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.
42 #include "../nb_kernel.h"
43 #include "gromacs/legacyheaders/types/simple.h"
44 #include "gromacs/math/vec.h"
45 #include "gromacs/legacyheaders/nrnb.h"
47 #include "kernelutil_sparc64_hpc_ace_double.h"
50 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomW3P1_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: ReactionField
52 * VdW interaction: None
53 * Geometry: Water3-Particle
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecRF_VdwNone_GeomW3P1_VF_sparc64_hpc_ace_double
58 (t_nblist * gmx_restrict nlist,
59 rvec * gmx_restrict xx,
60 rvec * gmx_restrict ff,
61 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 double precision SIMD, 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 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
80 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
82 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
84 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
85 int vdwjidx0A,vdwjidx0B;
86 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
87 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
88 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
89 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
90 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
93 _fjsp_v2r8 dummy_mask,cutoff_mask;
94 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
95 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
96 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
103 jindex = nlist->jindex;
105 shiftidx = nlist->shift;
107 shiftvec = fr->shift_vec[0];
108 fshift = fr->fshift[0];
109 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
110 charge = mdatoms->chargeA;
111 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
112 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
113 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
115 /* Setup water-specific parameters */
116 inr = nlist->iinr[0];
117 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
118 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
119 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
121 /* Avoid stupid compiler warnings */
129 /* Start outer loop over neighborlists */
130 for(iidx=0; iidx<nri; iidx++)
132 /* Load shift vector for this list */
133 i_shift_offset = DIM*shiftidx[iidx];
135 /* Load limits for loop over neighbors */
136 j_index_start = jindex[iidx];
137 j_index_end = jindex[iidx+1];
139 /* Get outer coordinate index */
141 i_coord_offset = DIM*inr;
143 /* Load i particle coords and add shift vector */
144 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
145 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
147 fix0 = _fjsp_setzero_v2r8();
148 fiy0 = _fjsp_setzero_v2r8();
149 fiz0 = _fjsp_setzero_v2r8();
150 fix1 = _fjsp_setzero_v2r8();
151 fiy1 = _fjsp_setzero_v2r8();
152 fiz1 = _fjsp_setzero_v2r8();
153 fix2 = _fjsp_setzero_v2r8();
154 fiy2 = _fjsp_setzero_v2r8();
155 fiz2 = _fjsp_setzero_v2r8();
157 /* Reset potential sums */
158 velecsum = _fjsp_setzero_v2r8();
160 /* Start inner kernel loop */
161 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
164 /* Get j neighbor index, and coordinate index */
167 j_coord_offsetA = DIM*jnrA;
168 j_coord_offsetB = DIM*jnrB;
170 /* load j atom coordinates */
171 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
174 /* Calculate displacement vector */
175 dx00 = _fjsp_sub_v2r8(ix0,jx0);
176 dy00 = _fjsp_sub_v2r8(iy0,jy0);
177 dz00 = _fjsp_sub_v2r8(iz0,jz0);
178 dx10 = _fjsp_sub_v2r8(ix1,jx0);
179 dy10 = _fjsp_sub_v2r8(iy1,jy0);
180 dz10 = _fjsp_sub_v2r8(iz1,jz0);
181 dx20 = _fjsp_sub_v2r8(ix2,jx0);
182 dy20 = _fjsp_sub_v2r8(iy2,jy0);
183 dz20 = _fjsp_sub_v2r8(iz2,jz0);
185 /* Calculate squared distance and things based on it */
186 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
187 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
188 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
190 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
191 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
192 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
194 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
195 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
196 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
198 /* Load parameters for j particles */
199 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
201 fjx0 = _fjsp_setzero_v2r8();
202 fjy0 = _fjsp_setzero_v2r8();
203 fjz0 = _fjsp_setzero_v2r8();
205 /**************************
206 * CALCULATE INTERACTIONS *
207 **************************/
209 /* Compute parameters for interactions between i and j atoms */
210 qq00 = _fjsp_mul_v2r8(iq0,jq0);
212 /* REACTION-FIELD ELECTROSTATICS */
213 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
214 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
216 /* Update potential sum for this i atom from the interaction with this j atom. */
217 velecsum = _fjsp_add_v2r8(velecsum,velec);
221 /* Update vectorial force */
222 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
223 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
224 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
226 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
227 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
228 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
230 /**************************
231 * CALCULATE INTERACTIONS *
232 **************************/
234 /* Compute parameters for interactions between i and j atoms */
235 qq10 = _fjsp_mul_v2r8(iq1,jq0);
237 /* REACTION-FIELD ELECTROSTATICS */
238 velec = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
239 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
241 /* Update potential sum for this i atom from the interaction with this j atom. */
242 velecsum = _fjsp_add_v2r8(velecsum,velec);
246 /* Update vectorial force */
247 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
248 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
249 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
251 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
252 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
253 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
255 /**************************
256 * CALCULATE INTERACTIONS *
257 **************************/
259 /* Compute parameters for interactions between i and j atoms */
260 qq20 = _fjsp_mul_v2r8(iq2,jq0);
262 /* REACTION-FIELD ELECTROSTATICS */
263 velec = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
264 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
266 /* Update potential sum for this i atom from the interaction with this j atom. */
267 velecsum = _fjsp_add_v2r8(velecsum,velec);
271 /* Update vectorial force */
272 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
273 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
274 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
276 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
277 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
278 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
280 gmx_fjsp_decrement_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
282 /* Inner loop uses 108 flops */
289 j_coord_offsetA = DIM*jnrA;
291 /* load j atom coordinates */
292 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
295 /* Calculate displacement vector */
296 dx00 = _fjsp_sub_v2r8(ix0,jx0);
297 dy00 = _fjsp_sub_v2r8(iy0,jy0);
298 dz00 = _fjsp_sub_v2r8(iz0,jz0);
299 dx10 = _fjsp_sub_v2r8(ix1,jx0);
300 dy10 = _fjsp_sub_v2r8(iy1,jy0);
301 dz10 = _fjsp_sub_v2r8(iz1,jz0);
302 dx20 = _fjsp_sub_v2r8(ix2,jx0);
303 dy20 = _fjsp_sub_v2r8(iy2,jy0);
304 dz20 = _fjsp_sub_v2r8(iz2,jz0);
306 /* Calculate squared distance and things based on it */
307 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
308 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
309 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
311 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
312 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
313 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
315 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
316 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
317 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
319 /* Load parameters for j particles */
320 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
322 fjx0 = _fjsp_setzero_v2r8();
323 fjy0 = _fjsp_setzero_v2r8();
324 fjz0 = _fjsp_setzero_v2r8();
326 /**************************
327 * CALCULATE INTERACTIONS *
328 **************************/
330 /* Compute parameters for interactions between i and j atoms */
331 qq00 = _fjsp_mul_v2r8(iq0,jq0);
333 /* REACTION-FIELD ELECTROSTATICS */
334 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
335 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
337 /* Update potential sum for this i atom from the interaction with this j atom. */
338 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
339 velecsum = _fjsp_add_v2r8(velecsum,velec);
343 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
345 /* Update vectorial force */
346 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
347 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
348 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
350 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
351 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
352 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
354 /**************************
355 * CALCULATE INTERACTIONS *
356 **************************/
358 /* Compute parameters for interactions between i and j atoms */
359 qq10 = _fjsp_mul_v2r8(iq1,jq0);
361 /* REACTION-FIELD ELECTROSTATICS */
362 velec = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
363 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
365 /* Update potential sum for this i atom from the interaction with this j atom. */
366 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
367 velecsum = _fjsp_add_v2r8(velecsum,velec);
371 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
373 /* Update vectorial force */
374 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
375 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
376 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
378 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
379 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
380 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
382 /**************************
383 * CALCULATE INTERACTIONS *
384 **************************/
386 /* Compute parameters for interactions between i and j atoms */
387 qq20 = _fjsp_mul_v2r8(iq2,jq0);
389 /* REACTION-FIELD ELECTROSTATICS */
390 velec = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
391 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
393 /* Update potential sum for this i atom from the interaction with this j atom. */
394 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
395 velecsum = _fjsp_add_v2r8(velecsum,velec);
399 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
401 /* Update vectorial force */
402 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
403 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
404 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
406 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
407 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
408 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
410 gmx_fjsp_decrement_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0);
412 /* Inner loop uses 108 flops */
415 /* End of innermost loop */
417 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
418 f+i_coord_offset,fshift+i_shift_offset);
421 /* Update potential energies */
422 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
424 /* Increment number of inner iterations */
425 inneriter += j_index_end - j_index_start;
427 /* Outer loop uses 19 flops */
430 /* Increment number of outer iterations */
433 /* Update outer/inner flops */
435 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3_VF,outeriter*19 + inneriter*108);
438 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomW3P1_F_sparc64_hpc_ace_double
439 * Electrostatics interaction: ReactionField
440 * VdW interaction: None
441 * Geometry: Water3-Particle
442 * Calculate force/pot: Force
445 nb_kernel_ElecRF_VdwNone_GeomW3P1_F_sparc64_hpc_ace_double
446 (t_nblist * gmx_restrict nlist,
447 rvec * gmx_restrict xx,
448 rvec * gmx_restrict ff,
449 t_forcerec * gmx_restrict fr,
450 t_mdatoms * gmx_restrict mdatoms,
451 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
452 t_nrnb * gmx_restrict nrnb)
454 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
455 * just 0 for non-waters.
456 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
457 * jnr indices corresponding to data put in the four positions in the SIMD register.
459 int i_shift_offset,i_coord_offset,outeriter,inneriter;
460 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
462 int j_coord_offsetA,j_coord_offsetB;
463 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
465 real *shiftvec,*fshift,*x,*f;
466 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
468 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
470 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
472 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
473 int vdwjidx0A,vdwjidx0B;
474 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
475 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
476 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
477 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
478 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
481 _fjsp_v2r8 dummy_mask,cutoff_mask;
482 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
483 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
484 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
491 jindex = nlist->jindex;
493 shiftidx = nlist->shift;
495 shiftvec = fr->shift_vec[0];
496 fshift = fr->fshift[0];
497 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
498 charge = mdatoms->chargeA;
499 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
500 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
501 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
503 /* Setup water-specific parameters */
504 inr = nlist->iinr[0];
505 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
506 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
507 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
509 /* Avoid stupid compiler warnings */
517 /* Start outer loop over neighborlists */
518 for(iidx=0; iidx<nri; iidx++)
520 /* Load shift vector for this list */
521 i_shift_offset = DIM*shiftidx[iidx];
523 /* Load limits for loop over neighbors */
524 j_index_start = jindex[iidx];
525 j_index_end = jindex[iidx+1];
527 /* Get outer coordinate index */
529 i_coord_offset = DIM*inr;
531 /* Load i particle coords and add shift vector */
532 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
533 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
535 fix0 = _fjsp_setzero_v2r8();
536 fiy0 = _fjsp_setzero_v2r8();
537 fiz0 = _fjsp_setzero_v2r8();
538 fix1 = _fjsp_setzero_v2r8();
539 fiy1 = _fjsp_setzero_v2r8();
540 fiz1 = _fjsp_setzero_v2r8();
541 fix2 = _fjsp_setzero_v2r8();
542 fiy2 = _fjsp_setzero_v2r8();
543 fiz2 = _fjsp_setzero_v2r8();
545 /* Start inner kernel loop */
546 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
549 /* Get j neighbor index, and coordinate index */
552 j_coord_offsetA = DIM*jnrA;
553 j_coord_offsetB = DIM*jnrB;
555 /* load j atom coordinates */
556 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
559 /* Calculate displacement vector */
560 dx00 = _fjsp_sub_v2r8(ix0,jx0);
561 dy00 = _fjsp_sub_v2r8(iy0,jy0);
562 dz00 = _fjsp_sub_v2r8(iz0,jz0);
563 dx10 = _fjsp_sub_v2r8(ix1,jx0);
564 dy10 = _fjsp_sub_v2r8(iy1,jy0);
565 dz10 = _fjsp_sub_v2r8(iz1,jz0);
566 dx20 = _fjsp_sub_v2r8(ix2,jx0);
567 dy20 = _fjsp_sub_v2r8(iy2,jy0);
568 dz20 = _fjsp_sub_v2r8(iz2,jz0);
570 /* Calculate squared distance and things based on it */
571 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
572 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
573 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
575 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
576 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
577 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
579 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
580 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
581 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
583 /* Load parameters for j particles */
584 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
586 fjx0 = _fjsp_setzero_v2r8();
587 fjy0 = _fjsp_setzero_v2r8();
588 fjz0 = _fjsp_setzero_v2r8();
590 /**************************
591 * CALCULATE INTERACTIONS *
592 **************************/
594 /* Compute parameters for interactions between i and j atoms */
595 qq00 = _fjsp_mul_v2r8(iq0,jq0);
597 /* REACTION-FIELD ELECTROSTATICS */
598 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
602 /* Update vectorial force */
603 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
604 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
605 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
607 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
608 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
609 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
611 /**************************
612 * CALCULATE INTERACTIONS *
613 **************************/
615 /* Compute parameters for interactions between i and j atoms */
616 qq10 = _fjsp_mul_v2r8(iq1,jq0);
618 /* REACTION-FIELD ELECTROSTATICS */
619 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
623 /* Update vectorial force */
624 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
625 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
626 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
628 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
629 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
630 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
632 /**************************
633 * CALCULATE INTERACTIONS *
634 **************************/
636 /* Compute parameters for interactions between i and j atoms */
637 qq20 = _fjsp_mul_v2r8(iq2,jq0);
639 /* REACTION-FIELD ELECTROSTATICS */
640 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
644 /* Update vectorial force */
645 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
646 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
647 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
649 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
650 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
651 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
653 gmx_fjsp_decrement_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
655 /* Inner loop uses 93 flops */
662 j_coord_offsetA = DIM*jnrA;
664 /* load j atom coordinates */
665 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
668 /* Calculate displacement vector */
669 dx00 = _fjsp_sub_v2r8(ix0,jx0);
670 dy00 = _fjsp_sub_v2r8(iy0,jy0);
671 dz00 = _fjsp_sub_v2r8(iz0,jz0);
672 dx10 = _fjsp_sub_v2r8(ix1,jx0);
673 dy10 = _fjsp_sub_v2r8(iy1,jy0);
674 dz10 = _fjsp_sub_v2r8(iz1,jz0);
675 dx20 = _fjsp_sub_v2r8(ix2,jx0);
676 dy20 = _fjsp_sub_v2r8(iy2,jy0);
677 dz20 = _fjsp_sub_v2r8(iz2,jz0);
679 /* Calculate squared distance and things based on it */
680 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
681 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
682 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
684 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
685 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
686 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
688 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
689 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
690 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
692 /* Load parameters for j particles */
693 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
695 fjx0 = _fjsp_setzero_v2r8();
696 fjy0 = _fjsp_setzero_v2r8();
697 fjz0 = _fjsp_setzero_v2r8();
699 /**************************
700 * CALCULATE INTERACTIONS *
701 **************************/
703 /* Compute parameters for interactions between i and j atoms */
704 qq00 = _fjsp_mul_v2r8(iq0,jq0);
706 /* REACTION-FIELD ELECTROSTATICS */
707 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
711 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
713 /* Update vectorial force */
714 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
715 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
716 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
718 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
719 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
720 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
722 /**************************
723 * CALCULATE INTERACTIONS *
724 **************************/
726 /* Compute parameters for interactions between i and j atoms */
727 qq10 = _fjsp_mul_v2r8(iq1,jq0);
729 /* REACTION-FIELD ELECTROSTATICS */
730 felec = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
734 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
736 /* Update vectorial force */
737 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
738 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
739 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
741 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
742 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
743 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
745 /**************************
746 * CALCULATE INTERACTIONS *
747 **************************/
749 /* Compute parameters for interactions between i and j atoms */
750 qq20 = _fjsp_mul_v2r8(iq2,jq0);
752 /* REACTION-FIELD ELECTROSTATICS */
753 felec = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
757 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
759 /* Update vectorial force */
760 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
761 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
762 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
764 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
765 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
766 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
768 gmx_fjsp_decrement_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0);
770 /* Inner loop uses 93 flops */
773 /* End of innermost loop */
775 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
776 f+i_coord_offset,fshift+i_shift_offset);
778 /* Increment number of inner iterations */
779 inneriter += j_index_end - j_index_start;
781 /* Outer loop uses 18 flops */
784 /* Increment number of outer iterations */
787 /* Update outer/inner flops */
789 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3_F,outeriter*18 + inneriter*93);