2 * Note: this file was generated by the Gromacs sse2_single 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_sse2_single.h"
34 #include "kernelutil_x86_sse2_single.h"
37 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_sse2_single
38 * Electrostatics interaction: Ewald
39 * VdW interaction: LennardJones
40 * Geometry: Water4-Particle
41 * Calculate force/pot: PotentialAndForce
44 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_sse2_single
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,C,D refer to j loop unrolling done with SSE, e.g. for the four 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;
60 int jnrA,jnrB,jnrC,jnrD;
61 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
62 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
63 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
65 real *shiftvec,*fshift,*x,*f;
66 real *fjptrA,*fjptrB,*fjptrC,*fjptrD;
68 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
70 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
72 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
74 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
76 __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
77 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
78 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
79 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
80 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
81 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
82 __m128 dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
83 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
86 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
89 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
90 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
92 __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
94 __m128 rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
95 real rswitch_scalar,d_scalar;
96 __m128 dummy_mask,cutoff_mask;
97 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
98 __m128 one = _mm_set1_ps(1.0);
99 __m128 two = _mm_set1_ps(2.0);
105 jindex = nlist->jindex;
107 shiftidx = nlist->shift;
109 shiftvec = fr->shift_vec[0];
110 fshift = fr->fshift[0];
111 facel = _mm_set1_ps(fr->epsfac);
112 charge = mdatoms->chargeA;
113 nvdwtype = fr->ntype;
115 vdwtype = mdatoms->typeA;
117 sh_ewald = _mm_set1_ps(fr->ic->sh_ewald);
118 ewtab = fr->ic->tabq_coul_FDV0;
119 ewtabscale = _mm_set1_ps(fr->ic->tabq_scale);
120 ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale);
122 /* Setup water-specific parameters */
123 inr = nlist->iinr[0];
124 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
125 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
126 iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
127 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
129 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
130 rcutoff_scalar = fr->rcoulomb;
131 rcutoff = _mm_set1_ps(rcutoff_scalar);
132 rcutoff2 = _mm_mul_ps(rcutoff,rcutoff);
134 rswitch_scalar = fr->rcoulomb_switch;
135 rswitch = _mm_set1_ps(rswitch_scalar);
136 /* Setup switch parameters */
137 d_scalar = rcutoff_scalar-rswitch_scalar;
138 d = _mm_set1_ps(d_scalar);
139 swV3 = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
140 swV4 = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
141 swV5 = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
142 swF2 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
143 swF3 = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
144 swF4 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
146 /* Avoid stupid compiler warnings */
147 jnrA = jnrB = jnrC = jnrD = 0;
156 for(iidx=0;iidx<4*DIM;iidx++)
161 /* Start outer loop over neighborlists */
162 for(iidx=0; iidx<nri; iidx++)
164 /* Load shift vector for this list */
165 i_shift_offset = DIM*shiftidx[iidx];
167 /* Load limits for loop over neighbors */
168 j_index_start = jindex[iidx];
169 j_index_end = jindex[iidx+1];
171 /* Get outer coordinate index */
173 i_coord_offset = DIM*inr;
175 /* Load i particle coords and add shift vector */
176 gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
177 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
179 fix0 = _mm_setzero_ps();
180 fiy0 = _mm_setzero_ps();
181 fiz0 = _mm_setzero_ps();
182 fix1 = _mm_setzero_ps();
183 fiy1 = _mm_setzero_ps();
184 fiz1 = _mm_setzero_ps();
185 fix2 = _mm_setzero_ps();
186 fiy2 = _mm_setzero_ps();
187 fiz2 = _mm_setzero_ps();
188 fix3 = _mm_setzero_ps();
189 fiy3 = _mm_setzero_ps();
190 fiz3 = _mm_setzero_ps();
192 /* Reset potential sums */
193 velecsum = _mm_setzero_ps();
194 vvdwsum = _mm_setzero_ps();
196 /* Start inner kernel loop */
197 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
200 /* Get j neighbor index, and coordinate index */
205 j_coord_offsetA = DIM*jnrA;
206 j_coord_offsetB = DIM*jnrB;
207 j_coord_offsetC = DIM*jnrC;
208 j_coord_offsetD = DIM*jnrD;
210 /* load j atom coordinates */
211 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
212 x+j_coord_offsetC,x+j_coord_offsetD,
215 /* Calculate displacement vector */
216 dx00 = _mm_sub_ps(ix0,jx0);
217 dy00 = _mm_sub_ps(iy0,jy0);
218 dz00 = _mm_sub_ps(iz0,jz0);
219 dx10 = _mm_sub_ps(ix1,jx0);
220 dy10 = _mm_sub_ps(iy1,jy0);
221 dz10 = _mm_sub_ps(iz1,jz0);
222 dx20 = _mm_sub_ps(ix2,jx0);
223 dy20 = _mm_sub_ps(iy2,jy0);
224 dz20 = _mm_sub_ps(iz2,jz0);
225 dx30 = _mm_sub_ps(ix3,jx0);
226 dy30 = _mm_sub_ps(iy3,jy0);
227 dz30 = _mm_sub_ps(iz3,jz0);
229 /* Calculate squared distance and things based on it */
230 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
231 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
232 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
233 rsq30 = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
235 rinv00 = gmx_mm_invsqrt_ps(rsq00);
236 rinv10 = gmx_mm_invsqrt_ps(rsq10);
237 rinv20 = gmx_mm_invsqrt_ps(rsq20);
238 rinv30 = gmx_mm_invsqrt_ps(rsq30);
240 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
241 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
242 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
243 rinvsq30 = _mm_mul_ps(rinv30,rinv30);
245 /* Load parameters for j particles */
246 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
247 charge+jnrC+0,charge+jnrD+0);
248 vdwjidx0A = 2*vdwtype[jnrA+0];
249 vdwjidx0B = 2*vdwtype[jnrB+0];
250 vdwjidx0C = 2*vdwtype[jnrC+0];
251 vdwjidx0D = 2*vdwtype[jnrD+0];
253 fjx0 = _mm_setzero_ps();
254 fjy0 = _mm_setzero_ps();
255 fjz0 = _mm_setzero_ps();
257 /**************************
258 * CALCULATE INTERACTIONS *
259 **************************/
261 if (gmx_mm_any_lt(rsq00,rcutoff2))
264 r00 = _mm_mul_ps(rsq00,rinv00);
266 /* Compute parameters for interactions between i and j atoms */
267 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
268 vdwparam+vdwioffset0+vdwjidx0B,
269 vdwparam+vdwioffset0+vdwjidx0C,
270 vdwparam+vdwioffset0+vdwjidx0D,
273 /* LENNARD-JONES DISPERSION/REPULSION */
275 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
276 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
277 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
278 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
279 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
281 d = _mm_sub_ps(r00,rswitch);
282 d = _mm_max_ps(d,_mm_setzero_ps());
283 d2 = _mm_mul_ps(d,d);
284 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
286 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
288 /* Evaluate switch function */
289 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
290 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
291 vvdw = _mm_mul_ps(vvdw,sw);
292 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
294 /* Update potential sum for this i atom from the interaction with this j atom. */
295 vvdw = _mm_and_ps(vvdw,cutoff_mask);
296 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
300 fscal = _mm_and_ps(fscal,cutoff_mask);
302 /* Calculate temporary vectorial force */
303 tx = _mm_mul_ps(fscal,dx00);
304 ty = _mm_mul_ps(fscal,dy00);
305 tz = _mm_mul_ps(fscal,dz00);
307 /* Update vectorial force */
308 fix0 = _mm_add_ps(fix0,tx);
309 fiy0 = _mm_add_ps(fiy0,ty);
310 fiz0 = _mm_add_ps(fiz0,tz);
312 fjx0 = _mm_add_ps(fjx0,tx);
313 fjy0 = _mm_add_ps(fjy0,ty);
314 fjz0 = _mm_add_ps(fjz0,tz);
318 /**************************
319 * CALCULATE INTERACTIONS *
320 **************************/
322 if (gmx_mm_any_lt(rsq10,rcutoff2))
325 r10 = _mm_mul_ps(rsq10,rinv10);
327 /* Compute parameters for interactions between i and j atoms */
328 qq10 = _mm_mul_ps(iq1,jq0);
330 /* EWALD ELECTROSTATICS */
332 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
333 ewrt = _mm_mul_ps(r10,ewtabscale);
334 ewitab = _mm_cvttps_epi32(ewrt);
335 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
336 ewitab = _mm_slli_epi32(ewitab,2);
337 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
338 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
339 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
340 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
341 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
342 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
343 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
344 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
345 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
347 d = _mm_sub_ps(r10,rswitch);
348 d = _mm_max_ps(d,_mm_setzero_ps());
349 d2 = _mm_mul_ps(d,d);
350 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
352 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
354 /* Evaluate switch function */
355 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
356 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
357 velec = _mm_mul_ps(velec,sw);
358 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
360 /* Update potential sum for this i atom from the interaction with this j atom. */
361 velec = _mm_and_ps(velec,cutoff_mask);
362 velecsum = _mm_add_ps(velecsum,velec);
366 fscal = _mm_and_ps(fscal,cutoff_mask);
368 /* Calculate temporary vectorial force */
369 tx = _mm_mul_ps(fscal,dx10);
370 ty = _mm_mul_ps(fscal,dy10);
371 tz = _mm_mul_ps(fscal,dz10);
373 /* Update vectorial force */
374 fix1 = _mm_add_ps(fix1,tx);
375 fiy1 = _mm_add_ps(fiy1,ty);
376 fiz1 = _mm_add_ps(fiz1,tz);
378 fjx0 = _mm_add_ps(fjx0,tx);
379 fjy0 = _mm_add_ps(fjy0,ty);
380 fjz0 = _mm_add_ps(fjz0,tz);
384 /**************************
385 * CALCULATE INTERACTIONS *
386 **************************/
388 if (gmx_mm_any_lt(rsq20,rcutoff2))
391 r20 = _mm_mul_ps(rsq20,rinv20);
393 /* Compute parameters for interactions between i and j atoms */
394 qq20 = _mm_mul_ps(iq2,jq0);
396 /* EWALD ELECTROSTATICS */
398 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
399 ewrt = _mm_mul_ps(r20,ewtabscale);
400 ewitab = _mm_cvttps_epi32(ewrt);
401 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
402 ewitab = _mm_slli_epi32(ewitab,2);
403 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
404 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
405 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
406 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
407 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
408 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
409 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
410 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
411 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
413 d = _mm_sub_ps(r20,rswitch);
414 d = _mm_max_ps(d,_mm_setzero_ps());
415 d2 = _mm_mul_ps(d,d);
416 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
418 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
420 /* Evaluate switch function */
421 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
422 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
423 velec = _mm_mul_ps(velec,sw);
424 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
426 /* Update potential sum for this i atom from the interaction with this j atom. */
427 velec = _mm_and_ps(velec,cutoff_mask);
428 velecsum = _mm_add_ps(velecsum,velec);
432 fscal = _mm_and_ps(fscal,cutoff_mask);
434 /* Calculate temporary vectorial force */
435 tx = _mm_mul_ps(fscal,dx20);
436 ty = _mm_mul_ps(fscal,dy20);
437 tz = _mm_mul_ps(fscal,dz20);
439 /* Update vectorial force */
440 fix2 = _mm_add_ps(fix2,tx);
441 fiy2 = _mm_add_ps(fiy2,ty);
442 fiz2 = _mm_add_ps(fiz2,tz);
444 fjx0 = _mm_add_ps(fjx0,tx);
445 fjy0 = _mm_add_ps(fjy0,ty);
446 fjz0 = _mm_add_ps(fjz0,tz);
450 /**************************
451 * CALCULATE INTERACTIONS *
452 **************************/
454 if (gmx_mm_any_lt(rsq30,rcutoff2))
457 r30 = _mm_mul_ps(rsq30,rinv30);
459 /* Compute parameters for interactions between i and j atoms */
460 qq30 = _mm_mul_ps(iq3,jq0);
462 /* EWALD ELECTROSTATICS */
464 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
465 ewrt = _mm_mul_ps(r30,ewtabscale);
466 ewitab = _mm_cvttps_epi32(ewrt);
467 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
468 ewitab = _mm_slli_epi32(ewitab,2);
469 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
470 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
471 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
472 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
473 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
474 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
475 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
476 velec = _mm_mul_ps(qq30,_mm_sub_ps(rinv30,velec));
477 felec = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
479 d = _mm_sub_ps(r30,rswitch);
480 d = _mm_max_ps(d,_mm_setzero_ps());
481 d2 = _mm_mul_ps(d,d);
482 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
484 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
486 /* Evaluate switch function */
487 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
488 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
489 velec = _mm_mul_ps(velec,sw);
490 cutoff_mask = _mm_cmplt_ps(rsq30,rcutoff2);
492 /* Update potential sum for this i atom from the interaction with this j atom. */
493 velec = _mm_and_ps(velec,cutoff_mask);
494 velecsum = _mm_add_ps(velecsum,velec);
498 fscal = _mm_and_ps(fscal,cutoff_mask);
500 /* Calculate temporary vectorial force */
501 tx = _mm_mul_ps(fscal,dx30);
502 ty = _mm_mul_ps(fscal,dy30);
503 tz = _mm_mul_ps(fscal,dz30);
505 /* Update vectorial force */
506 fix3 = _mm_add_ps(fix3,tx);
507 fiy3 = _mm_add_ps(fiy3,ty);
508 fiz3 = _mm_add_ps(fiz3,tz);
510 fjx0 = _mm_add_ps(fjx0,tx);
511 fjy0 = _mm_add_ps(fjy0,ty);
512 fjz0 = _mm_add_ps(fjz0,tz);
516 fjptrA = f+j_coord_offsetA;
517 fjptrB = f+j_coord_offsetB;
518 fjptrC = f+j_coord_offsetC;
519 fjptrD = f+j_coord_offsetD;
521 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
523 /* Inner loop uses 254 flops */
529 /* Get j neighbor index, and coordinate index */
530 jnrlistA = jjnr[jidx];
531 jnrlistB = jjnr[jidx+1];
532 jnrlistC = jjnr[jidx+2];
533 jnrlistD = jjnr[jidx+3];
534 /* Sign of each element will be negative for non-real atoms.
535 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
536 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
538 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
539 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
540 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
541 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
542 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
543 j_coord_offsetA = DIM*jnrA;
544 j_coord_offsetB = DIM*jnrB;
545 j_coord_offsetC = DIM*jnrC;
546 j_coord_offsetD = DIM*jnrD;
548 /* load j atom coordinates */
549 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
550 x+j_coord_offsetC,x+j_coord_offsetD,
553 /* Calculate displacement vector */
554 dx00 = _mm_sub_ps(ix0,jx0);
555 dy00 = _mm_sub_ps(iy0,jy0);
556 dz00 = _mm_sub_ps(iz0,jz0);
557 dx10 = _mm_sub_ps(ix1,jx0);
558 dy10 = _mm_sub_ps(iy1,jy0);
559 dz10 = _mm_sub_ps(iz1,jz0);
560 dx20 = _mm_sub_ps(ix2,jx0);
561 dy20 = _mm_sub_ps(iy2,jy0);
562 dz20 = _mm_sub_ps(iz2,jz0);
563 dx30 = _mm_sub_ps(ix3,jx0);
564 dy30 = _mm_sub_ps(iy3,jy0);
565 dz30 = _mm_sub_ps(iz3,jz0);
567 /* Calculate squared distance and things based on it */
568 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
569 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
570 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
571 rsq30 = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
573 rinv00 = gmx_mm_invsqrt_ps(rsq00);
574 rinv10 = gmx_mm_invsqrt_ps(rsq10);
575 rinv20 = gmx_mm_invsqrt_ps(rsq20);
576 rinv30 = gmx_mm_invsqrt_ps(rsq30);
578 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
579 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
580 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
581 rinvsq30 = _mm_mul_ps(rinv30,rinv30);
583 /* Load parameters for j particles */
584 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
585 charge+jnrC+0,charge+jnrD+0);
586 vdwjidx0A = 2*vdwtype[jnrA+0];
587 vdwjidx0B = 2*vdwtype[jnrB+0];
588 vdwjidx0C = 2*vdwtype[jnrC+0];
589 vdwjidx0D = 2*vdwtype[jnrD+0];
591 fjx0 = _mm_setzero_ps();
592 fjy0 = _mm_setzero_ps();
593 fjz0 = _mm_setzero_ps();
595 /**************************
596 * CALCULATE INTERACTIONS *
597 **************************/
599 if (gmx_mm_any_lt(rsq00,rcutoff2))
602 r00 = _mm_mul_ps(rsq00,rinv00);
603 r00 = _mm_andnot_ps(dummy_mask,r00);
605 /* Compute parameters for interactions between i and j atoms */
606 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
607 vdwparam+vdwioffset0+vdwjidx0B,
608 vdwparam+vdwioffset0+vdwjidx0C,
609 vdwparam+vdwioffset0+vdwjidx0D,
612 /* LENNARD-JONES DISPERSION/REPULSION */
614 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
615 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
616 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
617 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
618 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
620 d = _mm_sub_ps(r00,rswitch);
621 d = _mm_max_ps(d,_mm_setzero_ps());
622 d2 = _mm_mul_ps(d,d);
623 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
625 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
627 /* Evaluate switch function */
628 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
629 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
630 vvdw = _mm_mul_ps(vvdw,sw);
631 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
633 /* Update potential sum for this i atom from the interaction with this j atom. */
634 vvdw = _mm_and_ps(vvdw,cutoff_mask);
635 vvdw = _mm_andnot_ps(dummy_mask,vvdw);
636 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
640 fscal = _mm_and_ps(fscal,cutoff_mask);
642 fscal = _mm_andnot_ps(dummy_mask,fscal);
644 /* Calculate temporary vectorial force */
645 tx = _mm_mul_ps(fscal,dx00);
646 ty = _mm_mul_ps(fscal,dy00);
647 tz = _mm_mul_ps(fscal,dz00);
649 /* Update vectorial force */
650 fix0 = _mm_add_ps(fix0,tx);
651 fiy0 = _mm_add_ps(fiy0,ty);
652 fiz0 = _mm_add_ps(fiz0,tz);
654 fjx0 = _mm_add_ps(fjx0,tx);
655 fjy0 = _mm_add_ps(fjy0,ty);
656 fjz0 = _mm_add_ps(fjz0,tz);
660 /**************************
661 * CALCULATE INTERACTIONS *
662 **************************/
664 if (gmx_mm_any_lt(rsq10,rcutoff2))
667 r10 = _mm_mul_ps(rsq10,rinv10);
668 r10 = _mm_andnot_ps(dummy_mask,r10);
670 /* Compute parameters for interactions between i and j atoms */
671 qq10 = _mm_mul_ps(iq1,jq0);
673 /* EWALD ELECTROSTATICS */
675 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
676 ewrt = _mm_mul_ps(r10,ewtabscale);
677 ewitab = _mm_cvttps_epi32(ewrt);
678 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
679 ewitab = _mm_slli_epi32(ewitab,2);
680 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
681 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
682 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
683 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
684 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
685 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
686 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
687 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
688 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
690 d = _mm_sub_ps(r10,rswitch);
691 d = _mm_max_ps(d,_mm_setzero_ps());
692 d2 = _mm_mul_ps(d,d);
693 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
695 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
697 /* Evaluate switch function */
698 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
699 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
700 velec = _mm_mul_ps(velec,sw);
701 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
703 /* Update potential sum for this i atom from the interaction with this j atom. */
704 velec = _mm_and_ps(velec,cutoff_mask);
705 velec = _mm_andnot_ps(dummy_mask,velec);
706 velecsum = _mm_add_ps(velecsum,velec);
710 fscal = _mm_and_ps(fscal,cutoff_mask);
712 fscal = _mm_andnot_ps(dummy_mask,fscal);
714 /* Calculate temporary vectorial force */
715 tx = _mm_mul_ps(fscal,dx10);
716 ty = _mm_mul_ps(fscal,dy10);
717 tz = _mm_mul_ps(fscal,dz10);
719 /* Update vectorial force */
720 fix1 = _mm_add_ps(fix1,tx);
721 fiy1 = _mm_add_ps(fiy1,ty);
722 fiz1 = _mm_add_ps(fiz1,tz);
724 fjx0 = _mm_add_ps(fjx0,tx);
725 fjy0 = _mm_add_ps(fjy0,ty);
726 fjz0 = _mm_add_ps(fjz0,tz);
730 /**************************
731 * CALCULATE INTERACTIONS *
732 **************************/
734 if (gmx_mm_any_lt(rsq20,rcutoff2))
737 r20 = _mm_mul_ps(rsq20,rinv20);
738 r20 = _mm_andnot_ps(dummy_mask,r20);
740 /* Compute parameters for interactions between i and j atoms */
741 qq20 = _mm_mul_ps(iq2,jq0);
743 /* EWALD ELECTROSTATICS */
745 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
746 ewrt = _mm_mul_ps(r20,ewtabscale);
747 ewitab = _mm_cvttps_epi32(ewrt);
748 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
749 ewitab = _mm_slli_epi32(ewitab,2);
750 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
751 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
752 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
753 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
754 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
755 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
756 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
757 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
758 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
760 d = _mm_sub_ps(r20,rswitch);
761 d = _mm_max_ps(d,_mm_setzero_ps());
762 d2 = _mm_mul_ps(d,d);
763 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
765 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
767 /* Evaluate switch function */
768 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
769 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
770 velec = _mm_mul_ps(velec,sw);
771 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
773 /* Update potential sum for this i atom from the interaction with this j atom. */
774 velec = _mm_and_ps(velec,cutoff_mask);
775 velec = _mm_andnot_ps(dummy_mask,velec);
776 velecsum = _mm_add_ps(velecsum,velec);
780 fscal = _mm_and_ps(fscal,cutoff_mask);
782 fscal = _mm_andnot_ps(dummy_mask,fscal);
784 /* Calculate temporary vectorial force */
785 tx = _mm_mul_ps(fscal,dx20);
786 ty = _mm_mul_ps(fscal,dy20);
787 tz = _mm_mul_ps(fscal,dz20);
789 /* Update vectorial force */
790 fix2 = _mm_add_ps(fix2,tx);
791 fiy2 = _mm_add_ps(fiy2,ty);
792 fiz2 = _mm_add_ps(fiz2,tz);
794 fjx0 = _mm_add_ps(fjx0,tx);
795 fjy0 = _mm_add_ps(fjy0,ty);
796 fjz0 = _mm_add_ps(fjz0,tz);
800 /**************************
801 * CALCULATE INTERACTIONS *
802 **************************/
804 if (gmx_mm_any_lt(rsq30,rcutoff2))
807 r30 = _mm_mul_ps(rsq30,rinv30);
808 r30 = _mm_andnot_ps(dummy_mask,r30);
810 /* Compute parameters for interactions between i and j atoms */
811 qq30 = _mm_mul_ps(iq3,jq0);
813 /* EWALD ELECTROSTATICS */
815 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
816 ewrt = _mm_mul_ps(r30,ewtabscale);
817 ewitab = _mm_cvttps_epi32(ewrt);
818 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
819 ewitab = _mm_slli_epi32(ewitab,2);
820 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
821 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
822 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
823 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
824 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
825 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
826 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
827 velec = _mm_mul_ps(qq30,_mm_sub_ps(rinv30,velec));
828 felec = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
830 d = _mm_sub_ps(r30,rswitch);
831 d = _mm_max_ps(d,_mm_setzero_ps());
832 d2 = _mm_mul_ps(d,d);
833 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
835 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
837 /* Evaluate switch function */
838 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
839 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
840 velec = _mm_mul_ps(velec,sw);
841 cutoff_mask = _mm_cmplt_ps(rsq30,rcutoff2);
843 /* Update potential sum for this i atom from the interaction with this j atom. */
844 velec = _mm_and_ps(velec,cutoff_mask);
845 velec = _mm_andnot_ps(dummy_mask,velec);
846 velecsum = _mm_add_ps(velecsum,velec);
850 fscal = _mm_and_ps(fscal,cutoff_mask);
852 fscal = _mm_andnot_ps(dummy_mask,fscal);
854 /* Calculate temporary vectorial force */
855 tx = _mm_mul_ps(fscal,dx30);
856 ty = _mm_mul_ps(fscal,dy30);
857 tz = _mm_mul_ps(fscal,dz30);
859 /* Update vectorial force */
860 fix3 = _mm_add_ps(fix3,tx);
861 fiy3 = _mm_add_ps(fiy3,ty);
862 fiz3 = _mm_add_ps(fiz3,tz);
864 fjx0 = _mm_add_ps(fjx0,tx);
865 fjy0 = _mm_add_ps(fjy0,ty);
866 fjz0 = _mm_add_ps(fjz0,tz);
870 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
871 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
872 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
873 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
875 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
877 /* Inner loop uses 258 flops */
880 /* End of innermost loop */
882 gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
883 f+i_coord_offset,fshift+i_shift_offset);
886 /* Update potential energies */
887 gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
888 gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
890 /* Increment number of inner iterations */
891 inneriter += j_index_end - j_index_start;
893 /* Outer loop uses 26 flops */
896 /* Increment number of outer iterations */
899 /* Update outer/inner flops */
901 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*26 + inneriter*258);
904 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_sse2_single
905 * Electrostatics interaction: Ewald
906 * VdW interaction: LennardJones
907 * Geometry: Water4-Particle
908 * Calculate force/pot: Force
911 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_sse2_single
912 (t_nblist * gmx_restrict nlist,
913 rvec * gmx_restrict xx,
914 rvec * gmx_restrict ff,
915 t_forcerec * gmx_restrict fr,
916 t_mdatoms * gmx_restrict mdatoms,
917 nb_kernel_data_t * gmx_restrict kernel_data,
918 t_nrnb * gmx_restrict nrnb)
920 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
921 * just 0 for non-waters.
922 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
923 * jnr indices corresponding to data put in the four positions in the SIMD register.
925 int i_shift_offset,i_coord_offset,outeriter,inneriter;
926 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
927 int jnrA,jnrB,jnrC,jnrD;
928 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
929 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
930 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
932 real *shiftvec,*fshift,*x,*f;
933 real *fjptrA,*fjptrB,*fjptrC,*fjptrD;
935 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
937 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
939 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
941 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
943 __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
944 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
945 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
946 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
947 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
948 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
949 __m128 dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
950 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
953 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
956 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
957 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
959 __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
961 __m128 rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
962 real rswitch_scalar,d_scalar;
963 __m128 dummy_mask,cutoff_mask;
964 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
965 __m128 one = _mm_set1_ps(1.0);
966 __m128 two = _mm_set1_ps(2.0);
972 jindex = nlist->jindex;
974 shiftidx = nlist->shift;
976 shiftvec = fr->shift_vec[0];
977 fshift = fr->fshift[0];
978 facel = _mm_set1_ps(fr->epsfac);
979 charge = mdatoms->chargeA;
980 nvdwtype = fr->ntype;
982 vdwtype = mdatoms->typeA;
984 sh_ewald = _mm_set1_ps(fr->ic->sh_ewald);
985 ewtab = fr->ic->tabq_coul_FDV0;
986 ewtabscale = _mm_set1_ps(fr->ic->tabq_scale);
987 ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale);
989 /* Setup water-specific parameters */
990 inr = nlist->iinr[0];
991 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
992 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
993 iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
994 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
996 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
997 rcutoff_scalar = fr->rcoulomb;
998 rcutoff = _mm_set1_ps(rcutoff_scalar);
999 rcutoff2 = _mm_mul_ps(rcutoff,rcutoff);
1001 rswitch_scalar = fr->rcoulomb_switch;
1002 rswitch = _mm_set1_ps(rswitch_scalar);
1003 /* Setup switch parameters */
1004 d_scalar = rcutoff_scalar-rswitch_scalar;
1005 d = _mm_set1_ps(d_scalar);
1006 swV3 = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
1007 swV4 = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
1008 swV5 = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
1009 swF2 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
1010 swF3 = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
1011 swF4 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
1013 /* Avoid stupid compiler warnings */
1014 jnrA = jnrB = jnrC = jnrD = 0;
1015 j_coord_offsetA = 0;
1016 j_coord_offsetB = 0;
1017 j_coord_offsetC = 0;
1018 j_coord_offsetD = 0;
1023 for(iidx=0;iidx<4*DIM;iidx++)
1025 scratch[iidx] = 0.0;
1028 /* Start outer loop over neighborlists */
1029 for(iidx=0; iidx<nri; iidx++)
1031 /* Load shift vector for this list */
1032 i_shift_offset = DIM*shiftidx[iidx];
1034 /* Load limits for loop over neighbors */
1035 j_index_start = jindex[iidx];
1036 j_index_end = jindex[iidx+1];
1038 /* Get outer coordinate index */
1040 i_coord_offset = DIM*inr;
1042 /* Load i particle coords and add shift vector */
1043 gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1044 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1046 fix0 = _mm_setzero_ps();
1047 fiy0 = _mm_setzero_ps();
1048 fiz0 = _mm_setzero_ps();
1049 fix1 = _mm_setzero_ps();
1050 fiy1 = _mm_setzero_ps();
1051 fiz1 = _mm_setzero_ps();
1052 fix2 = _mm_setzero_ps();
1053 fiy2 = _mm_setzero_ps();
1054 fiz2 = _mm_setzero_ps();
1055 fix3 = _mm_setzero_ps();
1056 fiy3 = _mm_setzero_ps();
1057 fiz3 = _mm_setzero_ps();
1059 /* Start inner kernel loop */
1060 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1063 /* Get j neighbor index, and coordinate index */
1065 jnrB = jjnr[jidx+1];
1066 jnrC = jjnr[jidx+2];
1067 jnrD = jjnr[jidx+3];
1068 j_coord_offsetA = DIM*jnrA;
1069 j_coord_offsetB = DIM*jnrB;
1070 j_coord_offsetC = DIM*jnrC;
1071 j_coord_offsetD = DIM*jnrD;
1073 /* load j atom coordinates */
1074 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1075 x+j_coord_offsetC,x+j_coord_offsetD,
1078 /* Calculate displacement vector */
1079 dx00 = _mm_sub_ps(ix0,jx0);
1080 dy00 = _mm_sub_ps(iy0,jy0);
1081 dz00 = _mm_sub_ps(iz0,jz0);
1082 dx10 = _mm_sub_ps(ix1,jx0);
1083 dy10 = _mm_sub_ps(iy1,jy0);
1084 dz10 = _mm_sub_ps(iz1,jz0);
1085 dx20 = _mm_sub_ps(ix2,jx0);
1086 dy20 = _mm_sub_ps(iy2,jy0);
1087 dz20 = _mm_sub_ps(iz2,jz0);
1088 dx30 = _mm_sub_ps(ix3,jx0);
1089 dy30 = _mm_sub_ps(iy3,jy0);
1090 dz30 = _mm_sub_ps(iz3,jz0);
1092 /* Calculate squared distance and things based on it */
1093 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1094 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1095 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1096 rsq30 = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
1098 rinv00 = gmx_mm_invsqrt_ps(rsq00);
1099 rinv10 = gmx_mm_invsqrt_ps(rsq10);
1100 rinv20 = gmx_mm_invsqrt_ps(rsq20);
1101 rinv30 = gmx_mm_invsqrt_ps(rsq30);
1103 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
1104 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
1105 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
1106 rinvsq30 = _mm_mul_ps(rinv30,rinv30);
1108 /* Load parameters for j particles */
1109 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1110 charge+jnrC+0,charge+jnrD+0);
1111 vdwjidx0A = 2*vdwtype[jnrA+0];
1112 vdwjidx0B = 2*vdwtype[jnrB+0];
1113 vdwjidx0C = 2*vdwtype[jnrC+0];
1114 vdwjidx0D = 2*vdwtype[jnrD+0];
1116 fjx0 = _mm_setzero_ps();
1117 fjy0 = _mm_setzero_ps();
1118 fjz0 = _mm_setzero_ps();
1120 /**************************
1121 * CALCULATE INTERACTIONS *
1122 **************************/
1124 if (gmx_mm_any_lt(rsq00,rcutoff2))
1127 r00 = _mm_mul_ps(rsq00,rinv00);
1129 /* Compute parameters for interactions between i and j atoms */
1130 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
1131 vdwparam+vdwioffset0+vdwjidx0B,
1132 vdwparam+vdwioffset0+vdwjidx0C,
1133 vdwparam+vdwioffset0+vdwjidx0D,
1136 /* LENNARD-JONES DISPERSION/REPULSION */
1138 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1139 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
1140 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
1141 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
1142 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
1144 d = _mm_sub_ps(r00,rswitch);
1145 d = _mm_max_ps(d,_mm_setzero_ps());
1146 d2 = _mm_mul_ps(d,d);
1147 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1149 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1151 /* Evaluate switch function */
1152 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1153 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
1154 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
1158 fscal = _mm_and_ps(fscal,cutoff_mask);
1160 /* Calculate temporary vectorial force */
1161 tx = _mm_mul_ps(fscal,dx00);
1162 ty = _mm_mul_ps(fscal,dy00);
1163 tz = _mm_mul_ps(fscal,dz00);
1165 /* Update vectorial force */
1166 fix0 = _mm_add_ps(fix0,tx);
1167 fiy0 = _mm_add_ps(fiy0,ty);
1168 fiz0 = _mm_add_ps(fiz0,tz);
1170 fjx0 = _mm_add_ps(fjx0,tx);
1171 fjy0 = _mm_add_ps(fjy0,ty);
1172 fjz0 = _mm_add_ps(fjz0,tz);
1176 /**************************
1177 * CALCULATE INTERACTIONS *
1178 **************************/
1180 if (gmx_mm_any_lt(rsq10,rcutoff2))
1183 r10 = _mm_mul_ps(rsq10,rinv10);
1185 /* Compute parameters for interactions between i and j atoms */
1186 qq10 = _mm_mul_ps(iq1,jq0);
1188 /* EWALD ELECTROSTATICS */
1190 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1191 ewrt = _mm_mul_ps(r10,ewtabscale);
1192 ewitab = _mm_cvttps_epi32(ewrt);
1193 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1194 ewitab = _mm_slli_epi32(ewitab,2);
1195 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1196 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1197 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1198 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1199 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1200 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1201 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1202 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
1203 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1205 d = _mm_sub_ps(r10,rswitch);
1206 d = _mm_max_ps(d,_mm_setzero_ps());
1207 d2 = _mm_mul_ps(d,d);
1208 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1210 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1212 /* Evaluate switch function */
1213 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1214 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
1215 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
1219 fscal = _mm_and_ps(fscal,cutoff_mask);
1221 /* Calculate temporary vectorial force */
1222 tx = _mm_mul_ps(fscal,dx10);
1223 ty = _mm_mul_ps(fscal,dy10);
1224 tz = _mm_mul_ps(fscal,dz10);
1226 /* Update vectorial force */
1227 fix1 = _mm_add_ps(fix1,tx);
1228 fiy1 = _mm_add_ps(fiy1,ty);
1229 fiz1 = _mm_add_ps(fiz1,tz);
1231 fjx0 = _mm_add_ps(fjx0,tx);
1232 fjy0 = _mm_add_ps(fjy0,ty);
1233 fjz0 = _mm_add_ps(fjz0,tz);
1237 /**************************
1238 * CALCULATE INTERACTIONS *
1239 **************************/
1241 if (gmx_mm_any_lt(rsq20,rcutoff2))
1244 r20 = _mm_mul_ps(rsq20,rinv20);
1246 /* Compute parameters for interactions between i and j atoms */
1247 qq20 = _mm_mul_ps(iq2,jq0);
1249 /* EWALD ELECTROSTATICS */
1251 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1252 ewrt = _mm_mul_ps(r20,ewtabscale);
1253 ewitab = _mm_cvttps_epi32(ewrt);
1254 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1255 ewitab = _mm_slli_epi32(ewitab,2);
1256 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1257 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1258 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1259 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1260 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1261 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1262 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1263 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
1264 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
1266 d = _mm_sub_ps(r20,rswitch);
1267 d = _mm_max_ps(d,_mm_setzero_ps());
1268 d2 = _mm_mul_ps(d,d);
1269 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1271 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1273 /* Evaluate switch function */
1274 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1275 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
1276 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
1280 fscal = _mm_and_ps(fscal,cutoff_mask);
1282 /* Calculate temporary vectorial force */
1283 tx = _mm_mul_ps(fscal,dx20);
1284 ty = _mm_mul_ps(fscal,dy20);
1285 tz = _mm_mul_ps(fscal,dz20);
1287 /* Update vectorial force */
1288 fix2 = _mm_add_ps(fix2,tx);
1289 fiy2 = _mm_add_ps(fiy2,ty);
1290 fiz2 = _mm_add_ps(fiz2,tz);
1292 fjx0 = _mm_add_ps(fjx0,tx);
1293 fjy0 = _mm_add_ps(fjy0,ty);
1294 fjz0 = _mm_add_ps(fjz0,tz);
1298 /**************************
1299 * CALCULATE INTERACTIONS *
1300 **************************/
1302 if (gmx_mm_any_lt(rsq30,rcutoff2))
1305 r30 = _mm_mul_ps(rsq30,rinv30);
1307 /* Compute parameters for interactions between i and j atoms */
1308 qq30 = _mm_mul_ps(iq3,jq0);
1310 /* EWALD ELECTROSTATICS */
1312 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1313 ewrt = _mm_mul_ps(r30,ewtabscale);
1314 ewitab = _mm_cvttps_epi32(ewrt);
1315 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1316 ewitab = _mm_slli_epi32(ewitab,2);
1317 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1318 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1319 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1320 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1321 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1322 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1323 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1324 velec = _mm_mul_ps(qq30,_mm_sub_ps(rinv30,velec));
1325 felec = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
1327 d = _mm_sub_ps(r30,rswitch);
1328 d = _mm_max_ps(d,_mm_setzero_ps());
1329 d2 = _mm_mul_ps(d,d);
1330 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1332 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1334 /* Evaluate switch function */
1335 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1336 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
1337 cutoff_mask = _mm_cmplt_ps(rsq30,rcutoff2);
1341 fscal = _mm_and_ps(fscal,cutoff_mask);
1343 /* Calculate temporary vectorial force */
1344 tx = _mm_mul_ps(fscal,dx30);
1345 ty = _mm_mul_ps(fscal,dy30);
1346 tz = _mm_mul_ps(fscal,dz30);
1348 /* Update vectorial force */
1349 fix3 = _mm_add_ps(fix3,tx);
1350 fiy3 = _mm_add_ps(fiy3,ty);
1351 fiz3 = _mm_add_ps(fiz3,tz);
1353 fjx0 = _mm_add_ps(fjx0,tx);
1354 fjy0 = _mm_add_ps(fjy0,ty);
1355 fjz0 = _mm_add_ps(fjz0,tz);
1359 fjptrA = f+j_coord_offsetA;
1360 fjptrB = f+j_coord_offsetB;
1361 fjptrC = f+j_coord_offsetC;
1362 fjptrD = f+j_coord_offsetD;
1364 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1366 /* Inner loop uses 242 flops */
1369 if(jidx<j_index_end)
1372 /* Get j neighbor index, and coordinate index */
1373 jnrlistA = jjnr[jidx];
1374 jnrlistB = jjnr[jidx+1];
1375 jnrlistC = jjnr[jidx+2];
1376 jnrlistD = jjnr[jidx+3];
1377 /* Sign of each element will be negative for non-real atoms.
1378 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1379 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1381 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1382 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
1383 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
1384 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
1385 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
1386 j_coord_offsetA = DIM*jnrA;
1387 j_coord_offsetB = DIM*jnrB;
1388 j_coord_offsetC = DIM*jnrC;
1389 j_coord_offsetD = DIM*jnrD;
1391 /* load j atom coordinates */
1392 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1393 x+j_coord_offsetC,x+j_coord_offsetD,
1396 /* Calculate displacement vector */
1397 dx00 = _mm_sub_ps(ix0,jx0);
1398 dy00 = _mm_sub_ps(iy0,jy0);
1399 dz00 = _mm_sub_ps(iz0,jz0);
1400 dx10 = _mm_sub_ps(ix1,jx0);
1401 dy10 = _mm_sub_ps(iy1,jy0);
1402 dz10 = _mm_sub_ps(iz1,jz0);
1403 dx20 = _mm_sub_ps(ix2,jx0);
1404 dy20 = _mm_sub_ps(iy2,jy0);
1405 dz20 = _mm_sub_ps(iz2,jz0);
1406 dx30 = _mm_sub_ps(ix3,jx0);
1407 dy30 = _mm_sub_ps(iy3,jy0);
1408 dz30 = _mm_sub_ps(iz3,jz0);
1410 /* Calculate squared distance and things based on it */
1411 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1412 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1413 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1414 rsq30 = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
1416 rinv00 = gmx_mm_invsqrt_ps(rsq00);
1417 rinv10 = gmx_mm_invsqrt_ps(rsq10);
1418 rinv20 = gmx_mm_invsqrt_ps(rsq20);
1419 rinv30 = gmx_mm_invsqrt_ps(rsq30);
1421 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
1422 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
1423 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
1424 rinvsq30 = _mm_mul_ps(rinv30,rinv30);
1426 /* Load parameters for j particles */
1427 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1428 charge+jnrC+0,charge+jnrD+0);
1429 vdwjidx0A = 2*vdwtype[jnrA+0];
1430 vdwjidx0B = 2*vdwtype[jnrB+0];
1431 vdwjidx0C = 2*vdwtype[jnrC+0];
1432 vdwjidx0D = 2*vdwtype[jnrD+0];
1434 fjx0 = _mm_setzero_ps();
1435 fjy0 = _mm_setzero_ps();
1436 fjz0 = _mm_setzero_ps();
1438 /**************************
1439 * CALCULATE INTERACTIONS *
1440 **************************/
1442 if (gmx_mm_any_lt(rsq00,rcutoff2))
1445 r00 = _mm_mul_ps(rsq00,rinv00);
1446 r00 = _mm_andnot_ps(dummy_mask,r00);
1448 /* Compute parameters for interactions between i and j atoms */
1449 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
1450 vdwparam+vdwioffset0+vdwjidx0B,
1451 vdwparam+vdwioffset0+vdwjidx0C,
1452 vdwparam+vdwioffset0+vdwjidx0D,
1455 /* LENNARD-JONES DISPERSION/REPULSION */
1457 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1458 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
1459 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
1460 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
1461 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
1463 d = _mm_sub_ps(r00,rswitch);
1464 d = _mm_max_ps(d,_mm_setzero_ps());
1465 d2 = _mm_mul_ps(d,d);
1466 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1468 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1470 /* Evaluate switch function */
1471 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1472 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
1473 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
1477 fscal = _mm_and_ps(fscal,cutoff_mask);
1479 fscal = _mm_andnot_ps(dummy_mask,fscal);
1481 /* Calculate temporary vectorial force */
1482 tx = _mm_mul_ps(fscal,dx00);
1483 ty = _mm_mul_ps(fscal,dy00);
1484 tz = _mm_mul_ps(fscal,dz00);
1486 /* Update vectorial force */
1487 fix0 = _mm_add_ps(fix0,tx);
1488 fiy0 = _mm_add_ps(fiy0,ty);
1489 fiz0 = _mm_add_ps(fiz0,tz);
1491 fjx0 = _mm_add_ps(fjx0,tx);
1492 fjy0 = _mm_add_ps(fjy0,ty);
1493 fjz0 = _mm_add_ps(fjz0,tz);
1497 /**************************
1498 * CALCULATE INTERACTIONS *
1499 **************************/
1501 if (gmx_mm_any_lt(rsq10,rcutoff2))
1504 r10 = _mm_mul_ps(rsq10,rinv10);
1505 r10 = _mm_andnot_ps(dummy_mask,r10);
1507 /* Compute parameters for interactions between i and j atoms */
1508 qq10 = _mm_mul_ps(iq1,jq0);
1510 /* EWALD ELECTROSTATICS */
1512 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1513 ewrt = _mm_mul_ps(r10,ewtabscale);
1514 ewitab = _mm_cvttps_epi32(ewrt);
1515 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1516 ewitab = _mm_slli_epi32(ewitab,2);
1517 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1518 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1519 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1520 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1521 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1522 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1523 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1524 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
1525 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1527 d = _mm_sub_ps(r10,rswitch);
1528 d = _mm_max_ps(d,_mm_setzero_ps());
1529 d2 = _mm_mul_ps(d,d);
1530 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1532 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1534 /* Evaluate switch function */
1535 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1536 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
1537 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
1541 fscal = _mm_and_ps(fscal,cutoff_mask);
1543 fscal = _mm_andnot_ps(dummy_mask,fscal);
1545 /* Calculate temporary vectorial force */
1546 tx = _mm_mul_ps(fscal,dx10);
1547 ty = _mm_mul_ps(fscal,dy10);
1548 tz = _mm_mul_ps(fscal,dz10);
1550 /* Update vectorial force */
1551 fix1 = _mm_add_ps(fix1,tx);
1552 fiy1 = _mm_add_ps(fiy1,ty);
1553 fiz1 = _mm_add_ps(fiz1,tz);
1555 fjx0 = _mm_add_ps(fjx0,tx);
1556 fjy0 = _mm_add_ps(fjy0,ty);
1557 fjz0 = _mm_add_ps(fjz0,tz);
1561 /**************************
1562 * CALCULATE INTERACTIONS *
1563 **************************/
1565 if (gmx_mm_any_lt(rsq20,rcutoff2))
1568 r20 = _mm_mul_ps(rsq20,rinv20);
1569 r20 = _mm_andnot_ps(dummy_mask,r20);
1571 /* Compute parameters for interactions between i and j atoms */
1572 qq20 = _mm_mul_ps(iq2,jq0);
1574 /* EWALD ELECTROSTATICS */
1576 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1577 ewrt = _mm_mul_ps(r20,ewtabscale);
1578 ewitab = _mm_cvttps_epi32(ewrt);
1579 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1580 ewitab = _mm_slli_epi32(ewitab,2);
1581 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1582 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1583 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1584 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1585 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1586 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1587 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1588 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
1589 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
1591 d = _mm_sub_ps(r20,rswitch);
1592 d = _mm_max_ps(d,_mm_setzero_ps());
1593 d2 = _mm_mul_ps(d,d);
1594 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1596 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1598 /* Evaluate switch function */
1599 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1600 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
1601 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
1605 fscal = _mm_and_ps(fscal,cutoff_mask);
1607 fscal = _mm_andnot_ps(dummy_mask,fscal);
1609 /* Calculate temporary vectorial force */
1610 tx = _mm_mul_ps(fscal,dx20);
1611 ty = _mm_mul_ps(fscal,dy20);
1612 tz = _mm_mul_ps(fscal,dz20);
1614 /* Update vectorial force */
1615 fix2 = _mm_add_ps(fix2,tx);
1616 fiy2 = _mm_add_ps(fiy2,ty);
1617 fiz2 = _mm_add_ps(fiz2,tz);
1619 fjx0 = _mm_add_ps(fjx0,tx);
1620 fjy0 = _mm_add_ps(fjy0,ty);
1621 fjz0 = _mm_add_ps(fjz0,tz);
1625 /**************************
1626 * CALCULATE INTERACTIONS *
1627 **************************/
1629 if (gmx_mm_any_lt(rsq30,rcutoff2))
1632 r30 = _mm_mul_ps(rsq30,rinv30);
1633 r30 = _mm_andnot_ps(dummy_mask,r30);
1635 /* Compute parameters for interactions between i and j atoms */
1636 qq30 = _mm_mul_ps(iq3,jq0);
1638 /* EWALD ELECTROSTATICS */
1640 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1641 ewrt = _mm_mul_ps(r30,ewtabscale);
1642 ewitab = _mm_cvttps_epi32(ewrt);
1643 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1644 ewitab = _mm_slli_epi32(ewitab,2);
1645 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1646 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1647 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1648 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1649 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1650 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1651 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1652 velec = _mm_mul_ps(qq30,_mm_sub_ps(rinv30,velec));
1653 felec = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
1655 d = _mm_sub_ps(r30,rswitch);
1656 d = _mm_max_ps(d,_mm_setzero_ps());
1657 d2 = _mm_mul_ps(d,d);
1658 sw = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_add_ps(swV3,_mm_mul_ps(d,_mm_add_ps(swV4,_mm_mul_ps(d,swV5)))))));
1660 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1662 /* Evaluate switch function */
1663 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1664 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
1665 cutoff_mask = _mm_cmplt_ps(rsq30,rcutoff2);
1669 fscal = _mm_and_ps(fscal,cutoff_mask);
1671 fscal = _mm_andnot_ps(dummy_mask,fscal);
1673 /* Calculate temporary vectorial force */
1674 tx = _mm_mul_ps(fscal,dx30);
1675 ty = _mm_mul_ps(fscal,dy30);
1676 tz = _mm_mul_ps(fscal,dz30);
1678 /* Update vectorial force */
1679 fix3 = _mm_add_ps(fix3,tx);
1680 fiy3 = _mm_add_ps(fiy3,ty);
1681 fiz3 = _mm_add_ps(fiz3,tz);
1683 fjx0 = _mm_add_ps(fjx0,tx);
1684 fjy0 = _mm_add_ps(fjy0,ty);
1685 fjz0 = _mm_add_ps(fjz0,tz);
1689 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1690 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1691 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1692 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1694 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1696 /* Inner loop uses 246 flops */
1699 /* End of innermost loop */
1701 gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1702 f+i_coord_offset,fshift+i_shift_offset);
1704 /* Increment number of inner iterations */
1705 inneriter += j_index_end - j_index_start;
1707 /* Outer loop uses 24 flops */
1710 /* Increment number of outer iterations */
1713 /* Update outer/inner flops */
1715 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*24 + inneriter*246);