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 "types/simple.h"
46 #include "gromacs/legacyheaders/vec.h"
49 #include "kernelutil_sparc64_hpc_ace_double.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwNone_GeomW4P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: Coulomb
54 * VdW interaction: None
55 * Geometry: Water4-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecCoul_VdwNone_GeomW4P1_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 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
84 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
86 _fjsp_v2r8 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
87 int vdwjidx0A,vdwjidx0B;
88 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
89 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
90 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
91 _fjsp_v2r8 dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
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;
114 /* Setup water-specific parameters */
115 inr = nlist->iinr[0];
116 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
117 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
118 iq3 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+3]));
120 /* Avoid stupid compiler warnings */
128 /* Start outer loop over neighborlists */
129 for(iidx=0; iidx<nri; iidx++)
131 /* Load shift vector for this list */
132 i_shift_offset = DIM*shiftidx[iidx];
134 /* Load limits for loop over neighbors */
135 j_index_start = jindex[iidx];
136 j_index_end = jindex[iidx+1];
138 /* Get outer coordinate index */
140 i_coord_offset = DIM*inr;
142 /* Load i particle coords and add shift vector */
143 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
144 &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
146 fix1 = _fjsp_setzero_v2r8();
147 fiy1 = _fjsp_setzero_v2r8();
148 fiz1 = _fjsp_setzero_v2r8();
149 fix2 = _fjsp_setzero_v2r8();
150 fiy2 = _fjsp_setzero_v2r8();
151 fiz2 = _fjsp_setzero_v2r8();
152 fix3 = _fjsp_setzero_v2r8();
153 fiy3 = _fjsp_setzero_v2r8();
154 fiz3 = _fjsp_setzero_v2r8();
156 /* Reset potential sums */
157 velecsum = _fjsp_setzero_v2r8();
159 /* Start inner kernel loop */
160 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
163 /* Get j neighbor index, and coordinate index */
166 j_coord_offsetA = DIM*jnrA;
167 j_coord_offsetB = DIM*jnrB;
169 /* load j atom coordinates */
170 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
173 /* Calculate displacement vector */
174 dx10 = _fjsp_sub_v2r8(ix1,jx0);
175 dy10 = _fjsp_sub_v2r8(iy1,jy0);
176 dz10 = _fjsp_sub_v2r8(iz1,jz0);
177 dx20 = _fjsp_sub_v2r8(ix2,jx0);
178 dy20 = _fjsp_sub_v2r8(iy2,jy0);
179 dz20 = _fjsp_sub_v2r8(iz2,jz0);
180 dx30 = _fjsp_sub_v2r8(ix3,jx0);
181 dy30 = _fjsp_sub_v2r8(iy3,jy0);
182 dz30 = _fjsp_sub_v2r8(iz3,jz0);
184 /* Calculate squared distance and things based on it */
185 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
186 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
187 rsq30 = gmx_fjsp_calc_rsq_v2r8(dx30,dy30,dz30);
189 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
190 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
191 rinv30 = gmx_fjsp_invsqrt_v2r8(rsq30);
193 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
194 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
195 rinvsq30 = _fjsp_mul_v2r8(rinv30,rinv30);
197 /* Load parameters for j particles */
198 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
200 fjx0 = _fjsp_setzero_v2r8();
201 fjy0 = _fjsp_setzero_v2r8();
202 fjz0 = _fjsp_setzero_v2r8();
204 /**************************
205 * CALCULATE INTERACTIONS *
206 **************************/
208 /* Compute parameters for interactions between i and j atoms */
209 qq10 = _fjsp_mul_v2r8(iq1,jq0);
211 /* COULOMB ELECTROSTATICS */
212 velec = _fjsp_mul_v2r8(qq10,rinv10);
213 felec = _fjsp_mul_v2r8(velec,rinvsq10);
215 /* Update potential sum for this i atom from the interaction with this j atom. */
216 velecsum = _fjsp_add_v2r8(velecsum,velec);
220 /* Update vectorial force */
221 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
222 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
223 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
225 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
226 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
227 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
229 /**************************
230 * CALCULATE INTERACTIONS *
231 **************************/
233 /* Compute parameters for interactions between i and j atoms */
234 qq20 = _fjsp_mul_v2r8(iq2,jq0);
236 /* COULOMB ELECTROSTATICS */
237 velec = _fjsp_mul_v2r8(qq20,rinv20);
238 felec = _fjsp_mul_v2r8(velec,rinvsq20);
240 /* Update potential sum for this i atom from the interaction with this j atom. */
241 velecsum = _fjsp_add_v2r8(velecsum,velec);
245 /* Update vectorial force */
246 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
247 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
248 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
250 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
251 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
252 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
254 /**************************
255 * CALCULATE INTERACTIONS *
256 **************************/
258 /* Compute parameters for interactions between i and j atoms */
259 qq30 = _fjsp_mul_v2r8(iq3,jq0);
261 /* COULOMB ELECTROSTATICS */
262 velec = _fjsp_mul_v2r8(qq30,rinv30);
263 felec = _fjsp_mul_v2r8(velec,rinvsq30);
265 /* Update potential sum for this i atom from the interaction with this j atom. */
266 velecsum = _fjsp_add_v2r8(velecsum,velec);
270 /* Update vectorial force */
271 fix3 = _fjsp_madd_v2r8(dx30,fscal,fix3);
272 fiy3 = _fjsp_madd_v2r8(dy30,fscal,fiy3);
273 fiz3 = _fjsp_madd_v2r8(dz30,fscal,fiz3);
275 fjx0 = _fjsp_madd_v2r8(dx30,fscal,fjx0);
276 fjy0 = _fjsp_madd_v2r8(dy30,fscal,fjy0);
277 fjz0 = _fjsp_madd_v2r8(dz30,fscal,fjz0);
279 gmx_fjsp_decrement_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
281 /* Inner loop uses 96 flops */
288 j_coord_offsetA = DIM*jnrA;
290 /* load j atom coordinates */
291 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
294 /* Calculate displacement vector */
295 dx10 = _fjsp_sub_v2r8(ix1,jx0);
296 dy10 = _fjsp_sub_v2r8(iy1,jy0);
297 dz10 = _fjsp_sub_v2r8(iz1,jz0);
298 dx20 = _fjsp_sub_v2r8(ix2,jx0);
299 dy20 = _fjsp_sub_v2r8(iy2,jy0);
300 dz20 = _fjsp_sub_v2r8(iz2,jz0);
301 dx30 = _fjsp_sub_v2r8(ix3,jx0);
302 dy30 = _fjsp_sub_v2r8(iy3,jy0);
303 dz30 = _fjsp_sub_v2r8(iz3,jz0);
305 /* Calculate squared distance and things based on it */
306 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
307 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
308 rsq30 = gmx_fjsp_calc_rsq_v2r8(dx30,dy30,dz30);
310 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
311 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
312 rinv30 = gmx_fjsp_invsqrt_v2r8(rsq30);
314 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
315 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
316 rinvsq30 = _fjsp_mul_v2r8(rinv30,rinv30);
318 /* Load parameters for j particles */
319 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
321 fjx0 = _fjsp_setzero_v2r8();
322 fjy0 = _fjsp_setzero_v2r8();
323 fjz0 = _fjsp_setzero_v2r8();
325 /**************************
326 * CALCULATE INTERACTIONS *
327 **************************/
329 /* Compute parameters for interactions between i and j atoms */
330 qq10 = _fjsp_mul_v2r8(iq1,jq0);
332 /* COULOMB ELECTROSTATICS */
333 velec = _fjsp_mul_v2r8(qq10,rinv10);
334 felec = _fjsp_mul_v2r8(velec,rinvsq10);
336 /* Update potential sum for this i atom from the interaction with this j atom. */
337 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
338 velecsum = _fjsp_add_v2r8(velecsum,velec);
342 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
344 /* Update vectorial force */
345 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
346 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
347 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
349 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
350 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
351 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
353 /**************************
354 * CALCULATE INTERACTIONS *
355 **************************/
357 /* Compute parameters for interactions between i and j atoms */
358 qq20 = _fjsp_mul_v2r8(iq2,jq0);
360 /* COULOMB ELECTROSTATICS */
361 velec = _fjsp_mul_v2r8(qq20,rinv20);
362 felec = _fjsp_mul_v2r8(velec,rinvsq20);
364 /* Update potential sum for this i atom from the interaction with this j atom. */
365 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
366 velecsum = _fjsp_add_v2r8(velecsum,velec);
370 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
372 /* Update vectorial force */
373 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
374 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
375 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
377 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
378 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
379 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
381 /**************************
382 * CALCULATE INTERACTIONS *
383 **************************/
385 /* Compute parameters for interactions between i and j atoms */
386 qq30 = _fjsp_mul_v2r8(iq3,jq0);
388 /* COULOMB ELECTROSTATICS */
389 velec = _fjsp_mul_v2r8(qq30,rinv30);
390 felec = _fjsp_mul_v2r8(velec,rinvsq30);
392 /* Update potential sum for this i atom from the interaction with this j atom. */
393 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
394 velecsum = _fjsp_add_v2r8(velecsum,velec);
398 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
400 /* Update vectorial force */
401 fix3 = _fjsp_madd_v2r8(dx30,fscal,fix3);
402 fiy3 = _fjsp_madd_v2r8(dy30,fscal,fiy3);
403 fiz3 = _fjsp_madd_v2r8(dz30,fscal,fiz3);
405 fjx0 = _fjsp_madd_v2r8(dx30,fscal,fjx0);
406 fjy0 = _fjsp_madd_v2r8(dy30,fscal,fjy0);
407 fjz0 = _fjsp_madd_v2r8(dz30,fscal,fjz0);
409 gmx_fjsp_decrement_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0);
411 /* Inner loop uses 96 flops */
414 /* End of innermost loop */
416 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
417 f+i_coord_offset+DIM,fshift+i_shift_offset);
420 /* Update potential energies */
421 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
423 /* Increment number of inner iterations */
424 inneriter += j_index_end - j_index_start;
426 /* Outer loop uses 19 flops */
429 /* Increment number of outer iterations */
432 /* Update outer/inner flops */
434 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4_VF,outeriter*19 + inneriter*96);
437 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwNone_GeomW4P1_F_sparc64_hpc_ace_double
438 * Electrostatics interaction: Coulomb
439 * VdW interaction: None
440 * Geometry: Water4-Particle
441 * Calculate force/pot: Force
444 nb_kernel_ElecCoul_VdwNone_GeomW4P1_F_sparc64_hpc_ace_double
445 (t_nblist * gmx_restrict nlist,
446 rvec * gmx_restrict xx,
447 rvec * gmx_restrict ff,
448 t_forcerec * gmx_restrict fr,
449 t_mdatoms * gmx_restrict mdatoms,
450 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
451 t_nrnb * gmx_restrict nrnb)
453 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
454 * just 0 for non-waters.
455 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
456 * jnr indices corresponding to data put in the four positions in the SIMD register.
458 int i_shift_offset,i_coord_offset,outeriter,inneriter;
459 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
461 int j_coord_offsetA,j_coord_offsetB;
462 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
464 real *shiftvec,*fshift,*x,*f;
465 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
467 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
469 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
471 _fjsp_v2r8 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
472 int vdwjidx0A,vdwjidx0B;
473 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
474 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
475 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
476 _fjsp_v2r8 dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
477 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
480 _fjsp_v2r8 dummy_mask,cutoff_mask;
481 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
482 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
483 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
490 jindex = nlist->jindex;
492 shiftidx = nlist->shift;
494 shiftvec = fr->shift_vec[0];
495 fshift = fr->fshift[0];
496 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
497 charge = mdatoms->chargeA;
499 /* Setup water-specific parameters */
500 inr = nlist->iinr[0];
501 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
502 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
503 iq3 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+3]));
505 /* Avoid stupid compiler warnings */
513 /* Start outer loop over neighborlists */
514 for(iidx=0; iidx<nri; iidx++)
516 /* Load shift vector for this list */
517 i_shift_offset = DIM*shiftidx[iidx];
519 /* Load limits for loop over neighbors */
520 j_index_start = jindex[iidx];
521 j_index_end = jindex[iidx+1];
523 /* Get outer coordinate index */
525 i_coord_offset = DIM*inr;
527 /* Load i particle coords and add shift vector */
528 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
529 &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
531 fix1 = _fjsp_setzero_v2r8();
532 fiy1 = _fjsp_setzero_v2r8();
533 fiz1 = _fjsp_setzero_v2r8();
534 fix2 = _fjsp_setzero_v2r8();
535 fiy2 = _fjsp_setzero_v2r8();
536 fiz2 = _fjsp_setzero_v2r8();
537 fix3 = _fjsp_setzero_v2r8();
538 fiy3 = _fjsp_setzero_v2r8();
539 fiz3 = _fjsp_setzero_v2r8();
541 /* Start inner kernel loop */
542 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
545 /* Get j neighbor index, and coordinate index */
548 j_coord_offsetA = DIM*jnrA;
549 j_coord_offsetB = DIM*jnrB;
551 /* load j atom coordinates */
552 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
555 /* Calculate displacement vector */
556 dx10 = _fjsp_sub_v2r8(ix1,jx0);
557 dy10 = _fjsp_sub_v2r8(iy1,jy0);
558 dz10 = _fjsp_sub_v2r8(iz1,jz0);
559 dx20 = _fjsp_sub_v2r8(ix2,jx0);
560 dy20 = _fjsp_sub_v2r8(iy2,jy0);
561 dz20 = _fjsp_sub_v2r8(iz2,jz0);
562 dx30 = _fjsp_sub_v2r8(ix3,jx0);
563 dy30 = _fjsp_sub_v2r8(iy3,jy0);
564 dz30 = _fjsp_sub_v2r8(iz3,jz0);
566 /* Calculate squared distance and things based on it */
567 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
568 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
569 rsq30 = gmx_fjsp_calc_rsq_v2r8(dx30,dy30,dz30);
571 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
572 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
573 rinv30 = gmx_fjsp_invsqrt_v2r8(rsq30);
575 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
576 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
577 rinvsq30 = _fjsp_mul_v2r8(rinv30,rinv30);
579 /* Load parameters for j particles */
580 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
582 fjx0 = _fjsp_setzero_v2r8();
583 fjy0 = _fjsp_setzero_v2r8();
584 fjz0 = _fjsp_setzero_v2r8();
586 /**************************
587 * CALCULATE INTERACTIONS *
588 **************************/
590 /* Compute parameters for interactions between i and j atoms */
591 qq10 = _fjsp_mul_v2r8(iq1,jq0);
593 /* COULOMB ELECTROSTATICS */
594 velec = _fjsp_mul_v2r8(qq10,rinv10);
595 felec = _fjsp_mul_v2r8(velec,rinvsq10);
599 /* Update vectorial force */
600 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
601 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
602 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
604 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
605 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
606 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
608 /**************************
609 * CALCULATE INTERACTIONS *
610 **************************/
612 /* Compute parameters for interactions between i and j atoms */
613 qq20 = _fjsp_mul_v2r8(iq2,jq0);
615 /* COULOMB ELECTROSTATICS */
616 velec = _fjsp_mul_v2r8(qq20,rinv20);
617 felec = _fjsp_mul_v2r8(velec,rinvsq20);
621 /* Update vectorial force */
622 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
623 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
624 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
626 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
627 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
628 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
630 /**************************
631 * CALCULATE INTERACTIONS *
632 **************************/
634 /* Compute parameters for interactions between i and j atoms */
635 qq30 = _fjsp_mul_v2r8(iq3,jq0);
637 /* COULOMB ELECTROSTATICS */
638 velec = _fjsp_mul_v2r8(qq30,rinv30);
639 felec = _fjsp_mul_v2r8(velec,rinvsq30);
643 /* Update vectorial force */
644 fix3 = _fjsp_madd_v2r8(dx30,fscal,fix3);
645 fiy3 = _fjsp_madd_v2r8(dy30,fscal,fiy3);
646 fiz3 = _fjsp_madd_v2r8(dz30,fscal,fiz3);
648 fjx0 = _fjsp_madd_v2r8(dx30,fscal,fjx0);
649 fjy0 = _fjsp_madd_v2r8(dy30,fscal,fjy0);
650 fjz0 = _fjsp_madd_v2r8(dz30,fscal,fjz0);
652 gmx_fjsp_decrement_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
654 /* Inner loop uses 93 flops */
661 j_coord_offsetA = DIM*jnrA;
663 /* load j atom coordinates */
664 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
667 /* Calculate displacement vector */
668 dx10 = _fjsp_sub_v2r8(ix1,jx0);
669 dy10 = _fjsp_sub_v2r8(iy1,jy0);
670 dz10 = _fjsp_sub_v2r8(iz1,jz0);
671 dx20 = _fjsp_sub_v2r8(ix2,jx0);
672 dy20 = _fjsp_sub_v2r8(iy2,jy0);
673 dz20 = _fjsp_sub_v2r8(iz2,jz0);
674 dx30 = _fjsp_sub_v2r8(ix3,jx0);
675 dy30 = _fjsp_sub_v2r8(iy3,jy0);
676 dz30 = _fjsp_sub_v2r8(iz3,jz0);
678 /* Calculate squared distance and things based on it */
679 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
680 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
681 rsq30 = gmx_fjsp_calc_rsq_v2r8(dx30,dy30,dz30);
683 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
684 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
685 rinv30 = gmx_fjsp_invsqrt_v2r8(rsq30);
687 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
688 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
689 rinvsq30 = _fjsp_mul_v2r8(rinv30,rinv30);
691 /* Load parameters for j particles */
692 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
694 fjx0 = _fjsp_setzero_v2r8();
695 fjy0 = _fjsp_setzero_v2r8();
696 fjz0 = _fjsp_setzero_v2r8();
698 /**************************
699 * CALCULATE INTERACTIONS *
700 **************************/
702 /* Compute parameters for interactions between i and j atoms */
703 qq10 = _fjsp_mul_v2r8(iq1,jq0);
705 /* COULOMB ELECTROSTATICS */
706 velec = _fjsp_mul_v2r8(qq10,rinv10);
707 felec = _fjsp_mul_v2r8(velec,rinvsq10);
711 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
713 /* Update vectorial force */
714 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
715 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
716 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
718 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
719 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
720 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
722 /**************************
723 * CALCULATE INTERACTIONS *
724 **************************/
726 /* Compute parameters for interactions between i and j atoms */
727 qq20 = _fjsp_mul_v2r8(iq2,jq0);
729 /* COULOMB ELECTROSTATICS */
730 velec = _fjsp_mul_v2r8(qq20,rinv20);
731 felec = _fjsp_mul_v2r8(velec,rinvsq20);
735 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
737 /* Update vectorial force */
738 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
739 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
740 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
742 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
743 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
744 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
746 /**************************
747 * CALCULATE INTERACTIONS *
748 **************************/
750 /* Compute parameters for interactions between i and j atoms */
751 qq30 = _fjsp_mul_v2r8(iq3,jq0);
753 /* COULOMB ELECTROSTATICS */
754 velec = _fjsp_mul_v2r8(qq30,rinv30);
755 felec = _fjsp_mul_v2r8(velec,rinvsq30);
759 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
761 /* Update vectorial force */
762 fix3 = _fjsp_madd_v2r8(dx30,fscal,fix3);
763 fiy3 = _fjsp_madd_v2r8(dy30,fscal,fiy3);
764 fiz3 = _fjsp_madd_v2r8(dz30,fscal,fiz3);
766 fjx0 = _fjsp_madd_v2r8(dx30,fscal,fjx0);
767 fjy0 = _fjsp_madd_v2r8(dy30,fscal,fjy0);
768 fjz0 = _fjsp_madd_v2r8(dz30,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(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
778 f+i_coord_offset+DIM,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_W4_F,outeriter*18 + inneriter*93);