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_ElecEw_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: Ewald
52 * VdW interaction: LennardJones
53 * Geometry: Particle-Particle
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecEw_VdwLJ_GeomP1P1_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;
81 int vdwjidx0A,vdwjidx0B;
82 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
83 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
84 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
87 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
90 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
91 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
92 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
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 nvdwtype = fr->ntype;
115 vdwtype = mdatoms->typeA;
117 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
118 ewtab = fr->ic->tabq_coul_FDV0;
119 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
120 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
122 /* Avoid stupid compiler warnings */
130 /* Start outer loop over neighborlists */
131 for(iidx=0; iidx<nri; iidx++)
133 /* Load shift vector for this list */
134 i_shift_offset = DIM*shiftidx[iidx];
136 /* Load limits for loop over neighbors */
137 j_index_start = jindex[iidx];
138 j_index_end = jindex[iidx+1];
140 /* Get outer coordinate index */
142 i_coord_offset = DIM*inr;
144 /* Load i particle coords and add shift vector */
145 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
147 fix0 = _fjsp_setzero_v2r8();
148 fiy0 = _fjsp_setzero_v2r8();
149 fiz0 = _fjsp_setzero_v2r8();
151 /* Load parameters for i particles */
152 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
153 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
155 /* Reset potential sums */
156 velecsum = _fjsp_setzero_v2r8();
157 vvdwsum = _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 dx00 = _fjsp_sub_v2r8(ix0,jx0);
175 dy00 = _fjsp_sub_v2r8(iy0,jy0);
176 dz00 = _fjsp_sub_v2r8(iz0,jz0);
178 /* Calculate squared distance and things based on it */
179 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
181 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
183 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
185 /* Load parameters for j particles */
186 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
187 vdwjidx0A = 2*vdwtype[jnrA+0];
188 vdwjidx0B = 2*vdwtype[jnrB+0];
190 /**************************
191 * CALCULATE INTERACTIONS *
192 **************************/
194 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
196 /* Compute parameters for interactions between i and j atoms */
197 qq00 = _fjsp_mul_v2r8(iq0,jq0);
198 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
199 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
201 /* EWALD ELECTROSTATICS */
203 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
204 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
205 itab_tmp = _fjsp_dtox_v2r8(ewrt);
206 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
207 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
209 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
210 ewtabD = _fjsp_load_v2r8( ewtab + 4*ewconv.i[1] );
211 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
212 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
213 ewtabFn = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[1] +2);
214 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
215 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
216 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
217 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(rinv00,velec));
218 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
220 /* LENNARD-JONES DISPERSION/REPULSION */
222 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
223 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
224 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
225 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
226 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
228 /* Update potential sum for this i atom from the interaction with this j atom. */
229 velecsum = _fjsp_add_v2r8(velecsum,velec);
230 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
232 fscal = _fjsp_add_v2r8(felec,fvdw);
234 /* Update vectorial force */
235 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
236 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
237 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
239 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
241 /* Inner loop uses 56 flops */
248 j_coord_offsetA = DIM*jnrA;
250 /* load j atom coordinates */
251 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
254 /* Calculate displacement vector */
255 dx00 = _fjsp_sub_v2r8(ix0,jx0);
256 dy00 = _fjsp_sub_v2r8(iy0,jy0);
257 dz00 = _fjsp_sub_v2r8(iz0,jz0);
259 /* Calculate squared distance and things based on it */
260 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
262 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
264 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
266 /* Load parameters for j particles */
267 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
268 vdwjidx0A = 2*vdwtype[jnrA+0];
270 /**************************
271 * CALCULATE INTERACTIONS *
272 **************************/
274 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
276 /* Compute parameters for interactions between i and j atoms */
277 qq00 = _fjsp_mul_v2r8(iq0,jq0);
278 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
280 /* EWALD ELECTROSTATICS */
282 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
283 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
284 itab_tmp = _fjsp_dtox_v2r8(ewrt);
285 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
286 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
288 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
289 ewtabD = _fjsp_setzero_v2r8();
290 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
291 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
292 ewtabFn = _fjsp_setzero_v2r8();
293 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
294 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
295 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
296 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(rinv00,velec));
297 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
299 /* LENNARD-JONES DISPERSION/REPULSION */
301 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
302 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
303 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
304 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
305 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
307 /* Update potential sum for this i atom from the interaction with this j atom. */
308 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
309 velecsum = _fjsp_add_v2r8(velecsum,velec);
310 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
311 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
313 fscal = _fjsp_add_v2r8(felec,fvdw);
315 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
317 /* Update vectorial force */
318 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
319 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
320 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
322 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
324 /* Inner loop uses 56 flops */
327 /* End of innermost loop */
329 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
330 f+i_coord_offset,fshift+i_shift_offset);
333 /* Update potential energies */
334 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
335 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
337 /* Increment number of inner iterations */
338 inneriter += j_index_end - j_index_start;
340 /* Outer loop uses 9 flops */
343 /* Increment number of outer iterations */
346 /* Update outer/inner flops */
348 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*9 + inneriter*56);
351 * Gromacs nonbonded kernel: nb_kernel_ElecEw_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
352 * Electrostatics interaction: Ewald
353 * VdW interaction: LennardJones
354 * Geometry: Particle-Particle
355 * Calculate force/pot: Force
358 nb_kernel_ElecEw_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
359 (t_nblist * gmx_restrict nlist,
360 rvec * gmx_restrict xx,
361 rvec * gmx_restrict ff,
362 t_forcerec * gmx_restrict fr,
363 t_mdatoms * gmx_restrict mdatoms,
364 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
365 t_nrnb * gmx_restrict nrnb)
367 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
368 * just 0 for non-waters.
369 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
370 * jnr indices corresponding to data put in the four positions in the SIMD register.
372 int i_shift_offset,i_coord_offset,outeriter,inneriter;
373 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
375 int j_coord_offsetA,j_coord_offsetB;
376 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
378 real *shiftvec,*fshift,*x,*f;
379 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
381 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
382 int vdwjidx0A,vdwjidx0B;
383 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
384 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
385 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
388 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
391 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
392 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
393 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
396 _fjsp_v2r8 dummy_mask,cutoff_mask;
397 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
398 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
399 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
406 jindex = nlist->jindex;
408 shiftidx = nlist->shift;
410 shiftvec = fr->shift_vec[0];
411 fshift = fr->fshift[0];
412 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
413 charge = mdatoms->chargeA;
414 nvdwtype = fr->ntype;
416 vdwtype = mdatoms->typeA;
418 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
419 ewtab = fr->ic->tabq_coul_F;
420 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
421 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
423 /* Avoid stupid compiler warnings */
431 /* Start outer loop over neighborlists */
432 for(iidx=0; iidx<nri; iidx++)
434 /* Load shift vector for this list */
435 i_shift_offset = DIM*shiftidx[iidx];
437 /* Load limits for loop over neighbors */
438 j_index_start = jindex[iidx];
439 j_index_end = jindex[iidx+1];
441 /* Get outer coordinate index */
443 i_coord_offset = DIM*inr;
445 /* Load i particle coords and add shift vector */
446 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
448 fix0 = _fjsp_setzero_v2r8();
449 fiy0 = _fjsp_setzero_v2r8();
450 fiz0 = _fjsp_setzero_v2r8();
452 /* Load parameters for i particles */
453 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
454 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
456 /* Start inner kernel loop */
457 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
460 /* Get j neighbor index, and coordinate index */
463 j_coord_offsetA = DIM*jnrA;
464 j_coord_offsetB = DIM*jnrB;
466 /* load j atom coordinates */
467 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
470 /* Calculate displacement vector */
471 dx00 = _fjsp_sub_v2r8(ix0,jx0);
472 dy00 = _fjsp_sub_v2r8(iy0,jy0);
473 dz00 = _fjsp_sub_v2r8(iz0,jz0);
475 /* Calculate squared distance and things based on it */
476 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
478 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
480 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
482 /* Load parameters for j particles */
483 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
484 vdwjidx0A = 2*vdwtype[jnrA+0];
485 vdwjidx0B = 2*vdwtype[jnrB+0];
487 /**************************
488 * CALCULATE INTERACTIONS *
489 **************************/
491 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
493 /* Compute parameters for interactions between i and j atoms */
494 qq00 = _fjsp_mul_v2r8(iq0,jq0);
495 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
496 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
498 /* EWALD ELECTROSTATICS */
500 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
501 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
502 itab_tmp = _fjsp_dtox_v2r8(ewrt);
503 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
504 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
506 gmx_fjsp_load_2pair_swizzle_v2r8(ewtab+ewconv.i[0],ewtab+ewconv.i[1],
508 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
509 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
511 /* LENNARD-JONES DISPERSION/REPULSION */
513 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
514 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
516 fscal = _fjsp_add_v2r8(felec,fvdw);
518 /* Update vectorial force */
519 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
520 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
521 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
523 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
525 /* Inner loop uses 46 flops */
532 j_coord_offsetA = DIM*jnrA;
534 /* load j atom coordinates */
535 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
538 /* Calculate displacement vector */
539 dx00 = _fjsp_sub_v2r8(ix0,jx0);
540 dy00 = _fjsp_sub_v2r8(iy0,jy0);
541 dz00 = _fjsp_sub_v2r8(iz0,jz0);
543 /* Calculate squared distance and things based on it */
544 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
546 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
548 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
550 /* Load parameters for j particles */
551 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
552 vdwjidx0A = 2*vdwtype[jnrA+0];
554 /**************************
555 * CALCULATE INTERACTIONS *
556 **************************/
558 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
560 /* Compute parameters for interactions between i and j atoms */
561 qq00 = _fjsp_mul_v2r8(iq0,jq0);
562 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
564 /* EWALD ELECTROSTATICS */
566 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
567 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
568 itab_tmp = _fjsp_dtox_v2r8(ewrt);
569 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
570 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
572 gmx_fjsp_load_1pair_swizzle_v2r8(ewtab+ewconv.i[0],&ewtabF,&ewtabFn);
573 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
574 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
576 /* LENNARD-JONES DISPERSION/REPULSION */
578 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
579 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
581 fscal = _fjsp_add_v2r8(felec,fvdw);
583 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
585 /* Update vectorial force */
586 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
587 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
588 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
590 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
592 /* Inner loop uses 46 flops */
595 /* End of innermost loop */
597 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
598 f+i_coord_offset,fshift+i_shift_offset);
600 /* Increment number of inner iterations */
601 inneriter += j_index_end - j_index_start;
603 /* Outer loop uses 7 flops */
606 /* Increment number of outer iterations */
609 /* Update outer/inner flops */
611 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*7 + inneriter*46);