
676 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
678 /* Calculate temporary vectorial force */
679 tx = _mm_mul_pd(fscal,dx10);
680 ty = _mm_mul_pd(fscal,dy10);
681 tz = _mm_mul_pd(fscal,dz10);
683 /* Update vectorial force */
684 fix1 = _mm_add_pd(fix1,tx);
685 fiy1 = _mm_add_pd(fiy1,ty);
686 fiz1 = _mm_add_pd(fiz1,tz);
688 fjx0 = _mm_add_pd(fjx0,tx);
689 fjy0 = _mm_add_pd(fjy0,ty);
690 fjz0 = _mm_add_pd(fjz0,tz);
694 /**************************
695 * CALCULATE INTERACTIONS *
696 **************************/
698 if (gmx_mm_any_lt(rsq20,rcutoff2))
701 r20 = _mm_mul_pd(rsq20,rinv20);
703 /* Compute parameters for interactions between i and j atoms */
704 qq20 = _mm_mul_pd(iq2,jq0);
706 /* EWALD ELECTROSTATICS */
708 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
709 ewrt = _mm_mul_pd(r20,ewtabscale);
710 ewitab = _mm_cvttpd_epi32(ewrt);
711 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
712 ewitab = _mm_slli_epi32(ewitab,2);
713 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
714 ewtabD = _mm_setzero_pd();
715 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
716 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
717 ewtabFn = _mm_setzero_pd();
718 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
719 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
720 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
721 velec = _mm_mul_pd(qq20,_mm_sub_pd(rinv20,velec));
722 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
724 d = _mm_sub_pd(r20,rswitch);
725 d = _mm_max_pd(d,_mm_setzero_pd());
726 d2 = _mm_mul_pd(d,d);
727 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
729 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
731 /* Evaluate switch function */
732 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
733 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv20,_mm_mul_pd(velec,dsw)) );
734 velec = _mm_mul_pd(velec,sw);
735 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
737 /* Update potential sum for this i atom from the interaction with this j atom. */
738 velec = _mm_and_pd(velec,cutoff_mask);
739 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
740 velecsum = _mm_add_pd(velecsum,velec);
744 fscal = _mm_and_pd(fscal,cutoff_mask);
746 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
748 /* Calculate temporary vectorial force */
749 tx = _mm_mul_pd(fscal,dx20);
750 ty = _mm_mul_pd(fscal,dy20);
751 tz = _mm_mul_pd(fscal,dz20);
753 /* Update vectorial force */
754 fix2 = _mm_add_pd(fix2,tx);
755 fiy2 = _mm_add_pd(fiy2,ty);
756 fiz2 = _mm_add_pd(fiz2,tz);
758 fjx0 = _mm_add_pd(fjx0,tx);
759 fjy0 = _mm_add_pd(fjy0,ty);
760 fjz0 = _mm_add_pd(fjz0,tz);
764 /**************************
765 * CALCULATE INTERACTIONS *
766 **************************/
768 if (gmx_mm_any_lt(rsq30,rcutoff2))
771 r30 = _mm_mul_pd(rsq30,rinv30);
773 /* Compute parameters for interactions between i and j atoms */
774 qq30 = _mm_mul_pd(iq3,jq0);
776 /* EWALD ELECTROSTATICS */
778 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
779 ewrt = _mm_mul_pd(r30,ewtabscale);
780 ewitab = _mm_cvttpd_epi32(ewrt);
781 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
782 ewitab = _mm_slli_epi32(ewitab,2);
783 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
784 ewtabD = _mm_setzero_pd();
785 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
786 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
787 ewtabFn = _mm_setzero_pd();
788 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
789 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
790 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
791 velec = _mm_mul_pd(qq30,_mm_sub_pd(rinv30,velec));
792 felec = _mm_mul_pd(_mm_mul_pd(qq30,rinv30),_mm_sub_pd(rinvsq30,felec));
794 d = _mm_sub_pd(r30,rswitch);
795 d = _mm_max_pd(d,_mm_setzero_pd());
796 d2 = _mm_mul_pd(d,d);
797 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
799 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
801 /* Evaluate switch function */
802 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
803 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv30,_mm_mul_pd(velec,dsw)) );
804 velec = _mm_mul_pd(velec,sw);
805 cutoff_mask = _mm_cmplt_pd(rsq30,rcutoff2);
807 /* Update potential sum for this i atom from the interaction with this j atom. */
808 velec = _mm_and_pd(velec,cutoff_mask);
809 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
810 velecsum = _mm_add_pd(velecsum,velec);
814 fscal = _mm_and_pd(fscal,cutoff_mask);
816 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
818 /* Calculate temporary vectorial force */
819 tx = _mm_mul_pd(fscal,dx30);
820 ty = _mm_mul_pd(fscal,dy30);
821 tz = _mm_mul_pd(fscal,dz30);
823 /* Update vectorial force */
824 fix3 = _mm_add_pd(fix3,tx);
825 fiy3 = _mm_add_pd(fiy3,ty);
826 fiz3 = _mm_add_pd(fiz3,tz);
828 fjx0 = _mm_add_pd(fjx0,tx);
829 fjy0 = _mm_add_pd(fjy0,ty);
830 fjz0 = _mm_add_pd(fjz0,tz);
834 gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
836 /* Inner loop uses 257 flops */
839 /* End of innermost loop */
841 gmx_mm_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
842 f+i_coord_offset,fshift+i_shift_offset);
845 /* Update potential energies */
846 gmx_mm_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
847 gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
849 /* Increment number of inner iterations */
850 inneriter += j_index_end - j_index_start;
852 /* Outer loop uses 26 flops */
855 /* Increment number of outer iterations */
858 /* Update outer/inner flops */
860 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*26 + inneriter*257);
863 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_sse2_double
864 * Electrostatics interaction: Ewald
865 * VdW interaction: LennardJones
866 * Geometry: Water4-Particle
867 * Calculate force/pot: Force
870 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_sse2_double
871 (t_nblist * gmx_restrict nlist,
872 rvec * gmx_restrict xx,
873 rvec * gmx_restrict ff,
874 struct t_forcerec * gmx_restrict fr,
875 t_mdatoms * gmx_restrict mdatoms,
876 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
877 t_nrnb * gmx_restrict nrnb)
879 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
880 * just 0 for non-waters.
881 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
882 * jnr indices corresponding to data put in the four positions in the SIMD register.
884 int i_shift_offset,i_coord_offset,outeriter,inneriter;
885 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
887 int j_coord_offsetA,j_coord_offsetB;
888 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
890 real *shiftvec,*fshift,*x,*f;
891 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
893 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
895 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
897 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
899 __m128d ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
900 int vdwjidx0A,vdwjidx0B;
901 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
902 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
903 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
904 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
905 __m128d dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
906 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
909 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
912 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
913 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
915 __m128d ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
917 __m128d rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
918 real rswitch_scalar,d_scalar;
919 __m128d dummy_mask,cutoff_mask;
920 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
921 __m128d one = _mm_set1_pd(1.0);
922 __m128d two = _mm_set1_pd(2.0);
928 jindex = nlist->jindex;
930 shiftidx = nlist->shift;
932 shiftvec = fr->shift_vec[0];
933 fshift = fr->fshift[0];
934 facel = _mm_set1_pd(fr->ic->epsfac);
935 charge = mdatoms->chargeA;
936 nvdwtype = fr->ntype;
938 vdwtype = mdatoms->typeA;
940 sh_ewald = _mm_set1_pd(fr->ic->sh_ewald);
941 ewtab = fr->ic->tabq_coul_FDV0;
942 ewtabscale = _mm_set1_pd(fr->ic->tabq_scale);
943 ewtabhalfspace = _mm_set1_pd(0.5/fr->ic->tabq_scale);
945 /* Setup water-specific parameters */
946 inr = nlist->iinr[0];
947 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
948 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
949 iq3 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+3]));
950 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
952 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
953 rcutoff_scalar = fr->ic->rcoulomb;
954 rcutoff = _mm_set1_pd(rcutoff_scalar);
955 rcutoff2 = _mm_mul_pd(rcutoff,rcutoff);
957 rswitch_scalar = fr->ic->rcoulomb_switch;
958 rswitch = _mm_set1_pd(rswitch_scalar);
959 /* Setup switch parameters */
960 d_scalar = rcutoff_scalar-rswitch_scalar;
961 d = _mm_set1_pd(d_scalar);
962 swV3 = _mm_set1_pd(-10.0/(d_scalar*d_scalar*d_scalar));
963 swV4 = _mm_set1_pd( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
964 swV5 = _mm_set1_pd( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
965 swF2 = _mm_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar));
966 swF3 = _mm_set1_pd( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
967 swF4 = _mm_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
969 /* Avoid stupid compiler warnings */
977 /* Start outer loop over neighborlists */
978 for(iidx=0; iidx<nri; iidx++)
980 /* Load shift vector for this list */
981 i_shift_offset = DIM*shiftidx[iidx];
983 /* Load limits for loop over neighbors */
984 j_index_start = jindex[iidx];
985 j_index_end = jindex[iidx+1];
987 /* Get outer coordinate index */
989 i_coord_offset = DIM*inr;
991 /* Load i particle coords and add shift vector */
992 gmx_mm_load_shift_and_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
993 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
995 fix0 = _mm_setzero_pd();
996 fiy0 = _mm_setzero_pd();
997 fiz0 = _mm_setzero_pd();
998 fix1 = _mm_setzero_pd();
999 fiy1 = _mm_setzero_pd();
1000 fiz1 = _mm_setzero_pd();
1001 fix2 = _mm_setzero_pd();
1002 fiy2 = _mm_setzero_pd();
1003 fiz2 = _mm_setzero_pd();
1004 fix3 = _mm_setzero_pd();
1005 fiy3 = _mm_setzero_pd();
1006 fiz3 = _mm_setzero_pd();
1008 /* Start inner kernel loop */
1009 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1012 /* Get j neighbor index, and coordinate index */
1014 jnrB = jjnr[jidx+1];
1015 j_coord_offsetA = DIM*jnrA;
1016 j_coord_offsetB = DIM*jnrB;
1018 /* load j atom coordinates */
1019 gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
1022 /* Calculate displacement vector */
1023 dx00 = _mm_sub_pd(ix0,jx0);
1024 dy00 = _mm_sub_pd(iy0,jy0);
1025 dz00 = _mm_sub_pd(iz0,jz0);
1026 dx10 = _mm_sub_pd(ix1,jx0);
1027 dy10 = _mm_sub_pd(iy1,jy0);
1028 dz10 = _mm_sub_pd(iz1,jz0);
1029 dx20 = _mm_sub_pd(ix2,jx0);
1030 dy20 = _mm_sub_pd(iy2,jy0);
1031 dz20 = _mm_sub_pd(iz2,jz0);
1032 dx30 = _mm_sub_pd(ix3,jx0);
1033 dy30 = _mm_sub_pd(iy3,jy0);
1034 dz30 = _mm_sub_pd(iz3,jz0);
1036 /* Calculate squared distance and things based on it */
1037 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
1038 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
1039 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
1040 rsq30 = gmx_mm_calc_rsq_pd(dx30,dy30,dz30);
1042 rinv00 = sse2_invsqrt_d(rsq00);
1043 rinv10 = sse2_invsqrt_d(rsq10);
1044 rinv20 = sse2_invsqrt_d(rsq20);
1045 rinv30 = sse2_invsqrt_d(rsq30);
1047 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
1048 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
1049 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
1050 rinvsq30 = _mm_mul_pd(rinv30,rinv30);
1052 /* Load parameters for j particles */
1053 jq0 = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
1054 vdwjidx0A = 2*vdwtype[jnrA+0];
1055 vdwjidx0B = 2*vdwtype[jnrB+0];
1057 fjx0 = _mm_setzero_pd();
1058 fjy0 = _mm_setzero_pd();
1059 fjz0 = _mm_setzero_pd();
1061 /**************************
1062 * CALCULATE INTERACTIONS *
1063 **************************/
1065 if (gmx_mm_any_lt(rsq00,rcutoff2))
1068 r00 = _mm_mul_pd(rsq00,rinv00);
1070 /* Compute parameters for interactions between i and j atoms */
1071 gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
1072 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
1074 /* LENNARD-JONES DISPERSION/REPULSION */
1076 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1077 vvdw6 = _mm_mul_pd(c6_00,rinvsix);
1078 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
1079 vvdw = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
1080 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
1082 d = _mm_sub_pd(r00,rswitch);
1083 d = _mm_max_pd(d,_mm_setzero_pd());
1084 d2 = _mm_mul_pd(d,d);
1085 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1087 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1089 /* Evaluate switch function */
1090 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1091 fvdw = _mm_sub_pd( _mm_mul_pd(fvdw,sw) , _mm_mul_pd(rinv00,_mm_mul_pd(vvdw,dsw)) );
1092 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
1096 fscal = _mm_and_pd(fscal,cutoff_mask);
1098 /* Calculate temporary vectorial force */
1099 tx = _mm_mul_pd(fscal,dx00);
1100 ty = _mm_mul_pd(fscal,dy00);
1101 tz = _mm_mul_pd(fscal,dz00);
1103 /* Update vectorial force */
1104 fix0 = _mm_add_pd(fix0,tx);
1105 fiy0 = _mm_add_pd(fiy0,ty);
1106 fiz0 = _mm_add_pd(fiz0,tz);
1108 fjx0 = _mm_add_pd(fjx0,tx);
1109 fjy0 = _mm_add_pd(fjy0,ty);
1110 fjz0 = _mm_add_pd(fjz0,tz);
1114 /**************************
1115 * CALCULATE INTERACTIONS *
1116 **************************/
1118 if (gmx_mm_any_lt(rsq10,rcutoff2))
1121 r10 = _mm_mul_pd(rsq10,rinv10);
1123 /* Compute parameters for interactions between i and j atoms */
1124 qq10 = _mm_mul_pd(iq1,jq0);
1126 /* EWALD ELECTROSTATICS */
1128 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1129 ewrt = _mm_mul_pd(r10,ewtabscale);
1130 ewitab = _mm_cvttpd_epi32(ewrt);
1131 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
1132 ewitab = _mm_slli_epi32(ewitab,2);
1133 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1134 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1135 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1136 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
1137 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
1138 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1139 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
1140 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
1141 velec = _mm_mul_pd(qq10,_mm_sub_pd(rinv10,velec));
1142 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
1144 d = _mm_sub_pd(r10,rswitch);
1145 d = _mm_max_pd(d,_mm_setzero_pd());
1146 d2 = _mm_mul_pd(d,d);
1147 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1149 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1151 /* Evaluate switch function */
1152 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1153 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv10,_mm_mul_pd(velec,dsw)) );
1154 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
1158 fscal = _mm_and_pd(fscal,cutoff_mask);
1160 /* Calculate temporary vectorial force */
1161 tx = _mm_mul_pd(fscal,dx10);
1162 ty = _mm_mul_pd(fscal,dy10);
1163 tz = _mm_mul_pd(fscal,dz10);
1165 /* Update vectorial force */
1166 fix1 = _mm_add_pd(fix1,tx);
1167 fiy1 = _mm_add_pd(fiy1,ty);
1168 fiz1 = _mm_add_pd(fiz1,tz);
1170 fjx0 = _mm_add_pd(fjx0,tx);
1171 fjy0 = _mm_add_pd(fjy0,ty);
1172 fjz0 = _mm_add_pd(fjz0,tz);
1176 /**************************
1177 * CALCULATE INTERACTIONS *
1178 **************************/
1180 if (gmx_mm_any_lt(rsq20,rcutoff2))
1183 r20 = _mm_mul_pd(rsq20,rinv20);
1185 /* Compute parameters for interactions between i and j atoms */
1186 qq20 = _mm_mul_pd(iq2,jq0);
1188 /* EWALD ELECTROSTATICS */
1190 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1191 ewrt = _mm_mul_pd(r20,ewtabscale);
1192 ewitab = _mm_cvttpd_epi32(ewrt);
1193 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
1194 ewitab = _mm_slli_epi32(ewitab,2);
1195 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1196 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1197 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1198 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
1199 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
1200 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1201 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
1202 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
1203 velec = _mm_mul_pd(qq20,_mm_sub_pd(rinv20,velec));
1204 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
1206 d = _mm_sub_pd(r20,rswitch);
1207 d = _mm_max_pd(d,_mm_setzero_pd());
1208 d2 = _mm_mul_pd(d,d);
1209 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1211 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1213 /* Evaluate switch function */
1214 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1215 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv20,_mm_mul_pd(velec,dsw)) );
1216 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
1220 fscal = _mm_and_pd(fscal,cutoff_mask);
1222 /* Calculate temporary vectorial force */
1223 tx = _mm_mul_pd(fscal,dx20);
1224 ty = _mm_mul_pd(fscal,dy20);
1225 tz = _mm_mul_pd(fscal,dz20);
1227 /* Update vectorial force */
1228 fix2 = _mm_add_pd(fix2,tx);
1229 fiy2 = _mm_add_pd(fiy2,ty);
1230 fiz2 = _mm_add_pd(fiz2,tz);
1232 fjx0 = _mm_add_pd(fjx0,tx);
1233 fjy0 = _mm_add_pd(fjy0,ty);
1234 fjz0 = _mm_add_pd(fjz0,tz);
1238 /**************************
1239 * CALCULATE INTERACTIONS *
1240 **************************/
1242 if (gmx_mm_any_lt(rsq30,rcutoff2))
1245 r30 = _mm_mul_pd(rsq30,rinv30);
1247 /* Compute parameters for interactions between i and j atoms */
1248 qq30 = _mm_mul_pd(iq3,jq0);
1250 /* EWALD ELECTROSTATICS */
1252 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1253 ewrt = _mm_mul_pd(r30,ewtabscale);
1254 ewitab = _mm_cvttpd_epi32(ewrt);
1255 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
1256 ewitab = _mm_slli_epi32(ewitab,2);
1257 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1258 ewtabD = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1259 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1260 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
1261 ewtabFn = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,1) +2);
1262 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1263 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
1264 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
1265 velec = _mm_mul_pd(qq30,_mm_sub_pd(rinv30,velec));
1266 felec = _mm_mul_pd(_mm_mul_pd(qq30,rinv30),_mm_sub_pd(rinvsq30,felec));
1268 d = _mm_sub_pd(r30,rswitch);
1269 d = _mm_max_pd(d,_mm_setzero_pd());
1270 d2 = _mm_mul_pd(d,d);
1271 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1273 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1275 /* Evaluate switch function */
1276 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1277 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv30,_mm_mul_pd(velec,dsw)) );
1278 cutoff_mask = _mm_cmplt_pd(rsq30,rcutoff2);
1282 fscal = _mm_and_pd(fscal,cutoff_mask);
1284 /* Calculate temporary vectorial force */
1285 tx = _mm_mul_pd(fscal,dx30);
1286 ty = _mm_mul_pd(fscal,dy30);
1287 tz = _mm_mul_pd(fscal,dz30);
1289 /* Update vectorial force */
1290 fix3 = _mm_add_pd(fix3,tx);
1291 fiy3 = _mm_add_pd(fiy3,ty);
1292 fiz3 = _mm_add_pd(fiz3,tz);
1294 fjx0 = _mm_add_pd(fjx0,tx);
1295 fjy0 = _mm_add_pd(fjy0,ty);
1296 fjz0 = _mm_add_pd(fjz0,tz);
1300 gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
1302 /* Inner loop uses 245 flops */
1305 if(jidx<j_index_end)
1309 j_coord_offsetA = DIM*jnrA;
1311 /* load j atom coordinates */
1312 gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
1315 /* Calculate displacement vector */
1316 dx00 = _mm_sub_pd(ix0,jx0);
1317 dy00 = _mm_sub_pd(iy0,jy0);
1318 dz00 = _mm_sub_pd(iz0,jz0);
1319 dx10 = _mm_sub_pd(ix1,jx0);
1320 dy10 = _mm_sub_pd(iy1,jy0);
1321 dz10 = _mm_sub_pd(iz1,jz0);
1322 dx20 = _mm_sub_pd(ix2,jx0);
1323 dy20 = _mm_sub_pd(iy2,jy0);
1324 dz20 = _mm_sub_pd(iz2,jz0);
1325 dx30 = _mm_sub_pd(ix3,jx0);
1326 dy30 = _mm_sub_pd(iy3,jy0);
1327 dz30 = _mm_sub_pd(iz3,jz0);
1329 /* Calculate squared distance and things based on it */
1330 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
1331 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
1332 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
1333 rsq30 = gmx_mm_calc_rsq_pd(dx30,dy30,dz30);
1335 rinv00 = sse2_invsqrt_d(rsq00);
1336 rinv10 = sse2_invsqrt_d(rsq10);
1337 rinv20 = sse2_invsqrt_d(rsq20);
1338 rinv30 = sse2_invsqrt_d(rsq30);
1340 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
1341 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
1342 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
1343 rinvsq30 = _mm_mul_pd(rinv30,rinv30);
1345 /* Load parameters for j particles */
1346 jq0 = _mm_load_sd(charge+jnrA+0);
1347 vdwjidx0A = 2*vdwtype[jnrA+0];
1349 fjx0 = _mm_setzero_pd();
1350 fjy0 = _mm_setzero_pd();
1351 fjz0 = _mm_setzero_pd();
1353 /**************************
1354 * CALCULATE INTERACTIONS *
1355 **************************/
1357 if (gmx_mm_any_lt(rsq00,rcutoff2))
1360 r00 = _mm_mul_pd(rsq00,rinv00);
1362 /* Compute parameters for interactions between i and j atoms */
1363 gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
1365 /* LENNARD-JONES DISPERSION/REPULSION */
1367 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1368 vvdw6 = _mm_mul_pd(c6_00,rinvsix);
1369 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
1370 vvdw = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
1371 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
1373 d = _mm_sub_pd(r00,rswitch);
1374 d = _mm_max_pd(d,_mm_setzero_pd());
1375 d2 = _mm_mul_pd(d,d);
1376 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1378 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1380 /* Evaluate switch function */
1381 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1382 fvdw = _mm_sub_pd( _mm_mul_pd(fvdw,sw) , _mm_mul_pd(rinv00,_mm_mul_pd(vvdw,dsw)) );
1383 cutoff_mask = _mm_cmplt_pd(rsq00,rcutoff2);
1387 fscal = _mm_and_pd(fscal,cutoff_mask);
1389 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1391 /* Calculate temporary vectorial force */
1392 tx = _mm_mul_pd(fscal,dx00);
1393 ty = _mm_mul_pd(fscal,dy00);
1394 tz = _mm_mul_pd(fscal,dz00);
1396 /* Update vectorial force */
1397 fix0 = _mm_add_pd(fix0,tx);
1398 fiy0 = _mm_add_pd(fiy0,ty);
1399 fiz0 = _mm_add_pd(fiz0,tz);
1401 fjx0 = _mm_add_pd(fjx0,tx);
1402 fjy0 = _mm_add_pd(fjy0,ty);
1403 fjz0 = _mm_add_pd(fjz0,tz);
1407 /**************************
1408 * CALCULATE INTERACTIONS *
1409 **************************/
1411 if (gmx_mm_any_lt(rsq10,rcutoff2))
1414 r10 = _mm_mul_pd(rsq10,rinv10);
1416 /* Compute parameters for interactions between i and j atoms */
1417 qq10 = _mm_mul_pd(iq1,jq0);
1419 /* EWALD ELECTROSTATICS */
1421 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1422 ewrt = _mm_mul_pd(r10,ewtabscale);
1423 ewitab = _mm_cvttpd_epi32(ewrt);
1424 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
1425 ewitab = _mm_slli_epi32(ewitab,2);
1426 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1427 ewtabD = _mm_setzero_pd();
1428 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1429 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
1430 ewtabFn = _mm_setzero_pd();
1431 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1432 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
1433 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
1434 velec = _mm_mul_pd(qq10,_mm_sub_pd(rinv10,velec));
1435 felec = _mm_mul_pd(_mm_mul_pd(qq10,rinv10),_mm_sub_pd(rinvsq10,felec));
1437 d = _mm_sub_pd(r10,rswitch);
1438 d = _mm_max_pd(d,_mm_setzero_pd());
1439 d2 = _mm_mul_pd(d,d);
1440 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1442 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1444 /* Evaluate switch function */
1445 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1446 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv10,_mm_mul_pd(velec,dsw)) );
1447 cutoff_mask = _mm_cmplt_pd(rsq10,rcutoff2);
1451 fscal = _mm_and_pd(fscal,cutoff_mask);
1453 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1455 /* Calculate temporary vectorial force */
1456 tx = _mm_mul_pd(fscal,dx10);
1457 ty = _mm_mul_pd(fscal,dy10);
1458 tz = _mm_mul_pd(fscal,dz10);
1460 /* Update vectorial force */
1461 fix1 = _mm_add_pd(fix1,tx);
1462 fiy1 = _mm_add_pd(fiy1,ty);
1463 fiz1 = _mm_add_pd(fiz1,tz);
1465 fjx0 = _mm_add_pd(fjx0,tx);
1466 fjy0 = _mm_add_pd(fjy0,ty);
1467 fjz0 = _mm_add_pd(fjz0,tz);
1471 /**************************
1472 * CALCULATE INTERACTIONS *
1473 **************************/
1475 if (gmx_mm_any_lt(rsq20,rcutoff2))
1478 r20 = _mm_mul_pd(rsq20,rinv20);
1480 /* Compute parameters for interactions between i and j atoms */
1481 qq20 = _mm_mul_pd(iq2,jq0);
1483 /* EWALD ELECTROSTATICS */
1485 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1486 ewrt = _mm_mul_pd(r20,ewtabscale);
1487 ewitab = _mm_cvttpd_epi32(ewrt);
1488 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
1489 ewitab = _mm_slli_epi32(ewitab,2);
1490 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1491 ewtabD = _mm_setzero_pd();
1492 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1493 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
1494 ewtabFn = _mm_setzero_pd();
1495 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1496 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
1497 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
1498 velec = _mm_mul_pd(qq20,_mm_sub_pd(rinv20,velec));
1499 felec = _mm_mul_pd(_mm_mul_pd(qq20,rinv20),_mm_sub_pd(rinvsq20,felec));
1501 d = _mm_sub_pd(r20,rswitch);
1502 d = _mm_max_pd(d,_mm_setzero_pd());
1503 d2 = _mm_mul_pd(d,d);
1504 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1506 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1508 /* Evaluate switch function */
1509 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1510 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv20,_mm_mul_pd(velec,dsw)) );
1511 cutoff_mask = _mm_cmplt_pd(rsq20,rcutoff2);
1515 fscal = _mm_and_pd(fscal,cutoff_mask);
1517 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1519 /* Calculate temporary vectorial force */
1520 tx = _mm_mul_pd(fscal,dx20);
1521 ty = _mm_mul_pd(fscal,dy20);
1522 tz = _mm_mul_pd(fscal,dz20);
1524 /* Update vectorial force */
1525 fix2 = _mm_add_pd(fix2,tx);
1526 fiy2 = _mm_add_pd(fiy2,ty);
1527 fiz2 = _mm_add_pd(fiz2,tz);
1529 fjx0 = _mm_add_pd(fjx0,tx);
1530 fjy0 = _mm_add_pd(fjy0,ty);
1531 fjz0 = _mm_add_pd(fjz0,tz);
1535 /**************************
1536 * CALCULATE INTERACTIONS *
1537 **************************/
1539 if (gmx_mm_any_lt(rsq30,rcutoff2))
1542 r30 = _mm_mul_pd(rsq30,rinv30);
1544 /* Compute parameters for interactions between i and j atoms */
1545 qq30 = _mm_mul_pd(iq3,jq0);
1547 /* EWALD ELECTROSTATICS */
1549 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1550 ewrt = _mm_mul_pd(r30,ewtabscale);
1551 ewitab = _mm_cvttpd_epi32(ewrt);
1552 eweps = _mm_sub_pd(ewrt,_mm_cvtepi32_pd(ewitab));
1553 ewitab = _mm_slli_epi32(ewitab,2);
1554 ewtabF = _mm_load_pd( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1555 ewtabD = _mm_setzero_pd();
1556 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1557 ewtabV = _mm_load_sd( ewtab + gmx_mm_extract_epi32(ewitab,0) +2);
1558 ewtabFn = _mm_setzero_pd();
1559 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1560 felec = _mm_add_pd(ewtabF,_mm_mul_pd(eweps,ewtabD));
1561 velec = _mm_sub_pd(ewtabV,_mm_mul_pd(_mm_mul_pd(ewtabhalfspace,eweps),_mm_add_pd(ewtabF,felec)));
1562 velec = _mm_mul_pd(qq30,_mm_sub_pd(rinv30,velec));
1563 felec = _mm_mul_pd(_mm_mul_pd(qq30,rinv30),_mm_sub_pd(rinvsq30,felec));
1565 d = _mm_sub_pd(r30,rswitch);
1566 d = _mm_max_pd(d,_mm_setzero_pd());
1567 d2 = _mm_mul_pd(d,d);
1568 sw = _mm_add_pd(one,_mm_mul_pd(d2,_mm_mul_pd(d,_mm_add_pd(swV3,_mm_mul_pd(d,_mm_add_pd(swV4,_mm_mul_pd(d,swV5)))))));
1570 dsw = _mm_mul_pd(d2,_mm_add_pd(swF2,_mm_mul_pd(d,_mm_add_pd(swF3,_mm_mul_pd(d,swF4)))));
1572 /* Evaluate switch function */
1573 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1574 felec = _mm_sub_pd( _mm_mul_pd(felec,sw) , _mm_mul_pd(rinv30,_mm_mul_pd(velec,dsw)) );
1575 cutoff_mask = _mm_cmplt_pd(rsq30,rcutoff2);
1579 fscal = _mm_and_pd(fscal,cutoff_mask);
1581 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1583 /* Calculate temporary vectorial force */
1584 tx = _mm_mul_pd(fscal,dx30);
1585 ty = _mm_mul_pd(fscal,dy30);
1586 tz = _mm_mul_pd(fscal,dz30);
1588 /* Update vectorial force */
1589 fix3 = _mm_add_pd(fix3,tx);
1590 fiy3 = _mm_add_pd(fiy3,ty);
1591 fiz3 = _mm_add_pd(fiz3,tz);
1593 fjx0 = _mm_add_pd(fjx0,tx);
1594 fjy0 = _mm_add_pd(fjy0,ty);
1595 fjz0 = _mm_add_pd(fjz0,tz);
1599 gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
1601 /* Inner loop uses 245 flops */
1604 /* End of innermost loop */
1606 gmx_mm_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1607 f+i_coord_offset,fshift+i_shift_offset);
1609 /* Increment number of inner iterations */
1610 inneriter += j_index_end - j_index_start;
1612 /* Outer loop uses 24 flops */
1615 /* Increment number of outer iterations */
1618 /* Update outer/inner flops */
1620 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*24 + inneriter*245);