2 * Note: this file was generated by the Gromacs avx_128_fma_double kernel generator.
4 * This source code is part of
8 * Copyright (c) 2001-2012, The GROMACS Development Team
10 * Gromacs is a library for molecular simulation and trajectory analysis,
11 * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
12 * a full list of developers and information, check out http://www.gromacs.org
14 * This program is free software; you can redistribute it and/or modify it under
15 * the terms of the GNU Lesser General Public License as published by the Free
16 * Software Foundation; either version 2 of the License, or (at your option) any
19 * To help fund GROMACS development, we humbly ask that you cite
20 * the papers people have written on it - you can find them on the website.
28 #include "../nb_kernel.h"
29 #include "types/simple.h"
33 #include "gmx_math_x86_avx_128_fma_double.h"
34 #include "kernelutil_x86_avx_128_fma_double.h"
37 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwCSTab_GeomW3P1_VF_avx_128_fma_double
38 * Electrostatics interaction: ReactionField
39 * VdW interaction: CubicSplineTable
40 * Geometry: Water3-Particle
41 * Calculate force/pot: PotentialAndForce
44 nb_kernel_ElecRF_VdwCSTab_GeomW3P1_VF_avx_128_fma_double
45 (t_nblist * gmx_restrict nlist,
46 rvec * gmx_restrict xx,
47 rvec * gmx_restrict ff,
48 t_forcerec * gmx_restrict fr,
49 t_mdatoms * gmx_restrict mdatoms,
50 nb_kernel_data_t * gmx_restrict kernel_data,
51 t_nrnb * gmx_restrict nrnb)
53 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
54 * just 0 for non-waters.
55 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
56 * jnr indices corresponding to data put in the four positions in the SIMD register.
58 int i_shift_offset,i_coord_offset,outeriter,inneriter;
59 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
61 int j_coord_offsetA,j_coord_offsetB;
62 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
64 real *shiftvec,*fshift,*x,*f;
65 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
67 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
69 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
71 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
72 int vdwjidx0A,vdwjidx0B;
73 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
74 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
75 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
76 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
77 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
80 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
83 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
84 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
86 __m128i ifour = _mm_set1_epi32(4);
87 __m128d rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
89 __m128d dummy_mask,cutoff_mask;
90 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
91 __m128d one = _mm_set1_pd(1.0);
92 __m128d two = _mm_set1_pd(2.0);
98 jindex = nlist->jindex;
100 shiftidx = nlist->shift;
102 shiftvec = fr->shift_vec[0];
103 fshift = fr->fshift[0];
104 facel = _mm_set1_pd(fr->epsfac);
105 charge = mdatoms->chargeA;
106 krf = _mm_set1_pd(fr->ic->k_rf);
107 krf2 = _mm_set1_pd(fr->ic->k_rf*2.0);
108 crf = _mm_set1_pd(fr->ic->c_rf);
109 nvdwtype = fr->ntype;
111 vdwtype = mdatoms->typeA;
113 vftab = kernel_data->table_vdw->data;
114 vftabscale = _mm_set1_pd(kernel_data->table_vdw->scale);
116 /* Setup water-specific parameters */
117 inr = nlist->iinr[0];
118 iq0 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
119 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
120 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
121 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
123 /* Avoid stupid compiler warnings */
131 /* Start outer loop over neighborlists */
132 for(iidx=0; iidx<nri; iidx++)
134 /* Load shift vector for this list */
135 i_shift_offset = DIM*shiftidx[iidx];
137 /* Load limits for loop over neighbors */
138 j_index_start = jindex[iidx];
139 j_index_end = jindex[iidx+1];
141 /* Get outer coordinate index */
143 i_coord_offset = DIM*inr;
145 /* Load i particle coords and add shift vector */
146 gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
147 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
149 fix0 = _mm_setzero_pd();
150 fiy0 = _mm_setzero_pd();
151 fiz0 = _mm_setzero_pd();
152 fix1 = _mm_setzero_pd();
153 fiy1 = _mm_setzero_pd();
154 fiz1 = _mm_setzero_pd();
155 fix2 = _mm_setzero_pd();
156 fiy2 = _mm_setzero_pd();
157 fiz2 = _mm_setzero_pd();
159 /* Reset potential sums */
160 velecsum = _mm_setzero_pd();
161 vvdwsum = _mm_setzero_pd();
163 /* Start inner kernel loop */
164 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
167 /* Get j neighbor index, and coordinate index */
170 j_coord_offsetA = DIM*jnrA;
171 j_coord_offsetB = DIM*jnrB;
173 /* load j atom coordinates */
174 gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
177 /* Calculate displacement vector */
178 dx00 = _mm_sub_pd(ix0,jx0);
179 dy00 = _mm_sub_pd(iy0,jy0);
180 dz00 = _mm_sub_pd(iz0,jz0);
181 dx10 = _mm_sub_pd(ix1,jx0);
182 dy10 = _mm_sub_pd(iy1,jy0);
183 dz10 = _mm_sub_pd(iz1,jz0);
184 dx20 = _mm_sub_pd(ix2,jx0);
185 dy20 = _mm_sub_pd(iy2,jy0);
186 dz20 = _mm_sub_pd(iz2,jz0);
188 /* Calculate squared distance and things based on it */
189 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
190 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
191 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
193 rinv00 = gmx_mm_invsqrt_pd(rsq00);
194 rinv10 = gmx_mm_invsqrt_pd(rsq10);
195 rinv20 = gmx_mm_invsqrt_pd(rsq20);
197 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
198 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
199 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
201 /* Load parameters for j particles */
202 jq0 = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
203 vdwjidx0A = 2*vdwtype[jnrA+0];
204 vdwjidx0B = 2*vdwtype[jnrB+0];
206 fjx0 = _mm_setzero_pd();
207 fjy0 = _mm_setzero_pd();
208 fjz0 = _mm_setzero_pd();
210 /**************************
211 * CALCULATE INTERACTIONS *
212 **************************/
214 r00 = _mm_mul_pd(rsq00,rinv00);
216 /* Compute parameters for interactions between i and j atoms */
217 qq00 = _mm_mul_pd(iq0,jq0);
218 gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
219 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
221 /* Calculate table index by multiplying r with table scale and truncate to integer */
222 rt = _mm_mul_pd(r00,vftabscale);
223 vfitab = _mm_cvttpd_epi32(rt);
225 vfeps = _mm_frcz_pd(rt);
227 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
229 twovfeps = _mm_add_pd(vfeps,vfeps);
230 vfitab = _mm_slli_epi32(vfitab,3);
232 /* REACTION-FIELD ELECTROSTATICS */
233 velec = _mm_mul_pd(qq00,_mm_sub_pd(_mm_macc_pd(krf,rsq00,rinv00),crf));
234 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
236 /* CUBIC SPLINE TABLE DISPERSION */
237 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
238 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
239 GMX_MM_TRANSPOSE2_PD(Y,F);
240 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
241 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
242 GMX_MM_TRANSPOSE2_PD(G,H);
243 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
244 VV = _mm_macc_pd(vfeps,Fp,Y);
245 vvdw6 = _mm_mul_pd(c6_00,VV);
246 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
247 fvdw6 = _mm_mul_pd(c6_00,FF);
249 /* CUBIC SPLINE TABLE REPULSION */
250 vfitab = _mm_add_epi32(vfitab,ifour);
251 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
252 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
253 GMX_MM_TRANSPOSE2_PD(Y,F);
254 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
255 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
256 GMX_MM_TRANSPOSE2_PD(G,H);
257 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
258 VV = _mm_macc_pd(vfeps,Fp,Y);
259 vvdw12 = _mm_mul_pd(c12_00,VV);
260 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
261 fvdw12 = _mm_mul_pd(c12_00,FF);
262 vvdw = _mm_add_pd(vvdw12,vvdw6);
263 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
265 /* Update potential sum for this i atom from the interaction with this j atom. */
266 velecsum = _mm_add_pd(velecsum,velec);
267 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
269 fscal = _mm_add_pd(felec,fvdw);
271 /* Update vectorial force */
272 fix0 = _mm_macc_pd(dx00,fscal,fix0);
273 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
274 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
276 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
277 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
278 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
280 /**************************
281 * CALCULATE INTERACTIONS *
282 **************************/
284 /* Compute parameters for interactions between i and j atoms */
285 qq10 = _mm_mul_pd(iq1,jq0);
287 /* REACTION-FIELD ELECTROSTATICS */
288 velec = _mm_mul_pd(qq10,_mm_sub_pd(_mm_macc_pd(krf,rsq10,rinv10),crf));
289 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
291 /* Update potential sum for this i atom from the interaction with this j atom. */
292 velecsum = _mm_add_pd(velecsum,velec);
296 /* Update vectorial force */
297 fix1 = _mm_macc_pd(dx10,fscal,fix1);
298 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
299 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
301 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
302 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
303 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
305 /**************************
306 * CALCULATE INTERACTIONS *
307 **************************/
309 /* Compute parameters for interactions between i and j atoms */
310 qq20 = _mm_mul_pd(iq2,jq0);
312 /* REACTION-FIELD ELECTROSTATICS */
313 velec = _mm_mul_pd(qq20,_mm_sub_pd(_mm_macc_pd(krf,rsq20,rinv20),crf));
314 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
316 /* Update potential sum for this i atom from the interaction with this j atom. */
317 velecsum = _mm_add_pd(velecsum,velec);
321 /* Update vectorial force */
322 fix2 = _mm_macc_pd(dx20,fscal,fix2);
323 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
324 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
326 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
327 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
328 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
330 gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
332 /* Inner loop uses 143 flops */
339 j_coord_offsetA = DIM*jnrA;
341 /* load j atom coordinates */
342 gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
345 /* Calculate displacement vector */
346 dx00 = _mm_sub_pd(ix0,jx0);
347 dy00 = _mm_sub_pd(iy0,jy0);
348 dz00 = _mm_sub_pd(iz0,jz0);
349 dx10 = _mm_sub_pd(ix1,jx0);
350 dy10 = _mm_sub_pd(iy1,jy0);
351 dz10 = _mm_sub_pd(iz1,jz0);
352 dx20 = _mm_sub_pd(ix2,jx0);
353 dy20 = _mm_sub_pd(iy2,jy0);
354 dz20 = _mm_sub_pd(iz2,jz0);
356 /* Calculate squared distance and things based on it */
357 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
358 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
359 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
361 rinv00 = gmx_mm_invsqrt_pd(rsq00);
362 rinv10 = gmx_mm_invsqrt_pd(rsq10);
363 rinv20 = gmx_mm_invsqrt_pd(rsq20);
365 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
366 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
367 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
369 /* Load parameters for j particles */
370 jq0 = _mm_load_sd(charge+jnrA+0);
371 vdwjidx0A = 2*vdwtype[jnrA+0];
373 fjx0 = _mm_setzero_pd();
374 fjy0 = _mm_setzero_pd();
375 fjz0 = _mm_setzero_pd();
377 /**************************
378 * CALCULATE INTERACTIONS *
379 **************************/
381 r00 = _mm_mul_pd(rsq00,rinv00);
383 /* Compute parameters for interactions between i and j atoms */
384 qq00 = _mm_mul_pd(iq0,jq0);
385 gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
387 /* Calculate table index by multiplying r with table scale and truncate to integer */
388 rt = _mm_mul_pd(r00,vftabscale);
389 vfitab = _mm_cvttpd_epi32(rt);
391 vfeps = _mm_frcz_pd(rt);
393 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
395 twovfeps = _mm_add_pd(vfeps,vfeps);
396 vfitab = _mm_slli_epi32(vfitab,3);
398 /* REACTION-FIELD ELECTROSTATICS */
399 velec = _mm_mul_pd(qq00,_mm_sub_pd(_mm_macc_pd(krf,rsq00,rinv00),crf));
400 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
402 /* CUBIC SPLINE TABLE DISPERSION */
403 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
404 F = _mm_setzero_pd();
405 GMX_MM_TRANSPOSE2_PD(Y,F);
406 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
407 H = _mm_setzero_pd();
408 GMX_MM_TRANSPOSE2_PD(G,H);
409 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
410 VV = _mm_macc_pd(vfeps,Fp,Y);
411 vvdw6 = _mm_mul_pd(c6_00,VV);
412 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
413 fvdw6 = _mm_mul_pd(c6_00,FF);
415 /* CUBIC SPLINE TABLE REPULSION */
416 vfitab = _mm_add_epi32(vfitab,ifour);
417 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
418 F = _mm_setzero_pd();
419 GMX_MM_TRANSPOSE2_PD(Y,F);
420 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
421 H = _mm_setzero_pd();
422 GMX_MM_TRANSPOSE2_PD(G,H);
423 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
424 VV = _mm_macc_pd(vfeps,Fp,Y);
425 vvdw12 = _mm_mul_pd(c12_00,VV);
426 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
427 fvdw12 = _mm_mul_pd(c12_00,FF);
428 vvdw = _mm_add_pd(vvdw12,vvdw6);
429 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
431 /* Update potential sum for this i atom from the interaction with this j atom. */
432 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
433 velecsum = _mm_add_pd(velecsum,velec);
434 vvdw = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
435 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
437 fscal = _mm_add_pd(felec,fvdw);
439 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
441 /* Update vectorial force */
442 fix0 = _mm_macc_pd(dx00,fscal,fix0);
443 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
444 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
446 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
447 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
448 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
450 /**************************
451 * CALCULATE INTERACTIONS *
452 **************************/
454 /* Compute parameters for interactions between i and j atoms */
455 qq10 = _mm_mul_pd(iq1,jq0);
457 /* REACTION-FIELD ELECTROSTATICS */
458 velec = _mm_mul_pd(qq10,_mm_sub_pd(_mm_macc_pd(krf,rsq10,rinv10),crf));
459 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
461 /* Update potential sum for this i atom from the interaction with this j atom. */
462 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
463 velecsum = _mm_add_pd(velecsum,velec);
467 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
469 /* Update vectorial force */
470 fix1 = _mm_macc_pd(dx10,fscal,fix1);
471 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
472 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
474 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
475 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
476 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
478 /**************************
479 * CALCULATE INTERACTIONS *
480 **************************/
482 /* Compute parameters for interactions between i and j atoms */
483 qq20 = _mm_mul_pd(iq2,jq0);
485 /* REACTION-FIELD ELECTROSTATICS */
486 velec = _mm_mul_pd(qq20,_mm_sub_pd(_mm_macc_pd(krf,rsq20,rinv20),crf));
487 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
489 /* Update potential sum for this i atom from the interaction with this j atom. */
490 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
491 velecsum = _mm_add_pd(velecsum,velec);
495 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
497 /* Update vectorial force */
498 fix2 = _mm_macc_pd(dx20,fscal,fix2);
499 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
500 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
502 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
503 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
504 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
506 gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
508 /* Inner loop uses 143 flops */
511 /* End of innermost loop */
513 gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
514 f+i_coord_offset,fshift+i_shift_offset);
517 /* Update potential energies */
518 gmx_mm_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
519 gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
521 /* Increment number of inner iterations */
522 inneriter += j_index_end - j_index_start;
524 /* Outer loop uses 20 flops */
527 /* Increment number of outer iterations */
530 /* Update outer/inner flops */
532 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*20 + inneriter*143);
535 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwCSTab_GeomW3P1_F_avx_128_fma_double
536 * Electrostatics interaction: ReactionField
537 * VdW interaction: CubicSplineTable
538 * Geometry: Water3-Particle
539 * Calculate force/pot: Force
542 nb_kernel_ElecRF_VdwCSTab_GeomW3P1_F_avx_128_fma_double
543 (t_nblist * gmx_restrict nlist,
544 rvec * gmx_restrict xx,
545 rvec * gmx_restrict ff,
546 t_forcerec * gmx_restrict fr,
547 t_mdatoms * gmx_restrict mdatoms,
548 nb_kernel_data_t * gmx_restrict kernel_data,
549 t_nrnb * gmx_restrict nrnb)
551 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
552 * just 0 for non-waters.
553 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
554 * jnr indices corresponding to data put in the four positions in the SIMD register.
556 int i_shift_offset,i_coord_offset,outeriter,inneriter;
557 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
559 int j_coord_offsetA,j_coord_offsetB;
560 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
562 real *shiftvec,*fshift,*x,*f;
563 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
565 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
567 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
569 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
570 int vdwjidx0A,vdwjidx0B;
571 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
572 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
573 __m128d dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
574 __m128d dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
575 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
578 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
581 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
582 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
584 __m128i ifour = _mm_set1_epi32(4);
585 __m128d rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
587 __m128d dummy_mask,cutoff_mask;
588 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
589 __m128d one = _mm_set1_pd(1.0);
590 __m128d two = _mm_set1_pd(2.0);
596 jindex = nlist->jindex;
598 shiftidx = nlist->shift;
600 shiftvec = fr->shift_vec[0];
601 fshift = fr->fshift[0];
602 facel = _mm_set1_pd(fr->epsfac);
603 charge = mdatoms->chargeA;
604 krf = _mm_set1_pd(fr->ic->k_rf);
605 krf2 = _mm_set1_pd(fr->ic->k_rf*2.0);
606 crf = _mm_set1_pd(fr->ic->c_rf);
607 nvdwtype = fr->ntype;
609 vdwtype = mdatoms->typeA;
611 vftab = kernel_data->table_vdw->data;
612 vftabscale = _mm_set1_pd(kernel_data->table_vdw->scale);
614 /* Setup water-specific parameters */
615 inr = nlist->iinr[0];
616 iq0 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
617 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
618 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
619 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
621 /* Avoid stupid compiler warnings */
629 /* Start outer loop over neighborlists */
630 for(iidx=0; iidx<nri; iidx++)
632 /* Load shift vector for this list */
633 i_shift_offset = DIM*shiftidx[iidx];
635 /* Load limits for loop over neighbors */
636 j_index_start = jindex[iidx];
637 j_index_end = jindex[iidx+1];
639 /* Get outer coordinate index */
641 i_coord_offset = DIM*inr;
643 /* Load i particle coords and add shift vector */
644 gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
645 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
647 fix0 = _mm_setzero_pd();
648 fiy0 = _mm_setzero_pd();
649 fiz0 = _mm_setzero_pd();
650 fix1 = _mm_setzero_pd();
651 fiy1 = _mm_setzero_pd();
652 fiz1 = _mm_setzero_pd();
653 fix2 = _mm_setzero_pd();
654 fiy2 = _mm_setzero_pd();
655 fiz2 = _mm_setzero_pd();
657 /* Start inner kernel loop */
658 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
661 /* Get j neighbor index, and coordinate index */
664 j_coord_offsetA = DIM*jnrA;
665 j_coord_offsetB = DIM*jnrB;
667 /* load j atom coordinates */
668 gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
671 /* Calculate displacement vector */
672 dx00 = _mm_sub_pd(ix0,jx0);
673 dy00 = _mm_sub_pd(iy0,jy0);
674 dz00 = _mm_sub_pd(iz0,jz0);
675 dx10 = _mm_sub_pd(ix1,jx0);
676 dy10 = _mm_sub_pd(iy1,jy0);
677 dz10 = _mm_sub_pd(iz1,jz0);
678 dx20 = _mm_sub_pd(ix2,jx0);
679 dy20 = _mm_sub_pd(iy2,jy0);
680 dz20 = _mm_sub_pd(iz2,jz0);
682 /* Calculate squared distance and things based on it */
683 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
684 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
685 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
687 rinv00 = gmx_mm_invsqrt_pd(rsq00);
688 rinv10 = gmx_mm_invsqrt_pd(rsq10);
689 rinv20 = gmx_mm_invsqrt_pd(rsq20);
691 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
692 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
693 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
695 /* Load parameters for j particles */
696 jq0 = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
697 vdwjidx0A = 2*vdwtype[jnrA+0];
698 vdwjidx0B = 2*vdwtype[jnrB+0];
700 fjx0 = _mm_setzero_pd();
701 fjy0 = _mm_setzero_pd();
702 fjz0 = _mm_setzero_pd();
704 /**************************
705 * CALCULATE INTERACTIONS *
706 **************************/
708 r00 = _mm_mul_pd(rsq00,rinv00);
710 /* Compute parameters for interactions between i and j atoms */
711 qq00 = _mm_mul_pd(iq0,jq0);
712 gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
713 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
715 /* Calculate table index by multiplying r with table scale and truncate to integer */
716 rt = _mm_mul_pd(r00,vftabscale);
717 vfitab = _mm_cvttpd_epi32(rt);
719 vfeps = _mm_frcz_pd(rt);
721 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
723 twovfeps = _mm_add_pd(vfeps,vfeps);
724 vfitab = _mm_slli_epi32(vfitab,3);
726 /* REACTION-FIELD ELECTROSTATICS */
727 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
729 /* CUBIC SPLINE TABLE DISPERSION */
730 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
731 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
732 GMX_MM_TRANSPOSE2_PD(Y,F);
733 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
734 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
735 GMX_MM_TRANSPOSE2_PD(G,H);
736 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
737 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
738 fvdw6 = _mm_mul_pd(c6_00,FF);
740 /* CUBIC SPLINE TABLE REPULSION */
741 vfitab = _mm_add_epi32(vfitab,ifour);
742 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
743 F = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
744 GMX_MM_TRANSPOSE2_PD(Y,F);
745 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
746 H = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
747 GMX_MM_TRANSPOSE2_PD(G,H);
748 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
749 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
750 fvdw12 = _mm_mul_pd(c12_00,FF);
751 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
753 fscal = _mm_add_pd(felec,fvdw);
755 /* Update vectorial force */
756 fix0 = _mm_macc_pd(dx00,fscal,fix0);
757 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
758 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
760 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
761 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
762 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
764 /**************************
765 * CALCULATE INTERACTIONS *
766 **************************/
768 /* Compute parameters for interactions between i and j atoms */
769 qq10 = _mm_mul_pd(iq1,jq0);
771 /* REACTION-FIELD ELECTROSTATICS */
772 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
776 /* Update vectorial force */
777 fix1 = _mm_macc_pd(dx10,fscal,fix1);
778 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
779 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
781 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
782 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
783 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
785 /**************************
786 * CALCULATE INTERACTIONS *
787 **************************/
789 /* Compute parameters for interactions between i and j atoms */
790 qq20 = _mm_mul_pd(iq2,jq0);
792 /* REACTION-FIELD ELECTROSTATICS */
793 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
797 /* Update vectorial force */
798 fix2 = _mm_macc_pd(dx20,fscal,fix2);
799 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
800 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
802 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
803 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
804 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
806 gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
808 /* Inner loop uses 120 flops */
815 j_coord_offsetA = DIM*jnrA;
817 /* load j atom coordinates */
818 gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
821 /* Calculate displacement vector */
822 dx00 = _mm_sub_pd(ix0,jx0);
823 dy00 = _mm_sub_pd(iy0,jy0);
824 dz00 = _mm_sub_pd(iz0,jz0);
825 dx10 = _mm_sub_pd(ix1,jx0);
826 dy10 = _mm_sub_pd(iy1,jy0);
827 dz10 = _mm_sub_pd(iz1,jz0);
828 dx20 = _mm_sub_pd(ix2,jx0);
829 dy20 = _mm_sub_pd(iy2,jy0);
830 dz20 = _mm_sub_pd(iz2,jz0);
832 /* Calculate squared distance and things based on it */
833 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
834 rsq10 = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
835 rsq20 = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
837 rinv00 = gmx_mm_invsqrt_pd(rsq00);
838 rinv10 = gmx_mm_invsqrt_pd(rsq10);
839 rinv20 = gmx_mm_invsqrt_pd(rsq20);
841 rinvsq00 = _mm_mul_pd(rinv00,rinv00);
842 rinvsq10 = _mm_mul_pd(rinv10,rinv10);
843 rinvsq20 = _mm_mul_pd(rinv20,rinv20);
845 /* Load parameters for j particles */
846 jq0 = _mm_load_sd(charge+jnrA+0);
847 vdwjidx0A = 2*vdwtype[jnrA+0];
849 fjx0 = _mm_setzero_pd();
850 fjy0 = _mm_setzero_pd();
851 fjz0 = _mm_setzero_pd();
853 /**************************
854 * CALCULATE INTERACTIONS *
855 **************************/
857 r00 = _mm_mul_pd(rsq00,rinv00);
859 /* Compute parameters for interactions between i and j atoms */
860 qq00 = _mm_mul_pd(iq0,jq0);
861 gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
863 /* Calculate table index by multiplying r with table scale and truncate to integer */
864 rt = _mm_mul_pd(r00,vftabscale);
865 vfitab = _mm_cvttpd_epi32(rt);
867 vfeps = _mm_frcz_pd(rt);
869 vfeps = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
871 twovfeps = _mm_add_pd(vfeps,vfeps);
872 vfitab = _mm_slli_epi32(vfitab,3);
874 /* REACTION-FIELD ELECTROSTATICS */
875 felec = _mm_mul_pd(qq00,_mm_msub_pd(rinv00,rinvsq00,krf2));
877 /* CUBIC SPLINE TABLE DISPERSION */
878 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
879 F = _mm_setzero_pd();
880 GMX_MM_TRANSPOSE2_PD(Y,F);
881 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
882 H = _mm_setzero_pd();
883 GMX_MM_TRANSPOSE2_PD(G,H);
884 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
885 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
886 fvdw6 = _mm_mul_pd(c6_00,FF);
888 /* CUBIC SPLINE TABLE REPULSION */
889 vfitab = _mm_add_epi32(vfitab,ifour);
890 Y = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
891 F = _mm_setzero_pd();
892 GMX_MM_TRANSPOSE2_PD(Y,F);
893 G = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
894 H = _mm_setzero_pd();
895 GMX_MM_TRANSPOSE2_PD(G,H);
896 Fp = _mm_macc_pd(vfeps,_mm_macc_pd(H,vfeps,G),F);
897 FF = _mm_macc_pd(vfeps,_mm_macc_pd(twovfeps,H,G),Fp);
898 fvdw12 = _mm_mul_pd(c12_00,FF);
899 fvdw = _mm_xor_pd(signbit,_mm_mul_pd(_mm_add_pd(fvdw6,fvdw12),_mm_mul_pd(vftabscale,rinv00)));
901 fscal = _mm_add_pd(felec,fvdw);
903 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
905 /* Update vectorial force */
906 fix0 = _mm_macc_pd(dx00,fscal,fix0);
907 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
908 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
910 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
911 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
912 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
914 /**************************
915 * CALCULATE INTERACTIONS *
916 **************************/
918 /* Compute parameters for interactions between i and j atoms */
919 qq10 = _mm_mul_pd(iq1,jq0);
921 /* REACTION-FIELD ELECTROSTATICS */
922 felec = _mm_mul_pd(qq10,_mm_msub_pd(rinv10,rinvsq10,krf2));
926 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
928 /* Update vectorial force */
929 fix1 = _mm_macc_pd(dx10,fscal,fix1);
930 fiy1 = _mm_macc_pd(dy10,fscal,fiy1);
931 fiz1 = _mm_macc_pd(dz10,fscal,fiz1);
933 fjx0 = _mm_macc_pd(dx10,fscal,fjx0);
934 fjy0 = _mm_macc_pd(dy10,fscal,fjy0);
935 fjz0 = _mm_macc_pd(dz10,fscal,fjz0);
937 /**************************
938 * CALCULATE INTERACTIONS *
939 **************************/
941 /* Compute parameters for interactions between i and j atoms */
942 qq20 = _mm_mul_pd(iq2,jq0);
944 /* REACTION-FIELD ELECTROSTATICS */
945 felec = _mm_mul_pd(qq20,_mm_msub_pd(rinv20,rinvsq20,krf2));
949 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
951 /* Update vectorial force */
952 fix2 = _mm_macc_pd(dx20,fscal,fix2);
953 fiy2 = _mm_macc_pd(dy20,fscal,fiy2);
954 fiz2 = _mm_macc_pd(dz20,fscal,fiz2);
956 fjx0 = _mm_macc_pd(dx20,fscal,fjx0);
957 fjy0 = _mm_macc_pd(dy20,fscal,fjy0);
958 fjz0 = _mm_macc_pd(dz20,fscal,fjz0);
960 gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
962 /* Inner loop uses 120 flops */
965 /* End of innermost loop */
967 gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
968 f+i_coord_offset,fshift+i_shift_offset);
970 /* Increment number of inner iterations */
971 inneriter += j_index_end - j_index_start;
973 /* Outer loop uses 18 flops */
976 /* Increment number of outer iterations */
979 /* Update outer/inner flops */
981 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*18 + inneriter*120);