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 avx_128_fma_double kernel generator.
42 #include "../nb_kernel.h"
43 #include "types/simple.h"
44 #include "gromacs/math/vec.h"
47 #include "gromacs/simd/math_x86_avx_128_fma_double.h"
48 #include "kernelutil_x86_avx_128_fma_double.h"
51 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwCSTab_GeomW3W3_VF_avx_128_fma_double
52 * Electrostatics interaction: ReactionField
53 * VdW interaction: CubicSplineTable
54 * Geometry: Water3-Water3
55 * Calculate force/pot: PotentialAndForce
58 nb_kernel_ElecRF_VdwCSTab_GeomW3W3_VF_avx_128_fma_double
59 (t_nblist * gmx_restrict nlist,
60 rvec * gmx_restrict xx,
61 rvec * gmx_restrict ff,
62 t_forcerec * gmx_restrict fr,
63 t_mdatoms * gmx_restrict mdatoms,
64 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
65 t_nrnb * gmx_restrict nrnb)
67 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
68 * just 0 for non-waters.
69 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
70 * jnr indices corresponding to data put in the four positions in the SIMD register.
72 int i_shift_offset,i_coord_offset,outeriter,inneriter;
73 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
75 int j_coord_offsetA,j_coord_offsetB;
76 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
78 real *shiftvec,*fshift,*x,*f;
79 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
81 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
83 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
85 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
86 int vdwjidx0A,vdwjidx0B;
87 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
88 int vdwjidx1A,vdwjidx1B;
89 __m128d jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
90 int vdwjidx2A,vdwjidx2B;
91 __m128d jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
92 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
93 __m128d dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
94 __m128d dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
95 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
96 __m128d dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
97 __m128d dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
98 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
99 __m128d dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
100 __m128d dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
101 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
104 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
107 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
108 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
110 __m128i ifour = _mm_set1_epi32(4);
111 __m128d rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
113 __m128d dummy_mask,cutoff_mask;
114 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
115 __m128d one = _mm_set1_pd(1.0);
116 __m128d two = _mm_set1_pd(2.0);
122 jindex = nlist->jindex;
124 shiftidx = nlist->shift;
126 shiftvec = fr->shift_vec[0];
127 fshift = fr->fshift[0];
128 facel = _mm_set1_pd(fr->epsfac);
129 charge = mdatoms->chargeA;
130 krf = _mm_set1_pd(fr->ic->k_rf);
131 krf2 = _mm_set1_pd(fr->ic->k_rf*2.0);
132 crf = _mm_set1_pd(fr->ic->c_rf);
133 nvdwtype = fr->ntype;
135 vdwtype = mdatoms->typeA;
137 vftab = kernel_data->table_vdw->data;
138 vftabscale = _mm_set1_pd(kernel_data->table_vdw->scale);
140 /* Setup water-specific parameters */
141 inr = nlist->iinr[0];
142 iq0 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
143 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
144 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
145 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
147 jq0 = _mm_set1_pd(charge[inr+0]);
148 jq1 = _mm_set1_pd(charge[inr+1]);
149 jq2 = _mm_set1_pd(charge[inr+2]);
150 vdwjidx0A = 2*vdwtype[inr+0];
151 qq00 = _mm_mul_pd(iq0,jq0);
152 c6_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A]);
153 c12_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A+1]);
154 qq01 = _mm_mul_pd(iq0,jq1);
155 qq02 = _mm_mul_pd(iq0,jq2);
156 qq10 = _mm_mul_pd(iq1,jq0);
157 qq11 = _mm_mul_pd(iq1,jq1);
158 qq12 = _mm_mul_pd(iq1,jq2);
159 qq20 = _mm_mul_pd(iq2,jq0);
160 qq21 = _mm_mul_pd(iq2,jq1);
161 qq22 = _mm_mul_pd(iq2,jq2);
163 /* Avoid stupid compiler warnings */
171 /* Start outer loop over neighborlists */
172 for(iidx=0; iidx<nri; iidx++)
174 /* Load shift vector for this list */
175 i_shift_offset = DIM*shiftidx[iidx];
177 /* Load limits for loop over neighbors */
178 j_index_start = jindex[iidx];
179 j_index_end = jindex[iidx+1];
181 /* Get outer coordinate index */
183 i_coord_offset = DIM*inr;
185 /* Load i particle coords and add shift vector */
186 gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
187 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
189 fix0 = _mm_setzero_pd();
190 fiy0 = _mm_setzero_pd();
191 fiz0 = _mm_setzero_pd();
192 fix1 = _mm_setzero_pd();
193 fiy1 = _mm_setzero_pd();
194 fiz1 = _mm_setzero_pd();
195 fix2 = _mm_setzero_pd();
196 fiy2 = _mm_setzero_pd();
197 fiz2 = _mm_setzero_pd();
199 /* Reset potential sums */
200 velecsum = _mm_setzero_pd();
201 vvdwsum = _mm_setzero_pd();
203 /* Start inner kernel loop */
204 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
207 /* Get j neighbor index, and coordinate index */
210 j_coord_offsetA = DIM*jnrA;
211 j_coord_offsetB = DIM*jnrB;
213 /* load j atom coordinates */
214 gmx_mm_load_3rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
215 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
217 /* Calculate displacement vector */
218 dx00 = _mm_sub_pd(ix0,jx0);
219 dy00 = _mm_sub_pd(iy0,jy0);
220 dz00 = _mm_sub_pd(iz0,jz0);
221 dx01 = _mm_sub_pd(ix0,jx1);
222 dy01 = _mm_sub_pd(iy0,jy1);
223 dz01 = _mm_sub_pd(iz0,jz1);
224 dx02 = _mm_sub_pd(ix0,jx2);
225 dy02 = _mm_sub_pd(iy0,jy2);
226 dz02 = _mm_sub_pd(iz0,jz2);
227 dx10 = _mm_sub_pd(ix1,jx0);
228 dy10 = _mm_sub_pd(iy1,jy0);
229 dz10 = _mm_sub_pd(iz1,jz0);
230 dx11 = _mm_sub_pd(ix1,jx1);
231 dy11 = _mm_sub_pd(iy1,jy1);
232 dz11 = _mm_sub_pd(iz1,jz1);
233 dx12 = _mm_sub_pd(ix1,jx2);
234 dy12 = _mm_sub_pd(iy1,jy2);
235 dz12 = _mm_sub_pd(iz1,jz2);
236 dx20 = _mm_sub_pd(ix2,jx0);
237 dy20 = _mm_sub_pd(iy2,jy0);
238 dz20 = _mm_sub_pd(iz2,jz0);
239 dx21 = _mm_sub_pd(ix2,jx1);
240 dy21 = _mm_sub_pd(iy2,jy1);
241 dz21 = _mm_sub_pd(iz2,jz1);
242 dx22 = _mm_sub_pd(ix2,jx2);
243 dy22 = _mm_sub_pd(iy2,jy2);
244 dz22 = _mm_sub_pd(iz2,jz2);
246 /* Calculate squared distance and things based on it */
247 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
248 rsq01 = gmx_mm_calc_rsq_pd(dx01,dy01,dz01);
249 rsq02 = gmx_mm_calc_rsq_pd(dx02,dy02,dz02);
250 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
251 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
252 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
253 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
254 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
255 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
257 rinv00 = gmx_mm_invsqrt_pd(rsq00);
258 rinv01 = gmx_mm_invsqrt_pd(rsq01);
259 rinv02 = gmx_mm_invsqrt_pd(rsq02);
260 rinv10 = gmx_mm_invsqrt_pd(rsq10);
261 rinv11 = gmx_mm_invsqrt_pd(rsq11);
262 rinv12 = gmx_mm_invsqrt_pd(rsq12);
263 rinv20 = gmx_mm_invsqrt_pd(rsq20);
264 rinv21 = gmx_mm_invsqrt_pd(rsq21);
265 rinv22 = gmx_mm_invsqrt_pd(rsq22);
267 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
268 rinvsq01 = _mm_mul_pd(rinv01,rinv01);
269 rinvsq02 = _mm_mul_pd(rinv02,rinv02);
270 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
271 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
272 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
273 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
274 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
275 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
277 fjx0 = _mm_setzero_pd();
278 fjy0 = _mm_setzero_pd();
279 fjz0 = _mm_setzero_pd();
280 fjx1 = _mm_setzero_pd();
281 fjy1 = _mm_setzero_pd();
282 fjz1 = _mm_setzero_pd();
283 fjx2 = _mm_setzero_pd();
284 fjy2 = _mm_setzero_pd();
285 fjz2 = _mm_setzero_pd();
287 /**************************
288 * CALCULATE INTERACTIONS *
289 **************************/
291 r00 = _mm_mul_pd(rsq00,rinv00);
293 /* Calculate table index by multiplying r with table scale and truncate to integer */
294 rt = _mm_mul_pd(r00,vftabscale);
295 vfitab = _mm_cvttpd_epi32(rt);
297 vfeps = _mm_frcz_pd(rt);
299 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
301 twovfeps = _mm_add_pd(vfeps,vfeps);
302 vfitab = _mm_slli_epi32(vfitab,3);
304 /* REACTION-FIELD ELECTROSTATICS */
305 velec = _mm_mul_pd(qq00,_mm_sub_pd(_mm_macc_pd(krf,rsq00,rinv00),crf));
306 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
308 /* CUBIC SPLINE TABLE DISPERSION */
309 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
310 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
311 GMX_MM_TRANSPOSE2_PD(Y,F);
312 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
313 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
314 GMX_MM_TRANSPOSE2_PD(G,H);
315 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
316 VV = _mm_macc_pd(vfeps,Fp,Y);
317 vvdw6 = _mm_mul_pd(c6_00,VV);
318 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
319 fvdw6 = _mm_mul_pd(c6_00,FF);
321 /* CUBIC SPLINE TABLE REPULSION */
322 vfitab = _mm_add_epi32(vfitab,ifour);
323 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
324 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
325 GMX_MM_TRANSPOSE2_PD(Y,F);
326 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
327 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
328 GMX_MM_TRANSPOSE2_PD(G,H);
329 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
330 VV = _mm_macc_pd(vfeps,Fp,Y);
331 vvdw12 = _mm_mul_pd(c12_00,VV);
332 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
333 fvdw12 = _mm_mul_pd(c12_00,FF);
334 vvdw = _mm_add_pd(vvdw12,vvdw6);
335 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
337 /* Update potential sum for this i atom from the interaction with this j atom. */
338 velecsum = _mm_add_pd(velecsum,velec);
339 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
341 fscal = _mm_add_pd(felec,fvdw);
343 /* Update vectorial force */
344 fix0 = _mm_macc_pd(dx00,fscal,fix0);
345 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
346 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
348 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
349 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
350 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
352 /**************************
353 * CALCULATE INTERACTIONS *
354 **************************/
356 /* REACTION-FIELD ELECTROSTATICS */
357 velec = _mm_mul_pd(qq01,_mm_sub_pd(_mm_macc_pd(krf,rsq01,rinv01),crf));
358 felec = _mm_mul_pd(qq01,_mm_msub_pd(rinv01,rinvsq01,krf2));
360 /* Update potential sum for this i atom from the interaction with this j atom. */
361 velecsum = _mm_add_pd(velecsum,velec);
365 /* Update vectorial force */
366 fix0 = _mm_macc_pd(dx01,fscal,fix0);
367 fiy0 = _mm_macc_pd(dy01,fscal,fiy0);
368 fiz0 = _mm_macc_pd(dz01,fscal,fiz0);
370 fjx1 = _mm_macc_pd(dx01,fscal,fjx1);
371 fjy1 = _mm_macc_pd(dy01,fscal,fjy1);
372 fjz1 = _mm_macc_pd(dz01,fscal,fjz1);
374 /**************************
375 * CALCULATE INTERACTIONS *
376 **************************/
378 /* REACTION-FIELD ELECTROSTATICS */
379 velec = _mm_mul_pd(qq02,_mm_sub_pd(_mm_macc_pd(krf,rsq02,rinv02),crf));
380 felec = _mm_mul_pd(qq02,_mm_msub_pd(rinv02,rinvsq02,krf2));
382 /* Update potential sum for this i atom from the interaction with this j atom. */
383 velecsum = _mm_add_pd(velecsum,velec);
387 /* Update vectorial force */
388 fix0 = _mm_macc_pd(dx02,fscal,fix0);
389 fiy0 = _mm_macc_pd(dy02,fscal,fiy0);
390 fiz0 = _mm_macc_pd(dz02,fscal,fiz0);
392 fjx2 = _mm_macc_pd(dx02,fscal,fjx2);
393 fjy2 = _mm_macc_pd(dy02,fscal,fjy2);
394 fjz2 = _mm_macc_pd(dz02,fscal,fjz2);
396 /**************************
397 * CALCULATE INTERACTIONS *
398 **************************/
400 /* REACTION-FIELD ELECTROSTATICS */
401 velec = _mm_mul_pd(qq10,_mm_sub_pd(_mm_macc_pd(krf,rsq10,rinv10),crf));
402 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
404 /* Update potential sum for this i atom from the interaction with this j atom. */
405 velecsum = _mm_add_pd(velecsum,velec);
409 /* Update vectorial force */
410 fix1 = _mm_macc_pd(dx10,fscal,fix1);
411 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
412 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
414 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
415 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
416 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
418 /**************************
419 * CALCULATE INTERACTIONS *
420 **************************/
422 /* REACTION-FIELD ELECTROSTATICS */
423 velec = _mm_mul_pd(qq11,_mm_sub_pd(_mm_macc_pd(krf,rsq11,rinv11),crf));
424 felec = _mm_mul_pd(qq11,_mm_msub_pd(rinv11,rinvsq11,krf2));
426 /* Update potential sum for this i atom from the interaction with this j atom. */
427 velecsum = _mm_add_pd(velecsum,velec);
431 /* Update vectorial force */
432 fix1 = _mm_macc_pd(dx11,fscal,fix1);
433 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
434 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
436 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
437 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
438 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
440 /**************************
441 * CALCULATE INTERACTIONS *
442 **************************/
444 /* REACTION-FIELD ELECTROSTATICS */
445 velec = _mm_mul_pd(qq12,_mm_sub_pd(_mm_macc_pd(krf,rsq12,rinv12),crf));
446 felec = _mm_mul_pd(qq12,_mm_msub_pd(rinv12,rinvsq12,krf2));
448 /* Update potential sum for this i atom from the interaction with this j atom. */
449 velecsum = _mm_add_pd(velecsum,velec);
453 /* Update vectorial force */
454 fix1 = _mm_macc_pd(dx12,fscal,fix1);
455 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
456 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
458 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
459 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
460 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
462 /**************************
463 * CALCULATE INTERACTIONS *
464 **************************/
466 /* REACTION-FIELD ELECTROSTATICS */
467 velec = _mm_mul_pd(qq20,_mm_sub_pd(_mm_macc_pd(krf,rsq20,rinv20),crf));
468 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
470 /* Update potential sum for this i atom from the interaction with this j atom. */
471 velecsum = _mm_add_pd(velecsum,velec);
475 /* Update vectorial force */
476 fix2 = _mm_macc_pd(dx20,fscal,fix2);
477 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
478 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
480 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
481 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
482 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
484 /**************************
485 * CALCULATE INTERACTIONS *
486 **************************/
488 /* REACTION-FIELD ELECTROSTATICS */
489 velec = _mm_mul_pd(qq21,_mm_sub_pd(_mm_macc_pd(krf,rsq21,rinv21),crf));
490 felec = _mm_mul_pd(qq21,_mm_msub_pd(rinv21,rinvsq21,krf2));
492 /* Update potential sum for this i atom from the interaction with this j atom. */
493 velecsum = _mm_add_pd(velecsum,velec);
497 /* Update vectorial force */
498 fix2 = _mm_macc_pd(dx21,fscal,fix2);
499 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
500 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
502 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
503 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
504 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
506 /**************************
507 * CALCULATE INTERACTIONS *
508 **************************/
510 /* REACTION-FIELD ELECTROSTATICS */
511 velec = _mm_mul_pd(qq22,_mm_sub_pd(_mm_macc_pd(krf,rsq22,rinv22),crf));
512 felec = _mm_mul_pd(qq22,_mm_msub_pd(rinv22,rinvsq22,krf2));
514 /* Update potential sum for this i atom from the interaction with this j atom. */
515 velecsum = _mm_add_pd(velecsum,velec);
519 /* Update vectorial force */
520 fix2 = _mm_macc_pd(dx22,fscal,fix2);
521 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
522 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
524 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
525 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
526 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
528 gmx_mm_decrement_3rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
530 /* Inner loop uses 350 flops */
537 j_coord_offsetA = DIM*jnrA;
539 /* load j atom coordinates */
540 gmx_mm_load_3rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
541 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
543 /* Calculate displacement vector */
544 dx00 = _mm_sub_pd(ix0,jx0);
545 dy00 = _mm_sub_pd(iy0,jy0);
546 dz00 = _mm_sub_pd(iz0,jz0);
547 dx01 = _mm_sub_pd(ix0,jx1);
548 dy01 = _mm_sub_pd(iy0,jy1);
549 dz01 = _mm_sub_pd(iz0,jz1);
550 dx02 = _mm_sub_pd(ix0,jx2);
551 dy02 = _mm_sub_pd(iy0,jy2);
552 dz02 = _mm_sub_pd(iz0,jz2);
553 dx10 = _mm_sub_pd(ix1,jx0);
554 dy10 = _mm_sub_pd(iy1,jy0);
555 dz10 = _mm_sub_pd(iz1,jz0);
556 dx11 = _mm_sub_pd(ix1,jx1);
557 dy11 = _mm_sub_pd(iy1,jy1);
558 dz11 = _mm_sub_pd(iz1,jz1);
559 dx12 = _mm_sub_pd(ix1,jx2);
560 dy12 = _mm_sub_pd(iy1,jy2);
561 dz12 = _mm_sub_pd(iz1,jz2);
562 dx20 = _mm_sub_pd(ix2,jx0);
563 dy20 = _mm_sub_pd(iy2,jy0);
564 dz20 = _mm_sub_pd(iz2,jz0);
565 dx21 = _mm_sub_pd(ix2,jx1);
566 dy21 = _mm_sub_pd(iy2,jy1);
567 dz21 = _mm_sub_pd(iz2,jz1);
568 dx22 = _mm_sub_pd(ix2,jx2);
569 dy22 = _mm_sub_pd(iy2,jy2);
570 dz22 = _mm_sub_pd(iz2,jz2);
572 /* Calculate squared distance and things based on it */
573 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
574 rsq01 = gmx_mm_calc_rsq_pd(dx01,dy01,dz01);
575 rsq02 = gmx_mm_calc_rsq_pd(dx02,dy02,dz02);
576 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
577 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
578 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
579 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
580 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
581 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
583 rinv00 = gmx_mm_invsqrt_pd(rsq00);
584 rinv01 = gmx_mm_invsqrt_pd(rsq01);
585 rinv02 = gmx_mm_invsqrt_pd(rsq02);
586 rinv10 = gmx_mm_invsqrt_pd(rsq10);
587 rinv11 = gmx_mm_invsqrt_pd(rsq11);
588 rinv12 = gmx_mm_invsqrt_pd(rsq12);
589 rinv20 = gmx_mm_invsqrt_pd(rsq20);
590 rinv21 = gmx_mm_invsqrt_pd(rsq21);
591 rinv22 = gmx_mm_invsqrt_pd(rsq22);
593 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
594 rinvsq01 = _mm_mul_pd(rinv01,rinv01);
595 rinvsq02 = _mm_mul_pd(rinv02,rinv02);
596 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
597 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
598 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
599 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
600 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
601 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
603 fjx0 = _mm_setzero_pd();
604 fjy0 = _mm_setzero_pd();
605 fjz0 = _mm_setzero_pd();
606 fjx1 = _mm_setzero_pd();
607 fjy1 = _mm_setzero_pd();
608 fjz1 = _mm_setzero_pd();
609 fjx2 = _mm_setzero_pd();
610 fjy2 = _mm_setzero_pd();
611 fjz2 = _mm_setzero_pd();
613 /**************************
614 * CALCULATE INTERACTIONS *
615 **************************/
617 r00 = _mm_mul_pd(rsq00,rinv00);
619 /* Calculate table index by multiplying r with table scale and truncate to integer */
620 rt = _mm_mul_pd(r00,vftabscale);
621 vfitab = _mm_cvttpd_epi32(rt);
623 vfeps = _mm_frcz_pd(rt);
625 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
627 twovfeps = _mm_add_pd(vfeps,vfeps);
628 vfitab = _mm_slli_epi32(vfitab,3);
630 /* REACTION-FIELD ELECTROSTATICS */
631 velec = _mm_mul_pd(qq00,_mm_sub_pd(_mm_macc_pd(krf,rsq00,rinv00),crf));
632 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
634 /* CUBIC SPLINE TABLE DISPERSION */
635 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
636 F = _mm_setzero_pd();
637 GMX_MM_TRANSPOSE2_PD(Y,F);
638 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
639 H = _mm_setzero_pd();
640 GMX_MM_TRANSPOSE2_PD(G,H);
641 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
642 VV = _mm_macc_pd(vfeps,Fp,Y);
643 vvdw6 = _mm_mul_pd(c6_00,VV);
644 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
645 fvdw6 = _mm_mul_pd(c6_00,FF);
647 /* CUBIC SPLINE TABLE REPULSION */
648 vfitab = _mm_add_epi32(vfitab,ifour);
649 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
650 F = _mm_setzero_pd();
651 GMX_MM_TRANSPOSE2_PD(Y,F);
652 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
653 H = _mm_setzero_pd();
654 GMX_MM_TRANSPOSE2_PD(G,H);
655 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
656 VV = _mm_macc_pd(vfeps,Fp,Y);
657 vvdw12 = _mm_mul_pd(c12_00,VV);
658 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
659 fvdw12 = _mm_mul_pd(c12_00,FF);
660 vvdw = _mm_add_pd(vvdw12,vvdw6);
661 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
663 /* Update potential sum for this i atom from the interaction with this j atom. */
664 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
665 velecsum = _mm_add_pd(velecsum,velec);
666 vvdw = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
667 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
669 fscal = _mm_add_pd(felec,fvdw);
671 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
673 /* Update vectorial force */
674 fix0 = _mm_macc_pd(dx00,fscal,fix0);
675 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
676 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
678 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
679 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
680 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
682 /**************************
683 * CALCULATE INTERACTIONS *
684 **************************/
686 /* REACTION-FIELD ELECTROSTATICS */
687 velec = _mm_mul_pd(qq01,_mm_sub_pd(_mm_macc_pd(krf,rsq01,rinv01),crf));
688 felec = _mm_mul_pd(qq01,_mm_msub_pd(rinv01,rinvsq01,krf2));
690 /* Update potential sum for this i atom from the interaction with this j atom. */
691 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
692 velecsum = _mm_add_pd(velecsum,velec);
696 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
698 /* Update vectorial force */
699 fix0 = _mm_macc_pd(dx01,fscal,fix0);
700 fiy0 = _mm_macc_pd(dy01,fscal,fiy0);
701 fiz0 = _mm_macc_pd(dz01,fscal,fiz0);
703 fjx1 = _mm_macc_pd(dx01,fscal,fjx1);
704 fjy1 = _mm_macc_pd(dy01,fscal,fjy1);
705 fjz1 = _mm_macc_pd(dz01,fscal,fjz1);
707 /**************************
708 * CALCULATE INTERACTIONS *
709 **************************/
711 /* REACTION-FIELD ELECTROSTATICS */
712 velec = _mm_mul_pd(qq02,_mm_sub_pd(_mm_macc_pd(krf,rsq02,rinv02),crf));
713 felec = _mm_mul_pd(qq02,_mm_msub_pd(rinv02,rinvsq02,krf2));
715 /* Update potential sum for this i atom from the interaction with this j atom. */
716 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
717 velecsum = _mm_add_pd(velecsum,velec);
721 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
723 /* Update vectorial force */
724 fix0 = _mm_macc_pd(dx02,fscal,fix0);
725 fiy0 = _mm_macc_pd(dy02,fscal,fiy0);
726 fiz0 = _mm_macc_pd(dz02,fscal,fiz0);
728 fjx2 = _mm_macc_pd(dx02,fscal,fjx2);
729 fjy2 = _mm_macc_pd(dy02,fscal,fjy2);
730 fjz2 = _mm_macc_pd(dz02,fscal,fjz2);
732 /**************************
733 * CALCULATE INTERACTIONS *
734 **************************/
736 /* REACTION-FIELD ELECTROSTATICS */
737 velec = _mm_mul_pd(qq10,_mm_sub_pd(_mm_macc_pd(krf,rsq10,rinv10),crf));
738 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
740 /* Update potential sum for this i atom from the interaction with this j atom. */
741 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
742 velecsum = _mm_add_pd(velecsum,velec);
746 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
748 /* Update vectorial force */
749 fix1 = _mm_macc_pd(dx10,fscal,fix1);
750 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
751 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
753 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
754 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
755 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
757 /**************************
758 * CALCULATE INTERACTIONS *
759 **************************/
761 /* REACTION-FIELD ELECTROSTATICS */
762 velec = _mm_mul_pd(qq11,_mm_sub_pd(_mm_macc_pd(krf,rsq11,rinv11),crf));
763 felec = _mm_mul_pd(qq11,_mm_msub_pd(rinv11,rinvsq11,krf2));
765 /* Update potential sum for this i atom from the interaction with this j atom. */
766 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
767 velecsum = _mm_add_pd(velecsum,velec);
771 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
773 /* Update vectorial force */
774 fix1 = _mm_macc_pd(dx11,fscal,fix1);
775 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
776 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
778 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
779 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
780 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
782 /**************************
783 * CALCULATE INTERACTIONS *
784 **************************/
786 /* REACTION-FIELD ELECTROSTATICS */
787 velec = _mm_mul_pd(qq12,_mm_sub_pd(_mm_macc_pd(krf,rsq12,rinv12),crf));
788 felec = _mm_mul_pd(qq12,_mm_msub_pd(rinv12,rinvsq12,krf2));
790 /* Update potential sum for this i atom from the interaction with this j atom. */
791 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
792 velecsum = _mm_add_pd(velecsum,velec);
796 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
798 /* Update vectorial force */
799 fix1 = _mm_macc_pd(dx12,fscal,fix1);
800 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
801 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
803 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
804 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
805 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
807 /**************************
808 * CALCULATE INTERACTIONS *
809 **************************/
811 /* REACTION-FIELD ELECTROSTATICS */
812 velec = _mm_mul_pd(qq20,_mm_sub_pd(_mm_macc_pd(krf,rsq20,rinv20),crf));
813 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
815 /* Update potential sum for this i atom from the interaction with this j atom. */
816 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
817 velecsum = _mm_add_pd(velecsum,velec);
821 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
823 /* Update vectorial force */
824 fix2 = _mm_macc_pd(dx20,fscal,fix2);
825 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
826 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
828 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
829 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
830 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
832 /**************************
833 * CALCULATE INTERACTIONS *
834 **************************/
836 /* REACTION-FIELD ELECTROSTATICS */
837 velec = _mm_mul_pd(qq21,_mm_sub_pd(_mm_macc_pd(krf,rsq21,rinv21),crf));
838 felec = _mm_mul_pd(qq21,_mm_msub_pd(rinv21,rinvsq21,krf2));
840 /* Update potential sum for this i atom from the interaction with this j atom. */
841 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
842 velecsum = _mm_add_pd(velecsum,velec);
846 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
848 /* Update vectorial force */
849 fix2 = _mm_macc_pd(dx21,fscal,fix2);
850 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
851 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
853 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
854 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
855 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
857 /**************************
858 * CALCULATE INTERACTIONS *
859 **************************/
861 /* REACTION-FIELD ELECTROSTATICS */
862 velec = _mm_mul_pd(qq22,_mm_sub_pd(_mm_macc_pd(krf,rsq22,rinv22),crf));
863 felec = _mm_mul_pd(qq22,_mm_msub_pd(rinv22,rinvsq22,krf2));
865 /* Update potential sum for this i atom from the interaction with this j atom. */
866 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
867 velecsum = _mm_add_pd(velecsum,velec);
871 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
873 /* Update vectorial force */
874 fix2 = _mm_macc_pd(dx22,fscal,fix2);
875 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
876 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
878 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
879 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
880 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
882 gmx_mm_decrement_3rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
884 /* Inner loop uses 350 flops */
887 /* End of innermost loop */
889 gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
890 f+i_coord_offset,fshift+i_shift_offset);
893 /* Update potential energies */
894 gmx_mm_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
895 gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
897 /* Increment number of inner iterations */
898 inneriter += j_index_end - j_index_start;
900 /* Outer loop uses 20 flops */
903 /* Increment number of outer iterations */
906 /* Update outer/inner flops */
908 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*350);
911 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwCSTab_GeomW3W3_F_avx_128_fma_double
912 * Electrostatics interaction: ReactionField
913 * VdW interaction: CubicSplineTable
914 * Geometry: Water3-Water3
915 * Calculate force/pot: Force
918 nb_kernel_ElecRF_VdwCSTab_GeomW3W3_F_avx_128_fma_double
919 (t_nblist * gmx_restrict nlist,
920 rvec * gmx_restrict xx,
921 rvec * gmx_restrict ff,
922 t_forcerec * gmx_restrict fr,
923 t_mdatoms * gmx_restrict mdatoms,
924 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
925 t_nrnb * gmx_restrict nrnb)
927 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
928 * just 0 for non-waters.
929 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
930 * jnr indices corresponding to data put in the four positions in the SIMD register.
932 int i_shift_offset,i_coord_offset,outeriter,inneriter;
933 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
935 int j_coord_offsetA,j_coord_offsetB;
936 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
938 real *shiftvec,*fshift,*x,*f;
939 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
941 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
943 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
945 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
946 int vdwjidx0A,vdwjidx0B;
947 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
948 int vdwjidx1A,vdwjidx1B;
949 __m128d jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
950 int vdwjidx2A,vdwjidx2B;
951 __m128d jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
952 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
953 __m128d dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
954 __m128d dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
955 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
956 __m128d dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
957 __m128d dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
958 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
959 __m128d dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
960 __m128d dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
961 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
964 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
967 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
968 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
970 __m128i ifour = _mm_set1_epi32(4);
971 __m128d rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
973 __m128d dummy_mask,cutoff_mask;
974 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
975 __m128d one = _mm_set1_pd(1.0);
976 __m128d two = _mm_set1_pd(2.0);
982 jindex = nlist->jindex;
984 shiftidx = nlist->shift;
986 shiftvec = fr->shift_vec[0];
987 fshift = fr->fshift[0];
988 facel = _mm_set1_pd(fr->epsfac);
989 charge = mdatoms->chargeA;
990 krf = _mm_set1_pd(fr->ic->k_rf);
991 krf2 = _mm_set1_pd(fr->ic->k_rf*2.0);
992 crf = _mm_set1_pd(fr->ic->c_rf);
993 nvdwtype = fr->ntype;
995 vdwtype = mdatoms->typeA;
997 vftab = kernel_data->table_vdw->data;
998 vftabscale = _mm_set1_pd(kernel_data->table_vdw->scale);
1000 /* Setup water-specific parameters */
1001 inr = nlist->iinr[0];
1002 iq0 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
1003 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
1004 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
1005 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
1007 jq0 = _mm_set1_pd(charge[inr+0]);
1008 jq1 = _mm_set1_pd(charge[inr+1]);
1009 jq2 = _mm_set1_pd(charge[inr+2]);
1010 vdwjidx0A = 2*vdwtype[inr+0];
1011 qq00 = _mm_mul_pd(iq0,jq0);
1012 c6_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A]);
1013 c12_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A+1]);
1014 qq01 = _mm_mul_pd(iq0,jq1);
1015 qq02 = _mm_mul_pd(iq0,jq2);
1016 qq10 = _mm_mul_pd(iq1,jq0);
1017 qq11 = _mm_mul_pd(iq1,jq1);
1018 qq12 = _mm_mul_pd(iq1,jq2);
1019 qq20 = _mm_mul_pd(iq2,jq0);
1020 qq21 = _mm_mul_pd(iq2,jq1);
1021 qq22 = _mm_mul_pd(iq2,jq2);
1023 /* Avoid stupid compiler warnings */
1025 j_coord_offsetA = 0;
1026 j_coord_offsetB = 0;
1031 /* Start outer loop over neighborlists */
1032 for(iidx=0; iidx<nri; iidx++)
1034 /* Load shift vector for this list */
1035 i_shift_offset = DIM*shiftidx[iidx];
1037 /* Load limits for loop over neighbors */
1038 j_index_start = jindex[iidx];
1039 j_index_end = jindex[iidx+1];
1041 /* Get outer coordinate index */
1043 i_coord_offset = DIM*inr;
1045 /* Load i particle coords and add shift vector */
1046 gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
1047 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1049 fix0 = _mm_setzero_pd();
1050 fiy0 = _mm_setzero_pd();
1051 fiz0 = _mm_setzero_pd();
1052 fix1 = _mm_setzero_pd();
1053 fiy1 = _mm_setzero_pd();
1054 fiz1 = _mm_setzero_pd();
1055 fix2 = _mm_setzero_pd();
1056 fiy2 = _mm_setzero_pd();
1057 fiz2 = _mm_setzero_pd();
1059 /* Start inner kernel loop */
1060 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1063 /* Get j neighbor index, and coordinate index */
1065 jnrB = jjnr[jidx+1];
1066 j_coord_offsetA = DIM*jnrA;
1067 j_coord_offsetB = DIM*jnrB;
1069 /* load j atom coordinates */
1070 gmx_mm_load_3rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
1071 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1073 /* Calculate displacement vector */
1074 dx00 = _mm_sub_pd(ix0,jx0);
1075 dy00 = _mm_sub_pd(iy0,jy0);
1076 dz00 = _mm_sub_pd(iz0,jz0);
1077 dx01 = _mm_sub_pd(ix0,jx1);
1078 dy01 = _mm_sub_pd(iy0,jy1);
1079 dz01 = _mm_sub_pd(iz0,jz1);
1080 dx02 = _mm_sub_pd(ix0,jx2);
1081 dy02 = _mm_sub_pd(iy0,jy2);
1082 dz02 = _mm_sub_pd(iz0,jz2);
1083 dx10 = _mm_sub_pd(ix1,jx0);
1084 dy10 = _mm_sub_pd(iy1,jy0);
1085 dz10 = _mm_sub_pd(iz1,jz0);
1086 dx11 = _mm_sub_pd(ix1,jx1);
1087 dy11 = _mm_sub_pd(iy1,jy1);
1088 dz11 = _mm_sub_pd(iz1,jz1);
1089 dx12 = _mm_sub_pd(ix1,jx2);
1090 dy12 = _mm_sub_pd(iy1,jy2);
1091 dz12 = _mm_sub_pd(iz1,jz2);
1092 dx20 = _mm_sub_pd(ix2,jx0);
1093 dy20 = _mm_sub_pd(iy2,jy0);
1094 dz20 = _mm_sub_pd(iz2,jz0);
1095 dx21 = _mm_sub_pd(ix2,jx1);
1096 dy21 = _mm_sub_pd(iy2,jy1);
1097 dz21 = _mm_sub_pd(iz2,jz1);
1098 dx22 = _mm_sub_pd(ix2,jx2);
1099 dy22 = _mm_sub_pd(iy2,jy2);
1100 dz22 = _mm_sub_pd(iz2,jz2);
1102 /* Calculate squared distance and things based on it */
1103 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
1104 rsq01 = gmx_mm_calc_rsq_pd(dx01,dy01,dz01);
1105 rsq02 = gmx_mm_calc_rsq_pd(dx02,dy02,dz02);
1106 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
1107 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
1108 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
1109 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
1110 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
1111 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
1113 rinv00 = gmx_mm_invsqrt_pd(rsq00);
1114 rinv01 = gmx_mm_invsqrt_pd(rsq01);
1115 rinv02 = gmx_mm_invsqrt_pd(rsq02);
1116 rinv10 = gmx_mm_invsqrt_pd(rsq10);
1117 rinv11 = gmx_mm_invsqrt_pd(rsq11);
1118 rinv12 = gmx_mm_invsqrt_pd(rsq12);
1119 rinv20 = gmx_mm_invsqrt_pd(rsq20);
1120 rinv21 = gmx_mm_invsqrt_pd(rsq21);
1121 rinv22 = gmx_mm_invsqrt_pd(rsq22);
1123 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
1124 rinvsq01 = _mm_mul_pd(rinv01,rinv01);
1125 rinvsq02 = _mm_mul_pd(rinv02,rinv02);
1126 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
1127 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
1128 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
1129 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
1130 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
1131 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
1133 fjx0 = _mm_setzero_pd();
1134 fjy0 = _mm_setzero_pd();
1135 fjz0 = _mm_setzero_pd();
1136 fjx1 = _mm_setzero_pd();
1137 fjy1 = _mm_setzero_pd();
1138 fjz1 = _mm_setzero_pd();
1139 fjx2 = _mm_setzero_pd();
1140 fjy2 = _mm_setzero_pd();
1141 fjz2 = _mm_setzero_pd();
1143 /**************************
1144 * CALCULATE INTERACTIONS *
1145 **************************/
1147 r00 = _mm_mul_pd(rsq00,rinv00);
1149 /* Calculate table index by multiplying r with table scale and truncate to integer */
1150 rt = _mm_mul_pd(r00,vftabscale);
1151 vfitab = _mm_cvttpd_epi32(rt);
1153 vfeps = _mm_frcz_pd(rt);
1155 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
1157 twovfeps = _mm_add_pd(vfeps,vfeps);
1158 vfitab = _mm_slli_epi32(vfitab,3);
1160 /* REACTION-FIELD ELECTROSTATICS */
1161 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
1163 /* CUBIC SPLINE TABLE DISPERSION */
1164 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
1165 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
1166 GMX_MM_TRANSPOSE2_PD(Y,F);
1167 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
1168 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
1169 GMX_MM_TRANSPOSE2_PD(G,H);
1170 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
1171 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
1172 fvdw6 = _mm_mul_pd(c6_00,FF);
1174 /* CUBIC SPLINE TABLE REPULSION */
1175 vfitab = _mm_add_epi32(vfitab,ifour);
1176 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
1177 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
1178 GMX_MM_TRANSPOSE2_PD(Y,F);
1179 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
1180 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
1181 GMX_MM_TRANSPOSE2_PD(G,H);
1182 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
1183 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
1184 fvdw12 = _mm_mul_pd(c12_00,FF);
1185 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
1187 fscal = _mm_add_pd(felec,fvdw);
1189 /* Update vectorial force */
1190 fix0 = _mm_macc_pd(dx00,fscal,fix0);
1191 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
1192 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
1194 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
1195 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
1196 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
1198 /**************************
1199 * CALCULATE INTERACTIONS *
1200 **************************/
1202 /* REACTION-FIELD ELECTROSTATICS */
1203 felec = _mm_mul_pd(qq01,_mm_msub_pd(rinv01,rinvsq01,krf2));
1207 /* Update vectorial force */
1208 fix0 = _mm_macc_pd(dx01,fscal,fix0);
1209 fiy0 = _mm_macc_pd(dy01,fscal,fiy0);
1210 fiz0 = _mm_macc_pd(dz01,fscal,fiz0);
1212 fjx1 = _mm_macc_pd(dx01,fscal,fjx1);
1213 fjy1 = _mm_macc_pd(dy01,fscal,fjy1);
1214 fjz1 = _mm_macc_pd(dz01,fscal,fjz1);
1216 /**************************
1217 * CALCULATE INTERACTIONS *
1218 **************************/
1220 /* REACTION-FIELD ELECTROSTATICS */
1221 felec = _mm_mul_pd(qq02,_mm_msub_pd(rinv02,rinvsq02,krf2));
1225 /* Update vectorial force */
1226 fix0 = _mm_macc_pd(dx02,fscal,fix0);
1227 fiy0 = _mm_macc_pd(dy02,fscal,fiy0);
1228 fiz0 = _mm_macc_pd(dz02,fscal,fiz0);
1230 fjx2 = _mm_macc_pd(dx02,fscal,fjx2);
1231 fjy2 = _mm_macc_pd(dy02,fscal,fjy2);
1232 fjz2 = _mm_macc_pd(dz02,fscal,fjz2);
1234 /**************************
1235 * CALCULATE INTERACTIONS *
1236 **************************/
1238 /* REACTION-FIELD ELECTROSTATICS */
1239 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
1243 /* Update vectorial force */
1244 fix1 = _mm_macc_pd(dx10,fscal,fix1);
1245 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
1246 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
1248 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
1249 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
1250 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
1252 /**************************
1253 * CALCULATE INTERACTIONS *
1254 **************************/
1256 /* REACTION-FIELD ELECTROSTATICS */
1257 felec = _mm_mul_pd(qq11,_mm_msub_pd(rinv11,rinvsq11,krf2));
1261 /* Update vectorial force */
1262 fix1 = _mm_macc_pd(dx11,fscal,fix1);
1263 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
1264 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
1266 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
1267 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
1268 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
1270 /**************************
1271 * CALCULATE INTERACTIONS *
1272 **************************/
1274 /* REACTION-FIELD ELECTROSTATICS */
1275 felec = _mm_mul_pd(qq12,_mm_msub_pd(rinv12,rinvsq12,krf2));
1279 /* Update vectorial force */
1280 fix1 = _mm_macc_pd(dx12,fscal,fix1);
1281 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
1282 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
1284 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
1285 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
1286 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
1288 /**************************
1289 * CALCULATE INTERACTIONS *
1290 **************************/
1292 /* REACTION-FIELD ELECTROSTATICS */
1293 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
1297 /* Update vectorial force */
1298 fix2 = _mm_macc_pd(dx20,fscal,fix2);
1299 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
1300 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
1302 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
1303 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
1304 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
1306 /**************************
1307 * CALCULATE INTERACTIONS *
1308 **************************/
1310 /* REACTION-FIELD ELECTROSTATICS */
1311 felec = _mm_mul_pd(qq21,_mm_msub_pd(rinv21,rinvsq21,krf2));
1315 /* Update vectorial force */
1316 fix2 = _mm_macc_pd(dx21,fscal,fix2);
1317 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
1318 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
1320 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
1321 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
1322 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
1324 /**************************
1325 * CALCULATE INTERACTIONS *
1326 **************************/
1328 /* REACTION-FIELD ELECTROSTATICS */
1329 felec = _mm_mul_pd(qq22,_mm_msub_pd(rinv22,rinvsq22,krf2));
1333 /* Update vectorial force */
1334 fix2 = _mm_macc_pd(dx22,fscal,fix2);
1335 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
1336 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
1338 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
1339 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
1340 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
1342 gmx_mm_decrement_3rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1344 /* Inner loop uses 297 flops */
1347 if(jidx<j_index_end)
1351 j_coord_offsetA = DIM*jnrA;
1353 /* load j atom coordinates */
1354 gmx_mm_load_3rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
1355 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1357 /* Calculate displacement vector */
1358 dx00 = _mm_sub_pd(ix0,jx0);
1359 dy00 = _mm_sub_pd(iy0,jy0);
1360 dz00 = _mm_sub_pd(iz0,jz0);
1361 dx01 = _mm_sub_pd(ix0,jx1);
1362 dy01 = _mm_sub_pd(iy0,jy1);
1363 dz01 = _mm_sub_pd(iz0,jz1);
1364 dx02 = _mm_sub_pd(ix0,jx2);
1365 dy02 = _mm_sub_pd(iy0,jy2);
1366 dz02 = _mm_sub_pd(iz0,jz2);
1367 dx10 = _mm_sub_pd(ix1,jx0);
1368 dy10 = _mm_sub_pd(iy1,jy0);
1369 dz10 = _mm_sub_pd(iz1,jz0);
1370 dx11 = _mm_sub_pd(ix1,jx1);
1371 dy11 = _mm_sub_pd(iy1,jy1);
1372 dz11 = _mm_sub_pd(iz1,jz1);
1373 dx12 = _mm_sub_pd(ix1,jx2);
1374 dy12 = _mm_sub_pd(iy1,jy2);
1375 dz12 = _mm_sub_pd(iz1,jz2);
1376 dx20 = _mm_sub_pd(ix2,jx0);
1377 dy20 = _mm_sub_pd(iy2,jy0);
1378 dz20 = _mm_sub_pd(iz2,jz0);
1379 dx21 = _mm_sub_pd(ix2,jx1);
1380 dy21 = _mm_sub_pd(iy2,jy1);
1381 dz21 = _mm_sub_pd(iz2,jz1);
1382 dx22 = _mm_sub_pd(ix2,jx2);
1383 dy22 = _mm_sub_pd(iy2,jy2);
1384 dz22 = _mm_sub_pd(iz2,jz2);
1386 /* Calculate squared distance and things based on it */
1387 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
1388 rsq01 = gmx_mm_calc_rsq_pd(dx01,dy01,dz01);
1389 rsq02 = gmx_mm_calc_rsq_pd(dx02,dy02,dz02);
1390 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
1391 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
1392 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
1393 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
1394 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
1395 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
1397 rinv00 = gmx_mm_invsqrt_pd(rsq00);
1398 rinv01 = gmx_mm_invsqrt_pd(rsq01);
1399 rinv02 = gmx_mm_invsqrt_pd(rsq02);
1400 rinv10 = gmx_mm_invsqrt_pd(rsq10);
1401 rinv11 = gmx_mm_invsqrt_pd(rsq11);
1402 rinv12 = gmx_mm_invsqrt_pd(rsq12);
1403 rinv20 = gmx_mm_invsqrt_pd(rsq20);
1404 rinv21 = gmx_mm_invsqrt_pd(rsq21);
1405 rinv22 = gmx_mm_invsqrt_pd(rsq22);
1407 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
1408 rinvsq01 = _mm_mul_pd(rinv01,rinv01);
1409 rinvsq02 = _mm_mul_pd(rinv02,rinv02);
1410 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
1411 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
1412 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
1413 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
1414 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
1415 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
1417 fjx0 = _mm_setzero_pd();
1418 fjy0 = _mm_setzero_pd();
1419 fjz0 = _mm_setzero_pd();
1420 fjx1 = _mm_setzero_pd();
1421 fjy1 = _mm_setzero_pd();
1422 fjz1 = _mm_setzero_pd();
1423 fjx2 = _mm_setzero_pd();
1424 fjy2 = _mm_setzero_pd();
1425 fjz2 = _mm_setzero_pd();
1427 /**************************
1428 * CALCULATE INTERACTIONS *
1429 **************************/
1431 r00 = _mm_mul_pd(rsq00,rinv00);
1433 /* Calculate table index by multiplying r with table scale and truncate to integer */
1434 rt = _mm_mul_pd(r00,vftabscale);
1435 vfitab = _mm_cvttpd_epi32(rt);
1437 vfeps = _mm_frcz_pd(rt);
1439 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
1441 twovfeps = _mm_add_pd(vfeps,vfeps);
1442 vfitab = _mm_slli_epi32(vfitab,3);
1444 /* REACTION-FIELD ELECTROSTATICS */
1445 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
1447 /* CUBIC SPLINE TABLE DISPERSION */
1448 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
1449 F = _mm_setzero_pd();
1450 GMX_MM_TRANSPOSE2_PD(Y,F);
1451 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
1452 H = _mm_setzero_pd();
1453 GMX_MM_TRANSPOSE2_PD(G,H);
1454 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
1455 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
1456 fvdw6 = _mm_mul_pd(c6_00,FF);
1458 /* CUBIC SPLINE TABLE REPULSION */
1459 vfitab = _mm_add_epi32(vfitab,ifour);
1460 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
1461 F = _mm_setzero_pd();
1462 GMX_MM_TRANSPOSE2_PD(Y,F);
1463 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
1464 H = _mm_setzero_pd();
1465 GMX_MM_TRANSPOSE2_PD(G,H);
1466 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
1467 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
1468 fvdw12 = _mm_mul_pd(c12_00,FF);
1469 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
1471 fscal = _mm_add_pd(felec,fvdw);
1473 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1475 /* Update vectorial force */
1476 fix0 = _mm_macc_pd(dx00,fscal,fix0);
1477 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
1478 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
1480 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
1481 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
1482 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
1484 /**************************
1485 * CALCULATE INTERACTIONS *
1486 **************************/
1488 /* REACTION-FIELD ELECTROSTATICS */
1489 felec = _mm_mul_pd(qq01,_mm_msub_pd(rinv01,rinvsq01,krf2));
1493 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1495 /* Update vectorial force */
1496 fix0 = _mm_macc_pd(dx01,fscal,fix0);
1497 fiy0 = _mm_macc_pd(dy01,fscal,fiy0);
1498 fiz0 = _mm_macc_pd(dz01,fscal,fiz0);
1500 fjx1 = _mm_macc_pd(dx01,fscal,fjx1);
1501 fjy1 = _mm_macc_pd(dy01,fscal,fjy1);
1502 fjz1 = _mm_macc_pd(dz01,fscal,fjz1);
1504 /**************************
1505 * CALCULATE INTERACTIONS *
1506 **************************/
1508 /* REACTION-FIELD ELECTROSTATICS */
1509 felec = _mm_mul_pd(qq02,_mm_msub_pd(rinv02,rinvsq02,krf2));
1513 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1515 /* Update vectorial force */
1516 fix0 = _mm_macc_pd(dx02,fscal,fix0);
1517 fiy0 = _mm_macc_pd(dy02,fscal,fiy0);
1518 fiz0 = _mm_macc_pd(dz02,fscal,fiz0);
1520 fjx2 = _mm_macc_pd(dx02,fscal,fjx2);
1521 fjy2 = _mm_macc_pd(dy02,fscal,fjy2);
1522 fjz2 = _mm_macc_pd(dz02,fscal,fjz2);
1524 /**************************
1525 * CALCULATE INTERACTIONS *
1526 **************************/
1528 /* REACTION-FIELD ELECTROSTATICS */
1529 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
1533 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1535 /* Update vectorial force */
1536 fix1 = _mm_macc_pd(dx10,fscal,fix1);
1537 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
1538 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
1540 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
1541 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
1542 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
1544 /**************************
1545 * CALCULATE INTERACTIONS *
1546 **************************/
1548 /* REACTION-FIELD ELECTROSTATICS */
1549 felec = _mm_mul_pd(qq11,_mm_msub_pd(rinv11,rinvsq11,krf2));
1553 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1555 /* Update vectorial force */
1556 fix1 = _mm_macc_pd(dx11,fscal,fix1);
1557 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
1558 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
1560 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
1561 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
1562 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
1564 /**************************
1565 * CALCULATE INTERACTIONS *
1566 **************************/
1568 /* REACTION-FIELD ELECTROSTATICS */
1569 felec = _mm_mul_pd(qq12,_mm_msub_pd(rinv12,rinvsq12,krf2));
1573 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1575 /* Update vectorial force */
1576 fix1 = _mm_macc_pd(dx12,fscal,fix1);
1577 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
1578 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
1580 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
1581 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
1582 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
1584 /**************************
1585 * CALCULATE INTERACTIONS *
1586 **************************/
1588 /* REACTION-FIELD ELECTROSTATICS */
1589 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
1593 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1595 /* Update vectorial force */
1596 fix2 = _mm_macc_pd(dx20,fscal,fix2);
1597 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
1598 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
1600 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
1601 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
1602 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
1604 /**************************
1605 * CALCULATE INTERACTIONS *
1606 **************************/
1608 /* REACTION-FIELD ELECTROSTATICS */
1609 felec = _mm_mul_pd(qq21,_mm_msub_pd(rinv21,rinvsq21,krf2));
1613 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1615 /* Update vectorial force */
1616 fix2 = _mm_macc_pd(dx21,fscal,fix2);
1617 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
1618 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
1620 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
1621 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
1622 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
1624 /**************************
1625 * CALCULATE INTERACTIONS *
1626 **************************/
1628 /* REACTION-FIELD ELECTROSTATICS */
1629 felec = _mm_mul_pd(qq22,_mm_msub_pd(rinv22,rinvsq22,krf2));
1633 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1635 /* Update vectorial force */
1636 fix2 = _mm_macc_pd(dx22,fscal,fix2);
1637 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
1638 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
1640 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
1641 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
1642 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
1644 gmx_mm_decrement_3rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1646 /* Inner loop uses 297 flops */
1649 /* End of innermost loop */
1651 gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1652 f+i_coord_offset,fshift+i_shift_offset);
1654 /* Increment number of inner iterations */
1655 inneriter += j_index_end - j_index_start;
1657 /* Outer loop uses 18 flops */
1660 /* Increment number of outer iterations */
1663 /* Update outer/inner flops */
1665 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*297);