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_GeomW3P1_VF_sse2_single
38 * Electrostatics interaction: Ewald
39 * VdW interaction: LennardJones
40 * Geometry: Water3-Particle
41 * Calculate force/pot: PotentialAndForce
44 nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_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 j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
62 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
63 real shX,shY,shZ,rcutoff_scalar;
64 real *shiftvec,*fshift,*x,*f;
65 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
67 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
69 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
71 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
72 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
73 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
74 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
75 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
76 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
77 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
80 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
83 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
84 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
86 __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
88 __m128 rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
89 real rswitch_scalar,d_scalar;
90 __m128 dummy_mask,cutoff_mask;
91 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
92 __m128 one = _mm_set1_ps(1.0);
93 __m128 two = _mm_set1_ps(2.0);
99 jindex = nlist->jindex;
101 shiftidx = nlist->shift;
103 shiftvec = fr->shift_vec[0];
104 fshift = fr->fshift[0];
105 facel = _mm_set1_ps(fr->epsfac);
106 charge = mdatoms->chargeA;
107 nvdwtype = fr->ntype;
109 vdwtype = mdatoms->typeA;
111 sh_ewald = _mm_set1_ps(fr->ic->sh_ewald);
112 ewtab = fr->ic->tabq_coul_FDV0;
113 ewtabscale = _mm_set1_ps(fr->ic->tabq_scale);
114 ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale);
116 /* Setup water-specific parameters */
117 inr = nlist->iinr[0];
118 iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
119 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
120 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
121 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
123 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
124 rcutoff_scalar = fr->rcoulomb;
125 rcutoff = _mm_set1_ps(rcutoff_scalar);
126 rcutoff2 = _mm_mul_ps(rcutoff,rcutoff);
128 rswitch_scalar = fr->rcoulomb_switch;
129 rswitch = _mm_set1_ps(rswitch_scalar);
130 /* Setup switch parameters */
131 d_scalar = rcutoff_scalar-rswitch_scalar;
132 d = _mm_set1_ps(d_scalar);
133 swV3 = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
134 swV4 = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
135 swV5 = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
136 swF2 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
137 swF3 = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
138 swF4 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
140 /* Avoid stupid compiler warnings */
141 jnrA = jnrB = jnrC = jnrD = 0;
150 /* Start outer loop over neighborlists */
151 for(iidx=0; iidx<nri; iidx++)
153 /* Load shift vector for this list */
154 i_shift_offset = DIM*shiftidx[iidx];
155 shX = shiftvec[i_shift_offset+XX];
156 shY = shiftvec[i_shift_offset+YY];
157 shZ = shiftvec[i_shift_offset+ZZ];
159 /* Load limits for loop over neighbors */
160 j_index_start = jindex[iidx];
161 j_index_end = jindex[iidx+1];
163 /* Get outer coordinate index */
165 i_coord_offset = DIM*inr;
167 /* Load i particle coords and add shift vector */
168 ix0 = _mm_set1_ps(shX + x[i_coord_offset+DIM*0+XX]);
169 iy0 = _mm_set1_ps(shY + x[i_coord_offset+DIM*0+YY]);
170 iz0 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*0+ZZ]);
171 ix1 = _mm_set1_ps(shX + x[i_coord_offset+DIM*1+XX]);
172 iy1 = _mm_set1_ps(shY + x[i_coord_offset+DIM*1+YY]);
173 iz1 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*1+ZZ]);
174 ix2 = _mm_set1_ps(shX + x[i_coord_offset+DIM*2+XX]);
175 iy2 = _mm_set1_ps(shY + x[i_coord_offset+DIM*2+YY]);
176 iz2 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*2+ZZ]);
178 fix0 = _mm_setzero_ps();
179 fiy0 = _mm_setzero_ps();
180 fiz0 = _mm_setzero_ps();
181 fix1 = _mm_setzero_ps();
182 fiy1 = _mm_setzero_ps();
183 fiz1 = _mm_setzero_ps();
184 fix2 = _mm_setzero_ps();
185 fiy2 = _mm_setzero_ps();
186 fiz2 = _mm_setzero_ps();
188 /* Reset potential sums */
189 velecsum = _mm_setzero_ps();
190 vvdwsum = _mm_setzero_ps();
192 /* Start inner kernel loop */
193 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
196 /* Get j neighbor index, and coordinate index */
202 j_coord_offsetA = DIM*jnrA;
203 j_coord_offsetB = DIM*jnrB;
204 j_coord_offsetC = DIM*jnrC;
205 j_coord_offsetD = DIM*jnrD;
207 /* load j atom coordinates */
208 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
209 x+j_coord_offsetC,x+j_coord_offsetD,
212 /* Calculate displacement vector */
213 dx00 = _mm_sub_ps(ix0,jx0);
214 dy00 = _mm_sub_ps(iy0,jy0);
215 dz00 = _mm_sub_ps(iz0,jz0);
216 dx10 = _mm_sub_ps(ix1,jx0);
217 dy10 = _mm_sub_ps(iy1,jy0);
218 dz10 = _mm_sub_ps(iz1,jz0);
219 dx20 = _mm_sub_ps(ix2,jx0);
220 dy20 = _mm_sub_ps(iy2,jy0);
221 dz20 = _mm_sub_ps(iz2,jz0);
223 /* Calculate squared distance and things based on it */
224 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
225 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
226 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
228 rinv00 = gmx_mm_invsqrt_ps(rsq00);
229 rinv10 = gmx_mm_invsqrt_ps(rsq10);
230 rinv20 = gmx_mm_invsqrt_ps(rsq20);
232 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
233 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
234 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
236 /* Load parameters for j particles */
237 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
238 charge+jnrC+0,charge+jnrD+0);
239 vdwjidx0A = 2*vdwtype[jnrA+0];
240 vdwjidx0B = 2*vdwtype[jnrB+0];
241 vdwjidx0C = 2*vdwtype[jnrC+0];
242 vdwjidx0D = 2*vdwtype[jnrD+0];
244 /**************************
245 * CALCULATE INTERACTIONS *
246 **************************/
248 if (gmx_mm_any_lt(rsq00,rcutoff2))
251 r00 = _mm_mul_ps(rsq00,rinv00);
253 /* Compute parameters for interactions between i and j atoms */
254 qq00 = _mm_mul_ps(iq0,jq0);
255 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
256 vdwparam+vdwioffset0+vdwjidx0B,
257 vdwparam+vdwioffset0+vdwjidx0C,
258 vdwparam+vdwioffset0+vdwjidx0D,
261 /* EWALD ELECTROSTATICS */
263 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
264 ewrt = _mm_mul_ps(r00,ewtabscale);
265 ewitab = _mm_cvttps_epi32(ewrt);
266 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
267 ewitab = _mm_slli_epi32(ewitab,2);
268 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
269 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
270 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
271 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
272 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
273 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
274 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
275 velec = _mm_mul_ps(qq00,_mm_sub_ps(rinv00,velec));
276 felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
278 /* LENNARD-JONES DISPERSION/REPULSION */
280 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
281 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
282 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
283 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
284 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
286 d = _mm_sub_ps(r00,rswitch);
287 d = _mm_max_ps(d,_mm_setzero_ps());
288 d2 = _mm_mul_ps(d,d);
289 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)))))));
291 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
293 /* Evaluate switch function */
294 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
295 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
296 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
297 velec = _mm_mul_ps(velec,sw);
298 vvdw = _mm_mul_ps(vvdw,sw);
299 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
301 /* Update potential sum for this i atom from the interaction with this j atom. */
302 velec = _mm_and_ps(velec,cutoff_mask);
303 velecsum = _mm_add_ps(velecsum,velec);
304 vvdw = _mm_and_ps(vvdw,cutoff_mask);
305 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
307 fscal = _mm_add_ps(felec,fvdw);
309 fscal = _mm_and_ps(fscal,cutoff_mask);
311 /* Calculate temporary vectorial force */
312 tx = _mm_mul_ps(fscal,dx00);
313 ty = _mm_mul_ps(fscal,dy00);
314 tz = _mm_mul_ps(fscal,dz00);
316 /* Update vectorial force */
317 fix0 = _mm_add_ps(fix0,tx);
318 fiy0 = _mm_add_ps(fiy0,ty);
319 fiz0 = _mm_add_ps(fiz0,tz);
321 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
322 f+j_coord_offsetC,f+j_coord_offsetD,
327 /**************************
328 * CALCULATE INTERACTIONS *
329 **************************/
331 if (gmx_mm_any_lt(rsq10,rcutoff2))
334 r10 = _mm_mul_ps(rsq10,rinv10);
336 /* Compute parameters for interactions between i and j atoms */
337 qq10 = _mm_mul_ps(iq1,jq0);
339 /* EWALD ELECTROSTATICS */
341 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
342 ewrt = _mm_mul_ps(r10,ewtabscale);
343 ewitab = _mm_cvttps_epi32(ewrt);
344 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
345 ewitab = _mm_slli_epi32(ewitab,2);
346 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
347 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
348 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
349 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
350 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
351 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
352 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
353 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
354 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
356 d = _mm_sub_ps(r10,rswitch);
357 d = _mm_max_ps(d,_mm_setzero_ps());
358 d2 = _mm_mul_ps(d,d);
359 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)))))));
361 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
363 /* Evaluate switch function */
364 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
365 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
366 velec = _mm_mul_ps(velec,sw);
367 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
369 /* Update potential sum for this i atom from the interaction with this j atom. */
370 velec = _mm_and_ps(velec,cutoff_mask);
371 velecsum = _mm_add_ps(velecsum,velec);
375 fscal = _mm_and_ps(fscal,cutoff_mask);
377 /* Calculate temporary vectorial force */
378 tx = _mm_mul_ps(fscal,dx10);
379 ty = _mm_mul_ps(fscal,dy10);
380 tz = _mm_mul_ps(fscal,dz10);
382 /* Update vectorial force */
383 fix1 = _mm_add_ps(fix1,tx);
384 fiy1 = _mm_add_ps(fiy1,ty);
385 fiz1 = _mm_add_ps(fiz1,tz);
387 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
388 f+j_coord_offsetC,f+j_coord_offsetD,
393 /**************************
394 * CALCULATE INTERACTIONS *
395 **************************/
397 if (gmx_mm_any_lt(rsq20,rcutoff2))
400 r20 = _mm_mul_ps(rsq20,rinv20);
402 /* Compute parameters for interactions between i and j atoms */
403 qq20 = _mm_mul_ps(iq2,jq0);
405 /* EWALD ELECTROSTATICS */
407 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
408 ewrt = _mm_mul_ps(r20,ewtabscale);
409 ewitab = _mm_cvttps_epi32(ewrt);
410 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
411 ewitab = _mm_slli_epi32(ewitab,2);
412 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
413 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
414 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
415 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
416 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
417 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
418 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
419 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
420 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
422 d = _mm_sub_ps(r20,rswitch);
423 d = _mm_max_ps(d,_mm_setzero_ps());
424 d2 = _mm_mul_ps(d,d);
425 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)))))));
427 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
429 /* Evaluate switch function */
430 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
431 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
432 velec = _mm_mul_ps(velec,sw);
433 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
435 /* Update potential sum for this i atom from the interaction with this j atom. */
436 velec = _mm_and_ps(velec,cutoff_mask);
437 velecsum = _mm_add_ps(velecsum,velec);
441 fscal = _mm_and_ps(fscal,cutoff_mask);
443 /* Calculate temporary vectorial force */
444 tx = _mm_mul_ps(fscal,dx20);
445 ty = _mm_mul_ps(fscal,dy20);
446 tz = _mm_mul_ps(fscal,dz20);
448 /* Update vectorial force */
449 fix2 = _mm_add_ps(fix2,tx);
450 fiy2 = _mm_add_ps(fiy2,ty);
451 fiz2 = _mm_add_ps(fiz2,tz);
453 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
454 f+j_coord_offsetC,f+j_coord_offsetD,
459 /* Inner loop uses 213 flops */
465 /* Get j neighbor index, and coordinate index */
471 /* Sign of each element will be negative for non-real atoms.
472 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
473 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
475 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
476 jnrA = (jnrA>=0) ? jnrA : 0;
477 jnrB = (jnrB>=0) ? jnrB : 0;
478 jnrC = (jnrC>=0) ? jnrC : 0;
479 jnrD = (jnrD>=0) ? jnrD : 0;
481 j_coord_offsetA = DIM*jnrA;
482 j_coord_offsetB = DIM*jnrB;
483 j_coord_offsetC = DIM*jnrC;
484 j_coord_offsetD = DIM*jnrD;
486 /* load j atom coordinates */
487 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
488 x+j_coord_offsetC,x+j_coord_offsetD,
491 /* Calculate displacement vector */
492 dx00 = _mm_sub_ps(ix0,jx0);
493 dy00 = _mm_sub_ps(iy0,jy0);
494 dz00 = _mm_sub_ps(iz0,jz0);
495 dx10 = _mm_sub_ps(ix1,jx0);
496 dy10 = _mm_sub_ps(iy1,jy0);
497 dz10 = _mm_sub_ps(iz1,jz0);
498 dx20 = _mm_sub_ps(ix2,jx0);
499 dy20 = _mm_sub_ps(iy2,jy0);
500 dz20 = _mm_sub_ps(iz2,jz0);
502 /* Calculate squared distance and things based on it */
503 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
504 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
505 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
507 rinv00 = gmx_mm_invsqrt_ps(rsq00);
508 rinv10 = gmx_mm_invsqrt_ps(rsq10);
509 rinv20 = gmx_mm_invsqrt_ps(rsq20);
511 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
512 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
513 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
515 /* Load parameters for j particles */
516 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
517 charge+jnrC+0,charge+jnrD+0);
518 vdwjidx0A = 2*vdwtype[jnrA+0];
519 vdwjidx0B = 2*vdwtype[jnrB+0];
520 vdwjidx0C = 2*vdwtype[jnrC+0];
521 vdwjidx0D = 2*vdwtype[jnrD+0];
523 /**************************
524 * CALCULATE INTERACTIONS *
525 **************************/
527 if (gmx_mm_any_lt(rsq00,rcutoff2))
530 r00 = _mm_mul_ps(rsq00,rinv00);
531 r00 = _mm_andnot_ps(dummy_mask,r00);
533 /* Compute parameters for interactions between i and j atoms */
534 qq00 = _mm_mul_ps(iq0,jq0);
535 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
536 vdwparam+vdwioffset0+vdwjidx0B,
537 vdwparam+vdwioffset0+vdwjidx0C,
538 vdwparam+vdwioffset0+vdwjidx0D,
541 /* EWALD ELECTROSTATICS */
543 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
544 ewrt = _mm_mul_ps(r00,ewtabscale);
545 ewitab = _mm_cvttps_epi32(ewrt);
546 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
547 ewitab = _mm_slli_epi32(ewitab,2);
548 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
549 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
550 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
551 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
552 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
553 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
554 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
555 velec = _mm_mul_ps(qq00,_mm_sub_ps(rinv00,velec));
556 felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
558 /* LENNARD-JONES DISPERSION/REPULSION */
560 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
561 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
562 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
563 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
564 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
566 d = _mm_sub_ps(r00,rswitch);
567 d = _mm_max_ps(d,_mm_setzero_ps());
568 d2 = _mm_mul_ps(d,d);
569 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)))))));
571 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
573 /* Evaluate switch function */
574 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
575 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
576 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
577 velec = _mm_mul_ps(velec,sw);
578 vvdw = _mm_mul_ps(vvdw,sw);
579 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
581 /* Update potential sum for this i atom from the interaction with this j atom. */
582 velec = _mm_and_ps(velec,cutoff_mask);
583 velec = _mm_andnot_ps(dummy_mask,velec);
584 velecsum = _mm_add_ps(velecsum,velec);
585 vvdw = _mm_and_ps(vvdw,cutoff_mask);
586 vvdw = _mm_andnot_ps(dummy_mask,vvdw);
587 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
589 fscal = _mm_add_ps(felec,fvdw);
591 fscal = _mm_and_ps(fscal,cutoff_mask);
593 fscal = _mm_andnot_ps(dummy_mask,fscal);
595 /* Calculate temporary vectorial force */
596 tx = _mm_mul_ps(fscal,dx00);
597 ty = _mm_mul_ps(fscal,dy00);
598 tz = _mm_mul_ps(fscal,dz00);
600 /* Update vectorial force */
601 fix0 = _mm_add_ps(fix0,tx);
602 fiy0 = _mm_add_ps(fiy0,ty);
603 fiz0 = _mm_add_ps(fiz0,tz);
605 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
606 f+j_coord_offsetC,f+j_coord_offsetD,
611 /**************************
612 * CALCULATE INTERACTIONS *
613 **************************/
615 if (gmx_mm_any_lt(rsq10,rcutoff2))
618 r10 = _mm_mul_ps(rsq10,rinv10);
619 r10 = _mm_andnot_ps(dummy_mask,r10);
621 /* Compute parameters for interactions between i and j atoms */
622 qq10 = _mm_mul_ps(iq1,jq0);
624 /* EWALD ELECTROSTATICS */
626 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
627 ewrt = _mm_mul_ps(r10,ewtabscale);
628 ewitab = _mm_cvttps_epi32(ewrt);
629 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
630 ewitab = _mm_slli_epi32(ewitab,2);
631 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
632 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
633 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
634 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
635 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
636 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
637 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
638 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
639 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
641 d = _mm_sub_ps(r10,rswitch);
642 d = _mm_max_ps(d,_mm_setzero_ps());
643 d2 = _mm_mul_ps(d,d);
644 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)))))));
646 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
648 /* Evaluate switch function */
649 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
650 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
651 velec = _mm_mul_ps(velec,sw);
652 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
654 /* Update potential sum for this i atom from the interaction with this j atom. */
655 velec = _mm_and_ps(velec,cutoff_mask);
656 velec = _mm_andnot_ps(dummy_mask,velec);
657 velecsum = _mm_add_ps(velecsum,velec);
661 fscal = _mm_and_ps(fscal,cutoff_mask);
663 fscal = _mm_andnot_ps(dummy_mask,fscal);
665 /* Calculate temporary vectorial force */
666 tx = _mm_mul_ps(fscal,dx10);
667 ty = _mm_mul_ps(fscal,dy10);
668 tz = _mm_mul_ps(fscal,dz10);
670 /* Update vectorial force */
671 fix1 = _mm_add_ps(fix1,tx);
672 fiy1 = _mm_add_ps(fiy1,ty);
673 fiz1 = _mm_add_ps(fiz1,tz);
675 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
676 f+j_coord_offsetC,f+j_coord_offsetD,
681 /**************************
682 * CALCULATE INTERACTIONS *
683 **************************/
685 if (gmx_mm_any_lt(rsq20,rcutoff2))
688 r20 = _mm_mul_ps(rsq20,rinv20);
689 r20 = _mm_andnot_ps(dummy_mask,r20);
691 /* Compute parameters for interactions between i and j atoms */
692 qq20 = _mm_mul_ps(iq2,jq0);
694 /* EWALD ELECTROSTATICS */
696 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
697 ewrt = _mm_mul_ps(r20,ewtabscale);
698 ewitab = _mm_cvttps_epi32(ewrt);
699 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
700 ewitab = _mm_slli_epi32(ewitab,2);
701 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
702 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
703 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
704 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
705 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
706 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
707 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
708 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
709 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
711 d = _mm_sub_ps(r20,rswitch);
712 d = _mm_max_ps(d,_mm_setzero_ps());
713 d2 = _mm_mul_ps(d,d);
714 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)))))));
716 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
718 /* Evaluate switch function */
719 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
720 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
721 velec = _mm_mul_ps(velec,sw);
722 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
724 /* Update potential sum for this i atom from the interaction with this j atom. */
725 velec = _mm_and_ps(velec,cutoff_mask);
726 velec = _mm_andnot_ps(dummy_mask,velec);
727 velecsum = _mm_add_ps(velecsum,velec);
731 fscal = _mm_and_ps(fscal,cutoff_mask);
733 fscal = _mm_andnot_ps(dummy_mask,fscal);
735 /* Calculate temporary vectorial force */
736 tx = _mm_mul_ps(fscal,dx20);
737 ty = _mm_mul_ps(fscal,dy20);
738 tz = _mm_mul_ps(fscal,dz20);
740 /* Update vectorial force */
741 fix2 = _mm_add_ps(fix2,tx);
742 fiy2 = _mm_add_ps(fiy2,ty);
743 fiz2 = _mm_add_ps(fiz2,tz);
745 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
746 f+j_coord_offsetC,f+j_coord_offsetD,
751 /* Inner loop uses 216 flops */
754 /* End of innermost loop */
756 gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
757 f+i_coord_offset,fshift+i_shift_offset);
760 /* Update potential energies */
761 gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
762 gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
764 /* Increment number of inner iterations */
765 inneriter += j_index_end - j_index_start;
767 /* Outer loop uses 29 flops */
770 /* Increment number of outer iterations */
773 /* Update outer/inner flops */
775 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*29 + inneriter*216);
778 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_F_sse2_single
779 * Electrostatics interaction: Ewald
780 * VdW interaction: LennardJones
781 * Geometry: Water3-Particle
782 * Calculate force/pot: Force
785 nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_F_sse2_single
786 (t_nblist * gmx_restrict nlist,
787 rvec * gmx_restrict xx,
788 rvec * gmx_restrict ff,
789 t_forcerec * gmx_restrict fr,
790 t_mdatoms * gmx_restrict mdatoms,
791 nb_kernel_data_t * gmx_restrict kernel_data,
792 t_nrnb * gmx_restrict nrnb)
794 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
795 * just 0 for non-waters.
796 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
797 * jnr indices corresponding to data put in the four positions in the SIMD register.
799 int i_shift_offset,i_coord_offset,outeriter,inneriter;
800 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
801 int jnrA,jnrB,jnrC,jnrD;
802 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
803 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
804 real shX,shY,shZ,rcutoff_scalar;
805 real *shiftvec,*fshift,*x,*f;
806 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
808 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
810 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
812 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
813 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
814 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
815 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
816 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
817 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
818 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
821 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
824 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
825 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
827 __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
829 __m128 rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
830 real rswitch_scalar,d_scalar;
831 __m128 dummy_mask,cutoff_mask;
832 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
833 __m128 one = _mm_set1_ps(1.0);
834 __m128 two = _mm_set1_ps(2.0);
840 jindex = nlist->jindex;
842 shiftidx = nlist->shift;
844 shiftvec = fr->shift_vec[0];
845 fshift = fr->fshift[0];
846 facel = _mm_set1_ps(fr->epsfac);
847 charge = mdatoms->chargeA;
848 nvdwtype = fr->ntype;
850 vdwtype = mdatoms->typeA;
852 sh_ewald = _mm_set1_ps(fr->ic->sh_ewald);
853 ewtab = fr->ic->tabq_coul_FDV0;
854 ewtabscale = _mm_set1_ps(fr->ic->tabq_scale);
855 ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale);
857 /* Setup water-specific parameters */
858 inr = nlist->iinr[0];
859 iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
860 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
861 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
862 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
864 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
865 rcutoff_scalar = fr->rcoulomb;
866 rcutoff = _mm_set1_ps(rcutoff_scalar);
867 rcutoff2 = _mm_mul_ps(rcutoff,rcutoff);
869 rswitch_scalar = fr->rcoulomb_switch;
870 rswitch = _mm_set1_ps(rswitch_scalar);
871 /* Setup switch parameters */
872 d_scalar = rcutoff_scalar-rswitch_scalar;
873 d = _mm_set1_ps(d_scalar);
874 swV3 = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
875 swV4 = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
876 swV5 = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
877 swF2 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
878 swF3 = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
879 swF4 = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
881 /* Avoid stupid compiler warnings */
882 jnrA = jnrB = jnrC = jnrD = 0;
891 /* Start outer loop over neighborlists */
892 for(iidx=0; iidx<nri; iidx++)
894 /* Load shift vector for this list */
895 i_shift_offset = DIM*shiftidx[iidx];
896 shX = shiftvec[i_shift_offset+XX];
897 shY = shiftvec[i_shift_offset+YY];
898 shZ = shiftvec[i_shift_offset+ZZ];
900 /* Load limits for loop over neighbors */
901 j_index_start = jindex[iidx];
902 j_index_end = jindex[iidx+1];
904 /* Get outer coordinate index */
906 i_coord_offset = DIM*inr;
908 /* Load i particle coords and add shift vector */
909 ix0 = _mm_set1_ps(shX + x[i_coord_offset+DIM*0+XX]);
910 iy0 = _mm_set1_ps(shY + x[i_coord_offset+DIM*0+YY]);
911 iz0 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*0+ZZ]);
912 ix1 = _mm_set1_ps(shX + x[i_coord_offset+DIM*1+XX]);
913 iy1 = _mm_set1_ps(shY + x[i_coord_offset+DIM*1+YY]);
914 iz1 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*1+ZZ]);
915 ix2 = _mm_set1_ps(shX + x[i_coord_offset+DIM*2+XX]);
916 iy2 = _mm_set1_ps(shY + x[i_coord_offset+DIM*2+YY]);
917 iz2 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*2+ZZ]);
919 fix0 = _mm_setzero_ps();
920 fiy0 = _mm_setzero_ps();
921 fiz0 = _mm_setzero_ps();
922 fix1 = _mm_setzero_ps();
923 fiy1 = _mm_setzero_ps();
924 fiz1 = _mm_setzero_ps();
925 fix2 = _mm_setzero_ps();
926 fiy2 = _mm_setzero_ps();
927 fiz2 = _mm_setzero_ps();
929 /* Start inner kernel loop */
930 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
933 /* Get j neighbor index, and coordinate index */
939 j_coord_offsetA = DIM*jnrA;
940 j_coord_offsetB = DIM*jnrB;
941 j_coord_offsetC = DIM*jnrC;
942 j_coord_offsetD = DIM*jnrD;
944 /* load j atom coordinates */
945 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
946 x+j_coord_offsetC,x+j_coord_offsetD,
949 /* Calculate displacement vector */
950 dx00 = _mm_sub_ps(ix0,jx0);
951 dy00 = _mm_sub_ps(iy0,jy0);
952 dz00 = _mm_sub_ps(iz0,jz0);
953 dx10 = _mm_sub_ps(ix1,jx0);
954 dy10 = _mm_sub_ps(iy1,jy0);
955 dz10 = _mm_sub_ps(iz1,jz0);
956 dx20 = _mm_sub_ps(ix2,jx0);
957 dy20 = _mm_sub_ps(iy2,jy0);
958 dz20 = _mm_sub_ps(iz2,jz0);
960 /* Calculate squared distance and things based on it */
961 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
962 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
963 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
965 rinv00 = gmx_mm_invsqrt_ps(rsq00);
966 rinv10 = gmx_mm_invsqrt_ps(rsq10);
967 rinv20 = gmx_mm_invsqrt_ps(rsq20);
969 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
970 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
971 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
973 /* Load parameters for j particles */
974 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
975 charge+jnrC+0,charge+jnrD+0);
976 vdwjidx0A = 2*vdwtype[jnrA+0];
977 vdwjidx0B = 2*vdwtype[jnrB+0];
978 vdwjidx0C = 2*vdwtype[jnrC+0];
979 vdwjidx0D = 2*vdwtype[jnrD+0];
981 /**************************
982 * CALCULATE INTERACTIONS *
983 **************************/
985 if (gmx_mm_any_lt(rsq00,rcutoff2))
988 r00 = _mm_mul_ps(rsq00,rinv00);
990 /* Compute parameters for interactions between i and j atoms */
991 qq00 = _mm_mul_ps(iq0,jq0);
992 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
993 vdwparam+vdwioffset0+vdwjidx0B,
994 vdwparam+vdwioffset0+vdwjidx0C,
995 vdwparam+vdwioffset0+vdwjidx0D,
998 /* EWALD ELECTROSTATICS */
1000 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1001 ewrt = _mm_mul_ps(r00,ewtabscale);
1002 ewitab = _mm_cvttps_epi32(ewrt);
1003 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1004 ewitab = _mm_slli_epi32(ewitab,2);
1005 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1006 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1007 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1008 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1009 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1010 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1011 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1012 velec = _mm_mul_ps(qq00,_mm_sub_ps(rinv00,velec));
1013 felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
1015 /* LENNARD-JONES DISPERSION/REPULSION */
1017 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1018 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
1019 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
1020 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
1021 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
1023 d = _mm_sub_ps(r00,rswitch);
1024 d = _mm_max_ps(d,_mm_setzero_ps());
1025 d2 = _mm_mul_ps(d,d);
1026 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)))))));
1028 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1030 /* Evaluate switch function */
1031 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1032 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
1033 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
1034 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
1036 fscal = _mm_add_ps(felec,fvdw);
1038 fscal = _mm_and_ps(fscal,cutoff_mask);
1040 /* Calculate temporary vectorial force */
1041 tx = _mm_mul_ps(fscal,dx00);
1042 ty = _mm_mul_ps(fscal,dy00);
1043 tz = _mm_mul_ps(fscal,dz00);
1045 /* Update vectorial force */
1046 fix0 = _mm_add_ps(fix0,tx);
1047 fiy0 = _mm_add_ps(fiy0,ty);
1048 fiz0 = _mm_add_ps(fiz0,tz);
1050 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
1051 f+j_coord_offsetC,f+j_coord_offsetD,
1056 /**************************
1057 * CALCULATE INTERACTIONS *
1058 **************************/
1060 if (gmx_mm_any_lt(rsq10,rcutoff2))
1063 r10 = _mm_mul_ps(rsq10,rinv10);
1065 /* Compute parameters for interactions between i and j atoms */
1066 qq10 = _mm_mul_ps(iq1,jq0);
1068 /* EWALD ELECTROSTATICS */
1070 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1071 ewrt = _mm_mul_ps(r10,ewtabscale);
1072 ewitab = _mm_cvttps_epi32(ewrt);
1073 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1074 ewitab = _mm_slli_epi32(ewitab,2);
1075 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1076 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1077 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1078 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1079 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1080 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1081 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1082 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
1083 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1085 d = _mm_sub_ps(r10,rswitch);
1086 d = _mm_max_ps(d,_mm_setzero_ps());
1087 d2 = _mm_mul_ps(d,d);
1088 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)))))));
1090 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1092 /* Evaluate switch function */
1093 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1094 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
1095 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
1099 fscal = _mm_and_ps(fscal,cutoff_mask);
1101 /* Calculate temporary vectorial force */
1102 tx = _mm_mul_ps(fscal,dx10);
1103 ty = _mm_mul_ps(fscal,dy10);
1104 tz = _mm_mul_ps(fscal,dz10);
1106 /* Update vectorial force */
1107 fix1 = _mm_add_ps(fix1,tx);
1108 fiy1 = _mm_add_ps(fiy1,ty);
1109 fiz1 = _mm_add_ps(fiz1,tz);
1111 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
1112 f+j_coord_offsetC,f+j_coord_offsetD,
1117 /**************************
1118 * CALCULATE INTERACTIONS *
1119 **************************/
1121 if (gmx_mm_any_lt(rsq20,rcutoff2))
1124 r20 = _mm_mul_ps(rsq20,rinv20);
1126 /* Compute parameters for interactions between i and j atoms */
1127 qq20 = _mm_mul_ps(iq2,jq0);
1129 /* EWALD ELECTROSTATICS */
1131 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1132 ewrt = _mm_mul_ps(r20,ewtabscale);
1133 ewitab = _mm_cvttps_epi32(ewrt);
1134 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1135 ewitab = _mm_slli_epi32(ewitab,2);
1136 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1137 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1138 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1139 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1140 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1141 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1142 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1143 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
1144 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
1146 d = _mm_sub_ps(r20,rswitch);
1147 d = _mm_max_ps(d,_mm_setzero_ps());
1148 d2 = _mm_mul_ps(d,d);
1149 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)))))));
1151 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1153 /* Evaluate switch function */
1154 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1155 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
1156 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
1160 fscal = _mm_and_ps(fscal,cutoff_mask);
1162 /* Calculate temporary vectorial force */
1163 tx = _mm_mul_ps(fscal,dx20);
1164 ty = _mm_mul_ps(fscal,dy20);
1165 tz = _mm_mul_ps(fscal,dz20);
1167 /* Update vectorial force */
1168 fix2 = _mm_add_ps(fix2,tx);
1169 fiy2 = _mm_add_ps(fiy2,ty);
1170 fiz2 = _mm_add_ps(fiz2,tz);
1172 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
1173 f+j_coord_offsetC,f+j_coord_offsetD,
1178 /* Inner loop uses 201 flops */
1181 if(jidx<j_index_end)
1184 /* Get j neighbor index, and coordinate index */
1186 jnrB = jjnr[jidx+1];
1187 jnrC = jjnr[jidx+2];
1188 jnrD = jjnr[jidx+3];
1190 /* Sign of each element will be negative for non-real atoms.
1191 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1192 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1194 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1195 jnrA = (jnrA>=0) ? jnrA : 0;
1196 jnrB = (jnrB>=0) ? jnrB : 0;
1197 jnrC = (jnrC>=0) ? jnrC : 0;
1198 jnrD = (jnrD>=0) ? jnrD : 0;
1200 j_coord_offsetA = DIM*jnrA;
1201 j_coord_offsetB = DIM*jnrB;
1202 j_coord_offsetC = DIM*jnrC;
1203 j_coord_offsetD = DIM*jnrD;
1205 /* load j atom coordinates */
1206 gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1207 x+j_coord_offsetC,x+j_coord_offsetD,
1210 /* Calculate displacement vector */
1211 dx00 = _mm_sub_ps(ix0,jx0);
1212 dy00 = _mm_sub_ps(iy0,jy0);
1213 dz00 = _mm_sub_ps(iz0,jz0);
1214 dx10 = _mm_sub_ps(ix1,jx0);
1215 dy10 = _mm_sub_ps(iy1,jy0);
1216 dz10 = _mm_sub_ps(iz1,jz0);
1217 dx20 = _mm_sub_ps(ix2,jx0);
1218 dy20 = _mm_sub_ps(iy2,jy0);
1219 dz20 = _mm_sub_ps(iz2,jz0);
1221 /* Calculate squared distance and things based on it */
1222 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1223 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1224 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1226 rinv00 = gmx_mm_invsqrt_ps(rsq00);
1227 rinv10 = gmx_mm_invsqrt_ps(rsq10);
1228 rinv20 = gmx_mm_invsqrt_ps(rsq20);
1230 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
1231 rinvsq10 = _mm_mul_ps(rinv10,rinv10);
1232 rinvsq20 = _mm_mul_ps(rinv20,rinv20);
1234 /* Load parameters for j particles */
1235 jq0 = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1236 charge+jnrC+0,charge+jnrD+0);
1237 vdwjidx0A = 2*vdwtype[jnrA+0];
1238 vdwjidx0B = 2*vdwtype[jnrB+0];
1239 vdwjidx0C = 2*vdwtype[jnrC+0];
1240 vdwjidx0D = 2*vdwtype[jnrD+0];
1242 /**************************
1243 * CALCULATE INTERACTIONS *
1244 **************************/
1246 if (gmx_mm_any_lt(rsq00,rcutoff2))
1249 r00 = _mm_mul_ps(rsq00,rinv00);
1250 r00 = _mm_andnot_ps(dummy_mask,r00);
1252 /* Compute parameters for interactions between i and j atoms */
1253 qq00 = _mm_mul_ps(iq0,jq0);
1254 gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
1255 vdwparam+vdwioffset0+vdwjidx0B,
1256 vdwparam+vdwioffset0+vdwjidx0C,
1257 vdwparam+vdwioffset0+vdwjidx0D,
1260 /* EWALD ELECTROSTATICS */
1262 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1263 ewrt = _mm_mul_ps(r00,ewtabscale);
1264 ewitab = _mm_cvttps_epi32(ewrt);
1265 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1266 ewitab = _mm_slli_epi32(ewitab,2);
1267 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1268 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1269 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1270 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1271 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1272 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1273 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1274 velec = _mm_mul_ps(qq00,_mm_sub_ps(rinv00,velec));
1275 felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
1277 /* LENNARD-JONES DISPERSION/REPULSION */
1279 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1280 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
1281 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
1282 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
1283 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
1285 d = _mm_sub_ps(r00,rswitch);
1286 d = _mm_max_ps(d,_mm_setzero_ps());
1287 d2 = _mm_mul_ps(d,d);
1288 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)))))));
1290 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1292 /* Evaluate switch function */
1293 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1294 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
1295 fvdw = _mm_sub_ps( _mm_mul_ps(fvdw,sw) , _mm_mul_ps(rinv00,_mm_mul_ps(vvdw,dsw)) );
1296 cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2);
1298 fscal = _mm_add_ps(felec,fvdw);
1300 fscal = _mm_and_ps(fscal,cutoff_mask);
1302 fscal = _mm_andnot_ps(dummy_mask,fscal);
1304 /* Calculate temporary vectorial force */
1305 tx = _mm_mul_ps(fscal,dx00);
1306 ty = _mm_mul_ps(fscal,dy00);
1307 tz = _mm_mul_ps(fscal,dz00);
1309 /* Update vectorial force */
1310 fix0 = _mm_add_ps(fix0,tx);
1311 fiy0 = _mm_add_ps(fiy0,ty);
1312 fiz0 = _mm_add_ps(fiz0,tz);
1314 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
1315 f+j_coord_offsetC,f+j_coord_offsetD,
1320 /**************************
1321 * CALCULATE INTERACTIONS *
1322 **************************/
1324 if (gmx_mm_any_lt(rsq10,rcutoff2))
1327 r10 = _mm_mul_ps(rsq10,rinv10);
1328 r10 = _mm_andnot_ps(dummy_mask,r10);
1330 /* Compute parameters for interactions between i and j atoms */
1331 qq10 = _mm_mul_ps(iq1,jq0);
1333 /* EWALD ELECTROSTATICS */
1335 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1336 ewrt = _mm_mul_ps(r10,ewtabscale);
1337 ewitab = _mm_cvttps_epi32(ewrt);
1338 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1339 ewitab = _mm_slli_epi32(ewitab,2);
1340 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1341 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1342 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1343 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1344 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1345 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1346 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1347 velec = _mm_mul_ps(qq10,_mm_sub_ps(rinv10,velec));
1348 felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1350 d = _mm_sub_ps(r10,rswitch);
1351 d = _mm_max_ps(d,_mm_setzero_ps());
1352 d2 = _mm_mul_ps(d,d);
1353 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)))))));
1355 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1357 /* Evaluate switch function */
1358 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1359 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
1360 cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2);
1364 fscal = _mm_and_ps(fscal,cutoff_mask);
1366 fscal = _mm_andnot_ps(dummy_mask,fscal);
1368 /* Calculate temporary vectorial force */
1369 tx = _mm_mul_ps(fscal,dx10);
1370 ty = _mm_mul_ps(fscal,dy10);
1371 tz = _mm_mul_ps(fscal,dz10);
1373 /* Update vectorial force */
1374 fix1 = _mm_add_ps(fix1,tx);
1375 fiy1 = _mm_add_ps(fiy1,ty);
1376 fiz1 = _mm_add_ps(fiz1,tz);
1378 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
1379 f+j_coord_offsetC,f+j_coord_offsetD,
1384 /**************************
1385 * CALCULATE INTERACTIONS *
1386 **************************/
1388 if (gmx_mm_any_lt(rsq20,rcutoff2))
1391 r20 = _mm_mul_ps(rsq20,rinv20);
1392 r20 = _mm_andnot_ps(dummy_mask,r20);
1394 /* Compute parameters for interactions between i and j atoms */
1395 qq20 = _mm_mul_ps(iq2,jq0);
1397 /* EWALD ELECTROSTATICS */
1399 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1400 ewrt = _mm_mul_ps(r20,ewtabscale);
1401 ewitab = _mm_cvttps_epi32(ewrt);
1402 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1403 ewitab = _mm_slli_epi32(ewitab,2);
1404 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1405 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1406 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1407 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1408 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1409 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1410 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1411 velec = _mm_mul_ps(qq20,_mm_sub_ps(rinv20,velec));
1412 felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
1414 d = _mm_sub_ps(r20,rswitch);
1415 d = _mm_max_ps(d,_mm_setzero_ps());
1416 d2 = _mm_mul_ps(d,d);
1417 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)))))));
1419 dsw = _mm_mul_ps(d2,_mm_add_ps(swF2,_mm_mul_ps(d,_mm_add_ps(swF3,_mm_mul_ps(d,swF4)))));
1421 /* Evaluate switch function */
1422 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1423 felec = _mm_sub_ps( _mm_mul_ps(felec,sw) , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
1424 cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2);
1428 fscal = _mm_and_ps(fscal,cutoff_mask);
1430 fscal = _mm_andnot_ps(dummy_mask,fscal);
1432 /* Calculate temporary vectorial force */
1433 tx = _mm_mul_ps(fscal,dx20);
1434 ty = _mm_mul_ps(fscal,dy20);
1435 tz = _mm_mul_ps(fscal,dz20);
1437 /* Update vectorial force */
1438 fix2 = _mm_add_ps(fix2,tx);
1439 fiy2 = _mm_add_ps(fiy2,ty);
1440 fiz2 = _mm_add_ps(fiz2,tz);
1442 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(f+j_coord_offsetA,f+j_coord_offsetB,
1443 f+j_coord_offsetC,f+j_coord_offsetD,
1448 /* Inner loop uses 204 flops */
1451 /* End of innermost loop */
1453 gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1454 f+i_coord_offset,fshift+i_shift_offset);
1456 /* Increment number of inner iterations */
1457 inneriter += j_index_end - j_index_start;
1459 /* Outer loop uses 27 flops */
1462 /* Increment number of outer iterations */
1465 /* Update outer/inner flops */
1467 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*27 + inneriter*204);