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_ElecCoul_VdwCSTab_GeomW4W4_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: Coulomb
52 * VdW interaction: CubicSplineTable
53 * Geometry: Water4-Water4
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecCoul_VdwCSTab_GeomW4W4_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;
86 _fjsp_v2r8 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
87 int vdwjidx0A,vdwjidx0B;
88 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
89 int vdwjidx1A,vdwjidx1B;
90 _fjsp_v2r8 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
91 int vdwjidx2A,vdwjidx2B;
92 _fjsp_v2r8 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
93 int vdwjidx3A,vdwjidx3B;
94 _fjsp_v2r8 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
95 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
96 _fjsp_v2r8 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
97 _fjsp_v2r8 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
98 _fjsp_v2r8 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
99 _fjsp_v2r8 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
100 _fjsp_v2r8 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
101 _fjsp_v2r8 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
102 _fjsp_v2r8 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
103 _fjsp_v2r8 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
104 _fjsp_v2r8 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
105 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
108 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
111 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
112 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
113 _fjsp_v2r8 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
116 _fjsp_v2r8 dummy_mask,cutoff_mask;
117 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
118 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
119 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
126 jindex = nlist->jindex;
128 shiftidx = nlist->shift;
130 shiftvec = fr->shift_vec[0];
131 fshift = fr->fshift[0];
132 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
133 charge = mdatoms->chargeA;
134 nvdwtype = fr->ntype;
136 vdwtype = mdatoms->typeA;
138 vftab = kernel_data->table_vdw->data;
139 vftabscale = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
141 /* Setup water-specific parameters */
142 inr = nlist->iinr[0];
143 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
144 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
145 iq3 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+3]));
146 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
148 jq1 = gmx_fjsp_set1_v2r8(charge[inr+1]);
149 jq2 = gmx_fjsp_set1_v2r8(charge[inr+2]);
150 jq3 = gmx_fjsp_set1_v2r8(charge[inr+3]);
151 vdwjidx0A = 2*vdwtype[inr+0];
152 c6_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
153 c12_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
154 qq11 = _fjsp_mul_v2r8(iq1,jq1);
155 qq12 = _fjsp_mul_v2r8(iq1,jq2);
156 qq13 = _fjsp_mul_v2r8(iq1,jq3);
157 qq21 = _fjsp_mul_v2r8(iq2,jq1);
158 qq22 = _fjsp_mul_v2r8(iq2,jq2);
159 qq23 = _fjsp_mul_v2r8(iq2,jq3);
160 qq31 = _fjsp_mul_v2r8(iq3,jq1);
161 qq32 = _fjsp_mul_v2r8(iq3,jq2);
162 qq33 = _fjsp_mul_v2r8(iq3,jq3);
164 /* Avoid stupid compiler warnings */
172 /* Start outer loop over neighborlists */
173 for(iidx=0; iidx<nri; iidx++)
175 /* Load shift vector for this list */
176 i_shift_offset = DIM*shiftidx[iidx];
178 /* Load limits for loop over neighbors */
179 j_index_start = jindex[iidx];
180 j_index_end = jindex[iidx+1];
182 /* Get outer coordinate index */
184 i_coord_offset = DIM*inr;
186 /* Load i particle coords and add shift vector */
187 gmx_fjsp_load_shift_and_4rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
188 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
190 fix0 = _fjsp_setzero_v2r8();
191 fiy0 = _fjsp_setzero_v2r8();
192 fiz0 = _fjsp_setzero_v2r8();
193 fix1 = _fjsp_setzero_v2r8();
194 fiy1 = _fjsp_setzero_v2r8();
195 fiz1 = _fjsp_setzero_v2r8();
196 fix2 = _fjsp_setzero_v2r8();
197 fiy2 = _fjsp_setzero_v2r8();
198 fiz2 = _fjsp_setzero_v2r8();
199 fix3 = _fjsp_setzero_v2r8();
200 fiy3 = _fjsp_setzero_v2r8();
201 fiz3 = _fjsp_setzero_v2r8();
203 /* Reset potential sums */
204 velecsum = _fjsp_setzero_v2r8();
205 vvdwsum = _fjsp_setzero_v2r8();
207 /* Start inner kernel loop */
208 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
211 /* Get j neighbor index, and coordinate index */
214 j_coord_offsetA = DIM*jnrA;
215 j_coord_offsetB = DIM*jnrB;
217 /* load j atom coordinates */
218 gmx_fjsp_load_4rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
219 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
220 &jy2,&jz2,&jx3,&jy3,&jz3);
222 /* Calculate displacement vector */
223 dx00 = _fjsp_sub_v2r8(ix0,jx0);
224 dy00 = _fjsp_sub_v2r8(iy0,jy0);
225 dz00 = _fjsp_sub_v2r8(iz0,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 dx13 = _fjsp_sub_v2r8(ix1,jx3);
233 dy13 = _fjsp_sub_v2r8(iy1,jy3);
234 dz13 = _fjsp_sub_v2r8(iz1,jz3);
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);
241 dx23 = _fjsp_sub_v2r8(ix2,jx3);
242 dy23 = _fjsp_sub_v2r8(iy2,jy3);
243 dz23 = _fjsp_sub_v2r8(iz2,jz3);
244 dx31 = _fjsp_sub_v2r8(ix3,jx1);
245 dy31 = _fjsp_sub_v2r8(iy3,jy1);
246 dz31 = _fjsp_sub_v2r8(iz3,jz1);
247 dx32 = _fjsp_sub_v2r8(ix3,jx2);
248 dy32 = _fjsp_sub_v2r8(iy3,jy2);
249 dz32 = _fjsp_sub_v2r8(iz3,jz2);
250 dx33 = _fjsp_sub_v2r8(ix3,jx3);
251 dy33 = _fjsp_sub_v2r8(iy3,jy3);
252 dz33 = _fjsp_sub_v2r8(iz3,jz3);
254 /* Calculate squared distance and things based on it */
255 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
256 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
257 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
258 rsq13 = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
259 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
260 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
261 rsq23 = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
262 rsq31 = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
263 rsq32 = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
264 rsq33 = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
266 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
267 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
268 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
269 rinv13 = gmx_fjsp_invsqrt_v2r8(rsq13);
270 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
271 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
272 rinv23 = gmx_fjsp_invsqrt_v2r8(rsq23);
273 rinv31 = gmx_fjsp_invsqrt_v2r8(rsq31);
274 rinv32 = gmx_fjsp_invsqrt_v2r8(rsq32);
275 rinv33 = gmx_fjsp_invsqrt_v2r8(rsq33);
277 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
278 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
279 rinvsq13 = _fjsp_mul_v2r8(rinv13,rinv13);
280 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
281 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
282 rinvsq23 = _fjsp_mul_v2r8(rinv23,rinv23);
283 rinvsq31 = _fjsp_mul_v2r8(rinv31,rinv31);
284 rinvsq32 = _fjsp_mul_v2r8(rinv32,rinv32);
285 rinvsq33 = _fjsp_mul_v2r8(rinv33,rinv33);
287 fjx0 = _fjsp_setzero_v2r8();
288 fjy0 = _fjsp_setzero_v2r8();
289 fjz0 = _fjsp_setzero_v2r8();
290 fjx1 = _fjsp_setzero_v2r8();
291 fjy1 = _fjsp_setzero_v2r8();
292 fjz1 = _fjsp_setzero_v2r8();
293 fjx2 = _fjsp_setzero_v2r8();
294 fjy2 = _fjsp_setzero_v2r8();
295 fjz2 = _fjsp_setzero_v2r8();
296 fjx3 = _fjsp_setzero_v2r8();
297 fjy3 = _fjsp_setzero_v2r8();
298 fjz3 = _fjsp_setzero_v2r8();
300 /**************************
301 * CALCULATE INTERACTIONS *
302 **************************/
304 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
306 /* Calculate table index by multiplying r with table scale and truncate to integer */
307 rt = _fjsp_mul_v2r8(r00,vftabscale);
308 itab_tmp = _fjsp_dtox_v2r8(rt);
309 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
310 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
311 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
316 /* CUBIC SPLINE TABLE DISPERSION */
317 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
318 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
319 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
320 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
321 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
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 vvdw6 = _fjsp_mul_v2r8(c6_00,VV);
326 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
327 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
329 /* CUBIC SPLINE TABLE REPULSION */
330 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
331 F = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
332 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
333 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
334 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
335 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
336 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
337 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
338 vvdw12 = _fjsp_mul_v2r8(c12_00,VV);
339 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
340 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
341 vvdw = _fjsp_add_v2r8(vvdw12,vvdw6);
342 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
344 /* Update potential sum for this i atom from the interaction with this j atom. */
345 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
349 /* Update vectorial force */
350 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
351 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
352 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
354 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
355 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
356 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
358 /**************************
359 * CALCULATE INTERACTIONS *
360 **************************/
362 /* COULOMB ELECTROSTATICS */
363 velec = _fjsp_mul_v2r8(qq11,rinv11);
364 felec = _fjsp_mul_v2r8(velec,rinvsq11);
366 /* Update potential sum for this i atom from the interaction with this j atom. */
367 velecsum = _fjsp_add_v2r8(velecsum,velec);
371 /* Update vectorial force */
372 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
373 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
374 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
376 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
377 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
378 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
380 /**************************
381 * CALCULATE INTERACTIONS *
382 **************************/
384 /* COULOMB ELECTROSTATICS */
385 velec = _fjsp_mul_v2r8(qq12,rinv12);
386 felec = _fjsp_mul_v2r8(velec,rinvsq12);
388 /* Update potential sum for this i atom from the interaction with this j atom. */
389 velecsum = _fjsp_add_v2r8(velecsum,velec);
393 /* Update vectorial force */
394 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
395 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
396 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
398 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
399 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
400 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
402 /**************************
403 * CALCULATE INTERACTIONS *
404 **************************/
406 /* COULOMB ELECTROSTATICS */
407 velec = _fjsp_mul_v2r8(qq13,rinv13);
408 felec = _fjsp_mul_v2r8(velec,rinvsq13);
410 /* Update potential sum for this i atom from the interaction with this j atom. */
411 velecsum = _fjsp_add_v2r8(velecsum,velec);
415 /* Update vectorial force */
416 fix1 = _fjsp_madd_v2r8(dx13,fscal,fix1);
417 fiy1 = _fjsp_madd_v2r8(dy13,fscal,fiy1);
418 fiz1 = _fjsp_madd_v2r8(dz13,fscal,fiz1);
420 fjx3 = _fjsp_madd_v2r8(dx13,fscal,fjx3);
421 fjy3 = _fjsp_madd_v2r8(dy13,fscal,fjy3);
422 fjz3 = _fjsp_madd_v2r8(dz13,fscal,fjz3);
424 /**************************
425 * CALCULATE INTERACTIONS *
426 **************************/
428 /* COULOMB ELECTROSTATICS */
429 velec = _fjsp_mul_v2r8(qq21,rinv21);
430 felec = _fjsp_mul_v2r8(velec,rinvsq21);
432 /* Update potential sum for this i atom from the interaction with this j atom. */
433 velecsum = _fjsp_add_v2r8(velecsum,velec);
437 /* Update vectorial force */
438 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
439 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
440 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
442 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
443 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
444 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
446 /**************************
447 * CALCULATE INTERACTIONS *
448 **************************/
450 /* COULOMB ELECTROSTATICS */
451 velec = _fjsp_mul_v2r8(qq22,rinv22);
452 felec = _fjsp_mul_v2r8(velec,rinvsq22);
454 /* Update potential sum for this i atom from the interaction with this j atom. */
455 velecsum = _fjsp_add_v2r8(velecsum,velec);
459 /* Update vectorial force */
460 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
461 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
462 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
464 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
465 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
466 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
468 /**************************
469 * CALCULATE INTERACTIONS *
470 **************************/
472 /* COULOMB ELECTROSTATICS */
473 velec = _fjsp_mul_v2r8(qq23,rinv23);
474 felec = _fjsp_mul_v2r8(velec,rinvsq23);
476 /* Update potential sum for this i atom from the interaction with this j atom. */
477 velecsum = _fjsp_add_v2r8(velecsum,velec);
481 /* Update vectorial force */
482 fix2 = _fjsp_madd_v2r8(dx23,fscal,fix2);
483 fiy2 = _fjsp_madd_v2r8(dy23,fscal,fiy2);
484 fiz2 = _fjsp_madd_v2r8(dz23,fscal,fiz2);
486 fjx3 = _fjsp_madd_v2r8(dx23,fscal,fjx3);
487 fjy3 = _fjsp_madd_v2r8(dy23,fscal,fjy3);
488 fjz3 = _fjsp_madd_v2r8(dz23,fscal,fjz3);
490 /**************************
491 * CALCULATE INTERACTIONS *
492 **************************/
494 /* COULOMB ELECTROSTATICS */
495 velec = _fjsp_mul_v2r8(qq31,rinv31);
496 felec = _fjsp_mul_v2r8(velec,rinvsq31);
498 /* Update potential sum for this i atom from the interaction with this j atom. */
499 velecsum = _fjsp_add_v2r8(velecsum,velec);
503 /* Update vectorial force */
504 fix3 = _fjsp_madd_v2r8(dx31,fscal,fix3);
505 fiy3 = _fjsp_madd_v2r8(dy31,fscal,fiy3);
506 fiz3 = _fjsp_madd_v2r8(dz31,fscal,fiz3);
508 fjx1 = _fjsp_madd_v2r8(dx31,fscal,fjx1);
509 fjy1 = _fjsp_madd_v2r8(dy31,fscal,fjy1);
510 fjz1 = _fjsp_madd_v2r8(dz31,fscal,fjz1);
512 /**************************
513 * CALCULATE INTERACTIONS *
514 **************************/
516 /* COULOMB ELECTROSTATICS */
517 velec = _fjsp_mul_v2r8(qq32,rinv32);
518 felec = _fjsp_mul_v2r8(velec,rinvsq32);
520 /* Update potential sum for this i atom from the interaction with this j atom. */
521 velecsum = _fjsp_add_v2r8(velecsum,velec);
525 /* Update vectorial force */
526 fix3 = _fjsp_madd_v2r8(dx32,fscal,fix3);
527 fiy3 = _fjsp_madd_v2r8(dy32,fscal,fiy3);
528 fiz3 = _fjsp_madd_v2r8(dz32,fscal,fiz3);
530 fjx2 = _fjsp_madd_v2r8(dx32,fscal,fjx2);
531 fjy2 = _fjsp_madd_v2r8(dy32,fscal,fjy2);
532 fjz2 = _fjsp_madd_v2r8(dz32,fscal,fjz2);
534 /**************************
535 * CALCULATE INTERACTIONS *
536 **************************/
538 /* COULOMB ELECTROSTATICS */
539 velec = _fjsp_mul_v2r8(qq33,rinv33);
540 felec = _fjsp_mul_v2r8(velec,rinvsq33);
542 /* Update potential sum for this i atom from the interaction with this j atom. */
543 velecsum = _fjsp_add_v2r8(velecsum,velec);
547 /* Update vectorial force */
548 fix3 = _fjsp_madd_v2r8(dx33,fscal,fix3);
549 fiy3 = _fjsp_madd_v2r8(dy33,fscal,fiy3);
550 fiz3 = _fjsp_madd_v2r8(dz33,fscal,fiz3);
552 fjx3 = _fjsp_madd_v2r8(dx33,fscal,fjx3);
553 fjy3 = _fjsp_madd_v2r8(dy33,fscal,fjy3);
554 fjz3 = _fjsp_madd_v2r8(dz33,fscal,fjz3);
556 gmx_fjsp_decrement_4rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
558 /* Inner loop uses 341 flops */
565 j_coord_offsetA = DIM*jnrA;
567 /* load j atom coordinates */
568 gmx_fjsp_load_4rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
569 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
570 &jy2,&jz2,&jx3,&jy3,&jz3);
572 /* Calculate displacement vector */
573 dx00 = _fjsp_sub_v2r8(ix0,jx0);
574 dy00 = _fjsp_sub_v2r8(iy0,jy0);
575 dz00 = _fjsp_sub_v2r8(iz0,jz0);
576 dx11 = _fjsp_sub_v2r8(ix1,jx1);
577 dy11 = _fjsp_sub_v2r8(iy1,jy1);
578 dz11 = _fjsp_sub_v2r8(iz1,jz1);
579 dx12 = _fjsp_sub_v2r8(ix1,jx2);
580 dy12 = _fjsp_sub_v2r8(iy1,jy2);
581 dz12 = _fjsp_sub_v2r8(iz1,jz2);
582 dx13 = _fjsp_sub_v2r8(ix1,jx3);
583 dy13 = _fjsp_sub_v2r8(iy1,jy3);
584 dz13 = _fjsp_sub_v2r8(iz1,jz3);
585 dx21 = _fjsp_sub_v2r8(ix2,jx1);
586 dy21 = _fjsp_sub_v2r8(iy2,jy1);
587 dz21 = _fjsp_sub_v2r8(iz2,jz1);
588 dx22 = _fjsp_sub_v2r8(ix2,jx2);
589 dy22 = _fjsp_sub_v2r8(iy2,jy2);
590 dz22 = _fjsp_sub_v2r8(iz2,jz2);
591 dx23 = _fjsp_sub_v2r8(ix2,jx3);
592 dy23 = _fjsp_sub_v2r8(iy2,jy3);
593 dz23 = _fjsp_sub_v2r8(iz2,jz3);
594 dx31 = _fjsp_sub_v2r8(ix3,jx1);
595 dy31 = _fjsp_sub_v2r8(iy3,jy1);
596 dz31 = _fjsp_sub_v2r8(iz3,jz1);
597 dx32 = _fjsp_sub_v2r8(ix3,jx2);
598 dy32 = _fjsp_sub_v2r8(iy3,jy2);
599 dz32 = _fjsp_sub_v2r8(iz3,jz2);
600 dx33 = _fjsp_sub_v2r8(ix3,jx3);
601 dy33 = _fjsp_sub_v2r8(iy3,jy3);
602 dz33 = _fjsp_sub_v2r8(iz3,jz3);
604 /* Calculate squared distance and things based on it */
605 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
606 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
607 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
608 rsq13 = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
609 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
610 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
611 rsq23 = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
612 rsq31 = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
613 rsq32 = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
614 rsq33 = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
616 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
617 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
618 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
619 rinv13 = gmx_fjsp_invsqrt_v2r8(rsq13);
620 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
621 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
622 rinv23 = gmx_fjsp_invsqrt_v2r8(rsq23);
623 rinv31 = gmx_fjsp_invsqrt_v2r8(rsq31);
624 rinv32 = gmx_fjsp_invsqrt_v2r8(rsq32);
625 rinv33 = gmx_fjsp_invsqrt_v2r8(rsq33);
627 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
628 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
629 rinvsq13 = _fjsp_mul_v2r8(rinv13,rinv13);
630 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
631 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
632 rinvsq23 = _fjsp_mul_v2r8(rinv23,rinv23);
633 rinvsq31 = _fjsp_mul_v2r8(rinv31,rinv31);
634 rinvsq32 = _fjsp_mul_v2r8(rinv32,rinv32);
635 rinvsq33 = _fjsp_mul_v2r8(rinv33,rinv33);
637 fjx0 = _fjsp_setzero_v2r8();
638 fjy0 = _fjsp_setzero_v2r8();
639 fjz0 = _fjsp_setzero_v2r8();
640 fjx1 = _fjsp_setzero_v2r8();
641 fjy1 = _fjsp_setzero_v2r8();
642 fjz1 = _fjsp_setzero_v2r8();
643 fjx2 = _fjsp_setzero_v2r8();
644 fjy2 = _fjsp_setzero_v2r8();
645 fjz2 = _fjsp_setzero_v2r8();
646 fjx3 = _fjsp_setzero_v2r8();
647 fjy3 = _fjsp_setzero_v2r8();
648 fjz3 = _fjsp_setzero_v2r8();
650 /**************************
651 * CALCULATE INTERACTIONS *
652 **************************/
654 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
656 /* Calculate table index by multiplying r with table scale and truncate to integer */
657 rt = _fjsp_mul_v2r8(r00,vftabscale);
658 itab_tmp = _fjsp_dtox_v2r8(rt);
659 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
660 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
661 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
666 /* CUBIC SPLINE TABLE DISPERSION */
667 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
668 F = _fjsp_setzero_v2r8();
669 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
670 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
671 H = _fjsp_setzero_v2r8();
672 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
673 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
674 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
675 vvdw6 = _fjsp_mul_v2r8(c6_00,VV);
676 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
677 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
679 /* CUBIC SPLINE TABLE REPULSION */
680 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
681 F = _fjsp_setzero_v2r8();
682 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
683 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
684 H = _fjsp_setzero_v2r8();
685 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
686 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
687 VV = _fjsp_madd_v2r8(vfeps,Fp,Y);
688 vvdw12 = _fjsp_mul_v2r8(c12_00,VV);
689 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
690 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
691 vvdw = _fjsp_add_v2r8(vvdw12,vvdw6);
692 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
694 /* Update potential sum for this i atom from the interaction with this j atom. */
695 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
696 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
700 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
702 /* Update vectorial force */
703 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
704 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
705 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
707 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
708 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
709 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
711 /**************************
712 * CALCULATE INTERACTIONS *
713 **************************/
715 /* COULOMB ELECTROSTATICS */
716 velec = _fjsp_mul_v2r8(qq11,rinv11);
717 felec = _fjsp_mul_v2r8(velec,rinvsq11);
719 /* Update potential sum for this i atom from the interaction with this j atom. */
720 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
721 velecsum = _fjsp_add_v2r8(velecsum,velec);
725 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
727 /* Update vectorial force */
728 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
729 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
730 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
732 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
733 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
734 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
736 /**************************
737 * CALCULATE INTERACTIONS *
738 **************************/
740 /* COULOMB ELECTROSTATICS */
741 velec = _fjsp_mul_v2r8(qq12,rinv12);
742 felec = _fjsp_mul_v2r8(velec,rinvsq12);
744 /* Update potential sum for this i atom from the interaction with this j atom. */
745 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
746 velecsum = _fjsp_add_v2r8(velecsum,velec);
750 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
752 /* Update vectorial force */
753 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
754 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
755 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
757 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
758 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
759 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
761 /**************************
762 * CALCULATE INTERACTIONS *
763 **************************/
765 /* COULOMB ELECTROSTATICS */
766 velec = _fjsp_mul_v2r8(qq13,rinv13);
767 felec = _fjsp_mul_v2r8(velec,rinvsq13);
769 /* Update potential sum for this i atom from the interaction with this j atom. */
770 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
771 velecsum = _fjsp_add_v2r8(velecsum,velec);
775 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
777 /* Update vectorial force */
778 fix1 = _fjsp_madd_v2r8(dx13,fscal,fix1);
779 fiy1 = _fjsp_madd_v2r8(dy13,fscal,fiy1);
780 fiz1 = _fjsp_madd_v2r8(dz13,fscal,fiz1);
782 fjx3 = _fjsp_madd_v2r8(dx13,fscal,fjx3);
783 fjy3 = _fjsp_madd_v2r8(dy13,fscal,fjy3);
784 fjz3 = _fjsp_madd_v2r8(dz13,fscal,fjz3);
786 /**************************
787 * CALCULATE INTERACTIONS *
788 **************************/
790 /* COULOMB ELECTROSTATICS */
791 velec = _fjsp_mul_v2r8(qq21,rinv21);
792 felec = _fjsp_mul_v2r8(velec,rinvsq21);
794 /* Update potential sum for this i atom from the interaction with this j atom. */
795 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
796 velecsum = _fjsp_add_v2r8(velecsum,velec);
800 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
802 /* Update vectorial force */
803 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
804 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
805 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
807 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
808 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
809 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
811 /**************************
812 * CALCULATE INTERACTIONS *
813 **************************/
815 /* COULOMB ELECTROSTATICS */
816 velec = _fjsp_mul_v2r8(qq22,rinv22);
817 felec = _fjsp_mul_v2r8(velec,rinvsq22);
819 /* Update potential sum for this i atom from the interaction with this j atom. */
820 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
821 velecsum = _fjsp_add_v2r8(velecsum,velec);
825 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
827 /* Update vectorial force */
828 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
829 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
830 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
832 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
833 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
834 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
836 /**************************
837 * CALCULATE INTERACTIONS *
838 **************************/
840 /* COULOMB ELECTROSTATICS */
841 velec = _fjsp_mul_v2r8(qq23,rinv23);
842 felec = _fjsp_mul_v2r8(velec,rinvsq23);
844 /* Update potential sum for this i atom from the interaction with this j atom. */
845 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
846 velecsum = _fjsp_add_v2r8(velecsum,velec);
850 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
852 /* Update vectorial force */
853 fix2 = _fjsp_madd_v2r8(dx23,fscal,fix2);
854 fiy2 = _fjsp_madd_v2r8(dy23,fscal,fiy2);
855 fiz2 = _fjsp_madd_v2r8(dz23,fscal,fiz2);
857 fjx3 = _fjsp_madd_v2r8(dx23,fscal,fjx3);
858 fjy3 = _fjsp_madd_v2r8(dy23,fscal,fjy3);
859 fjz3 = _fjsp_madd_v2r8(dz23,fscal,fjz3);
861 /**************************
862 * CALCULATE INTERACTIONS *
863 **************************/
865 /* COULOMB ELECTROSTATICS */
866 velec = _fjsp_mul_v2r8(qq31,rinv31);
867 felec = _fjsp_mul_v2r8(velec,rinvsq31);
869 /* Update potential sum for this i atom from the interaction with this j atom. */
870 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
871 velecsum = _fjsp_add_v2r8(velecsum,velec);
875 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
877 /* Update vectorial force */
878 fix3 = _fjsp_madd_v2r8(dx31,fscal,fix3);
879 fiy3 = _fjsp_madd_v2r8(dy31,fscal,fiy3);
880 fiz3 = _fjsp_madd_v2r8(dz31,fscal,fiz3);
882 fjx1 = _fjsp_madd_v2r8(dx31,fscal,fjx1);
883 fjy1 = _fjsp_madd_v2r8(dy31,fscal,fjy1);
884 fjz1 = _fjsp_madd_v2r8(dz31,fscal,fjz1);
886 /**************************
887 * CALCULATE INTERACTIONS *
888 **************************/
890 /* COULOMB ELECTROSTATICS */
891 velec = _fjsp_mul_v2r8(qq32,rinv32);
892 felec = _fjsp_mul_v2r8(velec,rinvsq32);
894 /* Update potential sum for this i atom from the interaction with this j atom. */
895 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
896 velecsum = _fjsp_add_v2r8(velecsum,velec);
900 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
902 /* Update vectorial force */
903 fix3 = _fjsp_madd_v2r8(dx32,fscal,fix3);
904 fiy3 = _fjsp_madd_v2r8(dy32,fscal,fiy3);
905 fiz3 = _fjsp_madd_v2r8(dz32,fscal,fiz3);
907 fjx2 = _fjsp_madd_v2r8(dx32,fscal,fjx2);
908 fjy2 = _fjsp_madd_v2r8(dy32,fscal,fjy2);
909 fjz2 = _fjsp_madd_v2r8(dz32,fscal,fjz2);
911 /**************************
912 * CALCULATE INTERACTIONS *
913 **************************/
915 /* COULOMB ELECTROSTATICS */
916 velec = _fjsp_mul_v2r8(qq33,rinv33);
917 felec = _fjsp_mul_v2r8(velec,rinvsq33);
919 /* Update potential sum for this i atom from the interaction with this j atom. */
920 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
921 velecsum = _fjsp_add_v2r8(velecsum,velec);
925 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
927 /* Update vectorial force */
928 fix3 = _fjsp_madd_v2r8(dx33,fscal,fix3);
929 fiy3 = _fjsp_madd_v2r8(dy33,fscal,fiy3);
930 fiz3 = _fjsp_madd_v2r8(dz33,fscal,fiz3);
932 fjx3 = _fjsp_madd_v2r8(dx33,fscal,fjx3);
933 fjy3 = _fjsp_madd_v2r8(dy33,fscal,fjy3);
934 fjz3 = _fjsp_madd_v2r8(dz33,fscal,fjz3);
936 gmx_fjsp_decrement_4rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
938 /* Inner loop uses 341 flops */
941 /* End of innermost loop */
943 gmx_fjsp_update_iforce_4atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
944 f+i_coord_offset,fshift+i_shift_offset);
947 /* Update potential energies */
948 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
949 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
951 /* Increment number of inner iterations */
952 inneriter += j_index_end - j_index_start;
954 /* Outer loop uses 26 flops */
957 /* Increment number of outer iterations */
960 /* Update outer/inner flops */
962 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*26 + inneriter*341);
965 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwCSTab_GeomW4W4_F_sparc64_hpc_ace_double
966 * Electrostatics interaction: Coulomb
967 * VdW interaction: CubicSplineTable
968 * Geometry: Water4-Water4
969 * Calculate force/pot: Force
972 nb_kernel_ElecCoul_VdwCSTab_GeomW4W4_F_sparc64_hpc_ace_double
973 (t_nblist * gmx_restrict nlist,
974 rvec * gmx_restrict xx,
975 rvec * gmx_restrict ff,
976 t_forcerec * gmx_restrict fr,
977 t_mdatoms * gmx_restrict mdatoms,
978 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
979 t_nrnb * gmx_restrict nrnb)
981 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
982 * just 0 for non-waters.
983 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
984 * jnr indices corresponding to data put in the four positions in the SIMD register.
986 int i_shift_offset,i_coord_offset,outeriter,inneriter;
987 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
989 int j_coord_offsetA,j_coord_offsetB;
990 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
992 real *shiftvec,*fshift,*x,*f;
993 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
995 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
997 _fjsp_v2r8 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
999 _fjsp_v2r8 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1001 _fjsp_v2r8 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
1002 int vdwjidx0A,vdwjidx0B;
1003 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1004 int vdwjidx1A,vdwjidx1B;
1005 _fjsp_v2r8 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1006 int vdwjidx2A,vdwjidx2B;
1007 _fjsp_v2r8 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1008 int vdwjidx3A,vdwjidx3B;
1009 _fjsp_v2r8 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
1010 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1011 _fjsp_v2r8 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1012 _fjsp_v2r8 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1013 _fjsp_v2r8 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
1014 _fjsp_v2r8 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1015 _fjsp_v2r8 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1016 _fjsp_v2r8 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
1017 _fjsp_v2r8 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
1018 _fjsp_v2r8 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
1019 _fjsp_v2r8 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
1020 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
1023 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1026 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
1027 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
1028 _fjsp_v2r8 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
1030 _fjsp_v2r8 itab_tmp;
1031 _fjsp_v2r8 dummy_mask,cutoff_mask;
1032 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
1033 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
1034 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
1041 jindex = nlist->jindex;
1043 shiftidx = nlist->shift;
1045 shiftvec = fr->shift_vec[0];
1046 fshift = fr->fshift[0];
1047 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
1048 charge = mdatoms->chargeA;
1049 nvdwtype = fr->ntype;
1050 vdwparam = fr->nbfp;
1051 vdwtype = mdatoms->typeA;
1053 vftab = kernel_data->table_vdw->data;
1054 vftabscale = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
1056 /* Setup water-specific parameters */
1057 inr = nlist->iinr[0];
1058 iq1 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
1059 iq2 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
1060 iq3 = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+3]));
1061 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
1063 jq1 = gmx_fjsp_set1_v2r8(charge[inr+1]);
1064 jq2 = gmx_fjsp_set1_v2r8(charge[inr+2]);
1065 jq3 = gmx_fjsp_set1_v2r8(charge[inr+3]);
1066 vdwjidx0A = 2*vdwtype[inr+0];
1067 c6_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
1068 c12_00 = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
1069 qq11 = _fjsp_mul_v2r8(iq1,jq1);
1070 qq12 = _fjsp_mul_v2r8(iq1,jq2);
1071 qq13 = _fjsp_mul_v2r8(iq1,jq3);
1072 qq21 = _fjsp_mul_v2r8(iq2,jq1);
1073 qq22 = _fjsp_mul_v2r8(iq2,jq2);
1074 qq23 = _fjsp_mul_v2r8(iq2,jq3);
1075 qq31 = _fjsp_mul_v2r8(iq3,jq1);
1076 qq32 = _fjsp_mul_v2r8(iq3,jq2);
1077 qq33 = _fjsp_mul_v2r8(iq3,jq3);
1079 /* Avoid stupid compiler warnings */
1081 j_coord_offsetA = 0;
1082 j_coord_offsetB = 0;
1087 /* Start outer loop over neighborlists */
1088 for(iidx=0; iidx<nri; iidx++)
1090 /* Load shift vector for this list */
1091 i_shift_offset = DIM*shiftidx[iidx];
1093 /* Load limits for loop over neighbors */
1094 j_index_start = jindex[iidx];
1095 j_index_end = jindex[iidx+1];
1097 /* Get outer coordinate index */
1099 i_coord_offset = DIM*inr;
1101 /* Load i particle coords and add shift vector */
1102 gmx_fjsp_load_shift_and_4rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
1103 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1105 fix0 = _fjsp_setzero_v2r8();
1106 fiy0 = _fjsp_setzero_v2r8();
1107 fiz0 = _fjsp_setzero_v2r8();
1108 fix1 = _fjsp_setzero_v2r8();
1109 fiy1 = _fjsp_setzero_v2r8();
1110 fiz1 = _fjsp_setzero_v2r8();
1111 fix2 = _fjsp_setzero_v2r8();
1112 fiy2 = _fjsp_setzero_v2r8();
1113 fiz2 = _fjsp_setzero_v2r8();
1114 fix3 = _fjsp_setzero_v2r8();
1115 fiy3 = _fjsp_setzero_v2r8();
1116 fiz3 = _fjsp_setzero_v2r8();
1118 /* Start inner kernel loop */
1119 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1122 /* Get j neighbor index, and coordinate index */
1124 jnrB = jjnr[jidx+1];
1125 j_coord_offsetA = DIM*jnrA;
1126 j_coord_offsetB = DIM*jnrB;
1128 /* load j atom coordinates */
1129 gmx_fjsp_load_4rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
1130 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
1131 &jy2,&jz2,&jx3,&jy3,&jz3);
1133 /* Calculate displacement vector */
1134 dx00 = _fjsp_sub_v2r8(ix0,jx0);
1135 dy00 = _fjsp_sub_v2r8(iy0,jy0);
1136 dz00 = _fjsp_sub_v2r8(iz0,jz0);
1137 dx11 = _fjsp_sub_v2r8(ix1,jx1);
1138 dy11 = _fjsp_sub_v2r8(iy1,jy1);
1139 dz11 = _fjsp_sub_v2r8(iz1,jz1);
1140 dx12 = _fjsp_sub_v2r8(ix1,jx2);
1141 dy12 = _fjsp_sub_v2r8(iy1,jy2);
1142 dz12 = _fjsp_sub_v2r8(iz1,jz2);
1143 dx13 = _fjsp_sub_v2r8(ix1,jx3);
1144 dy13 = _fjsp_sub_v2r8(iy1,jy3);
1145 dz13 = _fjsp_sub_v2r8(iz1,jz3);
1146 dx21 = _fjsp_sub_v2r8(ix2,jx1);
1147 dy21 = _fjsp_sub_v2r8(iy2,jy1);
1148 dz21 = _fjsp_sub_v2r8(iz2,jz1);
1149 dx22 = _fjsp_sub_v2r8(ix2,jx2);
1150 dy22 = _fjsp_sub_v2r8(iy2,jy2);
1151 dz22 = _fjsp_sub_v2r8(iz2,jz2);
1152 dx23 = _fjsp_sub_v2r8(ix2,jx3);
1153 dy23 = _fjsp_sub_v2r8(iy2,jy3);
1154 dz23 = _fjsp_sub_v2r8(iz2,jz3);
1155 dx31 = _fjsp_sub_v2r8(ix3,jx1);
1156 dy31 = _fjsp_sub_v2r8(iy3,jy1);
1157 dz31 = _fjsp_sub_v2r8(iz3,jz1);
1158 dx32 = _fjsp_sub_v2r8(ix3,jx2);
1159 dy32 = _fjsp_sub_v2r8(iy3,jy2);
1160 dz32 = _fjsp_sub_v2r8(iz3,jz2);
1161 dx33 = _fjsp_sub_v2r8(ix3,jx3);
1162 dy33 = _fjsp_sub_v2r8(iy3,jy3);
1163 dz33 = _fjsp_sub_v2r8(iz3,jz3);
1165 /* Calculate squared distance and things based on it */
1166 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1167 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1168 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1169 rsq13 = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
1170 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1171 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1172 rsq23 = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
1173 rsq31 = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
1174 rsq32 = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
1175 rsq33 = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
1177 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
1178 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
1179 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
1180 rinv13 = gmx_fjsp_invsqrt_v2r8(rsq13);
1181 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
1182 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
1183 rinv23 = gmx_fjsp_invsqrt_v2r8(rsq23);
1184 rinv31 = gmx_fjsp_invsqrt_v2r8(rsq31);
1185 rinv32 = gmx_fjsp_invsqrt_v2r8(rsq32);
1186 rinv33 = gmx_fjsp_invsqrt_v2r8(rsq33);
1188 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
1189 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
1190 rinvsq13 = _fjsp_mul_v2r8(rinv13,rinv13);
1191 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
1192 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
1193 rinvsq23 = _fjsp_mul_v2r8(rinv23,rinv23);
1194 rinvsq31 = _fjsp_mul_v2r8(rinv31,rinv31);
1195 rinvsq32 = _fjsp_mul_v2r8(rinv32,rinv32);
1196 rinvsq33 = _fjsp_mul_v2r8(rinv33,rinv33);
1198 fjx0 = _fjsp_setzero_v2r8();
1199 fjy0 = _fjsp_setzero_v2r8();
1200 fjz0 = _fjsp_setzero_v2r8();
1201 fjx1 = _fjsp_setzero_v2r8();
1202 fjy1 = _fjsp_setzero_v2r8();
1203 fjz1 = _fjsp_setzero_v2r8();
1204 fjx2 = _fjsp_setzero_v2r8();
1205 fjy2 = _fjsp_setzero_v2r8();
1206 fjz2 = _fjsp_setzero_v2r8();
1207 fjx3 = _fjsp_setzero_v2r8();
1208 fjy3 = _fjsp_setzero_v2r8();
1209 fjz3 = _fjsp_setzero_v2r8();
1211 /**************************
1212 * CALCULATE INTERACTIONS *
1213 **************************/
1215 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
1217 /* Calculate table index by multiplying r with table scale and truncate to integer */
1218 rt = _fjsp_mul_v2r8(r00,vftabscale);
1219 itab_tmp = _fjsp_dtox_v2r8(rt);
1220 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1221 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1222 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1227 /* CUBIC SPLINE TABLE DISPERSION */
1228 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1229 F = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1230 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1231 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1232 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
1233 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1234 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1235 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1236 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
1238 /* CUBIC SPLINE TABLE REPULSION */
1239 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1240 F = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
1241 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1242 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1243 H = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
1244 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1245 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1246 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1247 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
1248 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1252 /* Update vectorial force */
1253 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
1254 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1255 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1257 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1258 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1259 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1261 /**************************
1262 * CALCULATE INTERACTIONS *
1263 **************************/
1265 /* COULOMB ELECTROSTATICS */
1266 velec = _fjsp_mul_v2r8(qq11,rinv11);
1267 felec = _fjsp_mul_v2r8(velec,rinvsq11);
1271 /* Update vectorial force */
1272 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
1273 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1274 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1276 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1277 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1278 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1280 /**************************
1281 * CALCULATE INTERACTIONS *
1282 **************************/
1284 /* COULOMB ELECTROSTATICS */
1285 velec = _fjsp_mul_v2r8(qq12,rinv12);
1286 felec = _fjsp_mul_v2r8(velec,rinvsq12);
1290 /* Update vectorial force */
1291 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
1292 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1293 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1295 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1296 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1297 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1299 /**************************
1300 * CALCULATE INTERACTIONS *
1301 **************************/
1303 /* COULOMB ELECTROSTATICS */
1304 velec = _fjsp_mul_v2r8(qq13,rinv13);
1305 felec = _fjsp_mul_v2r8(velec,rinvsq13);
1309 /* Update vectorial force */
1310 fix1 = _fjsp_madd_v2r8(dx13,fscal,fix1);
1311 fiy1 = _fjsp_madd_v2r8(dy13,fscal,fiy1);
1312 fiz1 = _fjsp_madd_v2r8(dz13,fscal,fiz1);
1314 fjx3 = _fjsp_madd_v2r8(dx13,fscal,fjx3);
1315 fjy3 = _fjsp_madd_v2r8(dy13,fscal,fjy3);
1316 fjz3 = _fjsp_madd_v2r8(dz13,fscal,fjz3);
1318 /**************************
1319 * CALCULATE INTERACTIONS *
1320 **************************/
1322 /* COULOMB ELECTROSTATICS */
1323 velec = _fjsp_mul_v2r8(qq21,rinv21);
1324 felec = _fjsp_mul_v2r8(velec,rinvsq21);
1328 /* Update vectorial force */
1329 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
1330 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1331 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1333 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1334 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1335 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1337 /**************************
1338 * CALCULATE INTERACTIONS *
1339 **************************/
1341 /* COULOMB ELECTROSTATICS */
1342 velec = _fjsp_mul_v2r8(qq22,rinv22);
1343 felec = _fjsp_mul_v2r8(velec,rinvsq22);
1347 /* Update vectorial force */
1348 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
1349 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1350 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1352 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1353 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1354 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1356 /**************************
1357 * CALCULATE INTERACTIONS *
1358 **************************/
1360 /* COULOMB ELECTROSTATICS */
1361 velec = _fjsp_mul_v2r8(qq23,rinv23);
1362 felec = _fjsp_mul_v2r8(velec,rinvsq23);
1366 /* Update vectorial force */
1367 fix2 = _fjsp_madd_v2r8(dx23,fscal,fix2);
1368 fiy2 = _fjsp_madd_v2r8(dy23,fscal,fiy2);
1369 fiz2 = _fjsp_madd_v2r8(dz23,fscal,fiz2);
1371 fjx3 = _fjsp_madd_v2r8(dx23,fscal,fjx3);
1372 fjy3 = _fjsp_madd_v2r8(dy23,fscal,fjy3);
1373 fjz3 = _fjsp_madd_v2r8(dz23,fscal,fjz3);
1375 /**************************
1376 * CALCULATE INTERACTIONS *
1377 **************************/
1379 /* COULOMB ELECTROSTATICS */
1380 velec = _fjsp_mul_v2r8(qq31,rinv31);
1381 felec = _fjsp_mul_v2r8(velec,rinvsq31);
1385 /* Update vectorial force */
1386 fix3 = _fjsp_madd_v2r8(dx31,fscal,fix3);
1387 fiy3 = _fjsp_madd_v2r8(dy31,fscal,fiy3);
1388 fiz3 = _fjsp_madd_v2r8(dz31,fscal,fiz3);
1390 fjx1 = _fjsp_madd_v2r8(dx31,fscal,fjx1);
1391 fjy1 = _fjsp_madd_v2r8(dy31,fscal,fjy1);
1392 fjz1 = _fjsp_madd_v2r8(dz31,fscal,fjz1);
1394 /**************************
1395 * CALCULATE INTERACTIONS *
1396 **************************/
1398 /* COULOMB ELECTROSTATICS */
1399 velec = _fjsp_mul_v2r8(qq32,rinv32);
1400 felec = _fjsp_mul_v2r8(velec,rinvsq32);
1404 /* Update vectorial force */
1405 fix3 = _fjsp_madd_v2r8(dx32,fscal,fix3);
1406 fiy3 = _fjsp_madd_v2r8(dy32,fscal,fiy3);
1407 fiz3 = _fjsp_madd_v2r8(dz32,fscal,fiz3);
1409 fjx2 = _fjsp_madd_v2r8(dx32,fscal,fjx2);
1410 fjy2 = _fjsp_madd_v2r8(dy32,fscal,fjy2);
1411 fjz2 = _fjsp_madd_v2r8(dz32,fscal,fjz2);
1413 /**************************
1414 * CALCULATE INTERACTIONS *
1415 **************************/
1417 /* COULOMB ELECTROSTATICS */
1418 velec = _fjsp_mul_v2r8(qq33,rinv33);
1419 felec = _fjsp_mul_v2r8(velec,rinvsq33);
1423 /* Update vectorial force */
1424 fix3 = _fjsp_madd_v2r8(dx33,fscal,fix3);
1425 fiy3 = _fjsp_madd_v2r8(dy33,fscal,fiy3);
1426 fiz3 = _fjsp_madd_v2r8(dz33,fscal,fiz3);
1428 fjx3 = _fjsp_madd_v2r8(dx33,fscal,fjx3);
1429 fjy3 = _fjsp_madd_v2r8(dy33,fscal,fjy3);
1430 fjz3 = _fjsp_madd_v2r8(dz33,fscal,fjz3);
1432 gmx_fjsp_decrement_4rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1434 /* Inner loop uses 324 flops */
1437 if(jidx<j_index_end)
1441 j_coord_offsetA = DIM*jnrA;
1443 /* load j atom coordinates */
1444 gmx_fjsp_load_4rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
1445 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
1446 &jy2,&jz2,&jx3,&jy3,&jz3);
1448 /* Calculate displacement vector */
1449 dx00 = _fjsp_sub_v2r8(ix0,jx0);
1450 dy00 = _fjsp_sub_v2r8(iy0,jy0);
1451 dz00 = _fjsp_sub_v2r8(iz0,jz0);
1452 dx11 = _fjsp_sub_v2r8(ix1,jx1);
1453 dy11 = _fjsp_sub_v2r8(iy1,jy1);
1454 dz11 = _fjsp_sub_v2r8(iz1,jz1);
1455 dx12 = _fjsp_sub_v2r8(ix1,jx2);
1456 dy12 = _fjsp_sub_v2r8(iy1,jy2);
1457 dz12 = _fjsp_sub_v2r8(iz1,jz2);
1458 dx13 = _fjsp_sub_v2r8(ix1,jx3);
1459 dy13 = _fjsp_sub_v2r8(iy1,jy3);
1460 dz13 = _fjsp_sub_v2r8(iz1,jz3);
1461 dx21 = _fjsp_sub_v2r8(ix2,jx1);
1462 dy21 = _fjsp_sub_v2r8(iy2,jy1);
1463 dz21 = _fjsp_sub_v2r8(iz2,jz1);
1464 dx22 = _fjsp_sub_v2r8(ix2,jx2);
1465 dy22 = _fjsp_sub_v2r8(iy2,jy2);
1466 dz22 = _fjsp_sub_v2r8(iz2,jz2);
1467 dx23 = _fjsp_sub_v2r8(ix2,jx3);
1468 dy23 = _fjsp_sub_v2r8(iy2,jy3);
1469 dz23 = _fjsp_sub_v2r8(iz2,jz3);
1470 dx31 = _fjsp_sub_v2r8(ix3,jx1);
1471 dy31 = _fjsp_sub_v2r8(iy3,jy1);
1472 dz31 = _fjsp_sub_v2r8(iz3,jz1);
1473 dx32 = _fjsp_sub_v2r8(ix3,jx2);
1474 dy32 = _fjsp_sub_v2r8(iy3,jy2);
1475 dz32 = _fjsp_sub_v2r8(iz3,jz2);
1476 dx33 = _fjsp_sub_v2r8(ix3,jx3);
1477 dy33 = _fjsp_sub_v2r8(iy3,jy3);
1478 dz33 = _fjsp_sub_v2r8(iz3,jz3);
1480 /* Calculate squared distance and things based on it */
1481 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1482 rsq11 = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1483 rsq12 = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1484 rsq13 = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
1485 rsq21 = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1486 rsq22 = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1487 rsq23 = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
1488 rsq31 = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
1489 rsq32 = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
1490 rsq33 = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
1492 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
1493 rinv11 = gmx_fjsp_invsqrt_v2r8(rsq11);
1494 rinv12 = gmx_fjsp_invsqrt_v2r8(rsq12);
1495 rinv13 = gmx_fjsp_invsqrt_v2r8(rsq13);
1496 rinv21 = gmx_fjsp_invsqrt_v2r8(rsq21);
1497 rinv22 = gmx_fjsp_invsqrt_v2r8(rsq22);
1498 rinv23 = gmx_fjsp_invsqrt_v2r8(rsq23);
1499 rinv31 = gmx_fjsp_invsqrt_v2r8(rsq31);
1500 rinv32 = gmx_fjsp_invsqrt_v2r8(rsq32);
1501 rinv33 = gmx_fjsp_invsqrt_v2r8(rsq33);
1503 rinvsq11 = _fjsp_mul_v2r8(rinv11,rinv11);
1504 rinvsq12 = _fjsp_mul_v2r8(rinv12,rinv12);
1505 rinvsq13 = _fjsp_mul_v2r8(rinv13,rinv13);
1506 rinvsq21 = _fjsp_mul_v2r8(rinv21,rinv21);
1507 rinvsq22 = _fjsp_mul_v2r8(rinv22,rinv22);
1508 rinvsq23 = _fjsp_mul_v2r8(rinv23,rinv23);
1509 rinvsq31 = _fjsp_mul_v2r8(rinv31,rinv31);
1510 rinvsq32 = _fjsp_mul_v2r8(rinv32,rinv32);
1511 rinvsq33 = _fjsp_mul_v2r8(rinv33,rinv33);
1513 fjx0 = _fjsp_setzero_v2r8();
1514 fjy0 = _fjsp_setzero_v2r8();
1515 fjz0 = _fjsp_setzero_v2r8();
1516 fjx1 = _fjsp_setzero_v2r8();
1517 fjy1 = _fjsp_setzero_v2r8();
1518 fjz1 = _fjsp_setzero_v2r8();
1519 fjx2 = _fjsp_setzero_v2r8();
1520 fjy2 = _fjsp_setzero_v2r8();
1521 fjz2 = _fjsp_setzero_v2r8();
1522 fjx3 = _fjsp_setzero_v2r8();
1523 fjy3 = _fjsp_setzero_v2r8();
1524 fjz3 = _fjsp_setzero_v2r8();
1526 /**************************
1527 * CALCULATE INTERACTIONS *
1528 **************************/
1530 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
1532 /* Calculate table index by multiplying r with table scale and truncate to integer */
1533 rt = _fjsp_mul_v2r8(r00,vftabscale);
1534 itab_tmp = _fjsp_dtox_v2r8(rt);
1535 vfeps = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1536 twovfeps = _fjsp_add_v2r8(vfeps,vfeps);
1537 _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1542 /* CUBIC SPLINE TABLE DISPERSION */
1543 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1544 F = _fjsp_setzero_v2r8();
1545 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1546 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1547 H = _fjsp_setzero_v2r8();
1548 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1549 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1550 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1551 fvdw6 = _fjsp_mul_v2r8(c6_00,FF);
1553 /* CUBIC SPLINE TABLE REPULSION */
1554 Y = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1555 F = _fjsp_setzero_v2r8();
1556 GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1557 G = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1558 H = _fjsp_setzero_v2r8();
1559 GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1560 Fp = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1561 FF = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1562 fvdw12 = _fjsp_mul_v2r8(c12_00,FF);
1563 fvdw = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1567 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1569 /* Update vectorial force */
1570 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
1571 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1572 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1574 fjx0 = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1575 fjy0 = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1576 fjz0 = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1578 /**************************
1579 * CALCULATE INTERACTIONS *
1580 **************************/
1582 /* COULOMB ELECTROSTATICS */
1583 velec = _fjsp_mul_v2r8(qq11,rinv11);
1584 felec = _fjsp_mul_v2r8(velec,rinvsq11);
1588 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1590 /* Update vectorial force */
1591 fix1 = _fjsp_madd_v2r8(dx11,fscal,fix1);
1592 fiy1 = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1593 fiz1 = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1595 fjx1 = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1596 fjy1 = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1597 fjz1 = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1599 /**************************
1600 * CALCULATE INTERACTIONS *
1601 **************************/
1603 /* COULOMB ELECTROSTATICS */
1604 velec = _fjsp_mul_v2r8(qq12,rinv12);
1605 felec = _fjsp_mul_v2r8(velec,rinvsq12);
1609 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1611 /* Update vectorial force */
1612 fix1 = _fjsp_madd_v2r8(dx12,fscal,fix1);
1613 fiy1 = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1614 fiz1 = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1616 fjx2 = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1617 fjy2 = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1618 fjz2 = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1620 /**************************
1621 * CALCULATE INTERACTIONS *
1622 **************************/
1624 /* COULOMB ELECTROSTATICS */
1625 velec = _fjsp_mul_v2r8(qq13,rinv13);
1626 felec = _fjsp_mul_v2r8(velec,rinvsq13);
1630 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1632 /* Update vectorial force */
1633 fix1 = _fjsp_madd_v2r8(dx13,fscal,fix1);
1634 fiy1 = _fjsp_madd_v2r8(dy13,fscal,fiy1);
1635 fiz1 = _fjsp_madd_v2r8(dz13,fscal,fiz1);
1637 fjx3 = _fjsp_madd_v2r8(dx13,fscal,fjx3);
1638 fjy3 = _fjsp_madd_v2r8(dy13,fscal,fjy3);
1639 fjz3 = _fjsp_madd_v2r8(dz13,fscal,fjz3);
1641 /**************************
1642 * CALCULATE INTERACTIONS *
1643 **************************/
1645 /* COULOMB ELECTROSTATICS */
1646 velec = _fjsp_mul_v2r8(qq21,rinv21);
1647 felec = _fjsp_mul_v2r8(velec,rinvsq21);
1651 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1653 /* Update vectorial force */
1654 fix2 = _fjsp_madd_v2r8(dx21,fscal,fix2);
1655 fiy2 = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1656 fiz2 = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1658 fjx1 = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1659 fjy1 = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1660 fjz1 = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1662 /**************************
1663 * CALCULATE INTERACTIONS *
1664 **************************/
1666 /* COULOMB ELECTROSTATICS */
1667 velec = _fjsp_mul_v2r8(qq22,rinv22);
1668 felec = _fjsp_mul_v2r8(velec,rinvsq22);
1672 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1674 /* Update vectorial force */
1675 fix2 = _fjsp_madd_v2r8(dx22,fscal,fix2);
1676 fiy2 = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1677 fiz2 = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1679 fjx2 = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1680 fjy2 = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1681 fjz2 = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1683 /**************************
1684 * CALCULATE INTERACTIONS *
1685 **************************/
1687 /* COULOMB ELECTROSTATICS */
1688 velec = _fjsp_mul_v2r8(qq23,rinv23);
1689 felec = _fjsp_mul_v2r8(velec,rinvsq23);
1693 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1695 /* Update vectorial force */
1696 fix2 = _fjsp_madd_v2r8(dx23,fscal,fix2);
1697 fiy2 = _fjsp_madd_v2r8(dy23,fscal,fiy2);
1698 fiz2 = _fjsp_madd_v2r8(dz23,fscal,fiz2);
1700 fjx3 = _fjsp_madd_v2r8(dx23,fscal,fjx3);
1701 fjy3 = _fjsp_madd_v2r8(dy23,fscal,fjy3);
1702 fjz3 = _fjsp_madd_v2r8(dz23,fscal,fjz3);
1704 /**************************
1705 * CALCULATE INTERACTIONS *
1706 **************************/
1708 /* COULOMB ELECTROSTATICS */
1709 velec = _fjsp_mul_v2r8(qq31,rinv31);
1710 felec = _fjsp_mul_v2r8(velec,rinvsq31);
1714 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1716 /* Update vectorial force */
1717 fix3 = _fjsp_madd_v2r8(dx31,fscal,fix3);
1718 fiy3 = _fjsp_madd_v2r8(dy31,fscal,fiy3);
1719 fiz3 = _fjsp_madd_v2r8(dz31,fscal,fiz3);
1721 fjx1 = _fjsp_madd_v2r8(dx31,fscal,fjx1);
1722 fjy1 = _fjsp_madd_v2r8(dy31,fscal,fjy1);
1723 fjz1 = _fjsp_madd_v2r8(dz31,fscal,fjz1);
1725 /**************************
1726 * CALCULATE INTERACTIONS *
1727 **************************/
1729 /* COULOMB ELECTROSTATICS */
1730 velec = _fjsp_mul_v2r8(qq32,rinv32);
1731 felec = _fjsp_mul_v2r8(velec,rinvsq32);
1735 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1737 /* Update vectorial force */
1738 fix3 = _fjsp_madd_v2r8(dx32,fscal,fix3);
1739 fiy3 = _fjsp_madd_v2r8(dy32,fscal,fiy3);
1740 fiz3 = _fjsp_madd_v2r8(dz32,fscal,fiz3);
1742 fjx2 = _fjsp_madd_v2r8(dx32,fscal,fjx2);
1743 fjy2 = _fjsp_madd_v2r8(dy32,fscal,fjy2);
1744 fjz2 = _fjsp_madd_v2r8(dz32,fscal,fjz2);
1746 /**************************
1747 * CALCULATE INTERACTIONS *
1748 **************************/
1750 /* COULOMB ELECTROSTATICS */
1751 velec = _fjsp_mul_v2r8(qq33,rinv33);
1752 felec = _fjsp_mul_v2r8(velec,rinvsq33);
1756 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1758 /* Update vectorial force */
1759 fix3 = _fjsp_madd_v2r8(dx33,fscal,fix3);
1760 fiy3 = _fjsp_madd_v2r8(dy33,fscal,fiy3);
1761 fiz3 = _fjsp_madd_v2r8(dz33,fscal,fiz3);
1763 fjx3 = _fjsp_madd_v2r8(dx33,fscal,fjx3);
1764 fjy3 = _fjsp_madd_v2r8(dy33,fscal,fjy3);
1765 fjz3 = _fjsp_madd_v2r8(dz33,fscal,fjz3);
1767 gmx_fjsp_decrement_4rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1769 /* Inner loop uses 324 flops */
1772 /* End of innermost loop */
1774 gmx_fjsp_update_iforce_4atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1775 f+i_coord_offset,fshift+i_shift_offset);
1777 /* Increment number of inner iterations */
1778 inneriter += j_index_end - j_index_start;
1780 /* Outer loop uses 24 flops */
1783 /* Increment number of outer iterations */
1786 /* Update outer/inner flops */
1788 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*24 + inneriter*324);