2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012, by the GROMACS development team, led by
5 * David van der Spoel, Berk Hess, Erik Lindahl, and including many
6 * others, as listed in the AUTHORS file in the top-level source
7 * 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"
49 #include "kernelutil_sparc64_hpc_ace_double.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwNone_GeomP1P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: Ewald
54 * VdW interaction: None
55 * Geometry: Particle-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecEwSh_VdwNone_GeomP1P1_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_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;
83 int vdwjidx0A,vdwjidx0B;
84 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
85 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
86 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
88 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
91 _fjsp_v2r8 dummy_mask,cutoff_mask;
92 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
93 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
94 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
101 jindex = nlist->jindex;
103 shiftidx = nlist->shift;
105 shiftvec = fr->shift_vec[0];
106 fshift = fr->fshift[0];
107 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
108 charge = mdatoms->chargeA;
110 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
111 ewtab = fr->ic->tabq_coul_FDV0;
112 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
113 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
115 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
116 rcutoff_scalar = fr->rcoulomb;
117 rcutoff = gmx_fjsp_set1_v2r8(rcutoff_scalar);
118 rcutoff2 = _fjsp_mul_v2r8(rcutoff,rcutoff);
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_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
145 fix0 = _fjsp_setzero_v2r8();
146 fiy0 = _fjsp_setzero_v2r8();
147 fiz0 = _fjsp_setzero_v2r8();
149 /* Load parameters for i particles */
150 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
152 /* Reset potential sums */
153 velecsum = _fjsp_setzero_v2r8();
155 /* Start inner kernel loop */
156 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
159 /* Get j neighbor index, and coordinate index */
162 j_coord_offsetA = DIM*jnrA;
163 j_coord_offsetB = DIM*jnrB;
165 /* load j atom coordinates */
166 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
169 /* Calculate displacement vector */
170 dx00 = _fjsp_sub_v2r8(ix0,jx0);
171 dy00 = _fjsp_sub_v2r8(iy0,jy0);
172 dz00 = _fjsp_sub_v2r8(iz0,jz0);
174 /* Calculate squared distance and things based on it */
175 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
177 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
179 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
181 /* Load parameters for j particles */
182 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
184 /**************************
185 * CALCULATE INTERACTIONS *
186 **************************/
188 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
191 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
193 /* Compute parameters for interactions between i and j atoms */
194 qq00 = _fjsp_mul_v2r8(iq0,jq0);
196 /* EWALD ELECTROSTATICS */
198 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
199 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
200 itab_tmp = _fjsp_dtox_v2r8(ewrt);
201 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
202 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
204 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
205 ewtabD = _fjsp_load_v2r8( ewtab + 4*ewconv.i[1] );
206 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
207 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
208 ewtabFn = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[1] +2);
209 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
210 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
211 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
212 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_sub_v2r8(rinv00,sh_ewald),velec));
213 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
215 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
217 /* Update potential sum for this i atom from the interaction with this j atom. */
218 velec = _fjsp_and_v2r8(velec,cutoff_mask);
219 velecsum = _fjsp_add_v2r8(velecsum,velec);
223 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
225 /* Update vectorial force */
226 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
227 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
228 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
230 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
234 /* Inner loop uses 49 flops */
241 j_coord_offsetA = DIM*jnrA;
243 /* load j atom coordinates */
244 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
247 /* Calculate displacement vector */
248 dx00 = _fjsp_sub_v2r8(ix0,jx0);
249 dy00 = _fjsp_sub_v2r8(iy0,jy0);
250 dz00 = _fjsp_sub_v2r8(iz0,jz0);
252 /* Calculate squared distance and things based on it */
253 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
255 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
257 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
259 /* Load parameters for j particles */
260 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
262 /**************************
263 * CALCULATE INTERACTIONS *
264 **************************/
266 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
269 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
271 /* Compute parameters for interactions between i and j atoms */
272 qq00 = _fjsp_mul_v2r8(iq0,jq0);
274 /* EWALD ELECTROSTATICS */
276 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
277 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
278 itab_tmp = _fjsp_dtox_v2r8(ewrt);
279 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
280 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
282 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
283 ewtabD = _fjsp_setzero_v2r8();
284 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
285 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
286 ewtabFn = _fjsp_setzero_v2r8();
287 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
288 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
289 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
290 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_sub_v2r8(rinv00,sh_ewald),velec));
291 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
293 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
295 /* Update potential sum for this i atom from the interaction with this j atom. */
296 velec = _fjsp_and_v2r8(velec,cutoff_mask);
297 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
298 velecsum = _fjsp_add_v2r8(velecsum,velec);
302 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
304 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
306 /* Update vectorial force */
307 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
308 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
309 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
311 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
315 /* Inner loop uses 49 flops */
318 /* End of innermost loop */
320 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
321 f+i_coord_offset,fshift+i_shift_offset);
324 /* Update potential energies */
325 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
327 /* Increment number of inner iterations */
328 inneriter += j_index_end - j_index_start;
330 /* Outer loop uses 8 flops */
333 /* Increment number of outer iterations */
336 /* Update outer/inner flops */
338 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VF,outeriter*8 + inneriter*49);
341 * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
342 * Electrostatics interaction: Ewald
343 * VdW interaction: None
344 * Geometry: Particle-Particle
345 * Calculate force/pot: Force
348 nb_kernel_ElecEwSh_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
349 (t_nblist * gmx_restrict nlist,
350 rvec * gmx_restrict xx,
351 rvec * gmx_restrict ff,
352 t_forcerec * gmx_restrict fr,
353 t_mdatoms * gmx_restrict mdatoms,
354 nb_kernel_data_t * gmx_restrict kernel_data,
355 t_nrnb * gmx_restrict nrnb)
357 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
358 * just 0 for non-waters.
359 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
360 * jnr indices corresponding to data put in the four positions in the SIMD register.
362 int i_shift_offset,i_coord_offset,outeriter,inneriter;
363 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
365 int j_coord_offsetA,j_coord_offsetB;
366 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
368 real *shiftvec,*fshift,*x,*f;
369 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
371 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
372 int vdwjidx0A,vdwjidx0B;
373 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
374 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
375 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
377 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
380 _fjsp_v2r8 dummy_mask,cutoff_mask;
381 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
382 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
383 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
390 jindex = nlist->jindex;
392 shiftidx = nlist->shift;
394 shiftvec = fr->shift_vec[0];
395 fshift = fr->fshift[0];
396 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
397 charge = mdatoms->chargeA;
399 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
400 ewtab = fr->ic->tabq_coul_F;
401 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
402 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
404 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
405 rcutoff_scalar = fr->rcoulomb;
406 rcutoff = gmx_fjsp_set1_v2r8(rcutoff_scalar);
407 rcutoff2 = _fjsp_mul_v2r8(rcutoff,rcutoff);
409 /* Avoid stupid compiler warnings */
417 /* Start outer loop over neighborlists */
418 for(iidx=0; iidx<nri; iidx++)
420 /* Load shift vector for this list */
421 i_shift_offset = DIM*shiftidx[iidx];
423 /* Load limits for loop over neighbors */
424 j_index_start = jindex[iidx];
425 j_index_end = jindex[iidx+1];
427 /* Get outer coordinate index */
429 i_coord_offset = DIM*inr;
431 /* Load i particle coords and add shift vector */
432 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
434 fix0 = _fjsp_setzero_v2r8();
435 fiy0 = _fjsp_setzero_v2r8();
436 fiz0 = _fjsp_setzero_v2r8();
438 /* Load parameters for i particles */
439 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
441 /* Start inner kernel loop */
442 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
445 /* Get j neighbor index, and coordinate index */
448 j_coord_offsetA = DIM*jnrA;
449 j_coord_offsetB = DIM*jnrB;
451 /* load j atom coordinates */
452 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
455 /* Calculate displacement vector */
456 dx00 = _fjsp_sub_v2r8(ix0,jx0);
457 dy00 = _fjsp_sub_v2r8(iy0,jy0);
458 dz00 = _fjsp_sub_v2r8(iz0,jz0);
460 /* Calculate squared distance and things based on it */
461 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
463 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
465 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
467 /* Load parameters for j particles */
468 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
470 /**************************
471 * CALCULATE INTERACTIONS *
472 **************************/
474 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
477 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
479 /* Compute parameters for interactions between i and j atoms */
480 qq00 = _fjsp_mul_v2r8(iq0,jq0);
482 /* EWALD ELECTROSTATICS */
484 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
485 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
486 itab_tmp = _fjsp_dtox_v2r8(ewrt);
487 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
488 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
490 gmx_fjsp_load_2pair_swizzle_v2r8(ewtab+ewconv.i[0],ewtab+ewconv.i[1],
492 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
493 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
495 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
499 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
501 /* Update vectorial force */
502 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
503 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
504 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
506 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
510 /* Inner loop uses 42 flops */
517 j_coord_offsetA = DIM*jnrA;
519 /* load j atom coordinates */
520 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
523 /* Calculate displacement vector */
524 dx00 = _fjsp_sub_v2r8(ix0,jx0);
525 dy00 = _fjsp_sub_v2r8(iy0,jy0);
526 dz00 = _fjsp_sub_v2r8(iz0,jz0);
528 /* Calculate squared distance and things based on it */
529 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
531 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
533 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
535 /* Load parameters for j particles */
536 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
538 /**************************
539 * CALCULATE INTERACTIONS *
540 **************************/
542 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
545 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
547 /* Compute parameters for interactions between i and j atoms */
548 qq00 = _fjsp_mul_v2r8(iq0,jq0);
550 /* EWALD ELECTROSTATICS */
552 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
553 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
554 itab_tmp = _fjsp_dtox_v2r8(ewrt);
555 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
556 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
558 gmx_fjsp_load_1pair_swizzle_v2r8(ewtab+ewconv.i[0],&ewtabF,&ewtabFn);
559 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
560 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
562 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
566 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
568 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
570 /* Update vectorial force */
571 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
572 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
573 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
575 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
579 /* Inner loop uses 42 flops */
582 /* End of innermost loop */
584 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
585 f+i_coord_offset,fshift+i_shift_offset);
587 /* Increment number of inner iterations */
588 inneriter += j_index_end - j_index_start;
590 /* Outer loop uses 7 flops */
593 /* Increment number of outer iterations */
596 /* Update outer/inner flops */
598 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_F,outeriter*7 + inneriter*42);