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_ElecEw_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: Ewald
54 * VdW interaction: LennardJones
55 * Geometry: Particle-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecEw_VdwLJ_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_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;
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;
89 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
92 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
93 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
94 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
97 _fjsp_v2r8 dummy_mask,cutoff_mask;
98 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
99 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
100 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
107 jindex = nlist->jindex;
109 shiftidx = nlist->shift;
111 shiftvec = fr->shift_vec[0];
112 fshift = fr->fshift[0];
113 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
114 charge = mdatoms->chargeA;
115 nvdwtype = fr->ntype;
117 vdwtype = mdatoms->typeA;
119 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
120 ewtab = fr->ic->tabq_coul_FDV0;
121 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
122 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
124 /* Avoid stupid compiler warnings */
132 /* Start outer loop over neighborlists */
133 for(iidx=0; iidx<nri; iidx++)
135 /* Load shift vector for this list */
136 i_shift_offset = DIM*shiftidx[iidx];
138 /* Load limits for loop over neighbors */
139 j_index_start = jindex[iidx];
140 j_index_end = jindex[iidx+1];
142 /* Get outer coordinate index */
144 i_coord_offset = DIM*inr;
146 /* Load i particle coords and add shift vector */
147 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
149 fix0 = _fjsp_setzero_v2r8();
150 fiy0 = _fjsp_setzero_v2r8();
151 fiz0 = _fjsp_setzero_v2r8();
153 /* Load parameters for i particles */
154 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
155 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
157 /* Reset potential sums */
158 velecsum = _fjsp_setzero_v2r8();
159 vvdwsum = _fjsp_setzero_v2r8();
161 /* Start inner kernel loop */
162 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
165 /* Get j neighbor index, and coordinate index */
168 j_coord_offsetA = DIM*jnrA;
169 j_coord_offsetB = DIM*jnrB;
171 /* load j atom coordinates */
172 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
175 /* Calculate displacement vector */
176 dx00 = _fjsp_sub_v2r8(ix0,jx0);
177 dy00 = _fjsp_sub_v2r8(iy0,jy0);
178 dz00 = _fjsp_sub_v2r8(iz0,jz0);
180 /* Calculate squared distance and things based on it */
181 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
183 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
185 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
187 /* Load parameters for j particles */
188 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
189 vdwjidx0A = 2*vdwtype[jnrA+0];
190 vdwjidx0B = 2*vdwtype[jnrB+0];
192 /**************************
193 * CALCULATE INTERACTIONS *
194 **************************/
196 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
198 /* Compute parameters for interactions between i and j atoms */
199 qq00 = _fjsp_mul_v2r8(iq0,jq0);
200 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
201 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
203 /* EWALD ELECTROSTATICS */
205 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
206 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
207 itab_tmp = _fjsp_dtox_v2r8(ewrt);
208 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
209 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
211 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
212 ewtabD = _fjsp_load_v2r8( ewtab + 4*ewconv.i[1] );
213 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
214 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
215 ewtabFn = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[1] +2);
216 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
217 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
218 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
219 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(rinv00,velec));
220 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
222 /* LENNARD-JONES DISPERSION/REPULSION */
224 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
225 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
226 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
227 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
228 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
230 /* Update potential sum for this i atom from the interaction with this j atom. */
231 velecsum = _fjsp_add_v2r8(velecsum,velec);
232 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
234 fscal = _fjsp_add_v2r8(felec,fvdw);
236 /* Update vectorial force */
237 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
238 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
239 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
241 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
243 /* Inner loop uses 56 flops */
250 j_coord_offsetA = DIM*jnrA;
252 /* load j atom coordinates */
253 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
256 /* Calculate displacement vector */
257 dx00 = _fjsp_sub_v2r8(ix0,jx0);
258 dy00 = _fjsp_sub_v2r8(iy0,jy0);
259 dz00 = _fjsp_sub_v2r8(iz0,jz0);
261 /* Calculate squared distance and things based on it */
262 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
264 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
266 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
268 /* Load parameters for j particles */
269 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
270 vdwjidx0A = 2*vdwtype[jnrA+0];
272 /**************************
273 * CALCULATE INTERACTIONS *
274 **************************/
276 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
278 /* Compute parameters for interactions between i and j atoms */
279 qq00 = _fjsp_mul_v2r8(iq0,jq0);
280 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
282 /* EWALD ELECTROSTATICS */
284 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
285 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
286 itab_tmp = _fjsp_dtox_v2r8(ewrt);
287 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
288 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
290 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
291 ewtabD = _fjsp_setzero_v2r8();
292 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
293 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
294 ewtabFn = _fjsp_setzero_v2r8();
295 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
296 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
297 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
298 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(rinv00,velec));
299 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
301 /* LENNARD-JONES DISPERSION/REPULSION */
303 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
304 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
305 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
306 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
307 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
309 /* Update potential sum for this i atom from the interaction with this j atom. */
310 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
311 velecsum = _fjsp_add_v2r8(velecsum,velec);
312 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
313 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
315 fscal = _fjsp_add_v2r8(felec,fvdw);
317 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
319 /* Update vectorial force */
320 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
321 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
322 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
324 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
326 /* Inner loop uses 56 flops */
329 /* End of innermost loop */
331 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
332 f+i_coord_offset,fshift+i_shift_offset);
335 /* Update potential energies */
336 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
337 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
339 /* Increment number of inner iterations */
340 inneriter += j_index_end - j_index_start;
342 /* Outer loop uses 9 flops */
345 /* Increment number of outer iterations */
348 /* Update outer/inner flops */
350 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*9 + inneriter*56);
353 * Gromacs nonbonded kernel: nb_kernel_ElecEw_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
354 * Electrostatics interaction: Ewald
355 * VdW interaction: LennardJones
356 * Geometry: Particle-Particle
357 * Calculate force/pot: Force
360 nb_kernel_ElecEw_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
361 (t_nblist * gmx_restrict nlist,
362 rvec * gmx_restrict xx,
363 rvec * gmx_restrict ff,
364 t_forcerec * gmx_restrict fr,
365 t_mdatoms * gmx_restrict mdatoms,
366 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
367 t_nrnb * gmx_restrict nrnb)
369 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
370 * just 0 for non-waters.
371 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
372 * jnr indices corresponding to data put in the four positions in the SIMD register.
374 int i_shift_offset,i_coord_offset,outeriter,inneriter;
375 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
377 int j_coord_offsetA,j_coord_offsetB;
378 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
380 real *shiftvec,*fshift,*x,*f;
381 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
383 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
384 int vdwjidx0A,vdwjidx0B;
385 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
386 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
387 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
390 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
393 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
394 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
395 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
398 _fjsp_v2r8 dummy_mask,cutoff_mask;
399 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
400 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
401 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
408 jindex = nlist->jindex;
410 shiftidx = nlist->shift;
412 shiftvec = fr->shift_vec[0];
413 fshift = fr->fshift[0];
414 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
415 charge = mdatoms->chargeA;
416 nvdwtype = fr->ntype;
418 vdwtype = mdatoms->typeA;
420 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
421 ewtab = fr->ic->tabq_coul_F;
422 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
423 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
425 /* Avoid stupid compiler warnings */
433 /* Start outer loop over neighborlists */
434 for(iidx=0; iidx<nri; iidx++)
436 /* Load shift vector for this list */
437 i_shift_offset = DIM*shiftidx[iidx];
439 /* Load limits for loop over neighbors */
440 j_index_start = jindex[iidx];
441 j_index_end = jindex[iidx+1];
443 /* Get outer coordinate index */
445 i_coord_offset = DIM*inr;
447 /* Load i particle coords and add shift vector */
448 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
450 fix0 = _fjsp_setzero_v2r8();
451 fiy0 = _fjsp_setzero_v2r8();
452 fiz0 = _fjsp_setzero_v2r8();
454 /* Load parameters for i particles */
455 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
456 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
458 /* Start inner kernel loop */
459 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
462 /* Get j neighbor index, and coordinate index */
465 j_coord_offsetA = DIM*jnrA;
466 j_coord_offsetB = DIM*jnrB;
468 /* load j atom coordinates */
469 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
472 /* Calculate displacement vector */
473 dx00 = _fjsp_sub_v2r8(ix0,jx0);
474 dy00 = _fjsp_sub_v2r8(iy0,jy0);
475 dz00 = _fjsp_sub_v2r8(iz0,jz0);
477 /* Calculate squared distance and things based on it */
478 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
480 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
482 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
484 /* Load parameters for j particles */
485 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
486 vdwjidx0A = 2*vdwtype[jnrA+0];
487 vdwjidx0B = 2*vdwtype[jnrB+0];
489 /**************************
490 * CALCULATE INTERACTIONS *
491 **************************/
493 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
495 /* Compute parameters for interactions between i and j atoms */
496 qq00 = _fjsp_mul_v2r8(iq0,jq0);
497 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
498 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
500 /* EWALD ELECTROSTATICS */
502 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
503 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
504 itab_tmp = _fjsp_dtox_v2r8(ewrt);
505 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
506 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
508 gmx_fjsp_load_2pair_swizzle_v2r8(ewtab+ewconv.i[0],ewtab+ewconv.i[1],
510 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
511 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
513 /* LENNARD-JONES DISPERSION/REPULSION */
515 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
516 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
518 fscal = _fjsp_add_v2r8(felec,fvdw);
520 /* Update vectorial force */
521 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
522 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
523 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
525 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
527 /* Inner loop uses 46 flops */
534 j_coord_offsetA = DIM*jnrA;
536 /* load j atom coordinates */
537 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
540 /* Calculate displacement vector */
541 dx00 = _fjsp_sub_v2r8(ix0,jx0);
542 dy00 = _fjsp_sub_v2r8(iy0,jy0);
543 dz00 = _fjsp_sub_v2r8(iz0,jz0);
545 /* Calculate squared distance and things based on it */
546 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
548 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
550 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
552 /* Load parameters for j particles */
553 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
554 vdwjidx0A = 2*vdwtype[jnrA+0];
556 /**************************
557 * CALCULATE INTERACTIONS *
558 **************************/
560 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
562 /* Compute parameters for interactions between i and j atoms */
563 qq00 = _fjsp_mul_v2r8(iq0,jq0);
564 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
566 /* EWALD ELECTROSTATICS */
568 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
569 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
570 itab_tmp = _fjsp_dtox_v2r8(ewrt);
571 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
572 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
574 gmx_fjsp_load_1pair_swizzle_v2r8(ewtab+ewconv.i[0],&ewtabF,&ewtabFn);
575 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
576 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
578 /* LENNARD-JONES DISPERSION/REPULSION */
580 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
581 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
583 fscal = _fjsp_add_v2r8(felec,fvdw);
585 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
587 /* Update vectorial force */
588 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
589 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
590 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
592 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
594 /* Inner loop uses 46 flops */
597 /* End of innermost loop */
599 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
600 f+i_coord_offset,fshift+i_shift_offset);
602 /* Increment number of inner iterations */
603 inneriter += j_index_end - j_index_start;
605 /* Outer loop uses 7 flops */
608 /* Increment number of outer iterations */
611 /* Update outer/inner flops */
613 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*7 + inneriter*46);