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_ElecCSTab_VdwLJ_GeomW3W3_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: CubicSplineTable
52 * VdW interaction: LennardJones
53 * Geometry: Water3-Water3
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecCSTab_VdwLJ_GeomW3W3_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 int vdwjidx1A,vdwjidx1B;
88 _fjsp_v2r8 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
89 int vdwjidx2A,vdwjidx2B;
90 _fjsp_v2r8 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
91 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
92 _fjsp_v2r8 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
93 _fjsp_v2r8 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
94 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
95 _fjsp_v2r8 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
96 _fjsp_v2r8 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
97 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
98 _fjsp_v2r8 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
99 _fjsp_v2r8 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
100 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
103 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
106 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
107 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
108 _fjsp_v2r8 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
111 _fjsp_v2r8 dummy_mask,cutoff_mask;
112 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
113 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
114 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
121 jindex = nlist->jindex;
123 shiftidx = nlist->shift;
125 shiftvec = fr->shift_vec[0];
126 fshift = fr->fshift[0];
127 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
128 charge = mdatoms->chargeA;
129 nvdwtype = fr->ntype;
131 vdwtype = mdatoms->typeA;
133 vftab = kernel_data->table_elec->data;
134 vftabscale = gmx_fjsp_set1_v2r8(kernel_data->table_elec->scale);
136 /* Setup water-specific parameters */
137 inr = nlist->iinr[0];
138 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
139 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
140 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
141 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
143 jq0 = gmx_fjsp_set1_v2r8(charge[inr+0]);
144 jq1 = gmx_fjsp_set1_v2r8(charge[inr+1]);
145 jq2 = gmx_fjsp_set1_v2r8(charge[inr+2]);
146 vdwjidx0A = 2*vdwtype[inr+0];
147 qq00 = _fjsp_mul_v2r8(iq0,jq0);
148 c6_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
149 c12_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
150 qq01 = _fjsp_mul_v2r8(iq0,jq1);
151 qq02 = _fjsp_mul_v2r8(iq0,jq2);
152 qq10 = _fjsp_mul_v2r8(iq1,jq0);
153 qq11 = _fjsp_mul_v2r8(iq1,jq1);
154 qq12 = _fjsp_mul_v2r8(iq1,jq2);
155 qq20 = _fjsp_mul_v2r8(iq2,jq0);
156 qq21 = _fjsp_mul_v2r8(iq2,jq1);
157 qq22 = _fjsp_mul_v2r8(iq2,jq2);
159 /* Avoid stupid compiler warnings */
167 /* Start outer loop over neighborlists */
168 for(iidx=0; iidx<nri; iidx++)
170 /* Load shift vector for this list */
171 i_shift_offset = DIM*shiftidx[iidx];
173 /* Load limits for loop over neighbors */
174 j_index_start = jindex[iidx];
175 j_index_end = jindex[iidx+1];
177 /* Get outer coordinate index */
179 i_coord_offset = DIM*inr;
181 /* Load i particle coords and add shift vector */
182 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
183 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
185 fix0 = _fjsp_setzero_v2r8();
186 fiy0 = _fjsp_setzero_v2r8();
187 fiz0 = _fjsp_setzero_v2r8();
188 fix1 = _fjsp_setzero_v2r8();
189 fiy1 = _fjsp_setzero_v2r8();
190 fiz1 = _fjsp_setzero_v2r8();
191 fix2 = _fjsp_setzero_v2r8();
192 fiy2 = _fjsp_setzero_v2r8();
193 fiz2 = _fjsp_setzero_v2r8();
195 /* Reset potential sums */
196 velecsum = _fjsp_setzero_v2r8();
197 vvdwsum = _fjsp_setzero_v2r8();
199 /* Start inner kernel loop */
200 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
203 /* Get j neighbor index, and coordinate index */
206 j_coord_offsetA = DIM*jnrA;
207 j_coord_offsetB = DIM*jnrB;
209 /* load j atom coordinates */
210 gmx_fjsp_load_3rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
211 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
213 /* Calculate displacement vector */
214 dx00 = _fjsp_sub_v2r8(ix0,jx0);
215 dy00 = _fjsp_sub_v2r8(iy0,jy0);
216 dz00 = _fjsp_sub_v2r8(iz0,jz0);
217 dx01 = _fjsp_sub_v2r8(ix0,jx1);
218 dy01 = _fjsp_sub_v2r8(iy0,jy1);
219 dz01 = _fjsp_sub_v2r8(iz0,jz1);
220 dx02 = _fjsp_sub_v2r8(ix0,jx2);
221 dy02 = _fjsp_sub_v2r8(iy0,jy2);
222 dz02 = _fjsp_sub_v2r8(iz0,jz2);
223 dx10 = _fjsp_sub_v2r8(ix1,jx0);
224 dy10 = _fjsp_sub_v2r8(iy1,jy0);
225 dz10 = _fjsp_sub_v2r8(iz1,jz0);
226 dx11 = _fjsp_sub_v2r8(ix1,jx1);
227 dy11 = _fjsp_sub_v2r8(iy1,jy1);
228 dz11 = _fjsp_sub_v2r8(iz1,jz1);
229 dx12 = _fjsp_sub_v2r8(ix1,jx2);
230 dy12 = _fjsp_sub_v2r8(iy1,jy2);
231 dz12 = _fjsp_sub_v2r8(iz1,jz2);
232 dx20 = _fjsp_sub_v2r8(ix2,jx0);
233 dy20 = _fjsp_sub_v2r8(iy2,jy0);
234 dz20 = _fjsp_sub_v2r8(iz2,jz0);
235 dx21 = _fjsp_sub_v2r8(ix2,jx1);
236 dy21 = _fjsp_sub_v2r8(iy2,jy1);
237 dz21 = _fjsp_sub_v2r8(iz2,jz1);
238 dx22 = _fjsp_sub_v2r8(ix2,jx2);
239 dy22 = _fjsp_sub_v2r8(iy2,jy2);
240 dz22 = _fjsp_sub_v2r8(iz2,jz2);
242 /* Calculate squared distance and things based on it */
243 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
244 rsq01 = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
245 rsq02 = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
246 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
247 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
248 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
249 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
250 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
251 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
253 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
254 rinv01 = gmx_fjsp_invsqrt_v2r8(rsq01);
255 rinv02 = gmx_fjsp_invsqrt_v2r8(rsq02);
256 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
257 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
258 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
259 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
260 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
261 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
263 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
265 fjx0 = _fjsp_setzero_v2r8();
266 fjy0 = _fjsp_setzero_v2r8();
267 fjz0 = _fjsp_setzero_v2r8();
268 fjx1 = _fjsp_setzero_v2r8();
269 fjy1 = _fjsp_setzero_v2r8();
270 fjz1 = _fjsp_setzero_v2r8();
271 fjx2 = _fjsp_setzero_v2r8();
272 fjy2 = _fjsp_setzero_v2r8();
273 fjz2 = _fjsp_setzero_v2r8();
275 /**************************
276 * CALCULATE INTERACTIONS *
277 **************************/
279 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
281 /* Calculate table index by multiplying r with table scale and truncate to integer */
282 rt = _fjsp_mul_v2r8(r00,vftabscale);
283 itab_tmp = _fjsp_dtox_v2r8(rt);
284 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
285 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
286 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
291 /* CUBIC SPLINE TABLE ELECTROSTATICS */
292 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
293 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
294 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
295 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
296 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
297 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
298 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
299 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
300 velec = _fjsp_mul_v2r8(qq00,VV);
301 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
302 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,FF),_fjsp_mul_v2r8(vftabscale,rinv00)));
304 /* LENNARD-JONES DISPERSION/REPULSION */
306 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
307 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
308 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
309 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
310 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
312 /* Update potential sum for this i atom from the interaction with this j atom. */
313 velecsum = _fjsp_add_v2r8(velecsum,velec);
314 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
316 fscal = _fjsp_add_v2r8(felec,fvdw);
318 /* Update vectorial force */
319 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
320 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
321 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
323 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
324 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
325 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
327 /**************************
328 * CALCULATE INTERACTIONS *
329 **************************/
331 r01 = _fjsp_mul_v2r8(rsq01,rinv01);
333 /* Calculate table index by multiplying r with table scale and truncate to integer */
334 rt = _fjsp_mul_v2r8(r01,vftabscale);
335 itab_tmp = _fjsp_dtox_v2r8(rt);
336 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
337 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
338 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
343 /* CUBIC SPLINE TABLE ELECTROSTATICS */
344 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
345 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
346 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
347 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
348 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
349 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
350 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
351 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
352 velec = _fjsp_mul_v2r8(qq01,VV);
353 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
354 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq01,FF),_fjsp_mul_v2r8(vftabscale,rinv01)));
356 /* Update potential sum for this i atom from the interaction with this j atom. */
357 velecsum = _fjsp_add_v2r8(velecsum,velec);
361 /* Update vectorial force */
362 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
363 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
364 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
366 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
367 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
368 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
370 /**************************
371 * CALCULATE INTERACTIONS *
372 **************************/
374 r02 = _fjsp_mul_v2r8(rsq02,rinv02);
376 /* Calculate table index by multiplying r with table scale and truncate to integer */
377 rt = _fjsp_mul_v2r8(r02,vftabscale);
378 itab_tmp = _fjsp_dtox_v2r8(rt);
379 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
380 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
381 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
386 /* CUBIC SPLINE TABLE ELECTROSTATICS */
387 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
388 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
389 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
390 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
391 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
392 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
393 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
394 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
395 velec = _fjsp_mul_v2r8(qq02,VV);
396 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
397 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq02,FF),_fjsp_mul_v2r8(vftabscale,rinv02)));
399 /* Update potential sum for this i atom from the interaction with this j atom. */
400 velecsum = _fjsp_add_v2r8(velecsum,velec);
404 /* Update vectorial force */
405 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
406 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
407 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
409 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
410 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
411 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
413 /**************************
414 * CALCULATE INTERACTIONS *
415 **************************/
417 r10 = _fjsp_mul_v2r8(rsq10,rinv10);
419 /* Calculate table index by multiplying r with table scale and truncate to integer */
420 rt = _fjsp_mul_v2r8(r10,vftabscale);
421 itab_tmp = _fjsp_dtox_v2r8(rt);
422 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
423 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
424 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
429 /* CUBIC SPLINE TABLE ELECTROSTATICS */
430 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
431 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
432 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
433 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
434 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
435 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
436 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
437 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
438 velec = _fjsp_mul_v2r8(qq10,VV);
439 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
440 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq10,FF),_fjsp_mul_v2r8(vftabscale,rinv10)));
442 /* Update potential sum for this i atom from the interaction with this j atom. */
443 velecsum = _fjsp_add_v2r8(velecsum,velec);
447 /* Update vectorial force */
448 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
449 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
450 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
452 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
453 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
454 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
456 /**************************
457 * CALCULATE INTERACTIONS *
458 **************************/
460 r11 = _fjsp_mul_v2r8(rsq11,rinv11);
462 /* Calculate table index by multiplying r with table scale and truncate to integer */
463 rt = _fjsp_mul_v2r8(r11,vftabscale);
464 itab_tmp = _fjsp_dtox_v2r8(rt);
465 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
466 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
467 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
472 /* CUBIC SPLINE TABLE ELECTROSTATICS */
473 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
474 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
475 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
476 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
477 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
478 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
479 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
480 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
481 velec = _fjsp_mul_v2r8(qq11,VV);
482 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
483 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq11,FF),_fjsp_mul_v2r8(vftabscale,rinv11)));
485 /* Update potential sum for this i atom from the interaction with this j atom. */
486 velecsum = _fjsp_add_v2r8(velecsum,velec);
490 /* Update vectorial force */
491 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
492 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
493 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
495 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
496 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
497 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
499 /**************************
500 * CALCULATE INTERACTIONS *
501 **************************/
503 r12 = _fjsp_mul_v2r8(rsq12,rinv12);
505 /* Calculate table index by multiplying r with table scale and truncate to integer */
506 rt = _fjsp_mul_v2r8(r12,vftabscale);
507 itab_tmp = _fjsp_dtox_v2r8(rt);
508 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
509 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
510 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
515 /* CUBIC SPLINE TABLE ELECTROSTATICS */
516 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
517 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
518 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
519 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
520 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
521 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
522 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
523 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
524 velec = _fjsp_mul_v2r8(qq12,VV);
525 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
526 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq12,FF),_fjsp_mul_v2r8(vftabscale,rinv12)));
528 /* Update potential sum for this i atom from the interaction with this j atom. */
529 velecsum = _fjsp_add_v2r8(velecsum,velec);
533 /* Update vectorial force */
534 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
535 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
536 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
538 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
539 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
540 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
542 /**************************
543 * CALCULATE INTERACTIONS *
544 **************************/
546 r20 = _fjsp_mul_v2r8(rsq20,rinv20);
548 /* Calculate table index by multiplying r with table scale and truncate to integer */
549 rt = _fjsp_mul_v2r8(r20,vftabscale);
550 itab_tmp = _fjsp_dtox_v2r8(rt);
551 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
552 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
553 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
558 /* CUBIC SPLINE TABLE ELECTROSTATICS */
559 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
560 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
561 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
562 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
563 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
564 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
565 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
566 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
567 velec = _fjsp_mul_v2r8(qq20,VV);
568 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
569 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq20,FF),_fjsp_mul_v2r8(vftabscale,rinv20)));
571 /* Update potential sum for this i atom from the interaction with this j atom. */
572 velecsum = _fjsp_add_v2r8(velecsum,velec);
576 /* Update vectorial force */
577 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
578 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
579 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
581 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
582 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
583 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
585 /**************************
586 * CALCULATE INTERACTIONS *
587 **************************/
589 r21 = _fjsp_mul_v2r8(rsq21,rinv21);
591 /* Calculate table index by multiplying r with table scale and truncate to integer */
592 rt = _fjsp_mul_v2r8(r21,vftabscale);
593 itab_tmp = _fjsp_dtox_v2r8(rt);
594 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
595 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
596 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
601 /* CUBIC SPLINE TABLE ELECTROSTATICS */
602 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
603 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
604 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
605 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
606 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
607 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
608 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
609 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
610 velec = _fjsp_mul_v2r8(qq21,VV);
611 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
612 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq21,FF),_fjsp_mul_v2r8(vftabscale,rinv21)));
614 /* Update potential sum for this i atom from the interaction with this j atom. */
615 velecsum = _fjsp_add_v2r8(velecsum,velec);
619 /* Update vectorial force */
620 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
621 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
622 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
624 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
625 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
626 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
628 /**************************
629 * CALCULATE INTERACTIONS *
630 **************************/
632 r22 = _fjsp_mul_v2r8(rsq22,rinv22);
634 /* Calculate table index by multiplying r with table scale and truncate to integer */
635 rt = _fjsp_mul_v2r8(r22,vftabscale);
636 itab_tmp = _fjsp_dtox_v2r8(rt);
637 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
638 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
639 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
644 /* CUBIC SPLINE TABLE ELECTROSTATICS */
645 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
646 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
647 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
648 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
649 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
650 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
651 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
652 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
653 velec = _fjsp_mul_v2r8(qq22,VV);
654 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
655 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq22,FF),_fjsp_mul_v2r8(vftabscale,rinv22)));
657 /* Update potential sum for this i atom from the interaction with this j atom. */
658 velecsum = _fjsp_add_v2r8(velecsum,velec);
662 /* Update vectorial force */
663 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
664 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
665 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
667 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
668 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
669 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
671 gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
673 /* Inner loop uses 427 flops */
680 j_coord_offsetA = DIM*jnrA;
682 /* load j atom coordinates */
683 gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
684 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
686 /* Calculate displacement vector */
687 dx00 = _fjsp_sub_v2r8(ix0,jx0);
688 dy00 = _fjsp_sub_v2r8(iy0,jy0);
689 dz00 = _fjsp_sub_v2r8(iz0,jz0);
690 dx01 = _fjsp_sub_v2r8(ix0,jx1);
691 dy01 = _fjsp_sub_v2r8(iy0,jy1);
692 dz01 = _fjsp_sub_v2r8(iz0,jz1);
693 dx02 = _fjsp_sub_v2r8(ix0,jx2);
694 dy02 = _fjsp_sub_v2r8(iy0,jy2);
695 dz02 = _fjsp_sub_v2r8(iz0,jz2);
696 dx10 = _fjsp_sub_v2r8(ix1,jx0);
697 dy10 = _fjsp_sub_v2r8(iy1,jy0);
698 dz10 = _fjsp_sub_v2r8(iz1,jz0);
699 dx11 = _fjsp_sub_v2r8(ix1,jx1);
700 dy11 = _fjsp_sub_v2r8(iy1,jy1);
701 dz11 = _fjsp_sub_v2r8(iz1,jz1);
702 dx12 = _fjsp_sub_v2r8(ix1,jx2);
703 dy12 = _fjsp_sub_v2r8(iy1,jy2);
704 dz12 = _fjsp_sub_v2r8(iz1,jz2);
705 dx20 = _fjsp_sub_v2r8(ix2,jx0);
706 dy20 = _fjsp_sub_v2r8(iy2,jy0);
707 dz20 = _fjsp_sub_v2r8(iz2,jz0);
708 dx21 = _fjsp_sub_v2r8(ix2,jx1);
709 dy21 = _fjsp_sub_v2r8(iy2,jy1);
710 dz21 = _fjsp_sub_v2r8(iz2,jz1);
711 dx22 = _fjsp_sub_v2r8(ix2,jx2);
712 dy22 = _fjsp_sub_v2r8(iy2,jy2);
713 dz22 = _fjsp_sub_v2r8(iz2,jz2);
715 /* Calculate squared distance and things based on it */
716 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
717 rsq01 = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
718 rsq02 = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
719 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
720 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
721 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
722 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
723 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
724 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
726 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
727 rinv01 = gmx_fjsp_invsqrt_v2r8(rsq01);
728 rinv02 = gmx_fjsp_invsqrt_v2r8(rsq02);
729 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
730 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
731 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
732 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
733 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
734 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
736 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
738 fjx0 = _fjsp_setzero_v2r8();
739 fjy0 = _fjsp_setzero_v2r8();
740 fjz0 = _fjsp_setzero_v2r8();
741 fjx1 = _fjsp_setzero_v2r8();
742 fjy1 = _fjsp_setzero_v2r8();
743 fjz1 = _fjsp_setzero_v2r8();
744 fjx2 = _fjsp_setzero_v2r8();
745 fjy2 = _fjsp_setzero_v2r8();
746 fjz2 = _fjsp_setzero_v2r8();
748 /**************************
749 * CALCULATE INTERACTIONS *
750 **************************/
752 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
754 /* Calculate table index by multiplying r with table scale and truncate to integer */
755 rt = _fjsp_mul_v2r8(r00,vftabscale);
756 itab_tmp = _fjsp_dtox_v2r8(rt);
757 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
758 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
759 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
764 /* CUBIC SPLINE TABLE ELECTROSTATICS */
765 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
766 F = _fjsp_setzero_v2r8();
767 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
768 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
769 H = _fjsp_setzero_v2r8();
770 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
771 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
772 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
773 velec = _fjsp_mul_v2r8(qq00,VV);
774 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
775 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,FF),_fjsp_mul_v2r8(vftabscale,rinv00)));
777 /* LENNARD-JONES DISPERSION/REPULSION */
779 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
780 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
781 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
782 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
783 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
785 /* Update potential sum for this i atom from the interaction with this j atom. */
786 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
787 velecsum = _fjsp_add_v2r8(velecsum,velec);
788 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
789 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
791 fscal = _fjsp_add_v2r8(felec,fvdw);
793 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
795 /* Update vectorial force */
796 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
797 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
798 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
800 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
801 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
802 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
804 /**************************
805 * CALCULATE INTERACTIONS *
806 **************************/
808 r01 = _fjsp_mul_v2r8(rsq01,rinv01);
810 /* Calculate table index by multiplying r with table scale and truncate to integer */
811 rt = _fjsp_mul_v2r8(r01,vftabscale);
812 itab_tmp = _fjsp_dtox_v2r8(rt);
813 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
814 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
815 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
820 /* CUBIC SPLINE TABLE ELECTROSTATICS */
821 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
822 F = _fjsp_setzero_v2r8();
823 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
824 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
825 H = _fjsp_setzero_v2r8();
826 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
827 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
828 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
829 velec = _fjsp_mul_v2r8(qq01,VV);
830 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
831 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq01,FF),_fjsp_mul_v2r8(vftabscale,rinv01)));
833 /* Update potential sum for this i atom from the interaction with this j atom. */
834 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
835 velecsum = _fjsp_add_v2r8(velecsum,velec);
839 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
841 /* Update vectorial force */
842 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
843 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
844 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
846 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
847 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
848 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
850 /**************************
851 * CALCULATE INTERACTIONS *
852 **************************/
854 r02 = _fjsp_mul_v2r8(rsq02,rinv02);
856 /* Calculate table index by multiplying r with table scale and truncate to integer */
857 rt = _fjsp_mul_v2r8(r02,vftabscale);
858 itab_tmp = _fjsp_dtox_v2r8(rt);
859 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
860 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
861 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
866 /* CUBIC SPLINE TABLE ELECTROSTATICS */
867 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
868 F = _fjsp_setzero_v2r8();
869 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
870 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
871 H = _fjsp_setzero_v2r8();
872 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
873 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
874 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
875 velec = _fjsp_mul_v2r8(qq02,VV);
876 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
877 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq02,FF),_fjsp_mul_v2r8(vftabscale,rinv02)));
879 /* Update potential sum for this i atom from the interaction with this j atom. */
880 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
881 velecsum = _fjsp_add_v2r8(velecsum,velec);
885 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
887 /* Update vectorial force */
888 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
889 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
890 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
892 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
893 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
894 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
896 /**************************
897 * CALCULATE INTERACTIONS *
898 **************************/
900 r10 = _fjsp_mul_v2r8(rsq10,rinv10);
902 /* Calculate table index by multiplying r with table scale and truncate to integer */
903 rt = _fjsp_mul_v2r8(r10,vftabscale);
904 itab_tmp = _fjsp_dtox_v2r8(rt);
905 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
906 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
907 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
912 /* CUBIC SPLINE TABLE ELECTROSTATICS */
913 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
914 F = _fjsp_setzero_v2r8();
915 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
916 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
917 H = _fjsp_setzero_v2r8();
918 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
919 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
920 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
921 velec = _fjsp_mul_v2r8(qq10,VV);
922 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
923 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq10,FF),_fjsp_mul_v2r8(vftabscale,rinv10)));
925 /* Update potential sum for this i atom from the interaction with this j atom. */
926 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
927 velecsum = _fjsp_add_v2r8(velecsum,velec);
931 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
933 /* Update vectorial force */
934 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
935 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
936 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
938 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
939 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
940 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
942 /**************************
943 * CALCULATE INTERACTIONS *
944 **************************/
946 r11 = _fjsp_mul_v2r8(rsq11,rinv11);
948 /* Calculate table index by multiplying r with table scale and truncate to integer */
949 rt = _fjsp_mul_v2r8(r11,vftabscale);
950 itab_tmp = _fjsp_dtox_v2r8(rt);
951 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
952 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
953 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
958 /* CUBIC SPLINE TABLE ELECTROSTATICS */
959 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
960 F = _fjsp_setzero_v2r8();
961 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
962 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
963 H = _fjsp_setzero_v2r8();
964 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
965 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
966 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
967 velec = _fjsp_mul_v2r8(qq11,VV);
968 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
969 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq11,FF),_fjsp_mul_v2r8(vftabscale,rinv11)));
971 /* Update potential sum for this i atom from the interaction with this j atom. */
972 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
973 velecsum = _fjsp_add_v2r8(velecsum,velec);
977 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
979 /* Update vectorial force */
980 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
981 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
982 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
984 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
985 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
986 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
988 /**************************
989 * CALCULATE INTERACTIONS *
990 **************************/
992 r12 = _fjsp_mul_v2r8(rsq12,rinv12);
994 /* Calculate table index by multiplying r with table scale and truncate to integer */
995 rt = _fjsp_mul_v2r8(r12,vftabscale);
996 itab_tmp = _fjsp_dtox_v2r8(rt);
997 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
998 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
999 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1004 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1005 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1006 F = _fjsp_setzero_v2r8();
1007 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1008 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1009 H = _fjsp_setzero_v2r8();
1010 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1011 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1012 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
1013 velec = _fjsp_mul_v2r8(qq12,VV);
1014 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1015 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq12,FF),_fjsp_mul_v2r8(vftabscale,rinv12)));
1017 /* Update potential sum for this i atom from the interaction with this j atom. */
1018 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
1019 velecsum = _fjsp_add_v2r8(velecsum,velec);
1023 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1025 /* Update vectorial force */
1026 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
1027 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1028 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1030 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1031 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1032 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1034 /**************************
1035 * CALCULATE INTERACTIONS *
1036 **************************/
1038 r20 = _fjsp_mul_v2r8(rsq20,rinv20);
1040 /* Calculate table index by multiplying r with table scale and truncate to integer */
1041 rt = _fjsp_mul_v2r8(r20,vftabscale);
1042 itab_tmp = _fjsp_dtox_v2r8(rt);
1043 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1044 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1045 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1050 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1051 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1052 F = _fjsp_setzero_v2r8();
1053 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1054 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1055 H = _fjsp_setzero_v2r8();
1056 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1057 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1058 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
1059 velec = _fjsp_mul_v2r8(qq20,VV);
1060 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1061 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq20,FF),_fjsp_mul_v2r8(vftabscale,rinv20)));
1063 /* Update potential sum for this i atom from the interaction with this j atom. */
1064 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
1065 velecsum = _fjsp_add_v2r8(velecsum,velec);
1069 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1071 /* Update vectorial force */
1072 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
1073 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1074 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1076 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1077 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1078 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1080 /**************************
1081 * CALCULATE INTERACTIONS *
1082 **************************/
1084 r21 = _fjsp_mul_v2r8(rsq21,rinv21);
1086 /* Calculate table index by multiplying r with table scale and truncate to integer */
1087 rt = _fjsp_mul_v2r8(r21,vftabscale);
1088 itab_tmp = _fjsp_dtox_v2r8(rt);
1089 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1090 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1091 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1096 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1097 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1098 F = _fjsp_setzero_v2r8();
1099 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1100 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1101 H = _fjsp_setzero_v2r8();
1102 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1103 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1104 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
1105 velec = _fjsp_mul_v2r8(qq21,VV);
1106 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1107 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq21,FF),_fjsp_mul_v2r8(vftabscale,rinv21)));
1109 /* Update potential sum for this i atom from the interaction with this j atom. */
1110 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
1111 velecsum = _fjsp_add_v2r8(velecsum,velec);
1115 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1117 /* Update vectorial force */
1118 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
1119 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1120 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1122 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1123 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1124 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1126 /**************************
1127 * CALCULATE INTERACTIONS *
1128 **************************/
1130 r22 = _fjsp_mul_v2r8(rsq22,rinv22);
1132 /* Calculate table index by multiplying r with table scale and truncate to integer */
1133 rt = _fjsp_mul_v2r8(r22,vftabscale);
1134 itab_tmp = _fjsp_dtox_v2r8(rt);
1135 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1136 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1137 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1142 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1143 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1144 F = _fjsp_setzero_v2r8();
1145 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1146 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1147 H = _fjsp_setzero_v2r8();
1148 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1149 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1150 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
1151 velec = _fjsp_mul_v2r8(qq22,VV);
1152 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1153 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq22,FF),_fjsp_mul_v2r8(vftabscale,rinv22)));
1155 /* Update potential sum for this i atom from the interaction with this j atom. */
1156 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
1157 velecsum = _fjsp_add_v2r8(velecsum,velec);
1161 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1163 /* Update vectorial force */
1164 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
1165 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1166 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1168 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1169 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1170 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1172 gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1174 /* Inner loop uses 427 flops */
1177 /* End of innermost loop */
1179 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1180 f+i_coord_offset,fshift+i_shift_offset);
1183 /* Update potential energies */
1184 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
1185 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
1187 /* Increment number of inner iterations */
1188 inneriter += j_index_end - j_index_start;
1190 /* Outer loop uses 20 flops */
1193 /* Increment number of outer iterations */
1196 /* Update outer/inner flops */
1198 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*427);
1201 * Gromacs nonbonded kernel: nb_kernel_ElecCSTab_VdwLJ_GeomW3W3_F_sparc64_hpc_ace_double
1202 * Electrostatics interaction: CubicSplineTable
1203 * VdW interaction: LennardJones
1204 * Geometry: Water3-Water3
1205 * Calculate force/pot: Force
1208 nb_kernel_ElecCSTab_VdwLJ_GeomW3W3_F_sparc64_hpc_ace_double
1209 (t_nblist * gmx_restrict nlist,
1210 rvec * gmx_restrict xx,
1211 rvec * gmx_restrict ff,
1212 t_forcerec * gmx_restrict fr,
1213 t_mdatoms * gmx_restrict mdatoms,
1214 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1215 t_nrnb * gmx_restrict nrnb)
1217 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1218 * just 0 for non-waters.
1219 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
1220 * jnr indices corresponding to data put in the four positions in the SIMD register.
1222 int i_shift_offset,i_coord_offset,outeriter,inneriter;
1223 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1225 int j_coord_offsetA,j_coord_offsetB;
1226 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
1227 real rcutoff_scalar;
1228 real *shiftvec,*fshift,*x,*f;
1229 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1231 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1233 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1235 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1236 int vdwjidx0A,vdwjidx0B;
1237 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1238 int vdwjidx1A,vdwjidx1B;
1239 _fjsp_v2r8 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1240 int vdwjidx2A,vdwjidx2B;
1241 _fjsp_v2r8 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1242 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1243 _fjsp_v2r8 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
1244 _fjsp_v2r8 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
1245 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
1246 _fjsp_v2r8 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1247 _fjsp_v2r8 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1248 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
1249 _fjsp_v2r8 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1250 _fjsp_v2r8 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1251 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
1254 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1257 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
1258 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
1259 _fjsp_v2r8 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
1261 _fjsp_v2r8 itab_tmp;
1262 _fjsp_v2r8 dummy_mask,cutoff_mask;
1263 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
1264 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
1265 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
1272 jindex = nlist->jindex;
1274 shiftidx = nlist->shift;
1276 shiftvec = fr->shift_vec[0];
1277 fshift = fr->fshift[0];
1278 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
1279 charge = mdatoms->chargeA;
1280 nvdwtype = fr->ntype;
1281 vdwparam = fr->nbfp;
1282 vdwtype = mdatoms->typeA;
1284 vftab = kernel_data->table_elec->data;
1285 vftabscale = gmx_fjsp_set1_v2r8(kernel_data->table_elec->scale);
1287 /* Setup water-specific parameters */
1288 inr = nlist->iinr[0];
1289 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
1290 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
1291 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
1292 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
1294 jq0 = gmx_fjsp_set1_v2r8(charge[inr+0]);
1295 jq1 = gmx_fjsp_set1_v2r8(charge[inr+1]);
1296 jq2 = gmx_fjsp_set1_v2r8(charge[inr+2]);
1297 vdwjidx0A = 2*vdwtype[inr+0];
1298 qq00 = _fjsp_mul_v2r8(iq0,jq0);
1299 c6_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
1300 c12_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
1301 qq01 = _fjsp_mul_v2r8(iq0,jq1);
1302 qq02 = _fjsp_mul_v2r8(iq0,jq2);
1303 qq10 = _fjsp_mul_v2r8(iq1,jq0);
1304 qq11 = _fjsp_mul_v2r8(iq1,jq1);
1305 qq12 = _fjsp_mul_v2r8(iq1,jq2);
1306 qq20 = _fjsp_mul_v2r8(iq2,jq0);
1307 qq21 = _fjsp_mul_v2r8(iq2,jq1);
1308 qq22 = _fjsp_mul_v2r8(iq2,jq2);
1310 /* Avoid stupid compiler warnings */
1312 j_coord_offsetA = 0;
1313 j_coord_offsetB = 0;
1318 /* Start outer loop over neighborlists */
1319 for(iidx=0; iidx<nri; iidx++)
1321 /* Load shift vector for this list */
1322 i_shift_offset = DIM*shiftidx[iidx];
1324 /* Load limits for loop over neighbors */
1325 j_index_start = jindex[iidx];
1326 j_index_end = jindex[iidx+1];
1328 /* Get outer coordinate index */
1330 i_coord_offset = DIM*inr;
1332 /* Load i particle coords and add shift vector */
1333 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
1334 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1336 fix0 = _fjsp_setzero_v2r8();
1337 fiy0 = _fjsp_setzero_v2r8();
1338 fiz0 = _fjsp_setzero_v2r8();
1339 fix1 = _fjsp_setzero_v2r8();
1340 fiy1 = _fjsp_setzero_v2r8();
1341 fiz1 = _fjsp_setzero_v2r8();
1342 fix2 = _fjsp_setzero_v2r8();
1343 fiy2 = _fjsp_setzero_v2r8();
1344 fiz2 = _fjsp_setzero_v2r8();
1346 /* Start inner kernel loop */
1347 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1350 /* Get j neighbor index, and coordinate index */
1352 jnrB = jjnr[jidx+1];
1353 j_coord_offsetA = DIM*jnrA;
1354 j_coord_offsetB = DIM*jnrB;
1356 /* load j atom coordinates */
1357 gmx_fjsp_load_3rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
1358 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1360 /* Calculate displacement vector */
1361 dx00 = _fjsp_sub_v2r8(ix0,jx0);
1362 dy00 = _fjsp_sub_v2r8(iy0,jy0);
1363 dz00 = _fjsp_sub_v2r8(iz0,jz0);
1364 dx01 = _fjsp_sub_v2r8(ix0,jx1);
1365 dy01 = _fjsp_sub_v2r8(iy0,jy1);
1366 dz01 = _fjsp_sub_v2r8(iz0,jz1);
1367 dx02 = _fjsp_sub_v2r8(ix0,jx2);
1368 dy02 = _fjsp_sub_v2r8(iy0,jy2);
1369 dz02 = _fjsp_sub_v2r8(iz0,jz2);
1370 dx10 = _fjsp_sub_v2r8(ix1,jx0);
1371 dy10 = _fjsp_sub_v2r8(iy1,jy0);
1372 dz10 = _fjsp_sub_v2r8(iz1,jz0);
1373 dx11 = _fjsp_sub_v2r8(ix1,jx1);
1374 dy11 = _fjsp_sub_v2r8(iy1,jy1);
1375 dz11 = _fjsp_sub_v2r8(iz1,jz1);
1376 dx12 = _fjsp_sub_v2r8(ix1,jx2);
1377 dy12 = _fjsp_sub_v2r8(iy1,jy2);
1378 dz12 = _fjsp_sub_v2r8(iz1,jz2);
1379 dx20 = _fjsp_sub_v2r8(ix2,jx0);
1380 dy20 = _fjsp_sub_v2r8(iy2,jy0);
1381 dz20 = _fjsp_sub_v2r8(iz2,jz0);
1382 dx21 = _fjsp_sub_v2r8(ix2,jx1);
1383 dy21 = _fjsp_sub_v2r8(iy2,jy1);
1384 dz21 = _fjsp_sub_v2r8(iz2,jz1);
1385 dx22 = _fjsp_sub_v2r8(ix2,jx2);
1386 dy22 = _fjsp_sub_v2r8(iy2,jy2);
1387 dz22 = _fjsp_sub_v2r8(iz2,jz2);
1389 /* Calculate squared distance and things based on it */
1390 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1391 rsq01 = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1392 rsq02 = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1393 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1394 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1395 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1396 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1397 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1398 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1400 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
1401 rinv01 = gmx_fjsp_invsqrt_v2r8(rsq01);
1402 rinv02 = gmx_fjsp_invsqrt_v2r8(rsq02);
1403 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
1404 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
1405 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
1406 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
1407 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
1408 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
1410 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
1412 fjx0 = _fjsp_setzero_v2r8();
1413 fjy0 = _fjsp_setzero_v2r8();
1414 fjz0 = _fjsp_setzero_v2r8();
1415 fjx1 = _fjsp_setzero_v2r8();
1416 fjy1 = _fjsp_setzero_v2r8();
1417 fjz1 = _fjsp_setzero_v2r8();
1418 fjx2 = _fjsp_setzero_v2r8();
1419 fjy2 = _fjsp_setzero_v2r8();
1420 fjz2 = _fjsp_setzero_v2r8();
1422 /**************************
1423 * CALCULATE INTERACTIONS *
1424 **************************/
1426 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
1428 /* Calculate table index by multiplying r with table scale and truncate to integer */
1429 rt = _fjsp_mul_v2r8(r00,vftabscale);
1430 itab_tmp = _fjsp_dtox_v2r8(rt);
1431 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1432 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1433 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1438 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1439 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1440 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1441 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1442 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1443 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1444 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1445 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1446 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1447 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,FF),_fjsp_mul_v2r8(vftabscale,rinv00)));
1449 /* LENNARD-JONES DISPERSION/REPULSION */
1451 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
1452 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
1454 fscal = _fjsp_add_v2r8(felec,fvdw);
1456 /* Update vectorial force */
1457 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
1458 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1459 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1461 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1462 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1463 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1465 /**************************
1466 * CALCULATE INTERACTIONS *
1467 **************************/
1469 r01 = _fjsp_mul_v2r8(rsq01,rinv01);
1471 /* Calculate table index by multiplying r with table scale and truncate to integer */
1472 rt = _fjsp_mul_v2r8(r01,vftabscale);
1473 itab_tmp = _fjsp_dtox_v2r8(rt);
1474 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1475 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1476 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1481 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1482 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1483 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1484 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1485 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1486 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1487 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1488 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1489 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1490 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq01,FF),_fjsp_mul_v2r8(vftabscale,rinv01)));
1494 /* Update vectorial force */
1495 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
1496 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1497 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1499 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1500 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1501 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1503 /**************************
1504 * CALCULATE INTERACTIONS *
1505 **************************/
1507 r02 = _fjsp_mul_v2r8(rsq02,rinv02);
1509 /* Calculate table index by multiplying r with table scale and truncate to integer */
1510 rt = _fjsp_mul_v2r8(r02,vftabscale);
1511 itab_tmp = _fjsp_dtox_v2r8(rt);
1512 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1513 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1514 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1519 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1520 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1521 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1522 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1523 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1524 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1525 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1526 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1527 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1528 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq02,FF),_fjsp_mul_v2r8(vftabscale,rinv02)));
1532 /* Update vectorial force */
1533 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
1534 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1535 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1537 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1538 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1539 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1541 /**************************
1542 * CALCULATE INTERACTIONS *
1543 **************************/
1545 r10 = _fjsp_mul_v2r8(rsq10,rinv10);
1547 /* Calculate table index by multiplying r with table scale and truncate to integer */
1548 rt = _fjsp_mul_v2r8(r10,vftabscale);
1549 itab_tmp = _fjsp_dtox_v2r8(rt);
1550 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1551 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1552 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1557 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1558 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1559 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1560 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1561 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1562 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1563 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1564 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1565 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1566 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq10,FF),_fjsp_mul_v2r8(vftabscale,rinv10)));
1570 /* Update vectorial force */
1571 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
1572 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
1573 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
1575 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
1576 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
1577 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
1579 /**************************
1580 * CALCULATE INTERACTIONS *
1581 **************************/
1583 r11 = _fjsp_mul_v2r8(rsq11,rinv11);
1585 /* Calculate table index by multiplying r with table scale and truncate to integer */
1586 rt = _fjsp_mul_v2r8(r11,vftabscale);
1587 itab_tmp = _fjsp_dtox_v2r8(rt);
1588 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1589 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1590 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1595 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1596 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1597 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1598 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1599 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1600 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1601 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1602 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1603 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1604 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq11,FF),_fjsp_mul_v2r8(vftabscale,rinv11)));
1608 /* Update vectorial force */
1609 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
1610 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1611 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1613 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1614 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1615 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1617 /**************************
1618 * CALCULATE INTERACTIONS *
1619 **************************/
1621 r12 = _fjsp_mul_v2r8(rsq12,rinv12);
1623 /* Calculate table index by multiplying r with table scale and truncate to integer */
1624 rt = _fjsp_mul_v2r8(r12,vftabscale);
1625 itab_tmp = _fjsp_dtox_v2r8(rt);
1626 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1627 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1628 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1633 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1634 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1635 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1636 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1637 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1638 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1639 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1640 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1641 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1642 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq12,FF),_fjsp_mul_v2r8(vftabscale,rinv12)));
1646 /* Update vectorial force */
1647 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
1648 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1649 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1651 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1652 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1653 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1655 /**************************
1656 * CALCULATE INTERACTIONS *
1657 **************************/
1659 r20 = _fjsp_mul_v2r8(rsq20,rinv20);
1661 /* Calculate table index by multiplying r with table scale and truncate to integer */
1662 rt = _fjsp_mul_v2r8(r20,vftabscale);
1663 itab_tmp = _fjsp_dtox_v2r8(rt);
1664 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1665 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1666 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1671 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1672 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1673 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1674 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1675 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1676 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1677 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1678 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1679 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1680 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq20,FF),_fjsp_mul_v2r8(vftabscale,rinv20)));
1684 /* Update vectorial force */
1685 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
1686 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1687 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1689 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1690 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1691 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1693 /**************************
1694 * CALCULATE INTERACTIONS *
1695 **************************/
1697 r21 = _fjsp_mul_v2r8(rsq21,rinv21);
1699 /* Calculate table index by multiplying r with table scale and truncate to integer */
1700 rt = _fjsp_mul_v2r8(r21,vftabscale);
1701 itab_tmp = _fjsp_dtox_v2r8(rt);
1702 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1703 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1704 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1709 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1710 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1711 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1712 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1713 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1714 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1715 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1716 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1717 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1718 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq21,FF),_fjsp_mul_v2r8(vftabscale,rinv21)));
1722 /* Update vectorial force */
1723 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
1724 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1725 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1727 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1728 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1729 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1731 /**************************
1732 * CALCULATE INTERACTIONS *
1733 **************************/
1735 r22 = _fjsp_mul_v2r8(rsq22,rinv22);
1737 /* Calculate table index by multiplying r with table scale and truncate to integer */
1738 rt = _fjsp_mul_v2r8(r22,vftabscale);
1739 itab_tmp = _fjsp_dtox_v2r8(rt);
1740 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1741 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1742 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1747 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1748 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1749 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1750 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1751 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1752 H = _fjsp_load_v2r8( vftab + vfconv.i[1] +2);
1753 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1754 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1755 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1756 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq22,FF),_fjsp_mul_v2r8(vftabscale,rinv22)));
1760 /* Update vectorial force */
1761 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
1762 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1763 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1765 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1766 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1767 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1769 gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1771 /* Inner loop uses 386 flops */
1774 if(jidx<j_index_end)
1778 j_coord_offsetA = DIM*jnrA;
1780 /* load j atom coordinates */
1781 gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
1782 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1784 /* Calculate displacement vector */
1785 dx00 = _fjsp_sub_v2r8(ix0,jx0);
1786 dy00 = _fjsp_sub_v2r8(iy0,jy0);
1787 dz00 = _fjsp_sub_v2r8(iz0,jz0);
1788 dx01 = _fjsp_sub_v2r8(ix0,jx1);
1789 dy01 = _fjsp_sub_v2r8(iy0,jy1);
1790 dz01 = _fjsp_sub_v2r8(iz0,jz1);
1791 dx02 = _fjsp_sub_v2r8(ix0,jx2);
1792 dy02 = _fjsp_sub_v2r8(iy0,jy2);
1793 dz02 = _fjsp_sub_v2r8(iz0,jz2);
1794 dx10 = _fjsp_sub_v2r8(ix1,jx0);
1795 dy10 = _fjsp_sub_v2r8(iy1,jy0);
1796 dz10 = _fjsp_sub_v2r8(iz1,jz0);
1797 dx11 = _fjsp_sub_v2r8(ix1,jx1);
1798 dy11 = _fjsp_sub_v2r8(iy1,jy1);
1799 dz11 = _fjsp_sub_v2r8(iz1,jz1);
1800 dx12 = _fjsp_sub_v2r8(ix1,jx2);
1801 dy12 = _fjsp_sub_v2r8(iy1,jy2);
1802 dz12 = _fjsp_sub_v2r8(iz1,jz2);
1803 dx20 = _fjsp_sub_v2r8(ix2,jx0);
1804 dy20 = _fjsp_sub_v2r8(iy2,jy0);
1805 dz20 = _fjsp_sub_v2r8(iz2,jz0);
1806 dx21 = _fjsp_sub_v2r8(ix2,jx1);
1807 dy21 = _fjsp_sub_v2r8(iy2,jy1);
1808 dz21 = _fjsp_sub_v2r8(iz2,jz1);
1809 dx22 = _fjsp_sub_v2r8(ix2,jx2);
1810 dy22 = _fjsp_sub_v2r8(iy2,jy2);
1811 dz22 = _fjsp_sub_v2r8(iz2,jz2);
1813 /* Calculate squared distance and things based on it */
1814 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1815 rsq01 = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1816 rsq02 = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1817 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1818 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1819 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1820 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1821 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1822 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1824 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
1825 rinv01 = gmx_fjsp_invsqrt_v2r8(rsq01);
1826 rinv02 = gmx_fjsp_invsqrt_v2r8(rsq02);
1827 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
1828 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
1829 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
1830 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
1831 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
1832 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
1834 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
1836 fjx0 = _fjsp_setzero_v2r8();
1837 fjy0 = _fjsp_setzero_v2r8();
1838 fjz0 = _fjsp_setzero_v2r8();
1839 fjx1 = _fjsp_setzero_v2r8();
1840 fjy1 = _fjsp_setzero_v2r8();
1841 fjz1 = _fjsp_setzero_v2r8();
1842 fjx2 = _fjsp_setzero_v2r8();
1843 fjy2 = _fjsp_setzero_v2r8();
1844 fjz2 = _fjsp_setzero_v2r8();
1846 /**************************
1847 * CALCULATE INTERACTIONS *
1848 **************************/
1850 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
1852 /* Calculate table index by multiplying r with table scale and truncate to integer */
1853 rt = _fjsp_mul_v2r8(r00,vftabscale);
1854 itab_tmp = _fjsp_dtox_v2r8(rt);
1855 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1856 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1857 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1862 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1863 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1864 F = _fjsp_setzero_v2r8();
1865 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1866 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1867 H = _fjsp_setzero_v2r8();
1868 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1869 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1870 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1871 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,FF),_fjsp_mul_v2r8(vftabscale,rinv00)));
1873 /* LENNARD-JONES DISPERSION/REPULSION */
1875 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
1876 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
1878 fscal = _fjsp_add_v2r8(felec,fvdw);
1880 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1882 /* Update vectorial force */
1883 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
1884 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1885 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1887 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1888 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1889 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1891 /**************************
1892 * CALCULATE INTERACTIONS *
1893 **************************/
1895 r01 = _fjsp_mul_v2r8(rsq01,rinv01);
1897 /* Calculate table index by multiplying r with table scale and truncate to integer */
1898 rt = _fjsp_mul_v2r8(r01,vftabscale);
1899 itab_tmp = _fjsp_dtox_v2r8(rt);
1900 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1901 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1902 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1907 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1908 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1909 F = _fjsp_setzero_v2r8();
1910 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1911 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1912 H = _fjsp_setzero_v2r8();
1913 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1914 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1915 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1916 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq01,FF),_fjsp_mul_v2r8(vftabscale,rinv01)));
1920 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1922 /* Update vectorial force */
1923 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
1924 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1925 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1927 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1928 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1929 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1931 /**************************
1932 * CALCULATE INTERACTIONS *
1933 **************************/
1935 r02 = _fjsp_mul_v2r8(rsq02,rinv02);
1937 /* Calculate table index by multiplying r with table scale and truncate to integer */
1938 rt = _fjsp_mul_v2r8(r02,vftabscale);
1939 itab_tmp = _fjsp_dtox_v2r8(rt);
1940 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1941 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1942 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1947 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1948 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1949 F = _fjsp_setzero_v2r8();
1950 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1951 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1952 H = _fjsp_setzero_v2r8();
1953 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1954 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1955 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1956 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq02,FF),_fjsp_mul_v2r8(vftabscale,rinv02)));
1960 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1962 /* Update vectorial force */
1963 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
1964 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1965 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1967 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1968 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1969 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1971 /**************************
1972 * CALCULATE INTERACTIONS *
1973 **************************/
1975 r10 = _fjsp_mul_v2r8(rsq10,rinv10);
1977 /* Calculate table index by multiplying r with table scale and truncate to integer */
1978 rt = _fjsp_mul_v2r8(r10,vftabscale);
1979 itab_tmp = _fjsp_dtox_v2r8(rt);
1980 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1981 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1982 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1987 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1988 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1989 F = _fjsp_setzero_v2r8();
1990 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1991 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
1992 H = _fjsp_setzero_v2r8();
1993 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1994 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
1995 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
1996 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq10,FF),_fjsp_mul_v2r8(vftabscale,rinv10)));
2000 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
2002 /* Update vectorial force */
2003 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
2004 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
2005 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
2007 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
2008 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
2009 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
2011 /**************************
2012 * CALCULATE INTERACTIONS *
2013 **************************/
2015 r11 = _fjsp_mul_v2r8(rsq11,rinv11);
2017 /* Calculate table index by multiplying r with table scale and truncate to integer */
2018 rt = _fjsp_mul_v2r8(r11,vftabscale);
2019 itab_tmp = _fjsp_dtox_v2r8(rt);
2020 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
2021 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
2022 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
2027 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2028 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
2029 F = _fjsp_setzero_v2r8();
2030 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
2031 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
2032 H = _fjsp_setzero_v2r8();
2033 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
2034 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
2035 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
2036 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq11,FF),_fjsp_mul_v2r8(vftabscale,rinv11)));
2040 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
2042 /* Update vectorial force */
2043 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
2044 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
2045 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
2047 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
2048 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
2049 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
2051 /**************************
2052 * CALCULATE INTERACTIONS *
2053 **************************/
2055 r12 = _fjsp_mul_v2r8(rsq12,rinv12);
2057 /* Calculate table index by multiplying r with table scale and truncate to integer */
2058 rt = _fjsp_mul_v2r8(r12,vftabscale);
2059 itab_tmp = _fjsp_dtox_v2r8(rt);
2060 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
2061 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
2062 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
2067 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2068 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
2069 F = _fjsp_setzero_v2r8();
2070 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
2071 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
2072 H = _fjsp_setzero_v2r8();
2073 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
2074 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
2075 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
2076 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq12,FF),_fjsp_mul_v2r8(vftabscale,rinv12)));
2080 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
2082 /* Update vectorial force */
2083 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
2084 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
2085 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
2087 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
2088 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
2089 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
2091 /**************************
2092 * CALCULATE INTERACTIONS *
2093 **************************/
2095 r20 = _fjsp_mul_v2r8(rsq20,rinv20);
2097 /* Calculate table index by multiplying r with table scale and truncate to integer */
2098 rt = _fjsp_mul_v2r8(r20,vftabscale);
2099 itab_tmp = _fjsp_dtox_v2r8(rt);
2100 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
2101 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
2102 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
2107 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2108 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
2109 F = _fjsp_setzero_v2r8();
2110 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
2111 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
2112 H = _fjsp_setzero_v2r8();
2113 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
2114 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
2115 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
2116 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq20,FF),_fjsp_mul_v2r8(vftabscale,rinv20)));
2120 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
2122 /* Update vectorial force */
2123 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
2124 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
2125 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
2127 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
2128 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
2129 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
2131 /**************************
2132 * CALCULATE INTERACTIONS *
2133 **************************/
2135 r21 = _fjsp_mul_v2r8(rsq21,rinv21);
2137 /* Calculate table index by multiplying r with table scale and truncate to integer */
2138 rt = _fjsp_mul_v2r8(r21,vftabscale);
2139 itab_tmp = _fjsp_dtox_v2r8(rt);
2140 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
2141 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
2142 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
2147 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2148 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
2149 F = _fjsp_setzero_v2r8();
2150 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
2151 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
2152 H = _fjsp_setzero_v2r8();
2153 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
2154 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
2155 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
2156 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq21,FF),_fjsp_mul_v2r8(vftabscale,rinv21)));
2160 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
2162 /* Update vectorial force */
2163 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
2164 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
2165 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
2167 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
2168 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
2169 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
2171 /**************************
2172 * CALCULATE INTERACTIONS *
2173 **************************/
2175 r22 = _fjsp_mul_v2r8(rsq22,rinv22);
2177 /* Calculate table index by multiplying r with table scale and truncate to integer */
2178 rt = _fjsp_mul_v2r8(r22,vftabscale);
2179 itab_tmp = _fjsp_dtox_v2r8(rt);
2180 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
2181 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
2182 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
2187 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2188 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
2189 F = _fjsp_setzero_v2r8();
2190 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
2191 G = _fjsp_load_v2r8( vftab + vfconv.i[0] +2);
2192 H = _fjsp_setzero_v2r8();
2193 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
2194 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(vfeps,H,G),F);
2195 FF = _fjsp_madd_v2r8(_fjsp_madd_v2r8(twovfeps,H,G),vfeps,Fp);
2196 felec = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_mul_v2r8(qq22,FF),_fjsp_mul_v2r8(vftabscale,rinv22)));
2200 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
2202 /* Update vectorial force */
2203 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
2204 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
2205 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
2207 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
2208 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
2209 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
2211 gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2213 /* Inner loop uses 386 flops */
2216 /* End of innermost loop */
2218 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
2219 f+i_coord_offset,fshift+i_shift_offset);
2221 /* Increment number of inner iterations */
2222 inneriter += j_index_end - j_index_start;
2224 /* Outer loop uses 18 flops */
2227 /* Increment number of outer iterations */
2230 /* Update outer/inner flops */
2232 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*386);