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 "types/simple.h"
44 #include "gromacs/math/vec.h"
47 #include "kernelutil_sparc64_hpc_ace_double.h"
50 * Gromacs nonbonded kernel: nb_kernel_ElecNone_VdwLJEw_GeomP1P1_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: None
52 * VdW interaction: LJEwald
53 * Geometry: Particle-Particle
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecNone_VdwLJEw_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;
85 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
88 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
89 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
92 _fjsp_v2r8 ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
93 _fjsp_v2r8 one_half = gmx_fjsp_set1_v2r8(0.5);
94 _fjsp_v2r8 minus_one = gmx_fjsp_set1_v2r8(-1.0);
96 _fjsp_v2r8 dummy_mask,cutoff_mask;
97 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
98 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
99 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
106 jindex = nlist->jindex;
108 shiftidx = nlist->shift;
110 shiftvec = fr->shift_vec[0];
111 fshift = fr->fshift[0];
112 nvdwtype = fr->ntype;
114 vdwtype = mdatoms->typeA;
115 vdwgridparam = fr->ljpme_c6grid;
116 sh_lj_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_lj_ewald);
117 ewclj = gmx_fjsp_set1_v2r8(fr->ewaldcoeff_lj);
118 ewclj2 = _fjsp_mul_v2r8(minus_one,_fjsp_mul_v2r8(ewclj,ewclj));
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 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
152 /* Reset potential sums */
153 vvdwsum = _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 vdwjidx0A = 2*vdwtype[jnrA+0];
183 vdwjidx0B = 2*vdwtype[jnrB+0];
185 /**************************
186 * CALCULATE INTERACTIONS *
187 **************************/
189 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
191 /* Compute parameters for interactions between i and j atoms */
192 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
193 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
195 c6grid_00 = gmx_fjsp_load_2real_swizzle_v2r8(vdwgridparam+vdwioffset0+vdwjidx0A,
196 vdwgridparam+vdwioffset0+vdwjidx0B);
198 /* Analytical LJ-PME */
199 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
200 ewcljrsq = _fjsp_mul_v2r8(ewclj2,rsq00);
201 ewclj6 = _fjsp_mul_v2r8(ewclj2,_fjsp_mul_v2r8(ewclj2,ewclj2));
202 exponent = gmx_simd_exp_d(-ewcljrsq);
203 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
204 poly = _fjsp_mul_v2r8(exponent,_fjsp_madd_v2r8(_fjsp_mul_v2r8(ewcljrsq,ewcljrsq),one_half,_fjsp_sub_v2r8(one,ewcljrsq)));
205 /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
206 vvdw6 = _fjsp_mul_v2r8(_fjsp_madd_v2r8(-c6grid_00,_fjsp_sub_v2r8(one,poly),c6_00),rinvsix);
207 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
208 vvdw = _fjsp_msub_v2r8(vvdw12,one_twelfth,_fjsp_mul_v2r8(vvdw6,one_sixth));
209 /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
210 fvdw = _fjsp_mul_v2r8(_fjsp_add_v2r8(vvdw12,_fjsp_msub_v2r8(_fjsp_mul_v2r8(c6grid_00,one_sixth),_fjsp_mul_v2r8(exponent,ewclj6),vvdw6)),rinvsq00);
212 /* Update potential sum for this i atom from the interaction with this j atom. */
213 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
217 /* Update vectorial force */
218 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
219 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
220 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
222 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
224 /* Inner loop uses 50 flops */
231 j_coord_offsetA = DIM*jnrA;
233 /* load j atom coordinates */
234 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
237 /* Calculate displacement vector */
238 dx00 = _fjsp_sub_v2r8(ix0,jx0);
239 dy00 = _fjsp_sub_v2r8(iy0,jy0);
240 dz00 = _fjsp_sub_v2r8(iz0,jz0);
242 /* Calculate squared distance and things based on it */
243 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
245 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
247 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
249 /* Load parameters for j particles */
250 vdwjidx0A = 2*vdwtype[jnrA+0];
252 /**************************
253 * CALCULATE INTERACTIONS *
254 **************************/
256 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
258 /* Compute parameters for interactions between i and j atoms */
259 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
261 c6grid_00 = gmx_fjsp_load_1real_swizzle_v2r8(vdwgridparam+vdwioffset0+vdwjidx0A);
263 /* Analytical LJ-PME */
264 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
265 ewcljrsq = _fjsp_mul_v2r8(ewclj2,rsq00);
266 ewclj6 = _fjsp_mul_v2r8(ewclj2,_fjsp_mul_v2r8(ewclj2,ewclj2));
267 exponent = gmx_simd_exp_d(-ewcljrsq);
268 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
269 poly = _fjsp_mul_v2r8(exponent,_fjsp_madd_v2r8(_fjsp_mul_v2r8(ewcljrsq,ewcljrsq),one_half,_fjsp_sub_v2r8(one,ewcljrsq)));
270 /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
271 vvdw6 = _fjsp_mul_v2r8(_fjsp_madd_v2r8(-c6grid_00,_fjsp_sub_v2r8(one,poly),c6_00),rinvsix);
272 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
273 vvdw = _fjsp_msub_v2r8(vvdw12,one_twelfth,_fjsp_mul_v2r8(vvdw6,one_sixth));
274 /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
275 fvdw = _fjsp_mul_v2r8(_fjsp_add_v2r8(vvdw12,_fjsp_msub_v2r8(_fjsp_mul_v2r8(c6grid_00,one_sixth),_fjsp_mul_v2r8(exponent,ewclj6),vvdw6)),rinvsq00);
277 /* Update potential sum for this i atom from the interaction with this j atom. */
278 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
279 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
283 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
285 /* Update vectorial force */
286 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
287 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
288 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
290 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
292 /* Inner loop uses 50 flops */
295 /* End of innermost loop */
297 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
298 f+i_coord_offset,fshift+i_shift_offset);
301 /* Update potential energies */
302 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
304 /* Increment number of inner iterations */
305 inneriter += j_index_end - j_index_start;
307 /* Outer loop uses 7 flops */
310 /* Increment number of outer iterations */
313 /* Update outer/inner flops */
315 inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*50);
318 * Gromacs nonbonded kernel: nb_kernel_ElecNone_VdwLJEw_GeomP1P1_F_sparc64_hpc_ace_double
319 * Electrostatics interaction: None
320 * VdW interaction: LJEwald
321 * Geometry: Particle-Particle
322 * Calculate force/pot: Force
325 nb_kernel_ElecNone_VdwLJEw_GeomP1P1_F_sparc64_hpc_ace_double
326 (t_nblist * gmx_restrict nlist,
327 rvec * gmx_restrict xx,
328 rvec * gmx_restrict ff,
329 t_forcerec * gmx_restrict fr,
330 t_mdatoms * gmx_restrict mdatoms,
331 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
332 t_nrnb * gmx_restrict nrnb)
334 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
335 * just 0 for non-waters.
336 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
337 * jnr indices corresponding to data put in the four positions in the SIMD register.
339 int i_shift_offset,i_coord_offset,outeriter,inneriter;
340 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
342 int j_coord_offsetA,j_coord_offsetB;
343 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
345 real *shiftvec,*fshift,*x,*f;
346 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
348 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
349 int vdwjidx0A,vdwjidx0B;
350 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
351 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
353 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
356 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
357 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
358 _fjsp_v2r8 c6grid_00;
360 _fjsp_v2r8 ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
361 _fjsp_v2r8 one_half = gmx_fjsp_set1_v2r8(0.5);
362 _fjsp_v2r8 minus_one = gmx_fjsp_set1_v2r8(-1.0);
364 _fjsp_v2r8 dummy_mask,cutoff_mask;
365 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
366 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
367 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
374 jindex = nlist->jindex;
376 shiftidx = nlist->shift;
378 shiftvec = fr->shift_vec[0];
379 fshift = fr->fshift[0];
380 nvdwtype = fr->ntype;
382 vdwtype = mdatoms->typeA;
383 vdwgridparam = fr->ljpme_c6grid;
384 sh_lj_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_lj_ewald);
385 ewclj = gmx_fjsp_set1_v2r8(fr->ewaldcoeff_lj);
386 ewclj2 = _fjsp_mul_v2r8(minus_one,_fjsp_mul_v2r8(ewclj,ewclj));
388 /* Avoid stupid compiler warnings */
396 /* Start outer loop over neighborlists */
397 for(iidx=0; iidx<nri; iidx++)
399 /* Load shift vector for this list */
400 i_shift_offset = DIM*shiftidx[iidx];
402 /* Load limits for loop over neighbors */
403 j_index_start = jindex[iidx];
404 j_index_end = jindex[iidx+1];
406 /* Get outer coordinate index */
408 i_coord_offset = DIM*inr;
410 /* Load i particle coords and add shift vector */
411 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
413 fix0 = _fjsp_setzero_v2r8();
414 fiy0 = _fjsp_setzero_v2r8();
415 fiz0 = _fjsp_setzero_v2r8();
417 /* Load parameters for i particles */
418 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
420 /* Start inner kernel loop */
421 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
424 /* Get j neighbor index, and coordinate index */
427 j_coord_offsetA = DIM*jnrA;
428 j_coord_offsetB = DIM*jnrB;
430 /* load j atom coordinates */
431 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
434 /* Calculate displacement vector */
435 dx00 = _fjsp_sub_v2r8(ix0,jx0);
436 dy00 = _fjsp_sub_v2r8(iy0,jy0);
437 dz00 = _fjsp_sub_v2r8(iz0,jz0);
439 /* Calculate squared distance and things based on it */
440 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
442 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
444 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
446 /* Load parameters for j particles */
447 vdwjidx0A = 2*vdwtype[jnrA+0];
448 vdwjidx0B = 2*vdwtype[jnrB+0];
450 /**************************
451 * CALCULATE INTERACTIONS *
452 **************************/
454 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
456 /* Compute parameters for interactions between i and j atoms */
457 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
458 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
460 c6grid_00 = gmx_fjsp_load_2real_swizzle_v2r8(vdwgridparam+vdwioffset0+vdwjidx0A,
461 vdwgridparam+vdwioffset0+vdwjidx0B);
463 /* Analytical LJ-PME */
464 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
465 ewcljrsq = _fjsp_mul_v2r8(ewclj2,rsq00);
466 ewclj6 = _fjsp_mul_v2r8(ewclj2,_fjsp_mul_v2r8(ewclj2,ewclj2));
467 exponent = gmx_simd_exp_d(-ewcljrsq);
468 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
469 poly = _fjsp_mul_v2r8(exponent,_fjsp_madd_v2r8(_fjsp_mul_v2r8(ewcljrsq,ewcljrsq),one_half,_fjsp_sub_v2r8(one,ewcljrsq)));
470 /* f6A = 6 * C6grid * (1 - poly) */
471 f6A = _fjsp_mul_v2r8(c6grid_00,_fjsp_msub_v2r8(one,poly));
472 /* f6B = C6grid * exponent * beta^6 */
473 f6B = _fjsp_mul_v2r8(_fjsp_mul_v2r8(c6grid_00,one_sixth),_fjsp_mul_v2r8(exponent,ewclj6));
474 /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
475 fvdw = _fjsp_mul_v2r8(_fjsp_madd_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,_fjsp_sub_v2r8(c6_00,f6A)),rinvsix,f6B),rinvsq00);
479 /* Update vectorial force */
480 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
481 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
482 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
484 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
486 /* Inner loop uses 48 flops */
493 j_coord_offsetA = DIM*jnrA;
495 /* load j atom coordinates */
496 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
499 /* Calculate displacement vector */
500 dx00 = _fjsp_sub_v2r8(ix0,jx0);
501 dy00 = _fjsp_sub_v2r8(iy0,jy0);
502 dz00 = _fjsp_sub_v2r8(iz0,jz0);
504 /* Calculate squared distance and things based on it */
505 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
507 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
509 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
511 /* Load parameters for j particles */
512 vdwjidx0A = 2*vdwtype[jnrA+0];
514 /**************************
515 * CALCULATE INTERACTIONS *
516 **************************/
518 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
520 /* Compute parameters for interactions between i and j atoms */
521 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
523 c6grid_00 = gmx_fjsp_load_1real_swizzle_v2r8(vdwgridparam+vdwioffset0+vdwjidx0A);
525 /* Analytical LJ-PME */
526 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
527 ewcljrsq = _fjsp_mul_v2r8(ewclj2,rsq00);
528 ewclj6 = _fjsp_mul_v2r8(ewclj2,_fjsp_mul_v2r8(ewclj2,ewclj2));
529 exponent = gmx_simd_exp_d(-ewcljrsq);
530 /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
531 poly = _fjsp_mul_v2r8(exponent,_fjsp_madd_v2r8(_fjsp_mul_v2r8(ewcljrsq,ewcljrsq),one_half,_fjsp_sub_v2r8(one,ewcljrsq)));
532 /* f6A = 6 * C6grid * (1 - poly) */
533 f6A = _fjsp_mul_v2r8(c6grid_00,_fjsp_msub_v2r8(one,poly));
534 /* f6B = C6grid * exponent * beta^6 */
535 f6B = _fjsp_mul_v2r8(_fjsp_mul_v2r8(c6grid_00,one_sixth),_fjsp_mul_v2r8(exponent,ewclj6));
536 /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
537 fvdw = _fjsp_mul_v2r8(_fjsp_madd_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,_fjsp_sub_v2r8(c6_00,f6A)),rinvsix,f6B),rinvsq00);
541 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
543 /* Update vectorial force */
544 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
545 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
546 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
548 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
550 /* Inner loop uses 48 flops */
553 /* End of innermost loop */
555 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
556 f+i_coord_offset,fshift+i_shift_offset);
558 /* Increment number of inner iterations */
559 inneriter += j_index_end - j_index_start;
561 /* Outer loop uses 6 flops */
564 /* Increment number of outer iterations */
567 /* Update outer/inner flops */
569 inc_nrnb(nrnb,eNR_NBKERNEL_VDW_F,outeriter*6 + inneriter*48);