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_ElecEw_VdwLJ_GeomW4W4_VF_avx_128_fma_double
38 * Electrostatics interaction: Ewald
39 * VdW interaction: LennardJones
40 * Geometry: Water4-Water4
41 * Calculate force/pot: PotentialAndForce
44 nb_kernel_ElecEw_VdwLJ_GeomW4W4_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;
73 __m128d ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
74 int vdwjidx0A,vdwjidx0B;
75 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
76 int vdwjidx1A,vdwjidx1B;
77 __m128d jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
78 int vdwjidx2A,vdwjidx2B;
79 __m128d jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
80 int vdwjidx3A,vdwjidx3B;
81 __m128d jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
82 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
83 __m128d dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
84 __m128d dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
85 __m128d dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
86 __m128d dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
87 __m128d dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
88 __m128d dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
89 __m128d dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
90 __m128d dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
91 __m128d dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
92 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
95 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
98 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
99 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
101 __m128d ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
103 __m128d dummy_mask,cutoff_mask;
104 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
105 __m128d one = _mm_set1_pd(1.0);
106 __m128d two = _mm_set1_pd(2.0);
112 jindex = nlist->jindex;
114 shiftidx = nlist->shift;
116 shiftvec = fr->shift_vec[0];
117 fshift = fr->fshift[0];
118 facel = _mm_set1_pd(fr->epsfac);
119 charge = mdatoms->chargeA;
120 nvdwtype = fr->ntype;
122 vdwtype = mdatoms->typeA;
124 sh_ewald = _mm_set1_pd(fr->ic->sh_ewald);
125 ewtab = fr->ic->tabq_coul_FDV0;
126 ewtabscale = _mm_set1_pd(fr->ic->tabq_scale);
127 ewtabhalfspace = _mm_set1_pd(0.5/fr->ic->tabq_scale);
129 /* Setup water-specific parameters */
130 inr = nlist->iinr[0];
131 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
132 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
133 iq3 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+3]));
134 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
136 jq1 = _mm_set1_pd(charge[inr+1]);
137 jq2 = _mm_set1_pd(charge[inr+2]);
138 jq3 = _mm_set1_pd(charge[inr+3]);
139 vdwjidx0A = 2*vdwtype[inr+0];
140 c6_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A]);
141 c12_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A+1]);
142 qq11 = _mm_mul_pd(iq1,jq1);
143 qq12 = _mm_mul_pd(iq1,jq2);
144 qq13 = _mm_mul_pd(iq1,jq3);
145 qq21 = _mm_mul_pd(iq2,jq1);
146 qq22 = _mm_mul_pd(iq2,jq2);
147 qq23 = _mm_mul_pd(iq2,jq3);
148 qq31 = _mm_mul_pd(iq3,jq1);
149 qq32 = _mm_mul_pd(iq3,jq2);
150 qq33 = _mm_mul_pd(iq3,jq3);
152 /* Avoid stupid compiler warnings */
160 /* Start outer loop over neighborlists */
161 for(iidx=0; iidx<nri; iidx++)
163 /* Load shift vector for this list */
164 i_shift_offset = DIM*shiftidx[iidx];
166 /* Load limits for loop over neighbors */
167 j_index_start = jindex[iidx];
168 j_index_end = jindex[iidx+1];
170 /* Get outer coordinate index */
172 i_coord_offset = DIM*inr;
174 /* Load i particle coords and add shift vector */
175 gmx_mm_load_shift_and_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
176 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
178 fix0 = _mm_setzero_pd();
179 fiy0 = _mm_setzero_pd();
180 fiz0 = _mm_setzero_pd();
181 fix1 = _mm_setzero_pd();
182 fiy1 = _mm_setzero_pd();
183 fiz1 = _mm_setzero_pd();
184 fix2 = _mm_setzero_pd();
185 fiy2 = _mm_setzero_pd();
186 fiz2 = _mm_setzero_pd();
187 fix3 = _mm_setzero_pd();
188 fiy3 = _mm_setzero_pd();
189 fiz3 = _mm_setzero_pd();
191 /* Reset potential sums */
192 velecsum = _mm_setzero_pd();
193 vvdwsum = _mm_setzero_pd();
195 /* Start inner kernel loop */
196 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
199 /* Get j neighbor index, and coordinate index */
202 j_coord_offsetA = DIM*jnrA;
203 j_coord_offsetB = DIM*jnrB;
205 /* load j atom coordinates */
206 gmx_mm_load_4rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
207 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
208 &jy2,&jz2,&jx3,&jy3,&jz3);
210 /* Calculate displacement vector */
211 dx00 = _mm_sub_pd(ix0,jx0);
212 dy00 = _mm_sub_pd(iy0,jy0);
213 dz00 = _mm_sub_pd(iz0,jz0);
214 dx11 = _mm_sub_pd(ix1,jx1);
215 dy11 = _mm_sub_pd(iy1,jy1);
216 dz11 = _mm_sub_pd(iz1,jz1);
217 dx12 = _mm_sub_pd(ix1,jx2);
218 dy12 = _mm_sub_pd(iy1,jy2);
219 dz12 = _mm_sub_pd(iz1,jz2);
220 dx13 = _mm_sub_pd(ix1,jx3);
221 dy13 = _mm_sub_pd(iy1,jy3);
222 dz13 = _mm_sub_pd(iz1,jz3);
223 dx21 = _mm_sub_pd(ix2,jx1);
224 dy21 = _mm_sub_pd(iy2,jy1);
225 dz21 = _mm_sub_pd(iz2,jz1);
226 dx22 = _mm_sub_pd(ix2,jx2);
227 dy22 = _mm_sub_pd(iy2,jy2);
228 dz22 = _mm_sub_pd(iz2,jz2);
229 dx23 = _mm_sub_pd(ix2,jx3);
230 dy23 = _mm_sub_pd(iy2,jy3);
231 dz23 = _mm_sub_pd(iz2,jz3);
232 dx31 = _mm_sub_pd(ix3,jx1);
233 dy31 = _mm_sub_pd(iy3,jy1);
234 dz31 = _mm_sub_pd(iz3,jz1);
235 dx32 = _mm_sub_pd(ix3,jx2);
236 dy32 = _mm_sub_pd(iy3,jy2);
237 dz32 = _mm_sub_pd(iz3,jz2);
238 dx33 = _mm_sub_pd(ix3,jx3);
239 dy33 = _mm_sub_pd(iy3,jy3);
240 dz33 = _mm_sub_pd(iz3,jz3);
242 /* Calculate squared distance and things based on it */
243 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
244 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
245 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
246 rsq13 = gmx_mm_calc_rsq_pd(dx13,dy13,dz13);
247 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
248 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
249 rsq23 = gmx_mm_calc_rsq_pd(dx23,dy23,dz23);
250 rsq31 = gmx_mm_calc_rsq_pd(dx31,dy31,dz31);
251 rsq32 = gmx_mm_calc_rsq_pd(dx32,dy32,dz32);
252 rsq33 = gmx_mm_calc_rsq_pd(dx33,dy33,dz33);
254 rinv11 = gmx_mm_invsqrt_pd(rsq11);
255 rinv12 = gmx_mm_invsqrt_pd(rsq12);
256 rinv13 = gmx_mm_invsqrt_pd(rsq13);
257 rinv21 = gmx_mm_invsqrt_pd(rsq21);
258 rinv22 = gmx_mm_invsqrt_pd(rsq22);
259 rinv23 = gmx_mm_invsqrt_pd(rsq23);
260 rinv31 = gmx_mm_invsqrt_pd(rsq31);
261 rinv32 = gmx_mm_invsqrt_pd(rsq32);
262 rinv33 = gmx_mm_invsqrt_pd(rsq33);
264 rinvsq00 = gmx_mm_inv_pd(rsq00);
265 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
266 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
267 rinvsq13 = _mm_mul_pd(rinv13,rinv13);
268 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
269 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
270 rinvsq23 = _mm_mul_pd(rinv23,rinv23);
271 rinvsq31 = _mm_mul_pd(rinv31,rinv31);
272 rinvsq32 = _mm_mul_pd(rinv32,rinv32);
273 rinvsq33 = _mm_mul_pd(rinv33,rinv33);
275 fjx0 = _mm_setzero_pd();
276 fjy0 = _mm_setzero_pd();
277 fjz0 = _mm_setzero_pd();
278 fjx1 = _mm_setzero_pd();
279 fjy1 = _mm_setzero_pd();
280 fjz1 = _mm_setzero_pd();
281 fjx2 = _mm_setzero_pd();
282 fjy2 = _mm_setzero_pd();
283 fjz2 = _mm_setzero_pd();
284 fjx3 = _mm_setzero_pd();
285 fjy3 = _mm_setzero_pd();
286 fjz3 = _mm_setzero_pd();
288 /**************************
289 * CALCULATE INTERACTIONS *
290 **************************/
292 /* LENNARD-JONES DISPERSION/REPULSION */
294 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
295 vvdw6 = _mm_mul_pd(c6_00,rinvsix);
296 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
297 vvdw = _mm_msub_pd( vvdw12,one_twelfth, _mm_mul_pd(vvdw6,one_sixth) );
298 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
300 /* Update potential sum for this i atom from the interaction with this j atom. */
301 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
305 /* Update vectorial force */
306 fix0 = _mm_macc_pd(dx00,fscal,fix0);
307 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
308 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
310 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
311 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
312 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
314 /**************************
315 * CALCULATE INTERACTIONS *
316 **************************/
318 r11 = _mm_mul_pd(rsq11,rinv11);
320 /* EWALD ELECTROSTATICS */
322 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
323 ewrt = _mm_mul_pd(r11,ewtabscale);
324 ewitab = _mm_cvttpd_epi32(ewrt);
326 eweps = _mm_frcz_pd(ewrt);
328 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
330 twoeweps = _mm_add_pd(eweps,eweps);
331 ewitab = _mm_slli_epi32(ewitab,2);
332 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
333 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
334 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
335 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
336 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
337 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
338 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
339 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
340 velec = _mm_mul_pd(qq11,_mm_sub_pd(rinv11,velec));
341 felec = _mm_mul_pd(_mm_mul_pd(qq11,rinv11),_mm_sub_pd(rinvsq11,felec));
343 /* Update potential sum for this i atom from the interaction with this j atom. */
344 velecsum = _mm_add_pd(velecsum,velec);
348 /* Update vectorial force */
349 fix1 = _mm_macc_pd(dx11,fscal,fix1);
350 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
351 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
353 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
354 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
355 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
357 /**************************
358 * CALCULATE INTERACTIONS *
359 **************************/
361 r12 = _mm_mul_pd(rsq12,rinv12);
363 /* EWALD ELECTROSTATICS */
365 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
366 ewrt = _mm_mul_pd(r12,ewtabscale);
367 ewitab = _mm_cvttpd_epi32(ewrt);
369 eweps = _mm_frcz_pd(ewrt);
371 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
373 twoeweps = _mm_add_pd(eweps,eweps);
374 ewitab = _mm_slli_epi32(ewitab,2);
375 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
376 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
377 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
378 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
379 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
380 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
381 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
382 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
383 velec = _mm_mul_pd(qq12,_mm_sub_pd(rinv12,velec));
384 felec = _mm_mul_pd(_mm_mul_pd(qq12,rinv12),_mm_sub_pd(rinvsq12,felec));
386 /* Update potential sum for this i atom from the interaction with this j atom. */
387 velecsum = _mm_add_pd(velecsum,velec);
391 /* Update vectorial force */
392 fix1 = _mm_macc_pd(dx12,fscal,fix1);
393 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
394 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
396 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
397 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
398 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
400 /**************************
401 * CALCULATE INTERACTIONS *
402 **************************/
404 r13 = _mm_mul_pd(rsq13,rinv13);
406 /* EWALD ELECTROSTATICS */
408 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
409 ewrt = _mm_mul_pd(r13,ewtabscale);
410 ewitab = _mm_cvttpd_epi32(ewrt);
412 eweps = _mm_frcz_pd(ewrt);
414 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
416 twoeweps = _mm_add_pd(eweps,eweps);
417 ewitab = _mm_slli_epi32(ewitab,2);
418 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
419 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
420 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
421 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
422 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
423 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
424 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
425 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
426 velec = _mm_mul_pd(qq13,_mm_sub_pd(rinv13,velec));
427 felec = _mm_mul_pd(_mm_mul_pd(qq13,rinv13),_mm_sub_pd(rinvsq13,felec));
429 /* Update potential sum for this i atom from the interaction with this j atom. */
430 velecsum = _mm_add_pd(velecsum,velec);
434 /* Update vectorial force */
435 fix1 = _mm_macc_pd(dx13,fscal,fix1);
436 fiy1 = _mm_macc_pd(dy13,fscal,fiy1);
437 fiz1 = _mm_macc_pd(dz13,fscal,fiz1);
439 fjx3 = _mm_macc_pd(dx13,fscal,fjx3);
440 fjy3 = _mm_macc_pd(dy13,fscal,fjy3);
441 fjz3 = _mm_macc_pd(dz13,fscal,fjz3);
443 /**************************
444 * CALCULATE INTERACTIONS *
445 **************************/
447 r21 = _mm_mul_pd(rsq21,rinv21);
449 /* EWALD ELECTROSTATICS */
451 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
452 ewrt = _mm_mul_pd(r21,ewtabscale);
453 ewitab = _mm_cvttpd_epi32(ewrt);
455 eweps = _mm_frcz_pd(ewrt);
457 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
459 twoeweps = _mm_add_pd(eweps,eweps);
460 ewitab = _mm_slli_epi32(ewitab,2);
461 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
462 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
463 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
464 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
465 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
466 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
467 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
468 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
469 velec = _mm_mul_pd(qq21,_mm_sub_pd(rinv21,velec));
470 felec = _mm_mul_pd(_mm_mul_pd(qq21,rinv21),_mm_sub_pd(rinvsq21,felec));
472 /* Update potential sum for this i atom from the interaction with this j atom. */
473 velecsum = _mm_add_pd(velecsum,velec);
477 /* Update vectorial force */
478 fix2 = _mm_macc_pd(dx21,fscal,fix2);
479 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
480 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
482 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
483 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
484 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
486 /**************************
487 * CALCULATE INTERACTIONS *
488 **************************/
490 r22 = _mm_mul_pd(rsq22,rinv22);
492 /* EWALD ELECTROSTATICS */
494 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
495 ewrt = _mm_mul_pd(r22,ewtabscale);
496 ewitab = _mm_cvttpd_epi32(ewrt);
498 eweps = _mm_frcz_pd(ewrt);
500 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
502 twoeweps = _mm_add_pd(eweps,eweps);
503 ewitab = _mm_slli_epi32(ewitab,2);
504 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
505 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
506 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
507 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
508 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
509 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
510 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
511 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
512 velec = _mm_mul_pd(qq22,_mm_sub_pd(rinv22,velec));
513 felec = _mm_mul_pd(_mm_mul_pd(qq22,rinv22),_mm_sub_pd(rinvsq22,felec));
515 /* Update potential sum for this i atom from the interaction with this j atom. */
516 velecsum = _mm_add_pd(velecsum,velec);
520 /* Update vectorial force */
521 fix2 = _mm_macc_pd(dx22,fscal,fix2);
522 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
523 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
525 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
526 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
527 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
529 /**************************
530 * CALCULATE INTERACTIONS *
531 **************************/
533 r23 = _mm_mul_pd(rsq23,rinv23);
535 /* EWALD ELECTROSTATICS */
537 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
538 ewrt = _mm_mul_pd(r23,ewtabscale);
539 ewitab = _mm_cvttpd_epi32(ewrt);
541 eweps = _mm_frcz_pd(ewrt);
543 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
545 twoeweps = _mm_add_pd(eweps,eweps);
546 ewitab = _mm_slli_epi32(ewitab,2);
547 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
548 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
549 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
550 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
551 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
552 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
553 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
554 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
555 velec = _mm_mul_pd(qq23,_mm_sub_pd(rinv23,velec));
556 felec = _mm_mul_pd(_mm_mul_pd(qq23,rinv23),_mm_sub_pd(rinvsq23,felec));
558 /* Update potential sum for this i atom from the interaction with this j atom. */
559 velecsum = _mm_add_pd(velecsum,velec);
563 /* Update vectorial force */
564 fix2 = _mm_macc_pd(dx23,fscal,fix2);
565 fiy2 = _mm_macc_pd(dy23,fscal,fiy2);
566 fiz2 = _mm_macc_pd(dz23,fscal,fiz2);
568 fjx3 = _mm_macc_pd(dx23,fscal,fjx3);
569 fjy3 = _mm_macc_pd(dy23,fscal,fjy3);
570 fjz3 = _mm_macc_pd(dz23,fscal,fjz3);
572 /**************************
573 * CALCULATE INTERACTIONS *
574 **************************/
576 r31 = _mm_mul_pd(rsq31,rinv31);
578 /* EWALD ELECTROSTATICS */
580 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
581 ewrt = _mm_mul_pd(r31,ewtabscale);
582 ewitab = _mm_cvttpd_epi32(ewrt);
584 eweps = _mm_frcz_pd(ewrt);
586 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
588 twoeweps = _mm_add_pd(eweps,eweps);
589 ewitab = _mm_slli_epi32(ewitab,2);
590 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
591 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
592 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
593 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
594 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
595 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
596 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
597 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
598 velec = _mm_mul_pd(qq31,_mm_sub_pd(rinv31,velec));
599 felec = _mm_mul_pd(_mm_mul_pd(qq31,rinv31),_mm_sub_pd(rinvsq31,felec));
601 /* Update potential sum for this i atom from the interaction with this j atom. */
602 velecsum = _mm_add_pd(velecsum,velec);
606 /* Update vectorial force */
607 fix3 = _mm_macc_pd(dx31,fscal,fix3);
608 fiy3 = _mm_macc_pd(dy31,fscal,fiy3);
609 fiz3 = _mm_macc_pd(dz31,fscal,fiz3);
611 fjx1 = _mm_macc_pd(dx31,fscal,fjx1);
612 fjy1 = _mm_macc_pd(dy31,fscal,fjy1);
613 fjz1 = _mm_macc_pd(dz31,fscal,fjz1);
615 /**************************
616 * CALCULATE INTERACTIONS *
617 **************************/
619 r32 = _mm_mul_pd(rsq32,rinv32);
621 /* EWALD ELECTROSTATICS */
623 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
624 ewrt = _mm_mul_pd(r32,ewtabscale);
625 ewitab = _mm_cvttpd_epi32(ewrt);
627 eweps = _mm_frcz_pd(ewrt);
629 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
631 twoeweps = _mm_add_pd(eweps,eweps);
632 ewitab = _mm_slli_epi32(ewitab,2);
633 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
634 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
635 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
636 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
637 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
638 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
639 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
640 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
641 velec = _mm_mul_pd(qq32,_mm_sub_pd(rinv32,velec));
642 felec = _mm_mul_pd(_mm_mul_pd(qq32,rinv32),_mm_sub_pd(rinvsq32,felec));
644 /* Update potential sum for this i atom from the interaction with this j atom. */
645 velecsum = _mm_add_pd(velecsum,velec);
649 /* Update vectorial force */
650 fix3 = _mm_macc_pd(dx32,fscal,fix3);
651 fiy3 = _mm_macc_pd(dy32,fscal,fiy3);
652 fiz3 = _mm_macc_pd(dz32,fscal,fiz3);
654 fjx2 = _mm_macc_pd(dx32,fscal,fjx2);
655 fjy2 = _mm_macc_pd(dy32,fscal,fjy2);
656 fjz2 = _mm_macc_pd(dz32,fscal,fjz2);
658 /**************************
659 * CALCULATE INTERACTIONS *
660 **************************/
662 r33 = _mm_mul_pd(rsq33,rinv33);
664 /* EWALD ELECTROSTATICS */
666 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
667 ewrt = _mm_mul_pd(r33,ewtabscale);
668 ewitab = _mm_cvttpd_epi32(ewrt);
670 eweps = _mm_frcz_pd(ewrt);
672 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
674 twoeweps = _mm_add_pd(eweps,eweps);
675 ewitab = _mm_slli_epi32(ewitab,2);
676 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
677 ewtabD = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
678 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
679 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
680 ewtabFn = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,1) +2);
681 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
682 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
683 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
684 velec = _mm_mul_pd(qq33,_mm_sub_pd(rinv33,velec));
685 felec = _mm_mul_pd(_mm_mul_pd(qq33,rinv33),_mm_sub_pd(rinvsq33,felec));
687 /* Update potential sum for this i atom from the interaction with this j atom. */
688 velecsum = _mm_add_pd(velecsum,velec);
692 /* Update vectorial force */
693 fix3 = _mm_macc_pd(dx33,fscal,fix3);
694 fiy3 = _mm_macc_pd(dy33,fscal,fiy3);
695 fiz3 = _mm_macc_pd(dz33,fscal,fiz3);
697 fjx3 = _mm_macc_pd(dx33,fscal,fjx3);
698 fjy3 = _mm_macc_pd(dy33,fscal,fjy3);
699 fjz3 = _mm_macc_pd(dz33,fscal,fjz3);
701 gmx_mm_decrement_4rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
703 /* Inner loop uses 434 flops */
710 j_coord_offsetA = DIM*jnrA;
712 /* load j atom coordinates */
713 gmx_mm_load_4rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
714 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
715 &jy2,&jz2,&jx3,&jy3,&jz3);
717 /* Calculate displacement vector */
718 dx00 = _mm_sub_pd(ix0,jx0);
719 dy00 = _mm_sub_pd(iy0,jy0);
720 dz00 = _mm_sub_pd(iz0,jz0);
721 dx11 = _mm_sub_pd(ix1,jx1);
722 dy11 = _mm_sub_pd(iy1,jy1);
723 dz11 = _mm_sub_pd(iz1,jz1);
724 dx12 = _mm_sub_pd(ix1,jx2);
725 dy12 = _mm_sub_pd(iy1,jy2);
726 dz12 = _mm_sub_pd(iz1,jz2);
727 dx13 = _mm_sub_pd(ix1,jx3);
728 dy13 = _mm_sub_pd(iy1,jy3);
729 dz13 = _mm_sub_pd(iz1,jz3);
730 dx21 = _mm_sub_pd(ix2,jx1);
731 dy21 = _mm_sub_pd(iy2,jy1);
732 dz21 = _mm_sub_pd(iz2,jz1);
733 dx22 = _mm_sub_pd(ix2,jx2);
734 dy22 = _mm_sub_pd(iy2,jy2);
735 dz22 = _mm_sub_pd(iz2,jz2);
736 dx23 = _mm_sub_pd(ix2,jx3);
737 dy23 = _mm_sub_pd(iy2,jy3);
738 dz23 = _mm_sub_pd(iz2,jz3);
739 dx31 = _mm_sub_pd(ix3,jx1);
740 dy31 = _mm_sub_pd(iy3,jy1);
741 dz31 = _mm_sub_pd(iz3,jz1);
742 dx32 = _mm_sub_pd(ix3,jx2);
743 dy32 = _mm_sub_pd(iy3,jy2);
744 dz32 = _mm_sub_pd(iz3,jz2);
745 dx33 = _mm_sub_pd(ix3,jx3);
746 dy33 = _mm_sub_pd(iy3,jy3);
747 dz33 = _mm_sub_pd(iz3,jz3);
749 /* Calculate squared distance and things based on it */
750 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
751 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
752 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
753 rsq13 = gmx_mm_calc_rsq_pd(dx13,dy13,dz13);
754 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
755 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
756 rsq23 = gmx_mm_calc_rsq_pd(dx23,dy23,dz23);
757 rsq31 = gmx_mm_calc_rsq_pd(dx31,dy31,dz31);
758 rsq32 = gmx_mm_calc_rsq_pd(dx32,dy32,dz32);
759 rsq33 = gmx_mm_calc_rsq_pd(dx33,dy33,dz33);
761 rinv11 = gmx_mm_invsqrt_pd(rsq11);
762 rinv12 = gmx_mm_invsqrt_pd(rsq12);
763 rinv13 = gmx_mm_invsqrt_pd(rsq13);
764 rinv21 = gmx_mm_invsqrt_pd(rsq21);
765 rinv22 = gmx_mm_invsqrt_pd(rsq22);
766 rinv23 = gmx_mm_invsqrt_pd(rsq23);
767 rinv31 = gmx_mm_invsqrt_pd(rsq31);
768 rinv32 = gmx_mm_invsqrt_pd(rsq32);
769 rinv33 = gmx_mm_invsqrt_pd(rsq33);
771 rinvsq00 = gmx_mm_inv_pd(rsq00);
772 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
773 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
774 rinvsq13 = _mm_mul_pd(rinv13,rinv13);
775 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
776 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
777 rinvsq23 = _mm_mul_pd(rinv23,rinv23);
778 rinvsq31 = _mm_mul_pd(rinv31,rinv31);
779 rinvsq32 = _mm_mul_pd(rinv32,rinv32);
780 rinvsq33 = _mm_mul_pd(rinv33,rinv33);
782 fjx0 = _mm_setzero_pd();
783 fjy0 = _mm_setzero_pd();
784 fjz0 = _mm_setzero_pd();
785 fjx1 = _mm_setzero_pd();
786 fjy1 = _mm_setzero_pd();
787 fjz1 = _mm_setzero_pd();
788 fjx2 = _mm_setzero_pd();
789 fjy2 = _mm_setzero_pd();
790 fjz2 = _mm_setzero_pd();
791 fjx3 = _mm_setzero_pd();
792 fjy3 = _mm_setzero_pd();
793 fjz3 = _mm_setzero_pd();
795 /**************************
796 * CALCULATE INTERACTIONS *
797 **************************/
799 /* LENNARD-JONES DISPERSION/REPULSION */
801 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
802 vvdw6 = _mm_mul_pd(c6_00,rinvsix);
803 vvdw12 = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
804 vvdw = _mm_msub_pd( vvdw12,one_twelfth, _mm_mul_pd(vvdw6,one_sixth) );
805 fvdw = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
807 /* Update potential sum for this i atom from the interaction with this j atom. */
808 vvdw = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
809 vvdwsum = _mm_add_pd(vvdwsum,vvdw);
813 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
815 /* Update vectorial force */
816 fix0 = _mm_macc_pd(dx00,fscal,fix0);
817 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
818 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
820 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
821 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
822 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
824 /**************************
825 * CALCULATE INTERACTIONS *
826 **************************/
828 r11 = _mm_mul_pd(rsq11,rinv11);
830 /* EWALD ELECTROSTATICS */
832 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
833 ewrt = _mm_mul_pd(r11,ewtabscale);
834 ewitab = _mm_cvttpd_epi32(ewrt);
836 eweps = _mm_frcz_pd(ewrt);
838 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
840 twoeweps = _mm_add_pd(eweps,eweps);
841 ewitab = _mm_slli_epi32(ewitab,2);
842 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
843 ewtabD = _mm_setzero_pd();
844 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
845 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
846 ewtabFn = _mm_setzero_pd();
847 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
848 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
849 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
850 velec = _mm_mul_pd(qq11,_mm_sub_pd(rinv11,velec));
851 felec = _mm_mul_pd(_mm_mul_pd(qq11,rinv11),_mm_sub_pd(rinvsq11,felec));
853 /* Update potential sum for this i atom from the interaction with this j atom. */
854 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
855 velecsum = _mm_add_pd(velecsum,velec);
859 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
861 /* Update vectorial force */
862 fix1 = _mm_macc_pd(dx11,fscal,fix1);
863 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
864 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
866 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
867 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
868 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
870 /**************************
871 * CALCULATE INTERACTIONS *
872 **************************/
874 r12 = _mm_mul_pd(rsq12,rinv12);
876 /* EWALD ELECTROSTATICS */
878 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
879 ewrt = _mm_mul_pd(r12,ewtabscale);
880 ewitab = _mm_cvttpd_epi32(ewrt);
882 eweps = _mm_frcz_pd(ewrt);
884 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
886 twoeweps = _mm_add_pd(eweps,eweps);
887 ewitab = _mm_slli_epi32(ewitab,2);
888 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
889 ewtabD = _mm_setzero_pd();
890 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
891 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
892 ewtabFn = _mm_setzero_pd();
893 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
894 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
895 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
896 velec = _mm_mul_pd(qq12,_mm_sub_pd(rinv12,velec));
897 felec = _mm_mul_pd(_mm_mul_pd(qq12,rinv12),_mm_sub_pd(rinvsq12,felec));
899 /* Update potential sum for this i atom from the interaction with this j atom. */
900 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
901 velecsum = _mm_add_pd(velecsum,velec);
905 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
907 /* Update vectorial force */
908 fix1 = _mm_macc_pd(dx12,fscal,fix1);
909 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
910 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
912 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
913 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
914 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
916 /**************************
917 * CALCULATE INTERACTIONS *
918 **************************/
920 r13 = _mm_mul_pd(rsq13,rinv13);
922 /* EWALD ELECTROSTATICS */
924 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
925 ewrt = _mm_mul_pd(r13,ewtabscale);
926 ewitab = _mm_cvttpd_epi32(ewrt);
928 eweps = _mm_frcz_pd(ewrt);
930 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
932 twoeweps = _mm_add_pd(eweps,eweps);
933 ewitab = _mm_slli_epi32(ewitab,2);
934 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
935 ewtabD = _mm_setzero_pd();
936 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
937 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
938 ewtabFn = _mm_setzero_pd();
939 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
940 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
941 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
942 velec = _mm_mul_pd(qq13,_mm_sub_pd(rinv13,velec));
943 felec = _mm_mul_pd(_mm_mul_pd(qq13,rinv13),_mm_sub_pd(rinvsq13,felec));
945 /* Update potential sum for this i atom from the interaction with this j atom. */
946 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
947 velecsum = _mm_add_pd(velecsum,velec);
951 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
953 /* Update vectorial force */
954 fix1 = _mm_macc_pd(dx13,fscal,fix1);
955 fiy1 = _mm_macc_pd(dy13,fscal,fiy1);
956 fiz1 = _mm_macc_pd(dz13,fscal,fiz1);
958 fjx3 = _mm_macc_pd(dx13,fscal,fjx3);
959 fjy3 = _mm_macc_pd(dy13,fscal,fjy3);
960 fjz3 = _mm_macc_pd(dz13,fscal,fjz3);
962 /**************************
963 * CALCULATE INTERACTIONS *
964 **************************/
966 r21 = _mm_mul_pd(rsq21,rinv21);
968 /* EWALD ELECTROSTATICS */
970 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
971 ewrt = _mm_mul_pd(r21,ewtabscale);
972 ewitab = _mm_cvttpd_epi32(ewrt);
974 eweps = _mm_frcz_pd(ewrt);
976 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
978 twoeweps = _mm_add_pd(eweps,eweps);
979 ewitab = _mm_slli_epi32(ewitab,2);
980 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
981 ewtabD = _mm_setzero_pd();
982 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
983 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
984 ewtabFn = _mm_setzero_pd();
985 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
986 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
987 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
988 velec = _mm_mul_pd(qq21,_mm_sub_pd(rinv21,velec));
989 felec = _mm_mul_pd(_mm_mul_pd(qq21,rinv21),_mm_sub_pd(rinvsq21,felec));
991 /* Update potential sum for this i atom from the interaction with this j atom. */
992 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
993 velecsum = _mm_add_pd(velecsum,velec);
997 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
999 /* Update vectorial force */
1000 fix2 = _mm_macc_pd(dx21,fscal,fix2);
1001 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
1002 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
1004 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
1005 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
1006 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
1008 /**************************
1009 * CALCULATE INTERACTIONS *
1010 **************************/
1012 r22 = _mm_mul_pd(rsq22,rinv22);
1014 /* EWALD ELECTROSTATICS */
1016 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1017 ewrt = _mm_mul_pd(r22,ewtabscale);
1018 ewitab = _mm_cvttpd_epi32(ewrt);
1020 eweps = _mm_frcz_pd(ewrt);
1022 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1024 twoeweps = _mm_add_pd(eweps,eweps);
1025 ewitab = _mm_slli_epi32(ewitab,2);
1026 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1027 ewtabD = _mm_setzero_pd();
1028 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1029 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
1030 ewtabFn = _mm_setzero_pd();
1031 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1032 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
1033 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
1034 velec = _mm_mul_pd(qq22,_mm_sub_pd(rinv22,velec));
1035 felec = _mm_mul_pd(_mm_mul_pd(qq22,rinv22),_mm_sub_pd(rinvsq22,felec));
1037 /* Update potential sum for this i atom from the interaction with this j atom. */
1038 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
1039 velecsum = _mm_add_pd(velecsum,velec);
1043 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1045 /* Update vectorial force */
1046 fix2 = _mm_macc_pd(dx22,fscal,fix2);
1047 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
1048 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
1050 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
1051 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
1052 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
1054 /**************************
1055 * CALCULATE INTERACTIONS *
1056 **************************/
1058 r23 = _mm_mul_pd(rsq23,rinv23);
1060 /* EWALD ELECTROSTATICS */
1062 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1063 ewrt = _mm_mul_pd(r23,ewtabscale);
1064 ewitab = _mm_cvttpd_epi32(ewrt);
1066 eweps = _mm_frcz_pd(ewrt);
1068 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1070 twoeweps = _mm_add_pd(eweps,eweps);
1071 ewitab = _mm_slli_epi32(ewitab,2);
1072 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1073 ewtabD = _mm_setzero_pd();
1074 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1075 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
1076 ewtabFn = _mm_setzero_pd();
1077 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1078 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
1079 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
1080 velec = _mm_mul_pd(qq23,_mm_sub_pd(rinv23,velec));
1081 felec = _mm_mul_pd(_mm_mul_pd(qq23,rinv23),_mm_sub_pd(rinvsq23,felec));
1083 /* Update potential sum for this i atom from the interaction with this j atom. */
1084 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
1085 velecsum = _mm_add_pd(velecsum,velec);
1089 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1091 /* Update vectorial force */
1092 fix2 = _mm_macc_pd(dx23,fscal,fix2);
1093 fiy2 = _mm_macc_pd(dy23,fscal,fiy2);
1094 fiz2 = _mm_macc_pd(dz23,fscal,fiz2);
1096 fjx3 = _mm_macc_pd(dx23,fscal,fjx3);
1097 fjy3 = _mm_macc_pd(dy23,fscal,fjy3);
1098 fjz3 = _mm_macc_pd(dz23,fscal,fjz3);
1100 /**************************
1101 * CALCULATE INTERACTIONS *
1102 **************************/
1104 r31 = _mm_mul_pd(rsq31,rinv31);
1106 /* EWALD ELECTROSTATICS */
1108 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1109 ewrt = _mm_mul_pd(r31,ewtabscale);
1110 ewitab = _mm_cvttpd_epi32(ewrt);
1112 eweps = _mm_frcz_pd(ewrt);
1114 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1116 twoeweps = _mm_add_pd(eweps,eweps);
1117 ewitab = _mm_slli_epi32(ewitab,2);
1118 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1119 ewtabD = _mm_setzero_pd();
1120 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1121 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
1122 ewtabFn = _mm_setzero_pd();
1123 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1124 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
1125 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
1126 velec = _mm_mul_pd(qq31,_mm_sub_pd(rinv31,velec));
1127 felec = _mm_mul_pd(_mm_mul_pd(qq31,rinv31),_mm_sub_pd(rinvsq31,felec));
1129 /* Update potential sum for this i atom from the interaction with this j atom. */
1130 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
1131 velecsum = _mm_add_pd(velecsum,velec);
1135 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1137 /* Update vectorial force */
1138 fix3 = _mm_macc_pd(dx31,fscal,fix3);
1139 fiy3 = _mm_macc_pd(dy31,fscal,fiy3);
1140 fiz3 = _mm_macc_pd(dz31,fscal,fiz3);
1142 fjx1 = _mm_macc_pd(dx31,fscal,fjx1);
1143 fjy1 = _mm_macc_pd(dy31,fscal,fjy1);
1144 fjz1 = _mm_macc_pd(dz31,fscal,fjz1);
1146 /**************************
1147 * CALCULATE INTERACTIONS *
1148 **************************/
1150 r32 = _mm_mul_pd(rsq32,rinv32);
1152 /* EWALD ELECTROSTATICS */
1154 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1155 ewrt = _mm_mul_pd(r32,ewtabscale);
1156 ewitab = _mm_cvttpd_epi32(ewrt);
1158 eweps = _mm_frcz_pd(ewrt);
1160 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1162 twoeweps = _mm_add_pd(eweps,eweps);
1163 ewitab = _mm_slli_epi32(ewitab,2);
1164 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1165 ewtabD = _mm_setzero_pd();
1166 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1167 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
1168 ewtabFn = _mm_setzero_pd();
1169 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1170 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
1171 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
1172 velec = _mm_mul_pd(qq32,_mm_sub_pd(rinv32,velec));
1173 felec = _mm_mul_pd(_mm_mul_pd(qq32,rinv32),_mm_sub_pd(rinvsq32,felec));
1175 /* Update potential sum for this i atom from the interaction with this j atom. */
1176 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
1177 velecsum = _mm_add_pd(velecsum,velec);
1181 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1183 /* Update vectorial force */
1184 fix3 = _mm_macc_pd(dx32,fscal,fix3);
1185 fiy3 = _mm_macc_pd(dy32,fscal,fiy3);
1186 fiz3 = _mm_macc_pd(dz32,fscal,fiz3);
1188 fjx2 = _mm_macc_pd(dx32,fscal,fjx2);
1189 fjy2 = _mm_macc_pd(dy32,fscal,fjy2);
1190 fjz2 = _mm_macc_pd(dz32,fscal,fjz2);
1192 /**************************
1193 * CALCULATE INTERACTIONS *
1194 **************************/
1196 r33 = _mm_mul_pd(rsq33,rinv33);
1198 /* EWALD ELECTROSTATICS */
1200 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1201 ewrt = _mm_mul_pd(r33,ewtabscale);
1202 ewitab = _mm_cvttpd_epi32(ewrt);
1204 eweps = _mm_frcz_pd(ewrt);
1206 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1208 twoeweps = _mm_add_pd(eweps,eweps);
1209 ewitab = _mm_slli_epi32(ewitab,2);
1210 ewtabF = _mm_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1211 ewtabD = _mm_setzero_pd();
1212 GMX_MM_TRANSPOSE2_PD(ewtabF,ewtabD);
1213 ewtabV = _mm_load_sd( ewtab + _mm_extract_epi32(ewitab,0) +2);
1214 ewtabFn = _mm_setzero_pd();
1215 GMX_MM_TRANSPOSE2_PD(ewtabV,ewtabFn);
1216 felec = _mm_macc_pd(eweps,ewtabD,ewtabF);
1217 velec = _mm_nmacc_pd(_mm_mul_pd(ewtabhalfspace,eweps) ,_mm_add_pd(ewtabF,felec), ewtabV);
1218 velec = _mm_mul_pd(qq33,_mm_sub_pd(rinv33,velec));
1219 felec = _mm_mul_pd(_mm_mul_pd(qq33,rinv33),_mm_sub_pd(rinvsq33,felec));
1221 /* Update potential sum for this i atom from the interaction with this j atom. */
1222 velec = _mm_unpacklo_pd(velec,_mm_setzero_pd());
1223 velecsum = _mm_add_pd(velecsum,velec);
1227 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1229 /* Update vectorial force */
1230 fix3 = _mm_macc_pd(dx33,fscal,fix3);
1231 fiy3 = _mm_macc_pd(dy33,fscal,fiy3);
1232 fiz3 = _mm_macc_pd(dz33,fscal,fiz3);
1234 fjx3 = _mm_macc_pd(dx33,fscal,fjx3);
1235 fjy3 = _mm_macc_pd(dy33,fscal,fjy3);
1236 fjz3 = _mm_macc_pd(dz33,fscal,fjz3);
1238 gmx_mm_decrement_4rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1240 /* Inner loop uses 434 flops */
1243 /* End of innermost loop */
1245 gmx_mm_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1246 f+i_coord_offset,fshift+i_shift_offset);
1249 /* Update potential energies */
1250 gmx_mm_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
1251 gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
1253 /* Increment number of inner iterations */
1254 inneriter += j_index_end - j_index_start;
1256 /* Outer loop uses 26 flops */
1259 /* Increment number of outer iterations */
1262 /* Update outer/inner flops */
1264 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*26 + inneriter*434);
1267 * Gromacs nonbonded kernel: nb_kernel_ElecEw_VdwLJ_GeomW4W4_F_avx_128_fma_double
1268 * Electrostatics interaction: Ewald
1269 * VdW interaction: LennardJones
1270 * Geometry: Water4-Water4
1271 * Calculate force/pot: Force
1274 nb_kernel_ElecEw_VdwLJ_GeomW4W4_F_avx_128_fma_double
1275 (t_nblist * gmx_restrict nlist,
1276 rvec * gmx_restrict xx,
1277 rvec * gmx_restrict ff,
1278 t_forcerec * gmx_restrict fr,
1279 t_mdatoms * gmx_restrict mdatoms,
1280 nb_kernel_data_t * gmx_restrict kernel_data,
1281 t_nrnb * gmx_restrict nrnb)
1283 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1284 * just 0 for non-waters.
1285 * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
1286 * jnr indices corresponding to data put in the four positions in the SIMD register.
1288 int i_shift_offset,i_coord_offset,outeriter,inneriter;
1289 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1291 int j_coord_offsetA,j_coord_offsetB;
1292 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
1293 real rcutoff_scalar;
1294 real *shiftvec,*fshift,*x,*f;
1295 __m128d tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1297 __m128d ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1299 __m128d ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1301 __m128d ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1303 __m128d ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
1304 int vdwjidx0A,vdwjidx0B;
1305 __m128d jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1306 int vdwjidx1A,vdwjidx1B;
1307 __m128d jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1308 int vdwjidx2A,vdwjidx2B;
1309 __m128d jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1310 int vdwjidx3A,vdwjidx3B;
1311 __m128d jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
1312 __m128d dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1313 __m128d dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1314 __m128d dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1315 __m128d dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
1316 __m128d dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1317 __m128d dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1318 __m128d dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
1319 __m128d dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
1320 __m128d dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
1321 __m128d dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
1322 __m128d velec,felec,velecsum,facel,crf,krf,krf2;
1325 __m128d rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1328 __m128d one_sixth = _mm_set1_pd(1.0/6.0);
1329 __m128d one_twelfth = _mm_set1_pd(1.0/12.0);
1331 __m128d ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1333 __m128d dummy_mask,cutoff_mask;
1334 __m128d signbit = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
1335 __m128d one = _mm_set1_pd(1.0);
1336 __m128d two = _mm_set1_pd(2.0);
1342 jindex = nlist->jindex;
1344 shiftidx = nlist->shift;
1346 shiftvec = fr->shift_vec[0];
1347 fshift = fr->fshift[0];
1348 facel = _mm_set1_pd(fr->epsfac);
1349 charge = mdatoms->chargeA;
1350 nvdwtype = fr->ntype;
1351 vdwparam = fr->nbfp;
1352 vdwtype = mdatoms->typeA;
1354 sh_ewald = _mm_set1_pd(fr->ic->sh_ewald);
1355 ewtab = fr->ic->tabq_coul_F;
1356 ewtabscale = _mm_set1_pd(fr->ic->tabq_scale);
1357 ewtabhalfspace = _mm_set1_pd(0.5/fr->ic->tabq_scale);
1359 /* Setup water-specific parameters */
1360 inr = nlist->iinr[0];
1361 iq1 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
1362 iq2 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
1363 iq3 = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+3]));
1364 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
1366 jq1 = _mm_set1_pd(charge[inr+1]);
1367 jq2 = _mm_set1_pd(charge[inr+2]);
1368 jq3 = _mm_set1_pd(charge[inr+3]);
1369 vdwjidx0A = 2*vdwtype[inr+0];
1370 c6_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A]);
1371 c12_00 = _mm_set1_pd(vdwparam[vdwioffset0+vdwjidx0A+1]);
1372 qq11 = _mm_mul_pd(iq1,jq1);
1373 qq12 = _mm_mul_pd(iq1,jq2);
1374 qq13 = _mm_mul_pd(iq1,jq3);
1375 qq21 = _mm_mul_pd(iq2,jq1);
1376 qq22 = _mm_mul_pd(iq2,jq2);
1377 qq23 = _mm_mul_pd(iq2,jq3);
1378 qq31 = _mm_mul_pd(iq3,jq1);
1379 qq32 = _mm_mul_pd(iq3,jq2);
1380 qq33 = _mm_mul_pd(iq3,jq3);
1382 /* Avoid stupid compiler warnings */
1384 j_coord_offsetA = 0;
1385 j_coord_offsetB = 0;
1390 /* Start outer loop over neighborlists */
1391 for(iidx=0; iidx<nri; iidx++)
1393 /* Load shift vector for this list */
1394 i_shift_offset = DIM*shiftidx[iidx];
1396 /* Load limits for loop over neighbors */
1397 j_index_start = jindex[iidx];
1398 j_index_end = jindex[iidx+1];
1400 /* Get outer coordinate index */
1402 i_coord_offset = DIM*inr;
1404 /* Load i particle coords and add shift vector */
1405 gmx_mm_load_shift_and_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
1406 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1408 fix0 = _mm_setzero_pd();
1409 fiy0 = _mm_setzero_pd();
1410 fiz0 = _mm_setzero_pd();
1411 fix1 = _mm_setzero_pd();
1412 fiy1 = _mm_setzero_pd();
1413 fiz1 = _mm_setzero_pd();
1414 fix2 = _mm_setzero_pd();
1415 fiy2 = _mm_setzero_pd();
1416 fiz2 = _mm_setzero_pd();
1417 fix3 = _mm_setzero_pd();
1418 fiy3 = _mm_setzero_pd();
1419 fiz3 = _mm_setzero_pd();
1421 /* Start inner kernel loop */
1422 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1425 /* Get j neighbor index, and coordinate index */
1427 jnrB = jjnr[jidx+1];
1428 j_coord_offsetA = DIM*jnrA;
1429 j_coord_offsetB = DIM*jnrB;
1431 /* load j atom coordinates */
1432 gmx_mm_load_4rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
1433 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
1434 &jy2,&jz2,&jx3,&jy3,&jz3);
1436 /* Calculate displacement vector */
1437 dx00 = _mm_sub_pd(ix0,jx0);
1438 dy00 = _mm_sub_pd(iy0,jy0);
1439 dz00 = _mm_sub_pd(iz0,jz0);
1440 dx11 = _mm_sub_pd(ix1,jx1);
1441 dy11 = _mm_sub_pd(iy1,jy1);
1442 dz11 = _mm_sub_pd(iz1,jz1);
1443 dx12 = _mm_sub_pd(ix1,jx2);
1444 dy12 = _mm_sub_pd(iy1,jy2);
1445 dz12 = _mm_sub_pd(iz1,jz2);
1446 dx13 = _mm_sub_pd(ix1,jx3);
1447 dy13 = _mm_sub_pd(iy1,jy3);
1448 dz13 = _mm_sub_pd(iz1,jz3);
1449 dx21 = _mm_sub_pd(ix2,jx1);
1450 dy21 = _mm_sub_pd(iy2,jy1);
1451 dz21 = _mm_sub_pd(iz2,jz1);
1452 dx22 = _mm_sub_pd(ix2,jx2);
1453 dy22 = _mm_sub_pd(iy2,jy2);
1454 dz22 = _mm_sub_pd(iz2,jz2);
1455 dx23 = _mm_sub_pd(ix2,jx3);
1456 dy23 = _mm_sub_pd(iy2,jy3);
1457 dz23 = _mm_sub_pd(iz2,jz3);
1458 dx31 = _mm_sub_pd(ix3,jx1);
1459 dy31 = _mm_sub_pd(iy3,jy1);
1460 dz31 = _mm_sub_pd(iz3,jz1);
1461 dx32 = _mm_sub_pd(ix3,jx2);
1462 dy32 = _mm_sub_pd(iy3,jy2);
1463 dz32 = _mm_sub_pd(iz3,jz2);
1464 dx33 = _mm_sub_pd(ix3,jx3);
1465 dy33 = _mm_sub_pd(iy3,jy3);
1466 dz33 = _mm_sub_pd(iz3,jz3);
1468 /* Calculate squared distance and things based on it */
1469 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
1470 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
1471 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
1472 rsq13 = gmx_mm_calc_rsq_pd(dx13,dy13,dz13);
1473 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
1474 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
1475 rsq23 = gmx_mm_calc_rsq_pd(dx23,dy23,dz23);
1476 rsq31 = gmx_mm_calc_rsq_pd(dx31,dy31,dz31);
1477 rsq32 = gmx_mm_calc_rsq_pd(dx32,dy32,dz32);
1478 rsq33 = gmx_mm_calc_rsq_pd(dx33,dy33,dz33);
1480 rinv11 = gmx_mm_invsqrt_pd(rsq11);
1481 rinv12 = gmx_mm_invsqrt_pd(rsq12);
1482 rinv13 = gmx_mm_invsqrt_pd(rsq13);
1483 rinv21 = gmx_mm_invsqrt_pd(rsq21);
1484 rinv22 = gmx_mm_invsqrt_pd(rsq22);
1485 rinv23 = gmx_mm_invsqrt_pd(rsq23);
1486 rinv31 = gmx_mm_invsqrt_pd(rsq31);
1487 rinv32 = gmx_mm_invsqrt_pd(rsq32);
1488 rinv33 = gmx_mm_invsqrt_pd(rsq33);
1490 rinvsq00 = gmx_mm_inv_pd(rsq00);
1491 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
1492 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
1493 rinvsq13 = _mm_mul_pd(rinv13,rinv13);
1494 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
1495 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
1496 rinvsq23 = _mm_mul_pd(rinv23,rinv23);
1497 rinvsq31 = _mm_mul_pd(rinv31,rinv31);
1498 rinvsq32 = _mm_mul_pd(rinv32,rinv32);
1499 rinvsq33 = _mm_mul_pd(rinv33,rinv33);
1501 fjx0 = _mm_setzero_pd();
1502 fjy0 = _mm_setzero_pd();
1503 fjz0 = _mm_setzero_pd();
1504 fjx1 = _mm_setzero_pd();
1505 fjy1 = _mm_setzero_pd();
1506 fjz1 = _mm_setzero_pd();
1507 fjx2 = _mm_setzero_pd();
1508 fjy2 = _mm_setzero_pd();
1509 fjz2 = _mm_setzero_pd();
1510 fjx3 = _mm_setzero_pd();
1511 fjy3 = _mm_setzero_pd();
1512 fjz3 = _mm_setzero_pd();
1514 /**************************
1515 * CALCULATE INTERACTIONS *
1516 **************************/
1518 /* LENNARD-JONES DISPERSION/REPULSION */
1520 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1521 fvdw = _mm_mul_pd(_mm_msub_pd(c12_00,rinvsix,c6_00),_mm_mul_pd(rinvsix,rinvsq00));
1525 /* Update vectorial force */
1526 fix0 = _mm_macc_pd(dx00,fscal,fix0);
1527 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
1528 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
1530 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
1531 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
1532 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
1534 /**************************
1535 * CALCULATE INTERACTIONS *
1536 **************************/
1538 r11 = _mm_mul_pd(rsq11,rinv11);
1540 /* EWALD ELECTROSTATICS */
1542 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1543 ewrt = _mm_mul_pd(r11,ewtabscale);
1544 ewitab = _mm_cvttpd_epi32(ewrt);
1546 eweps = _mm_frcz_pd(ewrt);
1548 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1550 twoeweps = _mm_add_pd(eweps,eweps);
1551 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1553 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1554 felec = _mm_mul_pd(_mm_mul_pd(qq11,rinv11),_mm_sub_pd(rinvsq11,felec));
1558 /* Update vectorial force */
1559 fix1 = _mm_macc_pd(dx11,fscal,fix1);
1560 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
1561 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
1563 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
1564 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
1565 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
1567 /**************************
1568 * CALCULATE INTERACTIONS *
1569 **************************/
1571 r12 = _mm_mul_pd(rsq12,rinv12);
1573 /* EWALD ELECTROSTATICS */
1575 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1576 ewrt = _mm_mul_pd(r12,ewtabscale);
1577 ewitab = _mm_cvttpd_epi32(ewrt);
1579 eweps = _mm_frcz_pd(ewrt);
1581 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1583 twoeweps = _mm_add_pd(eweps,eweps);
1584 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1586 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1587 felec = _mm_mul_pd(_mm_mul_pd(qq12,rinv12),_mm_sub_pd(rinvsq12,felec));
1591 /* Update vectorial force */
1592 fix1 = _mm_macc_pd(dx12,fscal,fix1);
1593 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
1594 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
1596 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
1597 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
1598 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
1600 /**************************
1601 * CALCULATE INTERACTIONS *
1602 **************************/
1604 r13 = _mm_mul_pd(rsq13,rinv13);
1606 /* EWALD ELECTROSTATICS */
1608 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1609 ewrt = _mm_mul_pd(r13,ewtabscale);
1610 ewitab = _mm_cvttpd_epi32(ewrt);
1612 eweps = _mm_frcz_pd(ewrt);
1614 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1616 twoeweps = _mm_add_pd(eweps,eweps);
1617 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1619 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1620 felec = _mm_mul_pd(_mm_mul_pd(qq13,rinv13),_mm_sub_pd(rinvsq13,felec));
1624 /* Update vectorial force */
1625 fix1 = _mm_macc_pd(dx13,fscal,fix1);
1626 fiy1 = _mm_macc_pd(dy13,fscal,fiy1);
1627 fiz1 = _mm_macc_pd(dz13,fscal,fiz1);
1629 fjx3 = _mm_macc_pd(dx13,fscal,fjx3);
1630 fjy3 = _mm_macc_pd(dy13,fscal,fjy3);
1631 fjz3 = _mm_macc_pd(dz13,fscal,fjz3);
1633 /**************************
1634 * CALCULATE INTERACTIONS *
1635 **************************/
1637 r21 = _mm_mul_pd(rsq21,rinv21);
1639 /* EWALD ELECTROSTATICS */
1641 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1642 ewrt = _mm_mul_pd(r21,ewtabscale);
1643 ewitab = _mm_cvttpd_epi32(ewrt);
1645 eweps = _mm_frcz_pd(ewrt);
1647 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1649 twoeweps = _mm_add_pd(eweps,eweps);
1650 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1652 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1653 felec = _mm_mul_pd(_mm_mul_pd(qq21,rinv21),_mm_sub_pd(rinvsq21,felec));
1657 /* Update vectorial force */
1658 fix2 = _mm_macc_pd(dx21,fscal,fix2);
1659 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
1660 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
1662 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
1663 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
1664 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
1666 /**************************
1667 * CALCULATE INTERACTIONS *
1668 **************************/
1670 r22 = _mm_mul_pd(rsq22,rinv22);
1672 /* EWALD ELECTROSTATICS */
1674 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1675 ewrt = _mm_mul_pd(r22,ewtabscale);
1676 ewitab = _mm_cvttpd_epi32(ewrt);
1678 eweps = _mm_frcz_pd(ewrt);
1680 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1682 twoeweps = _mm_add_pd(eweps,eweps);
1683 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1685 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1686 felec = _mm_mul_pd(_mm_mul_pd(qq22,rinv22),_mm_sub_pd(rinvsq22,felec));
1690 /* Update vectorial force */
1691 fix2 = _mm_macc_pd(dx22,fscal,fix2);
1692 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
1693 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
1695 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
1696 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
1697 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
1699 /**************************
1700 * CALCULATE INTERACTIONS *
1701 **************************/
1703 r23 = _mm_mul_pd(rsq23,rinv23);
1705 /* EWALD ELECTROSTATICS */
1707 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1708 ewrt = _mm_mul_pd(r23,ewtabscale);
1709 ewitab = _mm_cvttpd_epi32(ewrt);
1711 eweps = _mm_frcz_pd(ewrt);
1713 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1715 twoeweps = _mm_add_pd(eweps,eweps);
1716 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1718 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1719 felec = _mm_mul_pd(_mm_mul_pd(qq23,rinv23),_mm_sub_pd(rinvsq23,felec));
1723 /* Update vectorial force */
1724 fix2 = _mm_macc_pd(dx23,fscal,fix2);
1725 fiy2 = _mm_macc_pd(dy23,fscal,fiy2);
1726 fiz2 = _mm_macc_pd(dz23,fscal,fiz2);
1728 fjx3 = _mm_macc_pd(dx23,fscal,fjx3);
1729 fjy3 = _mm_macc_pd(dy23,fscal,fjy3);
1730 fjz3 = _mm_macc_pd(dz23,fscal,fjz3);
1732 /**************************
1733 * CALCULATE INTERACTIONS *
1734 **************************/
1736 r31 = _mm_mul_pd(rsq31,rinv31);
1738 /* EWALD ELECTROSTATICS */
1740 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1741 ewrt = _mm_mul_pd(r31,ewtabscale);
1742 ewitab = _mm_cvttpd_epi32(ewrt);
1744 eweps = _mm_frcz_pd(ewrt);
1746 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1748 twoeweps = _mm_add_pd(eweps,eweps);
1749 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1751 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1752 felec = _mm_mul_pd(_mm_mul_pd(qq31,rinv31),_mm_sub_pd(rinvsq31,felec));
1756 /* Update vectorial force */
1757 fix3 = _mm_macc_pd(dx31,fscal,fix3);
1758 fiy3 = _mm_macc_pd(dy31,fscal,fiy3);
1759 fiz3 = _mm_macc_pd(dz31,fscal,fiz3);
1761 fjx1 = _mm_macc_pd(dx31,fscal,fjx1);
1762 fjy1 = _mm_macc_pd(dy31,fscal,fjy1);
1763 fjz1 = _mm_macc_pd(dz31,fscal,fjz1);
1765 /**************************
1766 * CALCULATE INTERACTIONS *
1767 **************************/
1769 r32 = _mm_mul_pd(rsq32,rinv32);
1771 /* EWALD ELECTROSTATICS */
1773 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1774 ewrt = _mm_mul_pd(r32,ewtabscale);
1775 ewitab = _mm_cvttpd_epi32(ewrt);
1777 eweps = _mm_frcz_pd(ewrt);
1779 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1781 twoeweps = _mm_add_pd(eweps,eweps);
1782 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1784 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1785 felec = _mm_mul_pd(_mm_mul_pd(qq32,rinv32),_mm_sub_pd(rinvsq32,felec));
1789 /* Update vectorial force */
1790 fix3 = _mm_macc_pd(dx32,fscal,fix3);
1791 fiy3 = _mm_macc_pd(dy32,fscal,fiy3);
1792 fiz3 = _mm_macc_pd(dz32,fscal,fiz3);
1794 fjx2 = _mm_macc_pd(dx32,fscal,fjx2);
1795 fjy2 = _mm_macc_pd(dy32,fscal,fjy2);
1796 fjz2 = _mm_macc_pd(dz32,fscal,fjz2);
1798 /**************************
1799 * CALCULATE INTERACTIONS *
1800 **************************/
1802 r33 = _mm_mul_pd(rsq33,rinv33);
1804 /* EWALD ELECTROSTATICS */
1806 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1807 ewrt = _mm_mul_pd(r33,ewtabscale);
1808 ewitab = _mm_cvttpd_epi32(ewrt);
1810 eweps = _mm_frcz_pd(ewrt);
1812 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1814 twoeweps = _mm_add_pd(eweps,eweps);
1815 gmx_mm_load_2pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),ewtab+_mm_extract_epi32(ewitab,1),
1817 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1818 felec = _mm_mul_pd(_mm_mul_pd(qq33,rinv33),_mm_sub_pd(rinvsq33,felec));
1822 /* Update vectorial force */
1823 fix3 = _mm_macc_pd(dx33,fscal,fix3);
1824 fiy3 = _mm_macc_pd(dy33,fscal,fiy3);
1825 fiz3 = _mm_macc_pd(dz33,fscal,fiz3);
1827 fjx3 = _mm_macc_pd(dx33,fscal,fjx3);
1828 fjy3 = _mm_macc_pd(dy33,fscal,fjy3);
1829 fjz3 = _mm_macc_pd(dz33,fscal,fjz3);
1831 gmx_mm_decrement_4rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1833 /* Inner loop uses 384 flops */
1836 if(jidx<j_index_end)
1840 j_coord_offsetA = DIM*jnrA;
1842 /* load j atom coordinates */
1843 gmx_mm_load_4rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
1844 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
1845 &jy2,&jz2,&jx3,&jy3,&jz3);
1847 /* Calculate displacement vector */
1848 dx00 = _mm_sub_pd(ix0,jx0);
1849 dy00 = _mm_sub_pd(iy0,jy0);
1850 dz00 = _mm_sub_pd(iz0,jz0);
1851 dx11 = _mm_sub_pd(ix1,jx1);
1852 dy11 = _mm_sub_pd(iy1,jy1);
1853 dz11 = _mm_sub_pd(iz1,jz1);
1854 dx12 = _mm_sub_pd(ix1,jx2);
1855 dy12 = _mm_sub_pd(iy1,jy2);
1856 dz12 = _mm_sub_pd(iz1,jz2);
1857 dx13 = _mm_sub_pd(ix1,jx3);
1858 dy13 = _mm_sub_pd(iy1,jy3);
1859 dz13 = _mm_sub_pd(iz1,jz3);
1860 dx21 = _mm_sub_pd(ix2,jx1);
1861 dy21 = _mm_sub_pd(iy2,jy1);
1862 dz21 = _mm_sub_pd(iz2,jz1);
1863 dx22 = _mm_sub_pd(ix2,jx2);
1864 dy22 = _mm_sub_pd(iy2,jy2);
1865 dz22 = _mm_sub_pd(iz2,jz2);
1866 dx23 = _mm_sub_pd(ix2,jx3);
1867 dy23 = _mm_sub_pd(iy2,jy3);
1868 dz23 = _mm_sub_pd(iz2,jz3);
1869 dx31 = _mm_sub_pd(ix3,jx1);
1870 dy31 = _mm_sub_pd(iy3,jy1);
1871 dz31 = _mm_sub_pd(iz3,jz1);
1872 dx32 = _mm_sub_pd(ix3,jx2);
1873 dy32 = _mm_sub_pd(iy3,jy2);
1874 dz32 = _mm_sub_pd(iz3,jz2);
1875 dx33 = _mm_sub_pd(ix3,jx3);
1876 dy33 = _mm_sub_pd(iy3,jy3);
1877 dz33 = _mm_sub_pd(iz3,jz3);
1879 /* Calculate squared distance and things based on it */
1880 rsq00 = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
1881 rsq11 = gmx_mm_calc_rsq_pd(dx11,dy11,dz11);
1882 rsq12 = gmx_mm_calc_rsq_pd(dx12,dy12,dz12);
1883 rsq13 = gmx_mm_calc_rsq_pd(dx13,dy13,dz13);
1884 rsq21 = gmx_mm_calc_rsq_pd(dx21,dy21,dz21);
1885 rsq22 = gmx_mm_calc_rsq_pd(dx22,dy22,dz22);
1886 rsq23 = gmx_mm_calc_rsq_pd(dx23,dy23,dz23);
1887 rsq31 = gmx_mm_calc_rsq_pd(dx31,dy31,dz31);
1888 rsq32 = gmx_mm_calc_rsq_pd(dx32,dy32,dz32);
1889 rsq33 = gmx_mm_calc_rsq_pd(dx33,dy33,dz33);
1891 rinv11 = gmx_mm_invsqrt_pd(rsq11);
1892 rinv12 = gmx_mm_invsqrt_pd(rsq12);
1893 rinv13 = gmx_mm_invsqrt_pd(rsq13);
1894 rinv21 = gmx_mm_invsqrt_pd(rsq21);
1895 rinv22 = gmx_mm_invsqrt_pd(rsq22);
1896 rinv23 = gmx_mm_invsqrt_pd(rsq23);
1897 rinv31 = gmx_mm_invsqrt_pd(rsq31);
1898 rinv32 = gmx_mm_invsqrt_pd(rsq32);
1899 rinv33 = gmx_mm_invsqrt_pd(rsq33);
1901 rinvsq00 = gmx_mm_inv_pd(rsq00);
1902 rinvsq11 = _mm_mul_pd(rinv11,rinv11);
1903 rinvsq12 = _mm_mul_pd(rinv12,rinv12);
1904 rinvsq13 = _mm_mul_pd(rinv13,rinv13);
1905 rinvsq21 = _mm_mul_pd(rinv21,rinv21);
1906 rinvsq22 = _mm_mul_pd(rinv22,rinv22);
1907 rinvsq23 = _mm_mul_pd(rinv23,rinv23);
1908 rinvsq31 = _mm_mul_pd(rinv31,rinv31);
1909 rinvsq32 = _mm_mul_pd(rinv32,rinv32);
1910 rinvsq33 = _mm_mul_pd(rinv33,rinv33);
1912 fjx0 = _mm_setzero_pd();
1913 fjy0 = _mm_setzero_pd();
1914 fjz0 = _mm_setzero_pd();
1915 fjx1 = _mm_setzero_pd();
1916 fjy1 = _mm_setzero_pd();
1917 fjz1 = _mm_setzero_pd();
1918 fjx2 = _mm_setzero_pd();
1919 fjy2 = _mm_setzero_pd();
1920 fjz2 = _mm_setzero_pd();
1921 fjx3 = _mm_setzero_pd();
1922 fjy3 = _mm_setzero_pd();
1923 fjz3 = _mm_setzero_pd();
1925 /**************************
1926 * CALCULATE INTERACTIONS *
1927 **************************/
1929 /* LENNARD-JONES DISPERSION/REPULSION */
1931 rinvsix = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1932 fvdw = _mm_mul_pd(_mm_msub_pd(c12_00,rinvsix,c6_00),_mm_mul_pd(rinvsix,rinvsq00));
1936 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1938 /* Update vectorial force */
1939 fix0 = _mm_macc_pd(dx00,fscal,fix0);
1940 fiy0 = _mm_macc_pd(dy00,fscal,fiy0);
1941 fiz0 = _mm_macc_pd(dz00,fscal,fiz0);
1943 fjx0 = _mm_macc_pd(dx00,fscal,fjx0);
1944 fjy0 = _mm_macc_pd(dy00,fscal,fjy0);
1945 fjz0 = _mm_macc_pd(dz00,fscal,fjz0);
1947 /**************************
1948 * CALCULATE INTERACTIONS *
1949 **************************/
1951 r11 = _mm_mul_pd(rsq11,rinv11);
1953 /* EWALD ELECTROSTATICS */
1955 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1956 ewrt = _mm_mul_pd(r11,ewtabscale);
1957 ewitab = _mm_cvttpd_epi32(ewrt);
1959 eweps = _mm_frcz_pd(ewrt);
1961 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1963 twoeweps = _mm_add_pd(eweps,eweps);
1964 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
1965 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
1966 felec = _mm_mul_pd(_mm_mul_pd(qq11,rinv11),_mm_sub_pd(rinvsq11,felec));
1970 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1972 /* Update vectorial force */
1973 fix1 = _mm_macc_pd(dx11,fscal,fix1);
1974 fiy1 = _mm_macc_pd(dy11,fscal,fiy1);
1975 fiz1 = _mm_macc_pd(dz11,fscal,fiz1);
1977 fjx1 = _mm_macc_pd(dx11,fscal,fjx1);
1978 fjy1 = _mm_macc_pd(dy11,fscal,fjy1);
1979 fjz1 = _mm_macc_pd(dz11,fscal,fjz1);
1981 /**************************
1982 * CALCULATE INTERACTIONS *
1983 **************************/
1985 r12 = _mm_mul_pd(rsq12,rinv12);
1987 /* EWALD ELECTROSTATICS */
1989 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1990 ewrt = _mm_mul_pd(r12,ewtabscale);
1991 ewitab = _mm_cvttpd_epi32(ewrt);
1993 eweps = _mm_frcz_pd(ewrt);
1995 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
1997 twoeweps = _mm_add_pd(eweps,eweps);
1998 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
1999 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2000 felec = _mm_mul_pd(_mm_mul_pd(qq12,rinv12),_mm_sub_pd(rinvsq12,felec));
2004 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2006 /* Update vectorial force */
2007 fix1 = _mm_macc_pd(dx12,fscal,fix1);
2008 fiy1 = _mm_macc_pd(dy12,fscal,fiy1);
2009 fiz1 = _mm_macc_pd(dz12,fscal,fiz1);
2011 fjx2 = _mm_macc_pd(dx12,fscal,fjx2);
2012 fjy2 = _mm_macc_pd(dy12,fscal,fjy2);
2013 fjz2 = _mm_macc_pd(dz12,fscal,fjz2);
2015 /**************************
2016 * CALCULATE INTERACTIONS *
2017 **************************/
2019 r13 = _mm_mul_pd(rsq13,rinv13);
2021 /* EWALD ELECTROSTATICS */
2023 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2024 ewrt = _mm_mul_pd(r13,ewtabscale);
2025 ewitab = _mm_cvttpd_epi32(ewrt);
2027 eweps = _mm_frcz_pd(ewrt);
2029 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
2031 twoeweps = _mm_add_pd(eweps,eweps);
2032 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
2033 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2034 felec = _mm_mul_pd(_mm_mul_pd(qq13,rinv13),_mm_sub_pd(rinvsq13,felec));
2038 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2040 /* Update vectorial force */
2041 fix1 = _mm_macc_pd(dx13,fscal,fix1);
2042 fiy1 = _mm_macc_pd(dy13,fscal,fiy1);
2043 fiz1 = _mm_macc_pd(dz13,fscal,fiz1);
2045 fjx3 = _mm_macc_pd(dx13,fscal,fjx3);
2046 fjy3 = _mm_macc_pd(dy13,fscal,fjy3);
2047 fjz3 = _mm_macc_pd(dz13,fscal,fjz3);
2049 /**************************
2050 * CALCULATE INTERACTIONS *
2051 **************************/
2053 r21 = _mm_mul_pd(rsq21,rinv21);
2055 /* EWALD ELECTROSTATICS */
2057 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2058 ewrt = _mm_mul_pd(r21,ewtabscale);
2059 ewitab = _mm_cvttpd_epi32(ewrt);
2061 eweps = _mm_frcz_pd(ewrt);
2063 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
2065 twoeweps = _mm_add_pd(eweps,eweps);
2066 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
2067 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2068 felec = _mm_mul_pd(_mm_mul_pd(qq21,rinv21),_mm_sub_pd(rinvsq21,felec));
2072 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2074 /* Update vectorial force */
2075 fix2 = _mm_macc_pd(dx21,fscal,fix2);
2076 fiy2 = _mm_macc_pd(dy21,fscal,fiy2);
2077 fiz2 = _mm_macc_pd(dz21,fscal,fiz2);
2079 fjx1 = _mm_macc_pd(dx21,fscal,fjx1);
2080 fjy1 = _mm_macc_pd(dy21,fscal,fjy1);
2081 fjz1 = _mm_macc_pd(dz21,fscal,fjz1);
2083 /**************************
2084 * CALCULATE INTERACTIONS *
2085 **************************/
2087 r22 = _mm_mul_pd(rsq22,rinv22);
2089 /* EWALD ELECTROSTATICS */
2091 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2092 ewrt = _mm_mul_pd(r22,ewtabscale);
2093 ewitab = _mm_cvttpd_epi32(ewrt);
2095 eweps = _mm_frcz_pd(ewrt);
2097 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
2099 twoeweps = _mm_add_pd(eweps,eweps);
2100 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
2101 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2102 felec = _mm_mul_pd(_mm_mul_pd(qq22,rinv22),_mm_sub_pd(rinvsq22,felec));
2106 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2108 /* Update vectorial force */
2109 fix2 = _mm_macc_pd(dx22,fscal,fix2);
2110 fiy2 = _mm_macc_pd(dy22,fscal,fiy2);
2111 fiz2 = _mm_macc_pd(dz22,fscal,fiz2);
2113 fjx2 = _mm_macc_pd(dx22,fscal,fjx2);
2114 fjy2 = _mm_macc_pd(dy22,fscal,fjy2);
2115 fjz2 = _mm_macc_pd(dz22,fscal,fjz2);
2117 /**************************
2118 * CALCULATE INTERACTIONS *
2119 **************************/
2121 r23 = _mm_mul_pd(rsq23,rinv23);
2123 /* EWALD ELECTROSTATICS */
2125 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2126 ewrt = _mm_mul_pd(r23,ewtabscale);
2127 ewitab = _mm_cvttpd_epi32(ewrt);
2129 eweps = _mm_frcz_pd(ewrt);
2131 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
2133 twoeweps = _mm_add_pd(eweps,eweps);
2134 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
2135 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2136 felec = _mm_mul_pd(_mm_mul_pd(qq23,rinv23),_mm_sub_pd(rinvsq23,felec));
2140 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2142 /* Update vectorial force */
2143 fix2 = _mm_macc_pd(dx23,fscal,fix2);
2144 fiy2 = _mm_macc_pd(dy23,fscal,fiy2);
2145 fiz2 = _mm_macc_pd(dz23,fscal,fiz2);
2147 fjx3 = _mm_macc_pd(dx23,fscal,fjx3);
2148 fjy3 = _mm_macc_pd(dy23,fscal,fjy3);
2149 fjz3 = _mm_macc_pd(dz23,fscal,fjz3);
2151 /**************************
2152 * CALCULATE INTERACTIONS *
2153 **************************/
2155 r31 = _mm_mul_pd(rsq31,rinv31);
2157 /* EWALD ELECTROSTATICS */
2159 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2160 ewrt = _mm_mul_pd(r31,ewtabscale);
2161 ewitab = _mm_cvttpd_epi32(ewrt);
2163 eweps = _mm_frcz_pd(ewrt);
2165 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
2167 twoeweps = _mm_add_pd(eweps,eweps);
2168 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
2169 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2170 felec = _mm_mul_pd(_mm_mul_pd(qq31,rinv31),_mm_sub_pd(rinvsq31,felec));
2174 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2176 /* Update vectorial force */
2177 fix3 = _mm_macc_pd(dx31,fscal,fix3);
2178 fiy3 = _mm_macc_pd(dy31,fscal,fiy3);
2179 fiz3 = _mm_macc_pd(dz31,fscal,fiz3);
2181 fjx1 = _mm_macc_pd(dx31,fscal,fjx1);
2182 fjy1 = _mm_macc_pd(dy31,fscal,fjy1);
2183 fjz1 = _mm_macc_pd(dz31,fscal,fjz1);
2185 /**************************
2186 * CALCULATE INTERACTIONS *
2187 **************************/
2189 r32 = _mm_mul_pd(rsq32,rinv32);
2191 /* EWALD ELECTROSTATICS */
2193 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2194 ewrt = _mm_mul_pd(r32,ewtabscale);
2195 ewitab = _mm_cvttpd_epi32(ewrt);
2197 eweps = _mm_frcz_pd(ewrt);
2199 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
2201 twoeweps = _mm_add_pd(eweps,eweps);
2202 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
2203 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2204 felec = _mm_mul_pd(_mm_mul_pd(qq32,rinv32),_mm_sub_pd(rinvsq32,felec));
2208 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2210 /* Update vectorial force */
2211 fix3 = _mm_macc_pd(dx32,fscal,fix3);
2212 fiy3 = _mm_macc_pd(dy32,fscal,fiy3);
2213 fiz3 = _mm_macc_pd(dz32,fscal,fiz3);
2215 fjx2 = _mm_macc_pd(dx32,fscal,fjx2);
2216 fjy2 = _mm_macc_pd(dy32,fscal,fjy2);
2217 fjz2 = _mm_macc_pd(dz32,fscal,fjz2);
2219 /**************************
2220 * CALCULATE INTERACTIONS *
2221 **************************/
2223 r33 = _mm_mul_pd(rsq33,rinv33);
2225 /* EWALD ELECTROSTATICS */
2227 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2228 ewrt = _mm_mul_pd(r33,ewtabscale);
2229 ewitab = _mm_cvttpd_epi32(ewrt);
2231 eweps = _mm_frcz_pd(ewrt);
2233 eweps = _mm_sub_pd(ewrt,_mm_round_pd(ewrt, _MM_FROUND_FLOOR));
2235 twoeweps = _mm_add_pd(eweps,eweps);
2236 gmx_mm_load_1pair_swizzle_pd(ewtab+_mm_extract_epi32(ewitab,0),&ewtabF,&ewtabFn);
2237 felec = _mm_macc_pd(eweps,ewtabFn,_mm_mul_pd( _mm_sub_pd(one,eweps),ewtabF));
2238 felec = _mm_mul_pd(_mm_mul_pd(qq33,rinv33),_mm_sub_pd(rinvsq33,felec));
2242 fscal = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
2244 /* Update vectorial force */
2245 fix3 = _mm_macc_pd(dx33,fscal,fix3);
2246 fiy3 = _mm_macc_pd(dy33,fscal,fiy3);
2247 fiz3 = _mm_macc_pd(dz33,fscal,fiz3);
2249 fjx3 = _mm_macc_pd(dx33,fscal,fjx3);
2250 fjy3 = _mm_macc_pd(dy33,fscal,fjy3);
2251 fjz3 = _mm_macc_pd(dz33,fscal,fjz3);
2253 gmx_mm_decrement_4rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
2255 /* Inner loop uses 384 flops */
2258 /* End of innermost loop */
2260 gmx_mm_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
2261 f+i_coord_offset,fshift+i_shift_offset);
2263 /* Increment number of inner iterations */
2264 inneriter += j_index_end - j_index_start;
2266 /* Outer loop uses 24 flops */
2269 /* Increment number of outer iterations */
2272 /* Update outer/inner flops */
2274 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*24 + inneriter*384);