2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
36 * Note: this file was generated by the GROMACS sparc64_hpc_ace_double kernel generator.
42 #include "../nb_kernel.h"
43 #include "gromacs/legacyheaders/types/simple.h"
44 #include "gromacs/math/vec.h"
45 #include "gromacs/legacyheaders/nrnb.h"
47 #include "kernelutil_sparc64_hpc_ace_double.h"
50 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: Coulomb
52 * VdW interaction: CubicSplineTable
53 * Geometry: Water3-Water3
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecCoul_VdwCSTab_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_vdw->data;
134 vftabscale = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->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);
264 rinvsq01 = _fjsp_mul_v2r8(rinv01,rinv01);
265 rinvsq02 = _fjsp_mul_v2r8(rinv02,rinv02);
266 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
267 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
268 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
269 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
270 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
271 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
273 fjx0 = _fjsp_setzero_v2r8();
274 fjy0 = _fjsp_setzero_v2r8();
275 fjz0 = _fjsp_setzero_v2r8();
276 fjx1 = _fjsp_setzero_v2r8();
277 fjy1 = _fjsp_setzero_v2r8();
278 fjz1 = _fjsp_setzero_v2r8();
279 fjx2 = _fjsp_setzero_v2r8();
280 fjy2 = _fjsp_setzero_v2r8();
281 fjz2 = _fjsp_setzero_v2r8();
283 /**************************
284 * CALCULATE INTERACTIONS *
285 **************************/
287 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
289 /* Calculate table index by multiplying r with table scale and truncate to integer */
290 rt = _fjsp_mul_v2r8(r00,vftabscale);
291 itab_tmp = _fjsp_dtox_v2r8(rt);
292 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
293 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
294 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
299 /* COULOMB ELECTROSTATICS */
300 velec = _fjsp_mul_v2r8(qq00,rinv00);
301 felec = _fjsp_mul_v2r8(velec,rinvsq00);
303 /* CUBIC SPLINE TABLE DISPERSION */
304 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
305 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
306 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
307 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
308 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
309 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
310 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
311 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
312 vvdw6 = _fjsp_mul_v2r8(c6_00,VV);
313 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
314 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
316 /* CUBIC SPLINE TABLE REPULSION */
317 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
318 F = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
319 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
320 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
321 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
322 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
323 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
324 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
325 vvdw12 = _fjsp_mul_v2r8(c12_00,VV);
326 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
327 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
328 vvdw = _fjsp_add_v2r8(vvdw12,vvdw6);
329 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
331 /* Update potential sum for this i atom from the interaction with this j atom. */
332 velecsum = _fjsp_add_v2r8(velecsum,velec);
333 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
335 fscal = _fjsp_add_v2r8(felec,fvdw);
337 /* Update vectorial force */
338 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
339 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
340 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
342 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
343 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
344 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
346 /**************************
347 * CALCULATE INTERACTIONS *
348 **************************/
350 /* COULOMB ELECTROSTATICS */
351 velec = _fjsp_mul_v2r8(qq01,rinv01);
352 felec = _fjsp_mul_v2r8(velec,rinvsq01);
354 /* Update potential sum for this i atom from the interaction with this j atom. */
355 velecsum = _fjsp_add_v2r8(velecsum,velec);
359 /* Update vectorial force */
360 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
361 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
362 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
364 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
365 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
366 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
368 /**************************
369 * CALCULATE INTERACTIONS *
370 **************************/
372 /* COULOMB ELECTROSTATICS */
373 velec = _fjsp_mul_v2r8(qq02,rinv02);
374 felec = _fjsp_mul_v2r8(velec,rinvsq02);
376 /* Update potential sum for this i atom from the interaction with this j atom. */
377 velecsum = _fjsp_add_v2r8(velecsum,velec);
381 /* Update vectorial force */
382 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
383 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
384 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
386 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
387 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
388 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
390 /**************************
391 * CALCULATE INTERACTIONS *
392 **************************/
394 /* COULOMB ELECTROSTATICS */
395 velec = _fjsp_mul_v2r8(qq10,rinv10);
396 felec = _fjsp_mul_v2r8(velec,rinvsq10);
398 /* Update potential sum for this i atom from the interaction with this j atom. */
399 velecsum = _fjsp_add_v2r8(velecsum,velec);
403 /* Update vectorial force */
404 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
405 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
406 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
408 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
409 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
410 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
412 /**************************
413 * CALCULATE INTERACTIONS *
414 **************************/
416 /* COULOMB ELECTROSTATICS */
417 velec = _fjsp_mul_v2r8(qq11,rinv11);
418 felec = _fjsp_mul_v2r8(velec,rinvsq11);
420 /* Update potential sum for this i atom from the interaction with this j atom. */
421 velecsum = _fjsp_add_v2r8(velecsum,velec);
425 /* Update vectorial force */
426 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
427 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
428 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
430 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
431 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
432 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
434 /**************************
435 * CALCULATE INTERACTIONS *
436 **************************/
438 /* COULOMB ELECTROSTATICS */
439 velec = _fjsp_mul_v2r8(qq12,rinv12);
440 felec = _fjsp_mul_v2r8(velec,rinvsq12);
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(dx12,fscal,fix1);
449 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
450 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
452 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
453 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
454 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
456 /**************************
457 * CALCULATE INTERACTIONS *
458 **************************/
460 /* COULOMB ELECTROSTATICS */
461 velec = _fjsp_mul_v2r8(qq20,rinv20);
462 felec = _fjsp_mul_v2r8(velec,rinvsq20);
464 /* Update potential sum for this i atom from the interaction with this j atom. */
465 velecsum = _fjsp_add_v2r8(velecsum,velec);
469 /* Update vectorial force */
470 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
471 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
472 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
474 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
475 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
476 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
478 /**************************
479 * CALCULATE INTERACTIONS *
480 **************************/
482 /* COULOMB ELECTROSTATICS */
483 velec = _fjsp_mul_v2r8(qq21,rinv21);
484 felec = _fjsp_mul_v2r8(velec,rinvsq21);
486 /* Update potential sum for this i atom from the interaction with this j atom. */
487 velecsum = _fjsp_add_v2r8(velecsum,velec);
491 /* Update vectorial force */
492 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
493 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
494 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
496 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
497 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
498 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
500 /**************************
501 * CALCULATE INTERACTIONS *
502 **************************/
504 /* COULOMB ELECTROSTATICS */
505 velec = _fjsp_mul_v2r8(qq22,rinv22);
506 felec = _fjsp_mul_v2r8(velec,rinvsq22);
508 /* Update potential sum for this i atom from the interaction with this j atom. */
509 velecsum = _fjsp_add_v2r8(velecsum,velec);
513 /* Update vectorial force */
514 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
515 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
516 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
518 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
519 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
520 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
522 gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
524 /* Inner loop uses 314 flops */
531 j_coord_offsetA = DIM*jnrA;
533 /* load j atom coordinates */
534 gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
535 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
537 /* Calculate displacement vector */
538 dx00 = _fjsp_sub_v2r8(ix0,jx0);
539 dy00 = _fjsp_sub_v2r8(iy0,jy0);
540 dz00 = _fjsp_sub_v2r8(iz0,jz0);
541 dx01 = _fjsp_sub_v2r8(ix0,jx1);
542 dy01 = _fjsp_sub_v2r8(iy0,jy1);
543 dz01 = _fjsp_sub_v2r8(iz0,jz1);
544 dx02 = _fjsp_sub_v2r8(ix0,jx2);
545 dy02 = _fjsp_sub_v2r8(iy0,jy2);
546 dz02 = _fjsp_sub_v2r8(iz0,jz2);
547 dx10 = _fjsp_sub_v2r8(ix1,jx0);
548 dy10 = _fjsp_sub_v2r8(iy1,jy0);
549 dz10 = _fjsp_sub_v2r8(iz1,jz0);
550 dx11 = _fjsp_sub_v2r8(ix1,jx1);
551 dy11 = _fjsp_sub_v2r8(iy1,jy1);
552 dz11 = _fjsp_sub_v2r8(iz1,jz1);
553 dx12 = _fjsp_sub_v2r8(ix1,jx2);
554 dy12 = _fjsp_sub_v2r8(iy1,jy2);
555 dz12 = _fjsp_sub_v2r8(iz1,jz2);
556 dx20 = _fjsp_sub_v2r8(ix2,jx0);
557 dy20 = _fjsp_sub_v2r8(iy2,jy0);
558 dz20 = _fjsp_sub_v2r8(iz2,jz0);
559 dx21 = _fjsp_sub_v2r8(ix2,jx1);
560 dy21 = _fjsp_sub_v2r8(iy2,jy1);
561 dz21 = _fjsp_sub_v2r8(iz2,jz1);
562 dx22 = _fjsp_sub_v2r8(ix2,jx2);
563 dy22 = _fjsp_sub_v2r8(iy2,jy2);
564 dz22 = _fjsp_sub_v2r8(iz2,jz2);
566 /* Calculate squared distance and things based on it */
567 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
568 rsq01 = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
569 rsq02 = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
570 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
571 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
572 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
573 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
574 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
575 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
577 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
578 rinv01 = gmx_fjsp_invsqrt_v2r8(rsq01);
579 rinv02 = gmx_fjsp_invsqrt_v2r8(rsq02);
580 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
581 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
582 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
583 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
584 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
585 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
587 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
588 rinvsq01 = _fjsp_mul_v2r8(rinv01,rinv01);
589 rinvsq02 = _fjsp_mul_v2r8(rinv02,rinv02);
590 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
591 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
592 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
593 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
594 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
595 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
597 fjx0 = _fjsp_setzero_v2r8();
598 fjy0 = _fjsp_setzero_v2r8();
599 fjz0 = _fjsp_setzero_v2r8();
600 fjx1 = _fjsp_setzero_v2r8();
601 fjy1 = _fjsp_setzero_v2r8();
602 fjz1 = _fjsp_setzero_v2r8();
603 fjx2 = _fjsp_setzero_v2r8();
604 fjy2 = _fjsp_setzero_v2r8();
605 fjz2 = _fjsp_setzero_v2r8();
607 /**************************
608 * CALCULATE INTERACTIONS *
609 **************************/
611 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
613 /* Calculate table index by multiplying r with table scale and truncate to integer */
614 rt = _fjsp_mul_v2r8(r00,vftabscale);
615 itab_tmp = _fjsp_dtox_v2r8(rt);
616 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
617 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
618 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
623 /* COULOMB ELECTROSTATICS */
624 velec = _fjsp_mul_v2r8(qq00,rinv00);
625 felec = _fjsp_mul_v2r8(velec,rinvsq00);
627 /* CUBIC SPLINE TABLE DISPERSION */
628 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
629 F = _fjsp_setzero_v2r8();
630 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
631 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
632 H = _fjsp_setzero_v2r8();
633 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
634 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
635 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
636 vvdw6 = _fjsp_mul_v2r8(c6_00,VV);
637 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
638 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
640 /* CUBIC SPLINE TABLE REPULSION */
641 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
642 F = _fjsp_setzero_v2r8();
643 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
644 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
645 H = _fjsp_setzero_v2r8();
646 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
647 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
648 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
649 vvdw12 = _fjsp_mul_v2r8(c12_00,VV);
650 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
651 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
652 vvdw = _fjsp_add_v2r8(vvdw12,vvdw6);
653 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
655 /* Update potential sum for this i atom from the interaction with this j atom. */
656 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
657 velecsum = _fjsp_add_v2r8(velecsum,velec);
658 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
659 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
661 fscal = _fjsp_add_v2r8(felec,fvdw);
663 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
665 /* Update vectorial force */
666 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
667 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
668 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
670 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
671 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
672 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
674 /**************************
675 * CALCULATE INTERACTIONS *
676 **************************/
678 /* COULOMB ELECTROSTATICS */
679 velec = _fjsp_mul_v2r8(qq01,rinv01);
680 felec = _fjsp_mul_v2r8(velec,rinvsq01);
682 /* Update potential sum for this i atom from the interaction with this j atom. */
683 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
684 velecsum = _fjsp_add_v2r8(velecsum,velec);
688 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
690 /* Update vectorial force */
691 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
692 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
693 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
695 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
696 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
697 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
699 /**************************
700 * CALCULATE INTERACTIONS *
701 **************************/
703 /* COULOMB ELECTROSTATICS */
704 velec = _fjsp_mul_v2r8(qq02,rinv02);
705 felec = _fjsp_mul_v2r8(velec,rinvsq02);
707 /* Update potential sum for this i atom from the interaction with this j atom. */
708 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
709 velecsum = _fjsp_add_v2r8(velecsum,velec);
713 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
715 /* Update vectorial force */
716 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
717 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
718 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
720 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
721 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
722 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
724 /**************************
725 * CALCULATE INTERACTIONS *
726 **************************/
728 /* COULOMB ELECTROSTATICS */
729 velec = _fjsp_mul_v2r8(qq10,rinv10);
730 felec = _fjsp_mul_v2r8(velec,rinvsq10);
732 /* Update potential sum for this i atom from the interaction with this j atom. */
733 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
734 velecsum = _fjsp_add_v2r8(velecsum,velec);
738 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
740 /* Update vectorial force */
741 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
742 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
743 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
745 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
746 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
747 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
749 /**************************
750 * CALCULATE INTERACTIONS *
751 **************************/
753 /* COULOMB ELECTROSTATICS */
754 velec = _fjsp_mul_v2r8(qq11,rinv11);
755 felec = _fjsp_mul_v2r8(velec,rinvsq11);
757 /* Update potential sum for this i atom from the interaction with this j atom. */
758 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
759 velecsum = _fjsp_add_v2r8(velecsum,velec);
763 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
765 /* Update vectorial force */
766 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
767 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
768 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
770 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
771 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
772 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
774 /**************************
775 * CALCULATE INTERACTIONS *
776 **************************/
778 /* COULOMB ELECTROSTATICS */
779 velec = _fjsp_mul_v2r8(qq12,rinv12);
780 felec = _fjsp_mul_v2r8(velec,rinvsq12);
782 /* Update potential sum for this i atom from the interaction with this j atom. */
783 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
784 velecsum = _fjsp_add_v2r8(velecsum,velec);
788 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
790 /* Update vectorial force */
791 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
792 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
793 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
795 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
796 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
797 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
799 /**************************
800 * CALCULATE INTERACTIONS *
801 **************************/
803 /* COULOMB ELECTROSTATICS */
804 velec = _fjsp_mul_v2r8(qq20,rinv20);
805 felec = _fjsp_mul_v2r8(velec,rinvsq20);
807 /* Update potential sum for this i atom from the interaction with this j atom. */
808 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
809 velecsum = _fjsp_add_v2r8(velecsum,velec);
813 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
815 /* Update vectorial force */
816 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
817 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
818 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
820 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
821 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
822 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
824 /**************************
825 * CALCULATE INTERACTIONS *
826 **************************/
828 /* COULOMB ELECTROSTATICS */
829 velec = _fjsp_mul_v2r8(qq21,rinv21);
830 felec = _fjsp_mul_v2r8(velec,rinvsq21);
832 /* Update potential sum for this i atom from the interaction with this j atom. */
833 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
834 velecsum = _fjsp_add_v2r8(velecsum,velec);
838 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
840 /* Update vectorial force */
841 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
842 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
843 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
845 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
846 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
847 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
849 /**************************
850 * CALCULATE INTERACTIONS *
851 **************************/
853 /* COULOMB ELECTROSTATICS */
854 velec = _fjsp_mul_v2r8(qq22,rinv22);
855 felec = _fjsp_mul_v2r8(velec,rinvsq22);
857 /* Update potential sum for this i atom from the interaction with this j atom. */
858 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
859 velecsum = _fjsp_add_v2r8(velecsum,velec);
863 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
865 /* Update vectorial force */
866 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
867 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
868 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
870 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
871 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
872 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
874 gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
876 /* Inner loop uses 314 flops */
879 /* End of innermost loop */
881 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
882 f+i_coord_offset,fshift+i_shift_offset);
885 /* Update potential energies */
886 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
887 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
889 /* Increment number of inner iterations */
890 inneriter += j_index_end - j_index_start;
892 /* Outer loop uses 20 flops */
895 /* Increment number of outer iterations */
898 /* Update outer/inner flops */
900 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*314);
903 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_F_sparc64_hpc_ace_double
904 * Electrostatics interaction: Coulomb
905 * VdW interaction: CubicSplineTable
906 * Geometry: Water3-Water3
907 * Calculate force/pot: Force
910 nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_F_sparc64_hpc_ace_double
911 (t_nblist * gmx_restrict nlist,
912 rvec * gmx_restrict xx,
913 rvec * gmx_restrict ff,
914 t_forcerec * gmx_restrict fr,
915 t_mdatoms * gmx_restrict mdatoms,
916 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
917 t_nrnb * gmx_restrict nrnb)
919 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
920 * just 0 for non-waters.
921 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
922 * jnr indices corresponding to data put in the four positions in the SIMD register.
924 int i_shift_offset,i_coord_offset,outeriter,inneriter;
925 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
927 int j_coord_offsetA,j_coord_offsetB;
928 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
930 real *shiftvec,*fshift,*x,*f;
931 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
933 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
935 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
937 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
938 int vdwjidx0A,vdwjidx0B;
939 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
940 int vdwjidx1A,vdwjidx1B;
941 _fjsp_v2r8 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
942 int vdwjidx2A,vdwjidx2B;
943 _fjsp_v2r8 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
944 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
945 _fjsp_v2r8 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
946 _fjsp_v2r8 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
947 _fjsp_v2r8 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
948 _fjsp_v2r8 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
949 _fjsp_v2r8 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
950 _fjsp_v2r8 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
951 _fjsp_v2r8 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
952 _fjsp_v2r8 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
953 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
956 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
959 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
960 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
961 _fjsp_v2r8 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
964 _fjsp_v2r8 dummy_mask,cutoff_mask;
965 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
966 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
967 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
974 jindex = nlist->jindex;
976 shiftidx = nlist->shift;
978 shiftvec = fr->shift_vec[0];
979 fshift = fr->fshift[0];
980 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
981 charge = mdatoms->chargeA;
982 nvdwtype = fr->ntype;
984 vdwtype = mdatoms->typeA;
986 vftab = kernel_data->table_vdw->data;
987 vftabscale = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
989 /* Setup water-specific parameters */
990 inr = nlist->iinr[0];
991 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
992 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
993 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
994 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
996 jq0 = gmx_fjsp_set1_v2r8(charge[inr+0]);
997 jq1 = gmx_fjsp_set1_v2r8(charge[inr+1]);
998 jq2 = gmx_fjsp_set1_v2r8(charge[inr+2]);
999 vdwjidx0A = 2*vdwtype[inr+0];
1000 qq00 = _fjsp_mul_v2r8(iq0,jq0);
1001 c6_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
1002 c12_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
1003 qq01 = _fjsp_mul_v2r8(iq0,jq1);
1004 qq02 = _fjsp_mul_v2r8(iq0,jq2);
1005 qq10 = _fjsp_mul_v2r8(iq1,jq0);
1006 qq11 = _fjsp_mul_v2r8(iq1,jq1);
1007 qq12 = _fjsp_mul_v2r8(iq1,jq2);
1008 qq20 = _fjsp_mul_v2r8(iq2,jq0);
1009 qq21 = _fjsp_mul_v2r8(iq2,jq1);
1010 qq22 = _fjsp_mul_v2r8(iq2,jq2);
1012 /* Avoid stupid compiler warnings */
1014 j_coord_offsetA = 0;
1015 j_coord_offsetB = 0;
1020 /* Start outer loop over neighborlists */
1021 for(iidx=0; iidx<nri; iidx++)
1023 /* Load shift vector for this list */
1024 i_shift_offset = DIM*shiftidx[iidx];
1026 /* Load limits for loop over neighbors */
1027 j_index_start = jindex[iidx];
1028 j_index_end = jindex[iidx+1];
1030 /* Get outer coordinate index */
1032 i_coord_offset = DIM*inr;
1034 /* Load i particle coords and add shift vector */
1035 gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
1036 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1038 fix0 = _fjsp_setzero_v2r8();
1039 fiy0 = _fjsp_setzero_v2r8();
1040 fiz0 = _fjsp_setzero_v2r8();
1041 fix1 = _fjsp_setzero_v2r8();
1042 fiy1 = _fjsp_setzero_v2r8();
1043 fiz1 = _fjsp_setzero_v2r8();
1044 fix2 = _fjsp_setzero_v2r8();
1045 fiy2 = _fjsp_setzero_v2r8();
1046 fiz2 = _fjsp_setzero_v2r8();
1048 /* Start inner kernel loop */
1049 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1052 /* Get j neighbor index, and coordinate index */
1054 jnrB = jjnr[jidx+1];
1055 j_coord_offsetA = DIM*jnrA;
1056 j_coord_offsetB = DIM*jnrB;
1058 /* load j atom coordinates */
1059 gmx_fjsp_load_3rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
1060 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1062 /* Calculate displacement vector */
1063 dx00 = _fjsp_sub_v2r8(ix0,jx0);
1064 dy00 = _fjsp_sub_v2r8(iy0,jy0);
1065 dz00 = _fjsp_sub_v2r8(iz0,jz0);
1066 dx01 = _fjsp_sub_v2r8(ix0,jx1);
1067 dy01 = _fjsp_sub_v2r8(iy0,jy1);
1068 dz01 = _fjsp_sub_v2r8(iz0,jz1);
1069 dx02 = _fjsp_sub_v2r8(ix0,jx2);
1070 dy02 = _fjsp_sub_v2r8(iy0,jy2);
1071 dz02 = _fjsp_sub_v2r8(iz0,jz2);
1072 dx10 = _fjsp_sub_v2r8(ix1,jx0);
1073 dy10 = _fjsp_sub_v2r8(iy1,jy0);
1074 dz10 = _fjsp_sub_v2r8(iz1,jz0);
1075 dx11 = _fjsp_sub_v2r8(ix1,jx1);
1076 dy11 = _fjsp_sub_v2r8(iy1,jy1);
1077 dz11 = _fjsp_sub_v2r8(iz1,jz1);
1078 dx12 = _fjsp_sub_v2r8(ix1,jx2);
1079 dy12 = _fjsp_sub_v2r8(iy1,jy2);
1080 dz12 = _fjsp_sub_v2r8(iz1,jz2);
1081 dx20 = _fjsp_sub_v2r8(ix2,jx0);
1082 dy20 = _fjsp_sub_v2r8(iy2,jy0);
1083 dz20 = _fjsp_sub_v2r8(iz2,jz0);
1084 dx21 = _fjsp_sub_v2r8(ix2,jx1);
1085 dy21 = _fjsp_sub_v2r8(iy2,jy1);
1086 dz21 = _fjsp_sub_v2r8(iz2,jz1);
1087 dx22 = _fjsp_sub_v2r8(ix2,jx2);
1088 dy22 = _fjsp_sub_v2r8(iy2,jy2);
1089 dz22 = _fjsp_sub_v2r8(iz2,jz2);
1091 /* Calculate squared distance and things based on it */
1092 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1093 rsq01 = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1094 rsq02 = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1095 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1096 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1097 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1098 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1099 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1100 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1102 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
1103 rinv01 = gmx_fjsp_invsqrt_v2r8(rsq01);
1104 rinv02 = gmx_fjsp_invsqrt_v2r8(rsq02);
1105 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
1106 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
1107 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
1108 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
1109 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
1110 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
1112 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
1113 rinvsq01 = _fjsp_mul_v2r8(rinv01,rinv01);
1114 rinvsq02 = _fjsp_mul_v2r8(rinv02,rinv02);
1115 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
1116 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
1117 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
1118 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
1119 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
1120 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
1122 fjx0 = _fjsp_setzero_v2r8();
1123 fjy0 = _fjsp_setzero_v2r8();
1124 fjz0 = _fjsp_setzero_v2r8();
1125 fjx1 = _fjsp_setzero_v2r8();
1126 fjy1 = _fjsp_setzero_v2r8();
1127 fjz1 = _fjsp_setzero_v2r8();
1128 fjx2 = _fjsp_setzero_v2r8();
1129 fjy2 = _fjsp_setzero_v2r8();
1130 fjz2 = _fjsp_setzero_v2r8();
1132 /**************************
1133 * CALCULATE INTERACTIONS *
1134 **************************/
1136 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
1138 /* Calculate table index by multiplying r with table scale and truncate to integer */
1139 rt = _fjsp_mul_v2r8(r00,vftabscale);
1140 itab_tmp = _fjsp_dtox_v2r8(rt);
1141 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1142 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1143 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1148 /* COULOMB ELECTROSTATICS */
1149 velec = _fjsp_mul_v2r8(qq00,rinv00);
1150 felec = _fjsp_mul_v2r8(velec,rinvsq00);
1152 /* CUBIC SPLINE TABLE DISPERSION */
1153 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1154 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1155 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1156 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1157 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
1158 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1159 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1160 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1161 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
1163 /* CUBIC SPLINE TABLE REPULSION */
1164 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1165 F = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
1166 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1167 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1168 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
1169 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1170 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1171 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1172 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
1173 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1175 fscal = _fjsp_add_v2r8(felec,fvdw);
1177 /* Update vectorial force */
1178 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
1179 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1180 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1182 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1183 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1184 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1186 /**************************
1187 * CALCULATE INTERACTIONS *
1188 **************************/
1190 /* COULOMB ELECTROSTATICS */
1191 velec = _fjsp_mul_v2r8(qq01,rinv01);
1192 felec = _fjsp_mul_v2r8(velec,rinvsq01);
1196 /* Update vectorial force */
1197 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
1198 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1199 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1201 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1202 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1203 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1205 /**************************
1206 * CALCULATE INTERACTIONS *
1207 **************************/
1209 /* COULOMB ELECTROSTATICS */
1210 velec = _fjsp_mul_v2r8(qq02,rinv02);
1211 felec = _fjsp_mul_v2r8(velec,rinvsq02);
1215 /* Update vectorial force */
1216 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
1217 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1218 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1220 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1221 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1222 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1224 /**************************
1225 * CALCULATE INTERACTIONS *
1226 **************************/
1228 /* COULOMB ELECTROSTATICS */
1229 velec = _fjsp_mul_v2r8(qq10,rinv10);
1230 felec = _fjsp_mul_v2r8(velec,rinvsq10);
1234 /* Update vectorial force */
1235 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
1236 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
1237 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
1239 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
1240 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
1241 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
1243 /**************************
1244 * CALCULATE INTERACTIONS *
1245 **************************/
1247 /* COULOMB ELECTROSTATICS */
1248 velec = _fjsp_mul_v2r8(qq11,rinv11);
1249 felec = _fjsp_mul_v2r8(velec,rinvsq11);
1253 /* Update vectorial force */
1254 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
1255 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1256 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1258 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1259 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1260 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1262 /**************************
1263 * CALCULATE INTERACTIONS *
1264 **************************/
1266 /* COULOMB ELECTROSTATICS */
1267 velec = _fjsp_mul_v2r8(qq12,rinv12);
1268 felec = _fjsp_mul_v2r8(velec,rinvsq12);
1272 /* Update vectorial force */
1273 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
1274 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1275 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1277 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1278 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1279 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1281 /**************************
1282 * CALCULATE INTERACTIONS *
1283 **************************/
1285 /* COULOMB ELECTROSTATICS */
1286 velec = _fjsp_mul_v2r8(qq20,rinv20);
1287 felec = _fjsp_mul_v2r8(velec,rinvsq20);
1291 /* Update vectorial force */
1292 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
1293 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1294 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1296 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1297 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1298 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1300 /**************************
1301 * CALCULATE INTERACTIONS *
1302 **************************/
1304 /* COULOMB ELECTROSTATICS */
1305 velec = _fjsp_mul_v2r8(qq21,rinv21);
1306 felec = _fjsp_mul_v2r8(velec,rinvsq21);
1310 /* Update vectorial force */
1311 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
1312 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1313 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1315 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1316 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1317 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1319 /**************************
1320 * CALCULATE INTERACTIONS *
1321 **************************/
1323 /* COULOMB ELECTROSTATICS */
1324 velec = _fjsp_mul_v2r8(qq22,rinv22);
1325 felec = _fjsp_mul_v2r8(velec,rinvsq22);
1329 /* Update vectorial force */
1330 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
1331 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1332 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1334 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1335 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1336 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1338 gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1340 /* Inner loop uses 297 flops */
1343 if(jidx<j_index_end)
1347 j_coord_offsetA = DIM*jnrA;
1349 /* load j atom coordinates */
1350 gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
1351 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1353 /* Calculate displacement vector */
1354 dx00 = _fjsp_sub_v2r8(ix0,jx0);
1355 dy00 = _fjsp_sub_v2r8(iy0,jy0);
1356 dz00 = _fjsp_sub_v2r8(iz0,jz0);
1357 dx01 = _fjsp_sub_v2r8(ix0,jx1);
1358 dy01 = _fjsp_sub_v2r8(iy0,jy1);
1359 dz01 = _fjsp_sub_v2r8(iz0,jz1);
1360 dx02 = _fjsp_sub_v2r8(ix0,jx2);
1361 dy02 = _fjsp_sub_v2r8(iy0,jy2);
1362 dz02 = _fjsp_sub_v2r8(iz0,jz2);
1363 dx10 = _fjsp_sub_v2r8(ix1,jx0);
1364 dy10 = _fjsp_sub_v2r8(iy1,jy0);
1365 dz10 = _fjsp_sub_v2r8(iz1,jz0);
1366 dx11 = _fjsp_sub_v2r8(ix1,jx1);
1367 dy11 = _fjsp_sub_v2r8(iy1,jy1);
1368 dz11 = _fjsp_sub_v2r8(iz1,jz1);
1369 dx12 = _fjsp_sub_v2r8(ix1,jx2);
1370 dy12 = _fjsp_sub_v2r8(iy1,jy2);
1371 dz12 = _fjsp_sub_v2r8(iz1,jz2);
1372 dx20 = _fjsp_sub_v2r8(ix2,jx0);
1373 dy20 = _fjsp_sub_v2r8(iy2,jy0);
1374 dz20 = _fjsp_sub_v2r8(iz2,jz0);
1375 dx21 = _fjsp_sub_v2r8(ix2,jx1);
1376 dy21 = _fjsp_sub_v2r8(iy2,jy1);
1377 dz21 = _fjsp_sub_v2r8(iz2,jz1);
1378 dx22 = _fjsp_sub_v2r8(ix2,jx2);
1379 dy22 = _fjsp_sub_v2r8(iy2,jy2);
1380 dz22 = _fjsp_sub_v2r8(iz2,jz2);
1382 /* Calculate squared distance and things based on it */
1383 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1384 rsq01 = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1385 rsq02 = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1386 rsq10 = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1387 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1388 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1389 rsq20 = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1390 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1391 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1393 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
1394 rinv01 = gmx_fjsp_invsqrt_v2r8(rsq01);
1395 rinv02 = gmx_fjsp_invsqrt_v2r8(rsq02);
1396 rinv10 = gmx_fjsp_invsqrt_v2r8(rsq10);
1397 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
1398 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
1399 rinv20 = gmx_fjsp_invsqrt_v2r8(rsq20);
1400 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
1401 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
1403 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
1404 rinvsq01 = _fjsp_mul_v2r8(rinv01,rinv01);
1405 rinvsq02 = _fjsp_mul_v2r8(rinv02,rinv02);
1406 rinvsq10 = _fjsp_mul_v2r8(rinv10,rinv10);
1407 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
1408 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
1409 rinvsq20 = _fjsp_mul_v2r8(rinv20,rinv20);
1410 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
1411 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
1413 fjx0 = _fjsp_setzero_v2r8();
1414 fjy0 = _fjsp_setzero_v2r8();
1415 fjz0 = _fjsp_setzero_v2r8();
1416 fjx1 = _fjsp_setzero_v2r8();
1417 fjy1 = _fjsp_setzero_v2r8();
1418 fjz1 = _fjsp_setzero_v2r8();
1419 fjx2 = _fjsp_setzero_v2r8();
1420 fjy2 = _fjsp_setzero_v2r8();
1421 fjz2 = _fjsp_setzero_v2r8();
1423 /**************************
1424 * CALCULATE INTERACTIONS *
1425 **************************/
1427 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
1429 /* Calculate table index by multiplying r with table scale and truncate to integer */
1430 rt = _fjsp_mul_v2r8(r00,vftabscale);
1431 itab_tmp = _fjsp_dtox_v2r8(rt);
1432 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1433 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1434 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1439 /* COULOMB ELECTROSTATICS */
1440 velec = _fjsp_mul_v2r8(qq00,rinv00);
1441 felec = _fjsp_mul_v2r8(velec,rinvsq00);
1443 /* CUBIC SPLINE TABLE DISPERSION */
1444 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1445 F = _fjsp_setzero_v2r8();
1446 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1447 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1448 H = _fjsp_setzero_v2r8();
1449 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1450 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1451 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1452 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
1454 /* CUBIC SPLINE TABLE REPULSION */
1455 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1456 F = _fjsp_setzero_v2r8();
1457 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1458 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1459 H = _fjsp_setzero_v2r8();
1460 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1461 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1462 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1463 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
1464 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1466 fscal = _fjsp_add_v2r8(felec,fvdw);
1468 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1470 /* Update vectorial force */
1471 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
1472 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1473 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1475 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1476 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1477 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1479 /**************************
1480 * CALCULATE INTERACTIONS *
1481 **************************/
1483 /* COULOMB ELECTROSTATICS */
1484 velec = _fjsp_mul_v2r8(qq01,rinv01);
1485 felec = _fjsp_mul_v2r8(velec,rinvsq01);
1489 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1491 /* Update vectorial force */
1492 fix0 = _fjsp_madd_v2r8(dx01,fscal,fix0);
1493 fiy0 = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1494 fiz0 = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1496 fjx1 = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1497 fjy1 = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1498 fjz1 = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1500 /**************************
1501 * CALCULATE INTERACTIONS *
1502 **************************/
1504 /* COULOMB ELECTROSTATICS */
1505 velec = _fjsp_mul_v2r8(qq02,rinv02);
1506 felec = _fjsp_mul_v2r8(velec,rinvsq02);
1510 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1512 /* Update vectorial force */
1513 fix0 = _fjsp_madd_v2r8(dx02,fscal,fix0);
1514 fiy0 = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1515 fiz0 = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1517 fjx2 = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1518 fjy2 = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1519 fjz2 = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1521 /**************************
1522 * CALCULATE INTERACTIONS *
1523 **************************/
1525 /* COULOMB ELECTROSTATICS */
1526 velec = _fjsp_mul_v2r8(qq10,rinv10);
1527 felec = _fjsp_mul_v2r8(velec,rinvsq10);
1531 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1533 /* Update vectorial force */
1534 fix1 = _fjsp_madd_v2r8(dx10,fscal,fix1);
1535 fiy1 = _fjsp_madd_v2r8(dy10,fscal,fiy1);
1536 fiz1 = _fjsp_madd_v2r8(dz10,fscal,fiz1);
1538 fjx0 = _fjsp_madd_v2r8(dx10,fscal,fjx0);
1539 fjy0 = _fjsp_madd_v2r8(dy10,fscal,fjy0);
1540 fjz0 = _fjsp_madd_v2r8(dz10,fscal,fjz0);
1542 /**************************
1543 * CALCULATE INTERACTIONS *
1544 **************************/
1546 /* COULOMB ELECTROSTATICS */
1547 velec = _fjsp_mul_v2r8(qq11,rinv11);
1548 felec = _fjsp_mul_v2r8(velec,rinvsq11);
1552 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1554 /* Update vectorial force */
1555 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
1556 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1557 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1559 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1560 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1561 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1563 /**************************
1564 * CALCULATE INTERACTIONS *
1565 **************************/
1567 /* COULOMB ELECTROSTATICS */
1568 velec = _fjsp_mul_v2r8(qq12,rinv12);
1569 felec = _fjsp_mul_v2r8(velec,rinvsq12);
1573 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1575 /* Update vectorial force */
1576 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
1577 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1578 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1580 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1581 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1582 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1584 /**************************
1585 * CALCULATE INTERACTIONS *
1586 **************************/
1588 /* COULOMB ELECTROSTATICS */
1589 velec = _fjsp_mul_v2r8(qq20,rinv20);
1590 felec = _fjsp_mul_v2r8(velec,rinvsq20);
1594 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1596 /* Update vectorial force */
1597 fix2 = _fjsp_madd_v2r8(dx20,fscal,fix2);
1598 fiy2 = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1599 fiz2 = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1601 fjx0 = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1602 fjy0 = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1603 fjz0 = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1605 /**************************
1606 * CALCULATE INTERACTIONS *
1607 **************************/
1609 /* COULOMB ELECTROSTATICS */
1610 velec = _fjsp_mul_v2r8(qq21,rinv21);
1611 felec = _fjsp_mul_v2r8(velec,rinvsq21);
1615 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1617 /* Update vectorial force */
1618 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
1619 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1620 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1622 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1623 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1624 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1626 /**************************
1627 * CALCULATE INTERACTIONS *
1628 **************************/
1630 /* COULOMB ELECTROSTATICS */
1631 velec = _fjsp_mul_v2r8(qq22,rinv22);
1632 felec = _fjsp_mul_v2r8(velec,rinvsq22);
1636 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1638 /* Update vectorial force */
1639 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
1640 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1641 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1643 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1644 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1645 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1647 gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1649 /* Inner loop uses 297 flops */
1652 /* End of innermost loop */
1654 gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1655 f+i_coord_offset,fshift+i_shift_offset);
1657 /* Increment number of inner iterations */
1658 inneriter += j_index_end - j_index_start;
1660 /* Outer loop uses 18 flops */
1663 /* Increment number of outer iterations */
1666 /* Update outer/inner flops */
1668 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*297);