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 sse4_1_single kernel generator.
42 #include "../nb_kernel.h"
43 #include "gromacs/legacyheaders/types/simple.h"
44 #include "gromacs/math/vec.h"
45 #include "gromacs/legacyheaders/nrnb.h"
47 #include "gromacs/simd/math_x86_sse4_1_single.h"
48 #include "kernelutil_x86_sse4_1_single.h"
51 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_VF_sse4_1_single
52 * Electrostatics interaction: Coulomb
53 * VdW interaction: CubicSplineTable
54 * Geometry: Water3-Water3
55 * Calculate force/pot: PotentialAndForce
58 nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_VF_sse4_1_single
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,C,D refer to j loop unrolling done with SSE, e.g. for the four 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;
74 int jnrA,jnrB,jnrC,jnrD;
75 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
76 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
77 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
79 real *shiftvec,*fshift,*x,*f;
80 real *fjptrA,*fjptrB,*fjptrC,*fjptrD;
82 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
84 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
86 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
88 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
89 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
90 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
91 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
92 __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
93 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
94 __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
95 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
96 __m128 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
97 __m128 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
98 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
99 __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
100 __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
101 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
102 __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
103 __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
104 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
107 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
110 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
111 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
113 __m128i ifour = _mm_set1_epi32(4);
114 __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
116 __m128 dummy_mask,cutoff_mask;
117 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
118 __m128 one = _mm_set1_ps(1.0);
119 __m128 two = _mm_set1_ps(2.0);
125 jindex = nlist->jindex;
127 shiftidx = nlist->shift;
129 shiftvec = fr->shift_vec[0];
130 fshift = fr->fshift[0];
131 facel = _mm_set1_ps(fr->epsfac);
132 charge = mdatoms->chargeA;
133 nvdwtype = fr->ntype;
135 vdwtype = mdatoms->typeA;
137 vftab = kernel_data->table_vdw->data;
138 vftabscale = _mm_set1_ps(kernel_data->table_vdw->scale);
140 /* Setup water-specific parameters */
141 inr = nlist->iinr[0];
142 iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
143 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
144 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
145 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
147 jq0 = _mm_set1_ps(charge[inr+0]);
148 jq1 = _mm_set1_ps(charge[inr+1]);
149 jq2 = _mm_set1_ps(charge[inr+2]);
150 vdwjidx0A = 2*vdwtype[inr+0];
151 qq00 = _mm_mul_ps(iq0,jq0);
152 c6_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
153 c12_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
154 qq01 = _mm_mul_ps(iq0,jq1);
155 qq02 = _mm_mul_ps(iq0,jq2);
156 qq10 = _mm_mul_ps(iq1,jq0);
157 qq11 = _mm_mul_ps(iq1,jq1);
158 qq12 = _mm_mul_ps(iq1,jq2);
159 qq20 = _mm_mul_ps(iq2,jq0);
160 qq21 = _mm_mul_ps(iq2,jq1);
161 qq22 = _mm_mul_ps(iq2,jq2);
163 /* Avoid stupid compiler warnings */
164 jnrA = jnrB = jnrC = jnrD = 0;
173 for(iidx=0;iidx<4*DIM;iidx++)
178 /* Start outer loop over neighborlists */
179 for(iidx=0; iidx<nri; iidx++)
181 /* Load shift vector for this list */
182 i_shift_offset = DIM*shiftidx[iidx];
184 /* Load limits for loop over neighbors */
185 j_index_start = jindex[iidx];
186 j_index_end = jindex[iidx+1];
188 /* Get outer coordinate index */
190 i_coord_offset = DIM*inr;
192 /* Load i particle coords and add shift vector */
193 gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
194 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
196 fix0 = _mm_setzero_ps();
197 fiy0 = _mm_setzero_ps();
198 fiz0 = _mm_setzero_ps();
199 fix1 = _mm_setzero_ps();
200 fiy1 = _mm_setzero_ps();
201 fiz1 = _mm_setzero_ps();
202 fix2 = _mm_setzero_ps();
203 fiy2 = _mm_setzero_ps();
204 fiz2 = _mm_setzero_ps();
206 /* Reset potential sums */
207 velecsum = _mm_setzero_ps();
208 vvdwsum = _mm_setzero_ps();
210 /* Start inner kernel loop */
211 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
214 /* Get j neighbor index, and coordinate index */
219 j_coord_offsetA = DIM*jnrA;
220 j_coord_offsetB = DIM*jnrB;
221 j_coord_offsetC = DIM*jnrC;
222 j_coord_offsetD = DIM*jnrD;
224 /* load j atom coordinates */
225 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
226 x+j_coord_offsetC,x+j_coord_offsetD,
227 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
229 /* Calculate displacement vector */
230 dx00 = _mm_sub_ps(ix0,jx0);
231 dy00 = _mm_sub_ps(iy0,jy0);
232 dz00 = _mm_sub_ps(iz0,jz0);
233 dx01 = _mm_sub_ps(ix0,jx1);
234 dy01 = _mm_sub_ps(iy0,jy1);
235 dz01 = _mm_sub_ps(iz0,jz1);
236 dx02 = _mm_sub_ps(ix0,jx2);
237 dy02 = _mm_sub_ps(iy0,jy2);
238 dz02 = _mm_sub_ps(iz0,jz2);
239 dx10 = _mm_sub_ps(ix1,jx0);
240 dy10 = _mm_sub_ps(iy1,jy0);
241 dz10 = _mm_sub_ps(iz1,jz0);
242 dx11 = _mm_sub_ps(ix1,jx1);
243 dy11 = _mm_sub_ps(iy1,jy1);
244 dz11 = _mm_sub_ps(iz1,jz1);
245 dx12 = _mm_sub_ps(ix1,jx2);
246 dy12 = _mm_sub_ps(iy1,jy2);
247 dz12 = _mm_sub_ps(iz1,jz2);
248 dx20 = _mm_sub_ps(ix2,jx0);
249 dy20 = _mm_sub_ps(iy2,jy0);
250 dz20 = _mm_sub_ps(iz2,jz0);
251 dx21 = _mm_sub_ps(ix2,jx1);
252 dy21 = _mm_sub_ps(iy2,jy1);
253 dz21 = _mm_sub_ps(iz2,jz1);
254 dx22 = _mm_sub_ps(ix2,jx2);
255 dy22 = _mm_sub_ps(iy2,jy2);
256 dz22 = _mm_sub_ps(iz2,jz2);
258 /* Calculate squared distance and things based on it */
259 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
260 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
261 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
262 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
263 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
264 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
265 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
266 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
267 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
269 rinv00 = gmx_mm_invsqrt_ps(rsq00);
270 rinv01 = gmx_mm_invsqrt_ps(rsq01);
271 rinv02 = gmx_mm_invsqrt_ps(rsq02);
272 rinv10 = gmx_mm_invsqrt_ps(rsq10);
273 rinv11 = gmx_mm_invsqrt_ps(rsq11);
274 rinv12 = gmx_mm_invsqrt_ps(rsq12);
275 rinv20 = gmx_mm_invsqrt_ps(rsq20);
276 rinv21 = gmx_mm_invsqrt_ps(rsq21);
277 rinv22 = gmx_mm_invsqrt_ps(rsq22);
279 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
280 rinvsq01 = _mm_mul_ps(rinv01,rinv01);
281 rinvsq02 = _mm_mul_ps(rinv02,rinv02);
282 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
283 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
284 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
285 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
286 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
287 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
289 fjx0 = _mm_setzero_ps();
290 fjy0 = _mm_setzero_ps();
291 fjz0 = _mm_setzero_ps();
292 fjx1 = _mm_setzero_ps();
293 fjy1 = _mm_setzero_ps();
294 fjz1 = _mm_setzero_ps();
295 fjx2 = _mm_setzero_ps();
296 fjy2 = _mm_setzero_ps();
297 fjz2 = _mm_setzero_ps();
299 /**************************
300 * CALCULATE INTERACTIONS *
301 **************************/
303 r00 = _mm_mul_ps(rsq00,rinv00);
305 /* Calculate table index by multiplying r with table scale and truncate to integer */
306 rt = _mm_mul_ps(r00,vftabscale);
307 vfitab = _mm_cvttps_epi32(rt);
308 vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
309 vfitab = _mm_slli_epi32(vfitab,3);
311 /* COULOMB ELECTROSTATICS */
312 velec = _mm_mul_ps(qq00,rinv00);
313 felec = _mm_mul_ps(velec,rinvsq00);
315 /* CUBIC SPLINE TABLE DISPERSION */
316 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
317 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
318 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
319 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
320 _MM_TRANSPOSE4_PS(Y,F,G,H);
321 Heps = _mm_mul_ps(vfeps,H);
322 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
323 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
324 vvdw6 = _mm_mul_ps(c6_00,VV);
325 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
326 fvdw6 = _mm_mul_ps(c6_00,FF);
328 /* CUBIC SPLINE TABLE REPULSION */
329 vfitab = _mm_add_epi32(vfitab,ifour);
330 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
331 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
332 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
333 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
334 _MM_TRANSPOSE4_PS(Y,F,G,H);
335 Heps = _mm_mul_ps(vfeps,H);
336 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
337 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
338 vvdw12 = _mm_mul_ps(c12_00,VV);
339 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
340 fvdw12 = _mm_mul_ps(c12_00,FF);
341 vvdw = _mm_add_ps(vvdw12,vvdw6);
342 fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
344 /* Update potential sum for this i atom from the interaction with this j atom. */
345 velecsum = _mm_add_ps(velecsum,velec);
346 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
348 fscal = _mm_add_ps(felec,fvdw);
350 /* Calculate temporary vectorial force */
351 tx = _mm_mul_ps(fscal,dx00);
352 ty = _mm_mul_ps(fscal,dy00);
353 tz = _mm_mul_ps(fscal,dz00);
355 /* Update vectorial force */
356 fix0 = _mm_add_ps(fix0,tx);
357 fiy0 = _mm_add_ps(fiy0,ty);
358 fiz0 = _mm_add_ps(fiz0,tz);
360 fjx0 = _mm_add_ps(fjx0,tx);
361 fjy0 = _mm_add_ps(fjy0,ty);
362 fjz0 = _mm_add_ps(fjz0,tz);
364 /**************************
365 * CALCULATE INTERACTIONS *
366 **************************/
368 /* COULOMB ELECTROSTATICS */
369 velec = _mm_mul_ps(qq01,rinv01);
370 felec = _mm_mul_ps(velec,rinvsq01);
372 /* Update potential sum for this i atom from the interaction with this j atom. */
373 velecsum = _mm_add_ps(velecsum,velec);
377 /* Calculate temporary vectorial force */
378 tx = _mm_mul_ps(fscal,dx01);
379 ty = _mm_mul_ps(fscal,dy01);
380 tz = _mm_mul_ps(fscal,dz01);
382 /* Update vectorial force */
383 fix0 = _mm_add_ps(fix0,tx);
384 fiy0 = _mm_add_ps(fiy0,ty);
385 fiz0 = _mm_add_ps(fiz0,tz);
387 fjx1 = _mm_add_ps(fjx1,tx);
388 fjy1 = _mm_add_ps(fjy1,ty);
389 fjz1 = _mm_add_ps(fjz1,tz);
391 /**************************
392 * CALCULATE INTERACTIONS *
393 **************************/
395 /* COULOMB ELECTROSTATICS */
396 velec = _mm_mul_ps(qq02,rinv02);
397 felec = _mm_mul_ps(velec,rinvsq02);
399 /* Update potential sum for this i atom from the interaction with this j atom. */
400 velecsum = _mm_add_ps(velecsum,velec);
404 /* Calculate temporary vectorial force */
405 tx = _mm_mul_ps(fscal,dx02);
406 ty = _mm_mul_ps(fscal,dy02);
407 tz = _mm_mul_ps(fscal,dz02);
409 /* Update vectorial force */
410 fix0 = _mm_add_ps(fix0,tx);
411 fiy0 = _mm_add_ps(fiy0,ty);
412 fiz0 = _mm_add_ps(fiz0,tz);
414 fjx2 = _mm_add_ps(fjx2,tx);
415 fjy2 = _mm_add_ps(fjy2,ty);
416 fjz2 = _mm_add_ps(fjz2,tz);
418 /**************************
419 * CALCULATE INTERACTIONS *
420 **************************/
422 /* COULOMB ELECTROSTATICS */
423 velec = _mm_mul_ps(qq10,rinv10);
424 felec = _mm_mul_ps(velec,rinvsq10);
426 /* Update potential sum for this i atom from the interaction with this j atom. */
427 velecsum = _mm_add_ps(velecsum,velec);
431 /* Calculate temporary vectorial force */
432 tx = _mm_mul_ps(fscal,dx10);
433 ty = _mm_mul_ps(fscal,dy10);
434 tz = _mm_mul_ps(fscal,dz10);
436 /* Update vectorial force */
437 fix1 = _mm_add_ps(fix1,tx);
438 fiy1 = _mm_add_ps(fiy1,ty);
439 fiz1 = _mm_add_ps(fiz1,tz);
441 fjx0 = _mm_add_ps(fjx0,tx);
442 fjy0 = _mm_add_ps(fjy0,ty);
443 fjz0 = _mm_add_ps(fjz0,tz);
445 /**************************
446 * CALCULATE INTERACTIONS *
447 **************************/
449 /* COULOMB ELECTROSTATICS */
450 velec = _mm_mul_ps(qq11,rinv11);
451 felec = _mm_mul_ps(velec,rinvsq11);
453 /* Update potential sum for this i atom from the interaction with this j atom. */
454 velecsum = _mm_add_ps(velecsum,velec);
458 /* Calculate temporary vectorial force */
459 tx = _mm_mul_ps(fscal,dx11);
460 ty = _mm_mul_ps(fscal,dy11);
461 tz = _mm_mul_ps(fscal,dz11);
463 /* Update vectorial force */
464 fix1 = _mm_add_ps(fix1,tx);
465 fiy1 = _mm_add_ps(fiy1,ty);
466 fiz1 = _mm_add_ps(fiz1,tz);
468 fjx1 = _mm_add_ps(fjx1,tx);
469 fjy1 = _mm_add_ps(fjy1,ty);
470 fjz1 = _mm_add_ps(fjz1,tz);
472 /**************************
473 * CALCULATE INTERACTIONS *
474 **************************/
476 /* COULOMB ELECTROSTATICS */
477 velec = _mm_mul_ps(qq12,rinv12);
478 felec = _mm_mul_ps(velec,rinvsq12);
480 /* Update potential sum for this i atom from the interaction with this j atom. */
481 velecsum = _mm_add_ps(velecsum,velec);
485 /* Calculate temporary vectorial force */
486 tx = _mm_mul_ps(fscal,dx12);
487 ty = _mm_mul_ps(fscal,dy12);
488 tz = _mm_mul_ps(fscal,dz12);
490 /* Update vectorial force */
491 fix1 = _mm_add_ps(fix1,tx);
492 fiy1 = _mm_add_ps(fiy1,ty);
493 fiz1 = _mm_add_ps(fiz1,tz);
495 fjx2 = _mm_add_ps(fjx2,tx);
496 fjy2 = _mm_add_ps(fjy2,ty);
497 fjz2 = _mm_add_ps(fjz2,tz);
499 /**************************
500 * CALCULATE INTERACTIONS *
501 **************************/
503 /* COULOMB ELECTROSTATICS */
504 velec = _mm_mul_ps(qq20,rinv20);
505 felec = _mm_mul_ps(velec,rinvsq20);
507 /* Update potential sum for this i atom from the interaction with this j atom. */
508 velecsum = _mm_add_ps(velecsum,velec);
512 /* Calculate temporary vectorial force */
513 tx = _mm_mul_ps(fscal,dx20);
514 ty = _mm_mul_ps(fscal,dy20);
515 tz = _mm_mul_ps(fscal,dz20);
517 /* Update vectorial force */
518 fix2 = _mm_add_ps(fix2,tx);
519 fiy2 = _mm_add_ps(fiy2,ty);
520 fiz2 = _mm_add_ps(fiz2,tz);
522 fjx0 = _mm_add_ps(fjx0,tx);
523 fjy0 = _mm_add_ps(fjy0,ty);
524 fjz0 = _mm_add_ps(fjz0,tz);
526 /**************************
527 * CALCULATE INTERACTIONS *
528 **************************/
530 /* COULOMB ELECTROSTATICS */
531 velec = _mm_mul_ps(qq21,rinv21);
532 felec = _mm_mul_ps(velec,rinvsq21);
534 /* Update potential sum for this i atom from the interaction with this j atom. */
535 velecsum = _mm_add_ps(velecsum,velec);
539 /* Calculate temporary vectorial force */
540 tx = _mm_mul_ps(fscal,dx21);
541 ty = _mm_mul_ps(fscal,dy21);
542 tz = _mm_mul_ps(fscal,dz21);
544 /* Update vectorial force */
545 fix2 = _mm_add_ps(fix2,tx);
546 fiy2 = _mm_add_ps(fiy2,ty);
547 fiz2 = _mm_add_ps(fiz2,tz);
549 fjx1 = _mm_add_ps(fjx1,tx);
550 fjy1 = _mm_add_ps(fjy1,ty);
551 fjz1 = _mm_add_ps(fjz1,tz);
553 /**************************
554 * CALCULATE INTERACTIONS *
555 **************************/
557 /* COULOMB ELECTROSTATICS */
558 velec = _mm_mul_ps(qq22,rinv22);
559 felec = _mm_mul_ps(velec,rinvsq22);
561 /* Update potential sum for this i atom from the interaction with this j atom. */
562 velecsum = _mm_add_ps(velecsum,velec);
566 /* Calculate temporary vectorial force */
567 tx = _mm_mul_ps(fscal,dx22);
568 ty = _mm_mul_ps(fscal,dy22);
569 tz = _mm_mul_ps(fscal,dz22);
571 /* Update vectorial force */
572 fix2 = _mm_add_ps(fix2,tx);
573 fiy2 = _mm_add_ps(fiy2,ty);
574 fiz2 = _mm_add_ps(fiz2,tz);
576 fjx2 = _mm_add_ps(fjx2,tx);
577 fjy2 = _mm_add_ps(fjy2,ty);
578 fjz2 = _mm_add_ps(fjz2,tz);
580 fjptrA = f+j_coord_offsetA;
581 fjptrB = f+j_coord_offsetB;
582 fjptrC = f+j_coord_offsetC;
583 fjptrD = f+j_coord_offsetD;
585 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
586 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
588 /* Inner loop uses 287 flops */
594 /* Get j neighbor index, and coordinate index */
595 jnrlistA = jjnr[jidx];
596 jnrlistB = jjnr[jidx+1];
597 jnrlistC = jjnr[jidx+2];
598 jnrlistD = jjnr[jidx+3];
599 /* Sign of each element will be negative for non-real atoms.
600 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
601 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
603 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
604 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
605 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
606 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
607 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
608 j_coord_offsetA = DIM*jnrA;
609 j_coord_offsetB = DIM*jnrB;
610 j_coord_offsetC = DIM*jnrC;
611 j_coord_offsetD = DIM*jnrD;
613 /* load j atom coordinates */
614 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
615 x+j_coord_offsetC,x+j_coord_offsetD,
616 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
618 /* Calculate displacement vector */
619 dx00 = _mm_sub_ps(ix0,jx0);
620 dy00 = _mm_sub_ps(iy0,jy0);
621 dz00 = _mm_sub_ps(iz0,jz0);
622 dx01 = _mm_sub_ps(ix0,jx1);
623 dy01 = _mm_sub_ps(iy0,jy1);
624 dz01 = _mm_sub_ps(iz0,jz1);
625 dx02 = _mm_sub_ps(ix0,jx2);
626 dy02 = _mm_sub_ps(iy0,jy2);
627 dz02 = _mm_sub_ps(iz0,jz2);
628 dx10 = _mm_sub_ps(ix1,jx0);
629 dy10 = _mm_sub_ps(iy1,jy0);
630 dz10 = _mm_sub_ps(iz1,jz0);
631 dx11 = _mm_sub_ps(ix1,jx1);
632 dy11 = _mm_sub_ps(iy1,jy1);
633 dz11 = _mm_sub_ps(iz1,jz1);
634 dx12 = _mm_sub_ps(ix1,jx2);
635 dy12 = _mm_sub_ps(iy1,jy2);
636 dz12 = _mm_sub_ps(iz1,jz2);
637 dx20 = _mm_sub_ps(ix2,jx0);
638 dy20 = _mm_sub_ps(iy2,jy0);
639 dz20 = _mm_sub_ps(iz2,jz0);
640 dx21 = _mm_sub_ps(ix2,jx1);
641 dy21 = _mm_sub_ps(iy2,jy1);
642 dz21 = _mm_sub_ps(iz2,jz1);
643 dx22 = _mm_sub_ps(ix2,jx2);
644 dy22 = _mm_sub_ps(iy2,jy2);
645 dz22 = _mm_sub_ps(iz2,jz2);
647 /* Calculate squared distance and things based on it */
648 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
649 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
650 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
651 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
652 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
653 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
654 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
655 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
656 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
658 rinv00 = gmx_mm_invsqrt_ps(rsq00);
659 rinv01 = gmx_mm_invsqrt_ps(rsq01);
660 rinv02 = gmx_mm_invsqrt_ps(rsq02);
661 rinv10 = gmx_mm_invsqrt_ps(rsq10);
662 rinv11 = gmx_mm_invsqrt_ps(rsq11);
663 rinv12 = gmx_mm_invsqrt_ps(rsq12);
664 rinv20 = gmx_mm_invsqrt_ps(rsq20);
665 rinv21 = gmx_mm_invsqrt_ps(rsq21);
666 rinv22 = gmx_mm_invsqrt_ps(rsq22);
668 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
669 rinvsq01 = _mm_mul_ps(rinv01,rinv01);
670 rinvsq02 = _mm_mul_ps(rinv02,rinv02);
671 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
672 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
673 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
674 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
675 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
676 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
678 fjx0 = _mm_setzero_ps();
679 fjy0 = _mm_setzero_ps();
680 fjz0 = _mm_setzero_ps();
681 fjx1 = _mm_setzero_ps();
682 fjy1 = _mm_setzero_ps();
683 fjz1 = _mm_setzero_ps();
684 fjx2 = _mm_setzero_ps();
685 fjy2 = _mm_setzero_ps();
686 fjz2 = _mm_setzero_ps();
688 /**************************
689 * CALCULATE INTERACTIONS *
690 **************************/
692 r00 = _mm_mul_ps(rsq00,rinv00);
693 r00 = _mm_andnot_ps(dummy_mask,r00);
695 /* Calculate table index by multiplying r with table scale and truncate to integer */
696 rt = _mm_mul_ps(r00,vftabscale);
697 vfitab = _mm_cvttps_epi32(rt);
698 vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
699 vfitab = _mm_slli_epi32(vfitab,3);
701 /* COULOMB ELECTROSTATICS */
702 velec = _mm_mul_ps(qq00,rinv00);
703 felec = _mm_mul_ps(velec,rinvsq00);
705 /* CUBIC SPLINE TABLE DISPERSION */
706 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
707 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
708 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
709 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
710 _MM_TRANSPOSE4_PS(Y,F,G,H);
711 Heps = _mm_mul_ps(vfeps,H);
712 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
713 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
714 vvdw6 = _mm_mul_ps(c6_00,VV);
715 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
716 fvdw6 = _mm_mul_ps(c6_00,FF);
718 /* CUBIC SPLINE TABLE REPULSION */
719 vfitab = _mm_add_epi32(vfitab,ifour);
720 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
721 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
722 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
723 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
724 _MM_TRANSPOSE4_PS(Y,F,G,H);
725 Heps = _mm_mul_ps(vfeps,H);
726 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
727 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
728 vvdw12 = _mm_mul_ps(c12_00,VV);
729 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
730 fvdw12 = _mm_mul_ps(c12_00,FF);
731 vvdw = _mm_add_ps(vvdw12,vvdw6);
732 fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
734 /* Update potential sum for this i atom from the interaction with this j atom. */
735 velec = _mm_andnot_ps(dummy_mask,velec);
736 velecsum = _mm_add_ps(velecsum,velec);
737 vvdw = _mm_andnot_ps(dummy_mask,vvdw);
738 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
740 fscal = _mm_add_ps(felec,fvdw);
742 fscal = _mm_andnot_ps(dummy_mask,fscal);
744 /* Calculate temporary vectorial force */
745 tx = _mm_mul_ps(fscal,dx00);
746 ty = _mm_mul_ps(fscal,dy00);
747 tz = _mm_mul_ps(fscal,dz00);
749 /* Update vectorial force */
750 fix0 = _mm_add_ps(fix0,tx);
751 fiy0 = _mm_add_ps(fiy0,ty);
752 fiz0 = _mm_add_ps(fiz0,tz);
754 fjx0 = _mm_add_ps(fjx0,tx);
755 fjy0 = _mm_add_ps(fjy0,ty);
756 fjz0 = _mm_add_ps(fjz0,tz);
758 /**************************
759 * CALCULATE INTERACTIONS *
760 **************************/
762 /* COULOMB ELECTROSTATICS */
763 velec = _mm_mul_ps(qq01,rinv01);
764 felec = _mm_mul_ps(velec,rinvsq01);
766 /* Update potential sum for this i atom from the interaction with this j atom. */
767 velec = _mm_andnot_ps(dummy_mask,velec);
768 velecsum = _mm_add_ps(velecsum,velec);
772 fscal = _mm_andnot_ps(dummy_mask,fscal);
774 /* Calculate temporary vectorial force */
775 tx = _mm_mul_ps(fscal,dx01);
776 ty = _mm_mul_ps(fscal,dy01);
777 tz = _mm_mul_ps(fscal,dz01);
779 /* Update vectorial force */
780 fix0 = _mm_add_ps(fix0,tx);
781 fiy0 = _mm_add_ps(fiy0,ty);
782 fiz0 = _mm_add_ps(fiz0,tz);
784 fjx1 = _mm_add_ps(fjx1,tx);
785 fjy1 = _mm_add_ps(fjy1,ty);
786 fjz1 = _mm_add_ps(fjz1,tz);
788 /**************************
789 * CALCULATE INTERACTIONS *
790 **************************/
792 /* COULOMB ELECTROSTATICS */
793 velec = _mm_mul_ps(qq02,rinv02);
794 felec = _mm_mul_ps(velec,rinvsq02);
796 /* Update potential sum for this i atom from the interaction with this j atom. */
797 velec = _mm_andnot_ps(dummy_mask,velec);
798 velecsum = _mm_add_ps(velecsum,velec);
802 fscal = _mm_andnot_ps(dummy_mask,fscal);
804 /* Calculate temporary vectorial force */
805 tx = _mm_mul_ps(fscal,dx02);
806 ty = _mm_mul_ps(fscal,dy02);
807 tz = _mm_mul_ps(fscal,dz02);
809 /* Update vectorial force */
810 fix0 = _mm_add_ps(fix0,tx);
811 fiy0 = _mm_add_ps(fiy0,ty);
812 fiz0 = _mm_add_ps(fiz0,tz);
814 fjx2 = _mm_add_ps(fjx2,tx);
815 fjy2 = _mm_add_ps(fjy2,ty);
816 fjz2 = _mm_add_ps(fjz2,tz);
818 /**************************
819 * CALCULATE INTERACTIONS *
820 **************************/
822 /* COULOMB ELECTROSTATICS */
823 velec = _mm_mul_ps(qq10,rinv10);
824 felec = _mm_mul_ps(velec,rinvsq10);
826 /* Update potential sum for this i atom from the interaction with this j atom. */
827 velec = _mm_andnot_ps(dummy_mask,velec);
828 velecsum = _mm_add_ps(velecsum,velec);
832 fscal = _mm_andnot_ps(dummy_mask,fscal);
834 /* Calculate temporary vectorial force */
835 tx = _mm_mul_ps(fscal,dx10);
836 ty = _mm_mul_ps(fscal,dy10);
837 tz = _mm_mul_ps(fscal,dz10);
839 /* Update vectorial force */
840 fix1 = _mm_add_ps(fix1,tx);
841 fiy1 = _mm_add_ps(fiy1,ty);
842 fiz1 = _mm_add_ps(fiz1,tz);
844 fjx0 = _mm_add_ps(fjx0,tx);
845 fjy0 = _mm_add_ps(fjy0,ty);
846 fjz0 = _mm_add_ps(fjz0,tz);
848 /**************************
849 * CALCULATE INTERACTIONS *
850 **************************/
852 /* COULOMB ELECTROSTATICS */
853 velec = _mm_mul_ps(qq11,rinv11);
854 felec = _mm_mul_ps(velec,rinvsq11);
856 /* Update potential sum for this i atom from the interaction with this j atom. */
857 velec = _mm_andnot_ps(dummy_mask,velec);
858 velecsum = _mm_add_ps(velecsum,velec);
862 fscal = _mm_andnot_ps(dummy_mask,fscal);
864 /* Calculate temporary vectorial force */
865 tx = _mm_mul_ps(fscal,dx11);
866 ty = _mm_mul_ps(fscal,dy11);
867 tz = _mm_mul_ps(fscal,dz11);
869 /* Update vectorial force */
870 fix1 = _mm_add_ps(fix1,tx);
871 fiy1 = _mm_add_ps(fiy1,ty);
872 fiz1 = _mm_add_ps(fiz1,tz);
874 fjx1 = _mm_add_ps(fjx1,tx);
875 fjy1 = _mm_add_ps(fjy1,ty);
876 fjz1 = _mm_add_ps(fjz1,tz);
878 /**************************
879 * CALCULATE INTERACTIONS *
880 **************************/
882 /* COULOMB ELECTROSTATICS */
883 velec = _mm_mul_ps(qq12,rinv12);
884 felec = _mm_mul_ps(velec,rinvsq12);
886 /* Update potential sum for this i atom from the interaction with this j atom. */
887 velec = _mm_andnot_ps(dummy_mask,velec);
888 velecsum = _mm_add_ps(velecsum,velec);
892 fscal = _mm_andnot_ps(dummy_mask,fscal);
894 /* Calculate temporary vectorial force */
895 tx = _mm_mul_ps(fscal,dx12);
896 ty = _mm_mul_ps(fscal,dy12);
897 tz = _mm_mul_ps(fscal,dz12);
899 /* Update vectorial force */
900 fix1 = _mm_add_ps(fix1,tx);
901 fiy1 = _mm_add_ps(fiy1,ty);
902 fiz1 = _mm_add_ps(fiz1,tz);
904 fjx2 = _mm_add_ps(fjx2,tx);
905 fjy2 = _mm_add_ps(fjy2,ty);
906 fjz2 = _mm_add_ps(fjz2,tz);
908 /**************************
909 * CALCULATE INTERACTIONS *
910 **************************/
912 /* COULOMB ELECTROSTATICS */
913 velec = _mm_mul_ps(qq20,rinv20);
914 felec = _mm_mul_ps(velec,rinvsq20);
916 /* Update potential sum for this i atom from the interaction with this j atom. */
917 velec = _mm_andnot_ps(dummy_mask,velec);
918 velecsum = _mm_add_ps(velecsum,velec);
922 fscal = _mm_andnot_ps(dummy_mask,fscal);
924 /* Calculate temporary vectorial force */
925 tx = _mm_mul_ps(fscal,dx20);
926 ty = _mm_mul_ps(fscal,dy20);
927 tz = _mm_mul_ps(fscal,dz20);
929 /* Update vectorial force */
930 fix2 = _mm_add_ps(fix2,tx);
931 fiy2 = _mm_add_ps(fiy2,ty);
932 fiz2 = _mm_add_ps(fiz2,tz);
934 fjx0 = _mm_add_ps(fjx0,tx);
935 fjy0 = _mm_add_ps(fjy0,ty);
936 fjz0 = _mm_add_ps(fjz0,tz);
938 /**************************
939 * CALCULATE INTERACTIONS *
940 **************************/
942 /* COULOMB ELECTROSTATICS */
943 velec = _mm_mul_ps(qq21,rinv21);
944 felec = _mm_mul_ps(velec,rinvsq21);
946 /* Update potential sum for this i atom from the interaction with this j atom. */
947 velec = _mm_andnot_ps(dummy_mask,velec);
948 velecsum = _mm_add_ps(velecsum,velec);
952 fscal = _mm_andnot_ps(dummy_mask,fscal);
954 /* Calculate temporary vectorial force */
955 tx = _mm_mul_ps(fscal,dx21);
956 ty = _mm_mul_ps(fscal,dy21);
957 tz = _mm_mul_ps(fscal,dz21);
959 /* Update vectorial force */
960 fix2 = _mm_add_ps(fix2,tx);
961 fiy2 = _mm_add_ps(fiy2,ty);
962 fiz2 = _mm_add_ps(fiz2,tz);
964 fjx1 = _mm_add_ps(fjx1,tx);
965 fjy1 = _mm_add_ps(fjy1,ty);
966 fjz1 = _mm_add_ps(fjz1,tz);
968 /**************************
969 * CALCULATE INTERACTIONS *
970 **************************/
972 /* COULOMB ELECTROSTATICS */
973 velec = _mm_mul_ps(qq22,rinv22);
974 felec = _mm_mul_ps(velec,rinvsq22);
976 /* Update potential sum for this i atom from the interaction with this j atom. */
977 velec = _mm_andnot_ps(dummy_mask,velec);
978 velecsum = _mm_add_ps(velecsum,velec);
982 fscal = _mm_andnot_ps(dummy_mask,fscal);
984 /* Calculate temporary vectorial force */
985 tx = _mm_mul_ps(fscal,dx22);
986 ty = _mm_mul_ps(fscal,dy22);
987 tz = _mm_mul_ps(fscal,dz22);
989 /* Update vectorial force */
990 fix2 = _mm_add_ps(fix2,tx);
991 fiy2 = _mm_add_ps(fiy2,ty);
992 fiz2 = _mm_add_ps(fiz2,tz);
994 fjx2 = _mm_add_ps(fjx2,tx);
995 fjy2 = _mm_add_ps(fjy2,ty);
996 fjz2 = _mm_add_ps(fjz2,tz);
998 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
999 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1000 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1001 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1003 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1004 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1006 /* Inner loop uses 288 flops */
1009 /* End of innermost loop */
1011 gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1012 f+i_coord_offset,fshift+i_shift_offset);
1015 /* Update potential energies */
1016 gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1017 gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
1019 /* Increment number of inner iterations */
1020 inneriter += j_index_end - j_index_start;
1022 /* Outer loop uses 20 flops */
1025 /* Increment number of outer iterations */
1028 /* Update outer/inner flops */
1030 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*288);
1033 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_F_sse4_1_single
1034 * Electrostatics interaction: Coulomb
1035 * VdW interaction: CubicSplineTable
1036 * Geometry: Water3-Water3
1037 * Calculate force/pot: Force
1040 nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_F_sse4_1_single
1041 (t_nblist * gmx_restrict nlist,
1042 rvec * gmx_restrict xx,
1043 rvec * gmx_restrict ff,
1044 t_forcerec * gmx_restrict fr,
1045 t_mdatoms * gmx_restrict mdatoms,
1046 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1047 t_nrnb * gmx_restrict nrnb)
1049 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1050 * just 0 for non-waters.
1051 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
1052 * jnr indices corresponding to data put in the four positions in the SIMD register.
1054 int i_shift_offset,i_coord_offset,outeriter,inneriter;
1055 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1056 int jnrA,jnrB,jnrC,jnrD;
1057 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1058 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1059 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
1060 real rcutoff_scalar;
1061 real *shiftvec,*fshift,*x,*f;
1062 real *fjptrA,*fjptrB,*fjptrC,*fjptrD;
1063 real scratch[4*DIM];
1064 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1066 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1068 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1070 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1071 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
1072 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1073 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
1074 __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1075 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
1076 __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1077 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1078 __m128 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
1079 __m128 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
1080 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
1081 __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1082 __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1083 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
1084 __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1085 __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1086 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
1089 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1092 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
1093 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
1095 __m128i ifour = _mm_set1_epi32(4);
1096 __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
1098 __m128 dummy_mask,cutoff_mask;
1099 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
1100 __m128 one = _mm_set1_ps(1.0);
1101 __m128 two = _mm_set1_ps(2.0);
1107 jindex = nlist->jindex;
1109 shiftidx = nlist->shift;
1111 shiftvec = fr->shift_vec[0];
1112 fshift = fr->fshift[0];
1113 facel = _mm_set1_ps(fr->epsfac);
1114 charge = mdatoms->chargeA;
1115 nvdwtype = fr->ntype;
1116 vdwparam = fr->nbfp;
1117 vdwtype = mdatoms->typeA;
1119 vftab = kernel_data->table_vdw->data;
1120 vftabscale = _mm_set1_ps(kernel_data->table_vdw->scale);
1122 /* Setup water-specific parameters */
1123 inr = nlist->iinr[0];
1124 iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
1125 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
1126 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
1127 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
1129 jq0 = _mm_set1_ps(charge[inr+0]);
1130 jq1 = _mm_set1_ps(charge[inr+1]);
1131 jq2 = _mm_set1_ps(charge[inr+2]);
1132 vdwjidx0A = 2*vdwtype[inr+0];
1133 qq00 = _mm_mul_ps(iq0,jq0);
1134 c6_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
1135 c12_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
1136 qq01 = _mm_mul_ps(iq0,jq1);
1137 qq02 = _mm_mul_ps(iq0,jq2);
1138 qq10 = _mm_mul_ps(iq1,jq0);
1139 qq11 = _mm_mul_ps(iq1,jq1);
1140 qq12 = _mm_mul_ps(iq1,jq2);
1141 qq20 = _mm_mul_ps(iq2,jq0);
1142 qq21 = _mm_mul_ps(iq2,jq1);
1143 qq22 = _mm_mul_ps(iq2,jq2);
1145 /* Avoid stupid compiler warnings */
1146 jnrA = jnrB = jnrC = jnrD = 0;
1147 j_coord_offsetA = 0;
1148 j_coord_offsetB = 0;
1149 j_coord_offsetC = 0;
1150 j_coord_offsetD = 0;
1155 for(iidx=0;iidx<4*DIM;iidx++)
1157 scratch[iidx] = 0.0;
1160 /* Start outer loop over neighborlists */
1161 for(iidx=0; iidx<nri; iidx++)
1163 /* Load shift vector for this list */
1164 i_shift_offset = DIM*shiftidx[iidx];
1166 /* Load limits for loop over neighbors */
1167 j_index_start = jindex[iidx];
1168 j_index_end = jindex[iidx+1];
1170 /* Get outer coordinate index */
1172 i_coord_offset = DIM*inr;
1174 /* Load i particle coords and add shift vector */
1175 gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1176 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1178 fix0 = _mm_setzero_ps();
1179 fiy0 = _mm_setzero_ps();
1180 fiz0 = _mm_setzero_ps();
1181 fix1 = _mm_setzero_ps();
1182 fiy1 = _mm_setzero_ps();
1183 fiz1 = _mm_setzero_ps();
1184 fix2 = _mm_setzero_ps();
1185 fiy2 = _mm_setzero_ps();
1186 fiz2 = _mm_setzero_ps();
1188 /* Start inner kernel loop */
1189 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1192 /* Get j neighbor index, and coordinate index */
1194 jnrB = jjnr[jidx+1];
1195 jnrC = jjnr[jidx+2];
1196 jnrD = jjnr[jidx+3];
1197 j_coord_offsetA = DIM*jnrA;
1198 j_coord_offsetB = DIM*jnrB;
1199 j_coord_offsetC = DIM*jnrC;
1200 j_coord_offsetD = DIM*jnrD;
1202 /* load j atom coordinates */
1203 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1204 x+j_coord_offsetC,x+j_coord_offsetD,
1205 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1207 /* Calculate displacement vector */
1208 dx00 = _mm_sub_ps(ix0,jx0);
1209 dy00 = _mm_sub_ps(iy0,jy0);
1210 dz00 = _mm_sub_ps(iz0,jz0);
1211 dx01 = _mm_sub_ps(ix0,jx1);
1212 dy01 = _mm_sub_ps(iy0,jy1);
1213 dz01 = _mm_sub_ps(iz0,jz1);
1214 dx02 = _mm_sub_ps(ix0,jx2);
1215 dy02 = _mm_sub_ps(iy0,jy2);
1216 dz02 = _mm_sub_ps(iz0,jz2);
1217 dx10 = _mm_sub_ps(ix1,jx0);
1218 dy10 = _mm_sub_ps(iy1,jy0);
1219 dz10 = _mm_sub_ps(iz1,jz0);
1220 dx11 = _mm_sub_ps(ix1,jx1);
1221 dy11 = _mm_sub_ps(iy1,jy1);
1222 dz11 = _mm_sub_ps(iz1,jz1);
1223 dx12 = _mm_sub_ps(ix1,jx2);
1224 dy12 = _mm_sub_ps(iy1,jy2);
1225 dz12 = _mm_sub_ps(iz1,jz2);
1226 dx20 = _mm_sub_ps(ix2,jx0);
1227 dy20 = _mm_sub_ps(iy2,jy0);
1228 dz20 = _mm_sub_ps(iz2,jz0);
1229 dx21 = _mm_sub_ps(ix2,jx1);
1230 dy21 = _mm_sub_ps(iy2,jy1);
1231 dz21 = _mm_sub_ps(iz2,jz1);
1232 dx22 = _mm_sub_ps(ix2,jx2);
1233 dy22 = _mm_sub_ps(iy2,jy2);
1234 dz22 = _mm_sub_ps(iz2,jz2);
1236 /* Calculate squared distance and things based on it */
1237 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1238 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
1239 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
1240 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1241 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1242 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1243 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1244 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1245 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1247 rinv00 = gmx_mm_invsqrt_ps(rsq00);
1248 rinv01 = gmx_mm_invsqrt_ps(rsq01);
1249 rinv02 = gmx_mm_invsqrt_ps(rsq02);
1250 rinv10 = gmx_mm_invsqrt_ps(rsq10);
1251 rinv11 = gmx_mm_invsqrt_ps(rsq11);
1252 rinv12 = gmx_mm_invsqrt_ps(rsq12);
1253 rinv20 = gmx_mm_invsqrt_ps(rsq20);
1254 rinv21 = gmx_mm_invsqrt_ps(rsq21);
1255 rinv22 = gmx_mm_invsqrt_ps(rsq22);
1257 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
1258 rinvsq01 = _mm_mul_ps(rinv01,rinv01);
1259 rinvsq02 = _mm_mul_ps(rinv02,rinv02);
1260 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
1261 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
1262 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
1263 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
1264 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
1265 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
1267 fjx0 = _mm_setzero_ps();
1268 fjy0 = _mm_setzero_ps();
1269 fjz0 = _mm_setzero_ps();
1270 fjx1 = _mm_setzero_ps();
1271 fjy1 = _mm_setzero_ps();
1272 fjz1 = _mm_setzero_ps();
1273 fjx2 = _mm_setzero_ps();
1274 fjy2 = _mm_setzero_ps();
1275 fjz2 = _mm_setzero_ps();
1277 /**************************
1278 * CALCULATE INTERACTIONS *
1279 **************************/
1281 r00 = _mm_mul_ps(rsq00,rinv00);
1283 /* Calculate table index by multiplying r with table scale and truncate to integer */
1284 rt = _mm_mul_ps(r00,vftabscale);
1285 vfitab = _mm_cvttps_epi32(rt);
1286 vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
1287 vfitab = _mm_slli_epi32(vfitab,3);
1289 /* COULOMB ELECTROSTATICS */
1290 velec = _mm_mul_ps(qq00,rinv00);
1291 felec = _mm_mul_ps(velec,rinvsq00);
1293 /* CUBIC SPLINE TABLE DISPERSION */
1294 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1295 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1296 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1297 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1298 _MM_TRANSPOSE4_PS(Y,F,G,H);
1299 Heps = _mm_mul_ps(vfeps,H);
1300 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1301 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1302 fvdw6 = _mm_mul_ps(c6_00,FF);
1304 /* CUBIC SPLINE TABLE REPULSION */
1305 vfitab = _mm_add_epi32(vfitab,ifour);
1306 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1307 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1308 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1309 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1310 _MM_TRANSPOSE4_PS(Y,F,G,H);
1311 Heps = _mm_mul_ps(vfeps,H);
1312 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1313 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1314 fvdw12 = _mm_mul_ps(c12_00,FF);
1315 fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
1317 fscal = _mm_add_ps(felec,fvdw);
1319 /* Calculate temporary vectorial force */
1320 tx = _mm_mul_ps(fscal,dx00);
1321 ty = _mm_mul_ps(fscal,dy00);
1322 tz = _mm_mul_ps(fscal,dz00);
1324 /* Update vectorial force */
1325 fix0 = _mm_add_ps(fix0,tx);
1326 fiy0 = _mm_add_ps(fiy0,ty);
1327 fiz0 = _mm_add_ps(fiz0,tz);
1329 fjx0 = _mm_add_ps(fjx0,tx);
1330 fjy0 = _mm_add_ps(fjy0,ty);
1331 fjz0 = _mm_add_ps(fjz0,tz);
1333 /**************************
1334 * CALCULATE INTERACTIONS *
1335 **************************/
1337 /* COULOMB ELECTROSTATICS */
1338 velec = _mm_mul_ps(qq01,rinv01);
1339 felec = _mm_mul_ps(velec,rinvsq01);
1343 /* Calculate temporary vectorial force */
1344 tx = _mm_mul_ps(fscal,dx01);
1345 ty = _mm_mul_ps(fscal,dy01);
1346 tz = _mm_mul_ps(fscal,dz01);
1348 /* Update vectorial force */
1349 fix0 = _mm_add_ps(fix0,tx);
1350 fiy0 = _mm_add_ps(fiy0,ty);
1351 fiz0 = _mm_add_ps(fiz0,tz);
1353 fjx1 = _mm_add_ps(fjx1,tx);
1354 fjy1 = _mm_add_ps(fjy1,ty);
1355 fjz1 = _mm_add_ps(fjz1,tz);
1357 /**************************
1358 * CALCULATE INTERACTIONS *
1359 **************************/
1361 /* COULOMB ELECTROSTATICS */
1362 velec = _mm_mul_ps(qq02,rinv02);
1363 felec = _mm_mul_ps(velec,rinvsq02);
1367 /* Calculate temporary vectorial force */
1368 tx = _mm_mul_ps(fscal,dx02);
1369 ty = _mm_mul_ps(fscal,dy02);
1370 tz = _mm_mul_ps(fscal,dz02);
1372 /* Update vectorial force */
1373 fix0 = _mm_add_ps(fix0,tx);
1374 fiy0 = _mm_add_ps(fiy0,ty);
1375 fiz0 = _mm_add_ps(fiz0,tz);
1377 fjx2 = _mm_add_ps(fjx2,tx);
1378 fjy2 = _mm_add_ps(fjy2,ty);
1379 fjz2 = _mm_add_ps(fjz2,tz);
1381 /**************************
1382 * CALCULATE INTERACTIONS *
1383 **************************/
1385 /* COULOMB ELECTROSTATICS */
1386 velec = _mm_mul_ps(qq10,rinv10);
1387 felec = _mm_mul_ps(velec,rinvsq10);
1391 /* Calculate temporary vectorial force */
1392 tx = _mm_mul_ps(fscal,dx10);
1393 ty = _mm_mul_ps(fscal,dy10);
1394 tz = _mm_mul_ps(fscal,dz10);
1396 /* Update vectorial force */
1397 fix1 = _mm_add_ps(fix1,tx);
1398 fiy1 = _mm_add_ps(fiy1,ty);
1399 fiz1 = _mm_add_ps(fiz1,tz);
1401 fjx0 = _mm_add_ps(fjx0,tx);
1402 fjy0 = _mm_add_ps(fjy0,ty);
1403 fjz0 = _mm_add_ps(fjz0,tz);
1405 /**************************
1406 * CALCULATE INTERACTIONS *
1407 **************************/
1409 /* COULOMB ELECTROSTATICS */
1410 velec = _mm_mul_ps(qq11,rinv11);
1411 felec = _mm_mul_ps(velec,rinvsq11);
1415 /* Calculate temporary vectorial force */
1416 tx = _mm_mul_ps(fscal,dx11);
1417 ty = _mm_mul_ps(fscal,dy11);
1418 tz = _mm_mul_ps(fscal,dz11);
1420 /* Update vectorial force */
1421 fix1 = _mm_add_ps(fix1,tx);
1422 fiy1 = _mm_add_ps(fiy1,ty);
1423 fiz1 = _mm_add_ps(fiz1,tz);
1425 fjx1 = _mm_add_ps(fjx1,tx);
1426 fjy1 = _mm_add_ps(fjy1,ty);
1427 fjz1 = _mm_add_ps(fjz1,tz);
1429 /**************************
1430 * CALCULATE INTERACTIONS *
1431 **************************/
1433 /* COULOMB ELECTROSTATICS */
1434 velec = _mm_mul_ps(qq12,rinv12);
1435 felec = _mm_mul_ps(velec,rinvsq12);
1439 /* Calculate temporary vectorial force */
1440 tx = _mm_mul_ps(fscal,dx12);
1441 ty = _mm_mul_ps(fscal,dy12);
1442 tz = _mm_mul_ps(fscal,dz12);
1444 /* Update vectorial force */
1445 fix1 = _mm_add_ps(fix1,tx);
1446 fiy1 = _mm_add_ps(fiy1,ty);
1447 fiz1 = _mm_add_ps(fiz1,tz);
1449 fjx2 = _mm_add_ps(fjx2,tx);
1450 fjy2 = _mm_add_ps(fjy2,ty);
1451 fjz2 = _mm_add_ps(fjz2,tz);
1453 /**************************
1454 * CALCULATE INTERACTIONS *
1455 **************************/
1457 /* COULOMB ELECTROSTATICS */
1458 velec = _mm_mul_ps(qq20,rinv20);
1459 felec = _mm_mul_ps(velec,rinvsq20);
1463 /* Calculate temporary vectorial force */
1464 tx = _mm_mul_ps(fscal,dx20);
1465 ty = _mm_mul_ps(fscal,dy20);
1466 tz = _mm_mul_ps(fscal,dz20);
1468 /* Update vectorial force */
1469 fix2 = _mm_add_ps(fix2,tx);
1470 fiy2 = _mm_add_ps(fiy2,ty);
1471 fiz2 = _mm_add_ps(fiz2,tz);
1473 fjx0 = _mm_add_ps(fjx0,tx);
1474 fjy0 = _mm_add_ps(fjy0,ty);
1475 fjz0 = _mm_add_ps(fjz0,tz);
1477 /**************************
1478 * CALCULATE INTERACTIONS *
1479 **************************/
1481 /* COULOMB ELECTROSTATICS */
1482 velec = _mm_mul_ps(qq21,rinv21);
1483 felec = _mm_mul_ps(velec,rinvsq21);
1487 /* Calculate temporary vectorial force */
1488 tx = _mm_mul_ps(fscal,dx21);
1489 ty = _mm_mul_ps(fscal,dy21);
1490 tz = _mm_mul_ps(fscal,dz21);
1492 /* Update vectorial force */
1493 fix2 = _mm_add_ps(fix2,tx);
1494 fiy2 = _mm_add_ps(fiy2,ty);
1495 fiz2 = _mm_add_ps(fiz2,tz);
1497 fjx1 = _mm_add_ps(fjx1,tx);
1498 fjy1 = _mm_add_ps(fjy1,ty);
1499 fjz1 = _mm_add_ps(fjz1,tz);
1501 /**************************
1502 * CALCULATE INTERACTIONS *
1503 **************************/
1505 /* COULOMB ELECTROSTATICS */
1506 velec = _mm_mul_ps(qq22,rinv22);
1507 felec = _mm_mul_ps(velec,rinvsq22);
1511 /* Calculate temporary vectorial force */
1512 tx = _mm_mul_ps(fscal,dx22);
1513 ty = _mm_mul_ps(fscal,dy22);
1514 tz = _mm_mul_ps(fscal,dz22);
1516 /* Update vectorial force */
1517 fix2 = _mm_add_ps(fix2,tx);
1518 fiy2 = _mm_add_ps(fiy2,ty);
1519 fiz2 = _mm_add_ps(fiz2,tz);
1521 fjx2 = _mm_add_ps(fjx2,tx);
1522 fjy2 = _mm_add_ps(fjy2,ty);
1523 fjz2 = _mm_add_ps(fjz2,tz);
1525 fjptrA = f+j_coord_offsetA;
1526 fjptrB = f+j_coord_offsetB;
1527 fjptrC = f+j_coord_offsetC;
1528 fjptrD = f+j_coord_offsetD;
1530 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1531 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1533 /* Inner loop uses 270 flops */
1536 if(jidx<j_index_end)
1539 /* Get j neighbor index, and coordinate index */
1540 jnrlistA = jjnr[jidx];
1541 jnrlistB = jjnr[jidx+1];
1542 jnrlistC = jjnr[jidx+2];
1543 jnrlistD = jjnr[jidx+3];
1544 /* Sign of each element will be negative for non-real atoms.
1545 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1546 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1548 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1549 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
1550 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
1551 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
1552 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
1553 j_coord_offsetA = DIM*jnrA;
1554 j_coord_offsetB = DIM*jnrB;
1555 j_coord_offsetC = DIM*jnrC;
1556 j_coord_offsetD = DIM*jnrD;
1558 /* load j atom coordinates */
1559 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1560 x+j_coord_offsetC,x+j_coord_offsetD,
1561 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1563 /* Calculate displacement vector */
1564 dx00 = _mm_sub_ps(ix0,jx0);
1565 dy00 = _mm_sub_ps(iy0,jy0);
1566 dz00 = _mm_sub_ps(iz0,jz0);
1567 dx01 = _mm_sub_ps(ix0,jx1);
1568 dy01 = _mm_sub_ps(iy0,jy1);
1569 dz01 = _mm_sub_ps(iz0,jz1);
1570 dx02 = _mm_sub_ps(ix0,jx2);
1571 dy02 = _mm_sub_ps(iy0,jy2);
1572 dz02 = _mm_sub_ps(iz0,jz2);
1573 dx10 = _mm_sub_ps(ix1,jx0);
1574 dy10 = _mm_sub_ps(iy1,jy0);
1575 dz10 = _mm_sub_ps(iz1,jz0);
1576 dx11 = _mm_sub_ps(ix1,jx1);
1577 dy11 = _mm_sub_ps(iy1,jy1);
1578 dz11 = _mm_sub_ps(iz1,jz1);
1579 dx12 = _mm_sub_ps(ix1,jx2);
1580 dy12 = _mm_sub_ps(iy1,jy2);
1581 dz12 = _mm_sub_ps(iz1,jz2);
1582 dx20 = _mm_sub_ps(ix2,jx0);
1583 dy20 = _mm_sub_ps(iy2,jy0);
1584 dz20 = _mm_sub_ps(iz2,jz0);
1585 dx21 = _mm_sub_ps(ix2,jx1);
1586 dy21 = _mm_sub_ps(iy2,jy1);
1587 dz21 = _mm_sub_ps(iz2,jz1);
1588 dx22 = _mm_sub_ps(ix2,jx2);
1589 dy22 = _mm_sub_ps(iy2,jy2);
1590 dz22 = _mm_sub_ps(iz2,jz2);
1592 /* Calculate squared distance and things based on it */
1593 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1594 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
1595 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
1596 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1597 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1598 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1599 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1600 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1601 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1603 rinv00 = gmx_mm_invsqrt_ps(rsq00);
1604 rinv01 = gmx_mm_invsqrt_ps(rsq01);
1605 rinv02 = gmx_mm_invsqrt_ps(rsq02);
1606 rinv10 = gmx_mm_invsqrt_ps(rsq10);
1607 rinv11 = gmx_mm_invsqrt_ps(rsq11);
1608 rinv12 = gmx_mm_invsqrt_ps(rsq12);
1609 rinv20 = gmx_mm_invsqrt_ps(rsq20);
1610 rinv21 = gmx_mm_invsqrt_ps(rsq21);
1611 rinv22 = gmx_mm_invsqrt_ps(rsq22);
1613 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
1614 rinvsq01 = _mm_mul_ps(rinv01,rinv01);
1615 rinvsq02 = _mm_mul_ps(rinv02,rinv02);
1616 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
1617 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
1618 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
1619 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
1620 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
1621 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
1623 fjx0 = _mm_setzero_ps();
1624 fjy0 = _mm_setzero_ps();
1625 fjz0 = _mm_setzero_ps();
1626 fjx1 = _mm_setzero_ps();
1627 fjy1 = _mm_setzero_ps();
1628 fjz1 = _mm_setzero_ps();
1629 fjx2 = _mm_setzero_ps();
1630 fjy2 = _mm_setzero_ps();
1631 fjz2 = _mm_setzero_ps();
1633 /**************************
1634 * CALCULATE INTERACTIONS *
1635 **************************/
1637 r00 = _mm_mul_ps(rsq00,rinv00);
1638 r00 = _mm_andnot_ps(dummy_mask,r00);
1640 /* Calculate table index by multiplying r with table scale and truncate to integer */
1641 rt = _mm_mul_ps(r00,vftabscale);
1642 vfitab = _mm_cvttps_epi32(rt);
1643 vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
1644 vfitab = _mm_slli_epi32(vfitab,3);
1646 /* COULOMB ELECTROSTATICS */
1647 velec = _mm_mul_ps(qq00,rinv00);
1648 felec = _mm_mul_ps(velec,rinvsq00);
1650 /* CUBIC SPLINE TABLE DISPERSION */
1651 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1652 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1653 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1654 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1655 _MM_TRANSPOSE4_PS(Y,F,G,H);
1656 Heps = _mm_mul_ps(vfeps,H);
1657 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1658 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1659 fvdw6 = _mm_mul_ps(c6_00,FF);
1661 /* CUBIC SPLINE TABLE REPULSION */
1662 vfitab = _mm_add_epi32(vfitab,ifour);
1663 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1664 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1665 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1666 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1667 _MM_TRANSPOSE4_PS(Y,F,G,H);
1668 Heps = _mm_mul_ps(vfeps,H);
1669 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1670 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1671 fvdw12 = _mm_mul_ps(c12_00,FF);
1672 fvdw = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
1674 fscal = _mm_add_ps(felec,fvdw);
1676 fscal = _mm_andnot_ps(dummy_mask,fscal);
1678 /* Calculate temporary vectorial force */
1679 tx = _mm_mul_ps(fscal,dx00);
1680 ty = _mm_mul_ps(fscal,dy00);
1681 tz = _mm_mul_ps(fscal,dz00);
1683 /* Update vectorial force */
1684 fix0 = _mm_add_ps(fix0,tx);
1685 fiy0 = _mm_add_ps(fiy0,ty);
1686 fiz0 = _mm_add_ps(fiz0,tz);
1688 fjx0 = _mm_add_ps(fjx0,tx);
1689 fjy0 = _mm_add_ps(fjy0,ty);
1690 fjz0 = _mm_add_ps(fjz0,tz);
1692 /**************************
1693 * CALCULATE INTERACTIONS *
1694 **************************/
1696 /* COULOMB ELECTROSTATICS */
1697 velec = _mm_mul_ps(qq01,rinv01);
1698 felec = _mm_mul_ps(velec,rinvsq01);
1702 fscal = _mm_andnot_ps(dummy_mask,fscal);
1704 /* Calculate temporary vectorial force */
1705 tx = _mm_mul_ps(fscal,dx01);
1706 ty = _mm_mul_ps(fscal,dy01);
1707 tz = _mm_mul_ps(fscal,dz01);
1709 /* Update vectorial force */
1710 fix0 = _mm_add_ps(fix0,tx);
1711 fiy0 = _mm_add_ps(fiy0,ty);
1712 fiz0 = _mm_add_ps(fiz0,tz);
1714 fjx1 = _mm_add_ps(fjx1,tx);
1715 fjy1 = _mm_add_ps(fjy1,ty);
1716 fjz1 = _mm_add_ps(fjz1,tz);
1718 /**************************
1719 * CALCULATE INTERACTIONS *
1720 **************************/
1722 /* COULOMB ELECTROSTATICS */
1723 velec = _mm_mul_ps(qq02,rinv02);
1724 felec = _mm_mul_ps(velec,rinvsq02);
1728 fscal = _mm_andnot_ps(dummy_mask,fscal);
1730 /* Calculate temporary vectorial force */
1731 tx = _mm_mul_ps(fscal,dx02);
1732 ty = _mm_mul_ps(fscal,dy02);
1733 tz = _mm_mul_ps(fscal,dz02);
1735 /* Update vectorial force */
1736 fix0 = _mm_add_ps(fix0,tx);
1737 fiy0 = _mm_add_ps(fiy0,ty);
1738 fiz0 = _mm_add_ps(fiz0,tz);
1740 fjx2 = _mm_add_ps(fjx2,tx);
1741 fjy2 = _mm_add_ps(fjy2,ty);
1742 fjz2 = _mm_add_ps(fjz2,tz);
1744 /**************************
1745 * CALCULATE INTERACTIONS *
1746 **************************/
1748 /* COULOMB ELECTROSTATICS */
1749 velec = _mm_mul_ps(qq10,rinv10);
1750 felec = _mm_mul_ps(velec,rinvsq10);
1754 fscal = _mm_andnot_ps(dummy_mask,fscal);
1756 /* Calculate temporary vectorial force */
1757 tx = _mm_mul_ps(fscal,dx10);
1758 ty = _mm_mul_ps(fscal,dy10);
1759 tz = _mm_mul_ps(fscal,dz10);
1761 /* Update vectorial force */
1762 fix1 = _mm_add_ps(fix1,tx);
1763 fiy1 = _mm_add_ps(fiy1,ty);
1764 fiz1 = _mm_add_ps(fiz1,tz);
1766 fjx0 = _mm_add_ps(fjx0,tx);
1767 fjy0 = _mm_add_ps(fjy0,ty);
1768 fjz0 = _mm_add_ps(fjz0,tz);
1770 /**************************
1771 * CALCULATE INTERACTIONS *
1772 **************************/
1774 /* COULOMB ELECTROSTATICS */
1775 velec = _mm_mul_ps(qq11,rinv11);
1776 felec = _mm_mul_ps(velec,rinvsq11);
1780 fscal = _mm_andnot_ps(dummy_mask,fscal);
1782 /* Calculate temporary vectorial force */
1783 tx = _mm_mul_ps(fscal,dx11);
1784 ty = _mm_mul_ps(fscal,dy11);
1785 tz = _mm_mul_ps(fscal,dz11);
1787 /* Update vectorial force */
1788 fix1 = _mm_add_ps(fix1,tx);
1789 fiy1 = _mm_add_ps(fiy1,ty);
1790 fiz1 = _mm_add_ps(fiz1,tz);
1792 fjx1 = _mm_add_ps(fjx1,tx);
1793 fjy1 = _mm_add_ps(fjy1,ty);
1794 fjz1 = _mm_add_ps(fjz1,tz);
1796 /**************************
1797 * CALCULATE INTERACTIONS *
1798 **************************/
1800 /* COULOMB ELECTROSTATICS */
1801 velec = _mm_mul_ps(qq12,rinv12);
1802 felec = _mm_mul_ps(velec,rinvsq12);
1806 fscal = _mm_andnot_ps(dummy_mask,fscal);
1808 /* Calculate temporary vectorial force */
1809 tx = _mm_mul_ps(fscal,dx12);
1810 ty = _mm_mul_ps(fscal,dy12);
1811 tz = _mm_mul_ps(fscal,dz12);
1813 /* Update vectorial force */
1814 fix1 = _mm_add_ps(fix1,tx);
1815 fiy1 = _mm_add_ps(fiy1,ty);
1816 fiz1 = _mm_add_ps(fiz1,tz);
1818 fjx2 = _mm_add_ps(fjx2,tx);
1819 fjy2 = _mm_add_ps(fjy2,ty);
1820 fjz2 = _mm_add_ps(fjz2,tz);
1822 /**************************
1823 * CALCULATE INTERACTIONS *
1824 **************************/
1826 /* COULOMB ELECTROSTATICS */
1827 velec = _mm_mul_ps(qq20,rinv20);
1828 felec = _mm_mul_ps(velec,rinvsq20);
1832 fscal = _mm_andnot_ps(dummy_mask,fscal);
1834 /* Calculate temporary vectorial force */
1835 tx = _mm_mul_ps(fscal,dx20);
1836 ty = _mm_mul_ps(fscal,dy20);
1837 tz = _mm_mul_ps(fscal,dz20);
1839 /* Update vectorial force */
1840 fix2 = _mm_add_ps(fix2,tx);
1841 fiy2 = _mm_add_ps(fiy2,ty);
1842 fiz2 = _mm_add_ps(fiz2,tz);
1844 fjx0 = _mm_add_ps(fjx0,tx);
1845 fjy0 = _mm_add_ps(fjy0,ty);
1846 fjz0 = _mm_add_ps(fjz0,tz);
1848 /**************************
1849 * CALCULATE INTERACTIONS *
1850 **************************/
1852 /* COULOMB ELECTROSTATICS */
1853 velec = _mm_mul_ps(qq21,rinv21);
1854 felec = _mm_mul_ps(velec,rinvsq21);
1858 fscal = _mm_andnot_ps(dummy_mask,fscal);
1860 /* Calculate temporary vectorial force */
1861 tx = _mm_mul_ps(fscal,dx21);
1862 ty = _mm_mul_ps(fscal,dy21);
1863 tz = _mm_mul_ps(fscal,dz21);
1865 /* Update vectorial force */
1866 fix2 = _mm_add_ps(fix2,tx);
1867 fiy2 = _mm_add_ps(fiy2,ty);
1868 fiz2 = _mm_add_ps(fiz2,tz);
1870 fjx1 = _mm_add_ps(fjx1,tx);
1871 fjy1 = _mm_add_ps(fjy1,ty);
1872 fjz1 = _mm_add_ps(fjz1,tz);
1874 /**************************
1875 * CALCULATE INTERACTIONS *
1876 **************************/
1878 /* COULOMB ELECTROSTATICS */
1879 velec = _mm_mul_ps(qq22,rinv22);
1880 felec = _mm_mul_ps(velec,rinvsq22);
1884 fscal = _mm_andnot_ps(dummy_mask,fscal);
1886 /* Calculate temporary vectorial force */
1887 tx = _mm_mul_ps(fscal,dx22);
1888 ty = _mm_mul_ps(fscal,dy22);
1889 tz = _mm_mul_ps(fscal,dz22);
1891 /* Update vectorial force */
1892 fix2 = _mm_add_ps(fix2,tx);
1893 fiy2 = _mm_add_ps(fiy2,ty);
1894 fiz2 = _mm_add_ps(fiz2,tz);
1896 fjx2 = _mm_add_ps(fjx2,tx);
1897 fjy2 = _mm_add_ps(fjy2,ty);
1898 fjz2 = _mm_add_ps(fjz2,tz);
1900 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1901 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1902 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1903 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1905 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1906 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1908 /* Inner loop uses 271 flops */
1911 /* End of innermost loop */
1913 gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1914 f+i_coord_offset,fshift+i_shift_offset);
1916 /* Increment number of inner iterations */
1917 inneriter += j_index_end - j_index_start;
1919 /* Outer loop uses 18 flops */
1922 /* Increment number of outer iterations */
1925 /* Update outer/inner flops */
1927 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*271);