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 sparc64_hpc_ace_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 "kernelutil_sparc64_hpc_ace_double.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomW3P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: ReactionField
54 * VdW interaction: None
55 * Geometry: Water3-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecRF_VdwNone_GeomW3P1_VF_sparc64_hpc_ace_double
60 (t_nblist * gmx_restrict nlist,
61 rvec * gmx_restrict xx,
62 rvec * gmx_restrict ff,
63 t_forcerec * gmx_restrict fr,
64 t_mdatoms * gmx_restrict mdatoms,
65 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
66 t_nrnb * gmx_restrict nrnb)
68 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
69 * just 0 for non-waters.
70 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
71 * jnr indices corresponding to data put in the four positions in the SIMD register.
73 int i_shift_offset,i_coord_offset,outeriter,inneriter;
74 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
76 int j_coord_offsetA,j_coord_offsetB;
77 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
79 real *shiftvec,*fshift,*x,*f;
80 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
82 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
84 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
86 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
87 int vdwjidx0A,vdwjidx0B;
88 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
89 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
90 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
91 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
92 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
95 _fjsp_v2r8 dummy_mask,cutoff_mask;
96 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
97 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
98 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
105 jindex = nlist->jindex;
107 shiftidx = nlist->shift;
109 shiftvec = fr->shift_vec[0];
110 fshift = fr->fshift[0];
111 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
112 charge = mdatoms->chargeA;
113 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
114 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
115 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
117 /* Setup water-specific parameters */
118 inr = nlist->iinr[0];
119 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
120 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
121 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
123 /* Avoid stupid compiler warnings */
131 /* Start outer loop over neighborlists */
132 for(iidx=0; iidx<nri; iidx++)
134 /* Load shift vector for this list */
135 i_shift_offset = DIM*shiftidx[iidx];
137 /* Load limits for loop over neighbors */
138 j_index_start = jindex[iidx];
139 j_index_end = jindex[iidx+1];
141 /* Get outer coordinate index */
143 i_coord_offset = DIM*inr;
145 /* Load i particle coords and add shift vector */
146 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
147 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
149 fix0 = _fjsp_setzero_v2r8();
150 fiy0 = _fjsp_setzero_v2r8();
151 fiz0 = _fjsp_setzero_v2r8();
152 fix1 = _fjsp_setzero_v2r8();
153 fiy1 = _fjsp_setzero_v2r8();
154 fiz1 = _fjsp_setzero_v2r8();
155 fix2 = _fjsp_setzero_v2r8();
156 fiy2 = _fjsp_setzero_v2r8();
157 fiz2 = _fjsp_setzero_v2r8();
159 /* Reset potential sums */
160 velecsum = _fjsp_setzero_v2r8();
162 /* Start inner kernel loop */
163 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
166 /* Get j neighbor index, and coordinate index */
169 j_coord_offsetA = DIM*jnrA;
170 j_coord_offsetB = DIM*jnrB;
172 /* load j atom coordinates */
173 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
176 /* Calculate displacement vector */
177 dx00 = _fjsp_sub_v2r8(ix0,jx0);
178 dy00 = _fjsp_sub_v2r8(iy0,jy0);
179 dz00 = _fjsp_sub_v2r8(iz0,jz0);
180 dx10 = _fjsp_sub_v2r8(ix1,jx0);
181 dy10 = _fjsp_sub_v2r8(iy1,jy0);
182 dz10 = _fjsp_sub_v2r8(iz1,jz0);
183 dx20 = _fjsp_sub_v2r8(ix2,jx0);
184 dy20 = _fjsp_sub_v2r8(iy2,jy0);
185 dz20 = _fjsp_sub_v2r8(iz2,jz0);
187 /* Calculate squared distance and things based on it */
188 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
189 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
190 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
192 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
193 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
194 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
196 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
197 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
198 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
200 /* Load parameters for j particles */
201 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
203 fjx0 = _fjsp_setzero_v2r8();
204 fjy0 = _fjsp_setzero_v2r8();
205 fjz0 = _fjsp_setzero_v2r8();
207 /**************************
208 * CALCULATE INTERACTIONS *
209 **************************/
211 /* Compute parameters for interactions between i and j atoms */
212 qq00 = _fjsp_mul_v2r8(iq0,jq0);
214 /* REACTION-FIELD ELECTROSTATICS */
215 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
216 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
218 /* Update potential sum for this i atom from the interaction with this j atom. */
219 velecsum = _fjsp_add_v2r8(velecsum,velec);
223 /* Update vectorial force */
224 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
225 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
226 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
228 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
229 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
230 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
232 /**************************
233 * CALCULATE INTERACTIONS *
234 **************************/
236 /* Compute parameters for interactions between i and j atoms */
237 qq10 = _fjsp_mul_v2r8(iq1,jq0);
239 /* REACTION-FIELD ELECTROSTATICS */
240 velec = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
241 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
243 /* Update potential sum for this i atom from the interaction with this j atom. */
244 velecsum = _fjsp_add_v2r8(velecsum,velec);
248 /* Update vectorial force */
249 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
250 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
251 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
253 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
254 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
255 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
257 /**************************
258 * CALCULATE INTERACTIONS *
259 **************************/
261 /* Compute parameters for interactions between i and j atoms */
262 qq20 = _fjsp_mul_v2r8(iq2,jq0);
264 /* REACTION-FIELD ELECTROSTATICS */
265 velec = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
266 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
268 /* Update potential sum for this i atom from the interaction with this j atom. */
269 velecsum = _fjsp_add_v2r8(velecsum,velec);
273 /* Update vectorial force */
274 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
275 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
276 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
278 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
279 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
280 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
282 gmx_fjsp_decrement_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
284 /* Inner loop uses 108 flops */
291 j_coord_offsetA = DIM*jnrA;
293 /* load j atom coordinates */
294 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
297 /* Calculate displacement vector */
298 dx00 = _fjsp_sub_v2r8(ix0,jx0);
299 dy00 = _fjsp_sub_v2r8(iy0,jy0);
300 dz00 = _fjsp_sub_v2r8(iz0,jz0);
301 dx10 = _fjsp_sub_v2r8(ix1,jx0);
302 dy10 = _fjsp_sub_v2r8(iy1,jy0);
303 dz10 = _fjsp_sub_v2r8(iz1,jz0);
304 dx20 = _fjsp_sub_v2r8(ix2,jx0);
305 dy20 = _fjsp_sub_v2r8(iy2,jy0);
306 dz20 = _fjsp_sub_v2r8(iz2,jz0);
308 /* Calculate squared distance and things based on it */
309 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
310 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
311 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
313 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
314 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
315 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
317 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
318 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
319 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
321 /* Load parameters for j particles */
322 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
324 fjx0 = _fjsp_setzero_v2r8();
325 fjy0 = _fjsp_setzero_v2r8();
326 fjz0 = _fjsp_setzero_v2r8();
328 /**************************
329 * CALCULATE INTERACTIONS *
330 **************************/
332 /* Compute parameters for interactions between i and j atoms */
333 qq00 = _fjsp_mul_v2r8(iq0,jq0);
335 /* REACTION-FIELD ELECTROSTATICS */
336 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
337 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
339 /* Update potential sum for this i atom from the interaction with this j atom. */
340 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
341 velecsum = _fjsp_add_v2r8(velecsum,velec);
345 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
347 /* Update vectorial force */
348 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
349 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
350 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
352 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
353 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
354 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
356 /**************************
357 * CALCULATE INTERACTIONS *
358 **************************/
360 /* Compute parameters for interactions between i and j atoms */
361 qq10 = _fjsp_mul_v2r8(iq1,jq0);
363 /* REACTION-FIELD ELECTROSTATICS */
364 velec = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
365 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
367 /* Update potential sum for this i atom from the interaction with this j atom. */
368 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
369 velecsum = _fjsp_add_v2r8(velecsum,velec);
373 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
375 /* Update vectorial force */
376 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
377 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
378 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
380 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
381 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
382 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
384 /**************************
385 * CALCULATE INTERACTIONS *
386 **************************/
388 /* Compute parameters for interactions between i and j atoms */
389 qq20 = _fjsp_mul_v2r8(iq2,jq0);
391 /* REACTION-FIELD ELECTROSTATICS */
392 velec = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
393 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
395 /* Update potential sum for this i atom from the interaction with this j atom. */
396 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
397 velecsum = _fjsp_add_v2r8(velecsum,velec);
401 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
403 /* Update vectorial force */
404 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
405 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
406 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
408 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
409 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
410 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
412 gmx_fjsp_decrement_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0);
414 /* Inner loop uses 108 flops */
417 /* End of innermost loop */
419 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
420 f+i_coord_offset,fshift+i_shift_offset);
423 /* Update potential energies */
424 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
426 /* Increment number of inner iterations */
427 inneriter += j_index_end - j_index_start;
429 /* Outer loop uses 19 flops */
432 /* Increment number of outer iterations */
435 /* Update outer/inner flops */
437 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3_VF,outeriter*19 + inneriter*108);
440 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomW3P1_F_sparc64_hpc_ace_double
441 * Electrostatics interaction: ReactionField
442 * VdW interaction: None
443 * Geometry: Water3-Particle
444 * Calculate force/pot: Force
447 nb_kernel_ElecRF_VdwNone_GeomW3P1_F_sparc64_hpc_ace_double
448 (t_nblist * gmx_restrict nlist,
449 rvec * gmx_restrict xx,
450 rvec * gmx_restrict ff,
451 t_forcerec * gmx_restrict fr,
452 t_mdatoms * gmx_restrict mdatoms,
453 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
454 t_nrnb * gmx_restrict nrnb)
456 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
457 * just 0 for non-waters.
458 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
459 * jnr indices corresponding to data put in the four positions in the SIMD register.
461 int i_shift_offset,i_coord_offset,outeriter,inneriter;
462 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
464 int j_coord_offsetA,j_coord_offsetB;
465 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
467 real *shiftvec,*fshift,*x,*f;
468 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
470 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
472 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
474 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
475 int vdwjidx0A,vdwjidx0B;
476 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
477 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
478 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
479 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
480 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
483 _fjsp_v2r8 dummy_mask,cutoff_mask;
484 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
485 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
486 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
493 jindex = nlist->jindex;
495 shiftidx = nlist->shift;
497 shiftvec = fr->shift_vec[0];
498 fshift = fr->fshift[0];
499 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
500 charge = mdatoms->chargeA;
501 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
502 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
503 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
505 /* Setup water-specific parameters */
506 inr = nlist->iinr[0];
507 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
508 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
509 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
511 /* Avoid stupid compiler warnings */
519 /* Start outer loop over neighborlists */
520 for(iidx=0; iidx<nri; iidx++)
522 /* Load shift vector for this list */
523 i_shift_offset = DIM*shiftidx[iidx];
525 /* Load limits for loop over neighbors */
526 j_index_start = jindex[iidx];
527 j_index_end = jindex[iidx+1];
529 /* Get outer coordinate index */
531 i_coord_offset = DIM*inr;
533 /* Load i particle coords and add shift vector */
534 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
535 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
537 fix0 = _fjsp_setzero_v2r8();
538 fiy0 = _fjsp_setzero_v2r8();
539 fiz0 = _fjsp_setzero_v2r8();
540 fix1 = _fjsp_setzero_v2r8();
541 fiy1 = _fjsp_setzero_v2r8();
542 fiz1 = _fjsp_setzero_v2r8();
543 fix2 = _fjsp_setzero_v2r8();
544 fiy2 = _fjsp_setzero_v2r8();
545 fiz2 = _fjsp_setzero_v2r8();
547 /* Start inner kernel loop */
548 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
551 /* Get j neighbor index, and coordinate index */
554 j_coord_offsetA = DIM*jnrA;
555 j_coord_offsetB = DIM*jnrB;
557 /* load j atom coordinates */
558 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
561 /* Calculate displacement vector */
562 dx00 = _fjsp_sub_v2r8(ix0,jx0);
563 dy00 = _fjsp_sub_v2r8(iy0,jy0);
564 dz00 = _fjsp_sub_v2r8(iz0,jz0);
565 dx10 = _fjsp_sub_v2r8(ix1,jx0);
566 dy10 = _fjsp_sub_v2r8(iy1,jy0);
567 dz10 = _fjsp_sub_v2r8(iz1,jz0);
568 dx20 = _fjsp_sub_v2r8(ix2,jx0);
569 dy20 = _fjsp_sub_v2r8(iy2,jy0);
570 dz20 = _fjsp_sub_v2r8(iz2,jz0);
572 /* Calculate squared distance and things based on it */
573 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
574 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
575 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
577 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
578 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
579 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
581 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
582 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
583 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
585 /* Load parameters for j particles */
586 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
588 fjx0 = _fjsp_setzero_v2r8();
589 fjy0 = _fjsp_setzero_v2r8();
590 fjz0 = _fjsp_setzero_v2r8();
592 /**************************
593 * CALCULATE INTERACTIONS *
594 **************************/
596 /* Compute parameters for interactions between i and j atoms */
597 qq00 = _fjsp_mul_v2r8(iq0,jq0);
599 /* REACTION-FIELD ELECTROSTATICS */
600 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
604 /* Update vectorial force */
605 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
606 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
607 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
609 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
610 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
611 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
613 /**************************
614 * CALCULATE INTERACTIONS *
615 **************************/
617 /* Compute parameters for interactions between i and j atoms */
618 qq10 = _fjsp_mul_v2r8(iq1,jq0);
620 /* REACTION-FIELD ELECTROSTATICS */
621 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
625 /* Update vectorial force */
626 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
627 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
628 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
630 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
631 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
632 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
634 /**************************
635 * CALCULATE INTERACTIONS *
636 **************************/
638 /* Compute parameters for interactions between i and j atoms */
639 qq20 = _fjsp_mul_v2r8(iq2,jq0);
641 /* REACTION-FIELD ELECTROSTATICS */
642 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
646 /* Update vectorial force */
647 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
648 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
649 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
651 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
652 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
653 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
655 gmx_fjsp_decrement_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
657 /* Inner loop uses 93 flops */
664 j_coord_offsetA = DIM*jnrA;
666 /* load j atom coordinates */
667 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
670 /* Calculate displacement vector */
671 dx00 = _fjsp_sub_v2r8(ix0,jx0);
672 dy00 = _fjsp_sub_v2r8(iy0,jy0);
673 dz00 = _fjsp_sub_v2r8(iz0,jz0);
674 dx10 = _fjsp_sub_v2r8(ix1,jx0);
675 dy10 = _fjsp_sub_v2r8(iy1,jy0);
676 dz10 = _fjsp_sub_v2r8(iz1,jz0);
677 dx20 = _fjsp_sub_v2r8(ix2,jx0);
678 dy20 = _fjsp_sub_v2r8(iy2,jy0);
679 dz20 = _fjsp_sub_v2r8(iz2,jz0);
681 /* Calculate squared distance and things based on it */
682 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
683 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
684 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
686 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
687 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
688 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
690 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
691 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
692 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
694 /* Load parameters for j particles */
695 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
697 fjx0 = _fjsp_setzero_v2r8();
698 fjy0 = _fjsp_setzero_v2r8();
699 fjz0 = _fjsp_setzero_v2r8();
701 /**************************
702 * CALCULATE INTERACTIONS *
703 **************************/
705 /* Compute parameters for interactions between i and j atoms */
706 qq00 = _fjsp_mul_v2r8(iq0,jq0);
708 /* REACTION-FIELD ELECTROSTATICS */
709 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
713 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
715 /* Update vectorial force */
716 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
717 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
718 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
720 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
721 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
722 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
724 /**************************
725 * CALCULATE INTERACTIONS *
726 **************************/
728 /* Compute parameters for interactions between i and j atoms */
729 qq10 = _fjsp_mul_v2r8(iq1,jq0);
731 /* REACTION-FIELD ELECTROSTATICS */
732 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
736 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
738 /* Update vectorial force */
739 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
740 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
741 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
743 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
744 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
745 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
747 /**************************
748 * CALCULATE INTERACTIONS *
749 **************************/
751 /* Compute parameters for interactions between i and j atoms */
752 qq20 = _fjsp_mul_v2r8(iq2,jq0);
754 /* REACTION-FIELD ELECTROSTATICS */
755 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
759 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
761 /* Update vectorial force */
762 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
763 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
764 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
766 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
767 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
768 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
770 gmx_fjsp_decrement_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0);
772 /* Inner loop uses 93 flops */
775 /* End of innermost loop */
777 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
778 f+i_coord_offset,fshift+i_shift_offset);
780 /* Increment number of inner iterations */
781 inneriter += j_index_end - j_index_start;
783 /* Outer loop uses 18 flops */
786 /* Increment number of outer iterations */
789 /* Update outer/inner flops */
791 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3_F,outeriter*18 + inneriter*93);