2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014,2015,2017,2018, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
36 * Note: this file was generated by the GROMACS sse2_single kernel generator.
44 #include "../nb_kernel.h"
45 #include "gromacs/gmxlib/nrnb.h"
47 #include "kernelutil_x86_sse2_single.h"
50 * Gromacs nonbonded kernel: nb_kernel_ElecCSTab_VdwLJ_GeomW3W3_VF_sse2_single
51 * Electrostatics interaction: CubicSplineTable
52 * VdW interaction: LennardJones
53 * Geometry: Water3-Water3
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecCSTab_VdwLJ_GeomW3W3_VF_sse2_single
58 (t_nblist * gmx_restrict nlist,
59 rvec * gmx_restrict xx,
60 rvec * gmx_restrict ff,
61 struct t_forcerec * gmx_restrict fr,
62 t_mdatoms * gmx_restrict mdatoms,
63 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
64 t_nrnb * gmx_restrict nrnb)
66 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
67 * just 0 for non-waters.
68 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
69 * jnr indices corresponding to data put in the four positions in the SIMD register.
71 int i_shift_offset,i_coord_offset,outeriter,inneriter;
72 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
73 int jnrA,jnrB,jnrC,jnrD;
74 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
75 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
76 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
78 real *shiftvec,*fshift,*x,*f;
79 real *fjptrA,*fjptrB,*fjptrC,*fjptrD;
81 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
83 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
85 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
87 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
88 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
89 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
90 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
91 __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
92 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
93 __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
94 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
95 __m128 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
96 __m128 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
97 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
98 __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
99 __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
100 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
101 __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
102 __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
103 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
106 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
109 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
110 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
112 __m128i ifour = _mm_set1_epi32(4);
113 __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
115 __m128 dummy_mask,cutoff_mask;
116 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
117 __m128 one = _mm_set1_ps(1.0);
118 __m128 two = _mm_set1_ps(2.0);
124 jindex = nlist->jindex;
126 shiftidx = nlist->shift;
128 shiftvec = fr->shift_vec[0];
129 fshift = fr->fshift[0];
130 facel = _mm_set1_ps(fr->ic->epsfac);
131 charge = mdatoms->chargeA;
132 nvdwtype = fr->ntype;
134 vdwtype = mdatoms->typeA;
136 vftab = kernel_data->table_elec->data;
137 vftabscale = _mm_set1_ps(kernel_data->table_elec->scale);
139 /* Setup water-specific parameters */
140 inr = nlist->iinr[0];
141 iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
142 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
143 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
144 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
146 jq0 = _mm_set1_ps(charge[inr+0]);
147 jq1 = _mm_set1_ps(charge[inr+1]);
148 jq2 = _mm_set1_ps(charge[inr+2]);
149 vdwjidx0A = 2*vdwtype[inr+0];
150 qq00 = _mm_mul_ps(iq0,jq0);
151 c6_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
152 c12_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
153 qq01 = _mm_mul_ps(iq0,jq1);
154 qq02 = _mm_mul_ps(iq0,jq2);
155 qq10 = _mm_mul_ps(iq1,jq0);
156 qq11 = _mm_mul_ps(iq1,jq1);
157 qq12 = _mm_mul_ps(iq1,jq2);
158 qq20 = _mm_mul_ps(iq2,jq0);
159 qq21 = _mm_mul_ps(iq2,jq1);
160 qq22 = _mm_mul_ps(iq2,jq2);
162 /* Avoid stupid compiler warnings */
163 jnrA = jnrB = jnrC = jnrD = 0;
172 for(iidx=0;iidx<4*DIM;iidx++)
177 /* Start outer loop over neighborlists */
178 for(iidx=0; iidx<nri; iidx++)
180 /* Load shift vector for this list */
181 i_shift_offset = DIM*shiftidx[iidx];
183 /* Load limits for loop over neighbors */
184 j_index_start = jindex[iidx];
185 j_index_end = jindex[iidx+1];
187 /* Get outer coordinate index */
189 i_coord_offset = DIM*inr;
191 /* Load i particle coords and add shift vector */
192 gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
193 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
195 fix0 = _mm_setzero_ps();
196 fiy0 = _mm_setzero_ps();
197 fiz0 = _mm_setzero_ps();
198 fix1 = _mm_setzero_ps();
199 fiy1 = _mm_setzero_ps();
200 fiz1 = _mm_setzero_ps();
201 fix2 = _mm_setzero_ps();
202 fiy2 = _mm_setzero_ps();
203 fiz2 = _mm_setzero_ps();
205 /* Reset potential sums */
206 velecsum = _mm_setzero_ps();
207 vvdwsum = _mm_setzero_ps();
209 /* Start inner kernel loop */
210 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
213 /* Get j neighbor index, and coordinate index */
218 j_coord_offsetA = DIM*jnrA;
219 j_coord_offsetB = DIM*jnrB;
220 j_coord_offsetC = DIM*jnrC;
221 j_coord_offsetD = DIM*jnrD;
223 /* load j atom coordinates */
224 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
225 x+j_coord_offsetC,x+j_coord_offsetD,
226 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
228 /* Calculate displacement vector */
229 dx00 = _mm_sub_ps(ix0,jx0);
230 dy00 = _mm_sub_ps(iy0,jy0);
231 dz00 = _mm_sub_ps(iz0,jz0);
232 dx01 = _mm_sub_ps(ix0,jx1);
233 dy01 = _mm_sub_ps(iy0,jy1);
234 dz01 = _mm_sub_ps(iz0,jz1);
235 dx02 = _mm_sub_ps(ix0,jx2);
236 dy02 = _mm_sub_ps(iy0,jy2);
237 dz02 = _mm_sub_ps(iz0,jz2);
238 dx10 = _mm_sub_ps(ix1,jx0);
239 dy10 = _mm_sub_ps(iy1,jy0);
240 dz10 = _mm_sub_ps(iz1,jz0);
241 dx11 = _mm_sub_ps(ix1,jx1);
242 dy11 = _mm_sub_ps(iy1,jy1);
243 dz11 = _mm_sub_ps(iz1,jz1);
244 dx12 = _mm_sub_ps(ix1,jx2);
245 dy12 = _mm_sub_ps(iy1,jy2);
246 dz12 = _mm_sub_ps(iz1,jz2);
247 dx20 = _mm_sub_ps(ix2,jx0);
248 dy20 = _mm_sub_ps(iy2,jy0);
249 dz20 = _mm_sub_ps(iz2,jz0);
250 dx21 = _mm_sub_ps(ix2,jx1);
251 dy21 = _mm_sub_ps(iy2,jy1);
252 dz21 = _mm_sub_ps(iz2,jz1);
253 dx22 = _mm_sub_ps(ix2,jx2);
254 dy22 = _mm_sub_ps(iy2,jy2);
255 dz22 = _mm_sub_ps(iz2,jz2);
257 /* Calculate squared distance and things based on it */
258 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
259 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
260 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
261 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
262 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
263 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
264 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
265 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
266 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
268 rinv00 = sse2_invsqrt_f(rsq00);
269 rinv01 = sse2_invsqrt_f(rsq01);
270 rinv02 = sse2_invsqrt_f(rsq02);
271 rinv10 = sse2_invsqrt_f(rsq10);
272 rinv11 = sse2_invsqrt_f(rsq11);
273 rinv12 = sse2_invsqrt_f(rsq12);
274 rinv20 = sse2_invsqrt_f(rsq20);
275 rinv21 = sse2_invsqrt_f(rsq21);
276 rinv22 = sse2_invsqrt_f(rsq22);
278 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
280 fjx0 = _mm_setzero_ps();
281 fjy0 = _mm_setzero_ps();
282 fjz0 = _mm_setzero_ps();
283 fjx1 = _mm_setzero_ps();
284 fjy1 = _mm_setzero_ps();
285 fjz1 = _mm_setzero_ps();
286 fjx2 = _mm_setzero_ps();
287 fjy2 = _mm_setzero_ps();
288 fjz2 = _mm_setzero_ps();
290 /**************************
291 * CALCULATE INTERACTIONS *
292 **************************/
294 r00 = _mm_mul_ps(rsq00,rinv00);
296 /* Calculate table index by multiplying r with table scale and truncate to integer */
297 rt = _mm_mul_ps(r00,vftabscale);
298 vfitab = _mm_cvttps_epi32(rt);
299 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
300 vfitab = _mm_slli_epi32(vfitab,2);
302 /* CUBIC SPLINE TABLE ELECTROSTATICS */
303 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
304 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
305 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
306 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
307 _MM_TRANSPOSE4_PS(Y,F,G,H);
308 Heps = _mm_mul_ps(vfeps,H);
309 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
310 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
311 velec = _mm_mul_ps(qq00,VV);
312 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
313 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq00,FF),_mm_mul_ps(vftabscale,rinv00)));
315 /* LENNARD-JONES DISPERSION/REPULSION */
317 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
318 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
319 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
320 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
321 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
323 /* Update potential sum for this i atom from the interaction with this j atom. */
324 velecsum = _mm_add_ps(velecsum,velec);
325 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
327 fscal = _mm_add_ps(felec,fvdw);
329 /* Calculate temporary vectorial force */
330 tx = _mm_mul_ps(fscal,dx00);
331 ty = _mm_mul_ps(fscal,dy00);
332 tz = _mm_mul_ps(fscal,dz00);
334 /* Update vectorial force */
335 fix0 = _mm_add_ps(fix0,tx);
336 fiy0 = _mm_add_ps(fiy0,ty);
337 fiz0 = _mm_add_ps(fiz0,tz);
339 fjx0 = _mm_add_ps(fjx0,tx);
340 fjy0 = _mm_add_ps(fjy0,ty);
341 fjz0 = _mm_add_ps(fjz0,tz);
343 /**************************
344 * CALCULATE INTERACTIONS *
345 **************************/
347 r01 = _mm_mul_ps(rsq01,rinv01);
349 /* Calculate table index by multiplying r with table scale and truncate to integer */
350 rt = _mm_mul_ps(r01,vftabscale);
351 vfitab = _mm_cvttps_epi32(rt);
352 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
353 vfitab = _mm_slli_epi32(vfitab,2);
355 /* CUBIC SPLINE TABLE ELECTROSTATICS */
356 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
357 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
358 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
359 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
360 _MM_TRANSPOSE4_PS(Y,F,G,H);
361 Heps = _mm_mul_ps(vfeps,H);
362 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
363 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
364 velec = _mm_mul_ps(qq01,VV);
365 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
366 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq01,FF),_mm_mul_ps(vftabscale,rinv01)));
368 /* Update potential sum for this i atom from the interaction with this j atom. */
369 velecsum = _mm_add_ps(velecsum,velec);
373 /* Calculate temporary vectorial force */
374 tx = _mm_mul_ps(fscal,dx01);
375 ty = _mm_mul_ps(fscal,dy01);
376 tz = _mm_mul_ps(fscal,dz01);
378 /* Update vectorial force */
379 fix0 = _mm_add_ps(fix0,tx);
380 fiy0 = _mm_add_ps(fiy0,ty);
381 fiz0 = _mm_add_ps(fiz0,tz);
383 fjx1 = _mm_add_ps(fjx1,tx);
384 fjy1 = _mm_add_ps(fjy1,ty);
385 fjz1 = _mm_add_ps(fjz1,tz);
387 /**************************
388 * CALCULATE INTERACTIONS *
389 **************************/
391 r02 = _mm_mul_ps(rsq02,rinv02);
393 /* Calculate table index by multiplying r with table scale and truncate to integer */
394 rt = _mm_mul_ps(r02,vftabscale);
395 vfitab = _mm_cvttps_epi32(rt);
396 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
397 vfitab = _mm_slli_epi32(vfitab,2);
399 /* CUBIC SPLINE TABLE ELECTROSTATICS */
400 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
401 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
402 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
403 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
404 _MM_TRANSPOSE4_PS(Y,F,G,H);
405 Heps = _mm_mul_ps(vfeps,H);
406 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
407 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
408 velec = _mm_mul_ps(qq02,VV);
409 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
410 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq02,FF),_mm_mul_ps(vftabscale,rinv02)));
412 /* Update potential sum for this i atom from the interaction with this j atom. */
413 velecsum = _mm_add_ps(velecsum,velec);
417 /* Calculate temporary vectorial force */
418 tx = _mm_mul_ps(fscal,dx02);
419 ty = _mm_mul_ps(fscal,dy02);
420 tz = _mm_mul_ps(fscal,dz02);
422 /* Update vectorial force */
423 fix0 = _mm_add_ps(fix0,tx);
424 fiy0 = _mm_add_ps(fiy0,ty);
425 fiz0 = _mm_add_ps(fiz0,tz);
427 fjx2 = _mm_add_ps(fjx2,tx);
428 fjy2 = _mm_add_ps(fjy2,ty);
429 fjz2 = _mm_add_ps(fjz2,tz);
431 /**************************
432 * CALCULATE INTERACTIONS *
433 **************************/
435 r10 = _mm_mul_ps(rsq10,rinv10);
437 /* Calculate table index by multiplying r with table scale and truncate to integer */
438 rt = _mm_mul_ps(r10,vftabscale);
439 vfitab = _mm_cvttps_epi32(rt);
440 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
441 vfitab = _mm_slli_epi32(vfitab,2);
443 /* CUBIC SPLINE TABLE ELECTROSTATICS */
444 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
445 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
446 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
447 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
448 _MM_TRANSPOSE4_PS(Y,F,G,H);
449 Heps = _mm_mul_ps(vfeps,H);
450 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
451 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
452 velec = _mm_mul_ps(qq10,VV);
453 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
454 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq10,FF),_mm_mul_ps(vftabscale,rinv10)));
456 /* Update potential sum for this i atom from the interaction with this j atom. */
457 velecsum = _mm_add_ps(velecsum,velec);
461 /* Calculate temporary vectorial force */
462 tx = _mm_mul_ps(fscal,dx10);
463 ty = _mm_mul_ps(fscal,dy10);
464 tz = _mm_mul_ps(fscal,dz10);
466 /* Update vectorial force */
467 fix1 = _mm_add_ps(fix1,tx);
468 fiy1 = _mm_add_ps(fiy1,ty);
469 fiz1 = _mm_add_ps(fiz1,tz);
471 fjx0 = _mm_add_ps(fjx0,tx);
472 fjy0 = _mm_add_ps(fjy0,ty);
473 fjz0 = _mm_add_ps(fjz0,tz);
475 /**************************
476 * CALCULATE INTERACTIONS *
477 **************************/
479 r11 = _mm_mul_ps(rsq11,rinv11);
481 /* Calculate table index by multiplying r with table scale and truncate to integer */
482 rt = _mm_mul_ps(r11,vftabscale);
483 vfitab = _mm_cvttps_epi32(rt);
484 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
485 vfitab = _mm_slli_epi32(vfitab,2);
487 /* CUBIC SPLINE TABLE ELECTROSTATICS */
488 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
489 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
490 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
491 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
492 _MM_TRANSPOSE4_PS(Y,F,G,H);
493 Heps = _mm_mul_ps(vfeps,H);
494 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
495 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
496 velec = _mm_mul_ps(qq11,VV);
497 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
498 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11)));
500 /* Update potential sum for this i atom from the interaction with this j atom. */
501 velecsum = _mm_add_ps(velecsum,velec);
505 /* Calculate temporary vectorial force */
506 tx = _mm_mul_ps(fscal,dx11);
507 ty = _mm_mul_ps(fscal,dy11);
508 tz = _mm_mul_ps(fscal,dz11);
510 /* Update vectorial force */
511 fix1 = _mm_add_ps(fix1,tx);
512 fiy1 = _mm_add_ps(fiy1,ty);
513 fiz1 = _mm_add_ps(fiz1,tz);
515 fjx1 = _mm_add_ps(fjx1,tx);
516 fjy1 = _mm_add_ps(fjy1,ty);
517 fjz1 = _mm_add_ps(fjz1,tz);
519 /**************************
520 * CALCULATE INTERACTIONS *
521 **************************/
523 r12 = _mm_mul_ps(rsq12,rinv12);
525 /* Calculate table index by multiplying r with table scale and truncate to integer */
526 rt = _mm_mul_ps(r12,vftabscale);
527 vfitab = _mm_cvttps_epi32(rt);
528 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
529 vfitab = _mm_slli_epi32(vfitab,2);
531 /* CUBIC SPLINE TABLE ELECTROSTATICS */
532 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
533 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
534 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
535 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
536 _MM_TRANSPOSE4_PS(Y,F,G,H);
537 Heps = _mm_mul_ps(vfeps,H);
538 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
539 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
540 velec = _mm_mul_ps(qq12,VV);
541 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
542 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12)));
544 /* Update potential sum for this i atom from the interaction with this j atom. */
545 velecsum = _mm_add_ps(velecsum,velec);
549 /* Calculate temporary vectorial force */
550 tx = _mm_mul_ps(fscal,dx12);
551 ty = _mm_mul_ps(fscal,dy12);
552 tz = _mm_mul_ps(fscal,dz12);
554 /* Update vectorial force */
555 fix1 = _mm_add_ps(fix1,tx);
556 fiy1 = _mm_add_ps(fiy1,ty);
557 fiz1 = _mm_add_ps(fiz1,tz);
559 fjx2 = _mm_add_ps(fjx2,tx);
560 fjy2 = _mm_add_ps(fjy2,ty);
561 fjz2 = _mm_add_ps(fjz2,tz);
563 /**************************
564 * CALCULATE INTERACTIONS *
565 **************************/
567 r20 = _mm_mul_ps(rsq20,rinv20);
569 /* Calculate table index by multiplying r with table scale and truncate to integer */
570 rt = _mm_mul_ps(r20,vftabscale);
571 vfitab = _mm_cvttps_epi32(rt);
572 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
573 vfitab = _mm_slli_epi32(vfitab,2);
575 /* CUBIC SPLINE TABLE ELECTROSTATICS */
576 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
577 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
578 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
579 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
580 _MM_TRANSPOSE4_PS(Y,F,G,H);
581 Heps = _mm_mul_ps(vfeps,H);
582 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
583 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
584 velec = _mm_mul_ps(qq20,VV);
585 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
586 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq20,FF),_mm_mul_ps(vftabscale,rinv20)));
588 /* Update potential sum for this i atom from the interaction with this j atom. */
589 velecsum = _mm_add_ps(velecsum,velec);
593 /* Calculate temporary vectorial force */
594 tx = _mm_mul_ps(fscal,dx20);
595 ty = _mm_mul_ps(fscal,dy20);
596 tz = _mm_mul_ps(fscal,dz20);
598 /* Update vectorial force */
599 fix2 = _mm_add_ps(fix2,tx);
600 fiy2 = _mm_add_ps(fiy2,ty);
601 fiz2 = _mm_add_ps(fiz2,tz);
603 fjx0 = _mm_add_ps(fjx0,tx);
604 fjy0 = _mm_add_ps(fjy0,ty);
605 fjz0 = _mm_add_ps(fjz0,tz);
607 /**************************
608 * CALCULATE INTERACTIONS *
609 **************************/
611 r21 = _mm_mul_ps(rsq21,rinv21);
613 /* Calculate table index by multiplying r with table scale and truncate to integer */
614 rt = _mm_mul_ps(r21,vftabscale);
615 vfitab = _mm_cvttps_epi32(rt);
616 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
617 vfitab = _mm_slli_epi32(vfitab,2);
619 /* CUBIC SPLINE TABLE ELECTROSTATICS */
620 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
621 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
622 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
623 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
624 _MM_TRANSPOSE4_PS(Y,F,G,H);
625 Heps = _mm_mul_ps(vfeps,H);
626 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
627 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
628 velec = _mm_mul_ps(qq21,VV);
629 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
630 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21)));
632 /* Update potential sum for this i atom from the interaction with this j atom. */
633 velecsum = _mm_add_ps(velecsum,velec);
637 /* Calculate temporary vectorial force */
638 tx = _mm_mul_ps(fscal,dx21);
639 ty = _mm_mul_ps(fscal,dy21);
640 tz = _mm_mul_ps(fscal,dz21);
642 /* Update vectorial force */
643 fix2 = _mm_add_ps(fix2,tx);
644 fiy2 = _mm_add_ps(fiy2,ty);
645 fiz2 = _mm_add_ps(fiz2,tz);
647 fjx1 = _mm_add_ps(fjx1,tx);
648 fjy1 = _mm_add_ps(fjy1,ty);
649 fjz1 = _mm_add_ps(fjz1,tz);
651 /**************************
652 * CALCULATE INTERACTIONS *
653 **************************/
655 r22 = _mm_mul_ps(rsq22,rinv22);
657 /* Calculate table index by multiplying r with table scale and truncate to integer */
658 rt = _mm_mul_ps(r22,vftabscale);
659 vfitab = _mm_cvttps_epi32(rt);
660 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
661 vfitab = _mm_slli_epi32(vfitab,2);
663 /* CUBIC SPLINE TABLE ELECTROSTATICS */
664 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
665 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
666 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
667 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
668 _MM_TRANSPOSE4_PS(Y,F,G,H);
669 Heps = _mm_mul_ps(vfeps,H);
670 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
671 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
672 velec = _mm_mul_ps(qq22,VV);
673 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
674 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22)));
676 /* Update potential sum for this i atom from the interaction with this j atom. */
677 velecsum = _mm_add_ps(velecsum,velec);
681 /* Calculate temporary vectorial force */
682 tx = _mm_mul_ps(fscal,dx22);
683 ty = _mm_mul_ps(fscal,dy22);
684 tz = _mm_mul_ps(fscal,dz22);
686 /* Update vectorial force */
687 fix2 = _mm_add_ps(fix2,tx);
688 fiy2 = _mm_add_ps(fiy2,ty);
689 fiz2 = _mm_add_ps(fiz2,tz);
691 fjx2 = _mm_add_ps(fjx2,tx);
692 fjy2 = _mm_add_ps(fjy2,ty);
693 fjz2 = _mm_add_ps(fjz2,tz);
695 fjptrA = f+j_coord_offsetA;
696 fjptrB = f+j_coord_offsetB;
697 fjptrC = f+j_coord_offsetC;
698 fjptrD = f+j_coord_offsetD;
700 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
701 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
703 /* Inner loop uses 400 flops */
709 /* Get j neighbor index, and coordinate index */
710 jnrlistA = jjnr[jidx];
711 jnrlistB = jjnr[jidx+1];
712 jnrlistC = jjnr[jidx+2];
713 jnrlistD = jjnr[jidx+3];
714 /* Sign of each element will be negative for non-real atoms.
715 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
716 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
718 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
719 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
720 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
721 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
722 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
723 j_coord_offsetA = DIM*jnrA;
724 j_coord_offsetB = DIM*jnrB;
725 j_coord_offsetC = DIM*jnrC;
726 j_coord_offsetD = DIM*jnrD;
728 /* load j atom coordinates */
729 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
730 x+j_coord_offsetC,x+j_coord_offsetD,
731 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
733 /* Calculate displacement vector */
734 dx00 = _mm_sub_ps(ix0,jx0);
735 dy00 = _mm_sub_ps(iy0,jy0);
736 dz00 = _mm_sub_ps(iz0,jz0);
737 dx01 = _mm_sub_ps(ix0,jx1);
738 dy01 = _mm_sub_ps(iy0,jy1);
739 dz01 = _mm_sub_ps(iz0,jz1);
740 dx02 = _mm_sub_ps(ix0,jx2);
741 dy02 = _mm_sub_ps(iy0,jy2);
742 dz02 = _mm_sub_ps(iz0,jz2);
743 dx10 = _mm_sub_ps(ix1,jx0);
744 dy10 = _mm_sub_ps(iy1,jy0);
745 dz10 = _mm_sub_ps(iz1,jz0);
746 dx11 = _mm_sub_ps(ix1,jx1);
747 dy11 = _mm_sub_ps(iy1,jy1);
748 dz11 = _mm_sub_ps(iz1,jz1);
749 dx12 = _mm_sub_ps(ix1,jx2);
750 dy12 = _mm_sub_ps(iy1,jy2);
751 dz12 = _mm_sub_ps(iz1,jz2);
752 dx20 = _mm_sub_ps(ix2,jx0);
753 dy20 = _mm_sub_ps(iy2,jy0);
754 dz20 = _mm_sub_ps(iz2,jz0);
755 dx21 = _mm_sub_ps(ix2,jx1);
756 dy21 = _mm_sub_ps(iy2,jy1);
757 dz21 = _mm_sub_ps(iz2,jz1);
758 dx22 = _mm_sub_ps(ix2,jx2);
759 dy22 = _mm_sub_ps(iy2,jy2);
760 dz22 = _mm_sub_ps(iz2,jz2);
762 /* Calculate squared distance and things based on it */
763 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
764 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
765 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
766 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
767 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
768 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
769 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
770 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
771 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
773 rinv00 = sse2_invsqrt_f(rsq00);
774 rinv01 = sse2_invsqrt_f(rsq01);
775 rinv02 = sse2_invsqrt_f(rsq02);
776 rinv10 = sse2_invsqrt_f(rsq10);
777 rinv11 = sse2_invsqrt_f(rsq11);
778 rinv12 = sse2_invsqrt_f(rsq12);
779 rinv20 = sse2_invsqrt_f(rsq20);
780 rinv21 = sse2_invsqrt_f(rsq21);
781 rinv22 = sse2_invsqrt_f(rsq22);
783 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
785 fjx0 = _mm_setzero_ps();
786 fjy0 = _mm_setzero_ps();
787 fjz0 = _mm_setzero_ps();
788 fjx1 = _mm_setzero_ps();
789 fjy1 = _mm_setzero_ps();
790 fjz1 = _mm_setzero_ps();
791 fjx2 = _mm_setzero_ps();
792 fjy2 = _mm_setzero_ps();
793 fjz2 = _mm_setzero_ps();
795 /**************************
796 * CALCULATE INTERACTIONS *
797 **************************/
799 r00 = _mm_mul_ps(rsq00,rinv00);
800 r00 = _mm_andnot_ps(dummy_mask,r00);
802 /* Calculate table index by multiplying r with table scale and truncate to integer */
803 rt = _mm_mul_ps(r00,vftabscale);
804 vfitab = _mm_cvttps_epi32(rt);
805 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
806 vfitab = _mm_slli_epi32(vfitab,2);
808 /* CUBIC SPLINE TABLE ELECTROSTATICS */
809 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
810 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
811 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
812 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
813 _MM_TRANSPOSE4_PS(Y,F,G,H);
814 Heps = _mm_mul_ps(vfeps,H);
815 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
816 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
817 velec = _mm_mul_ps(qq00,VV);
818 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
819 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq00,FF),_mm_mul_ps(vftabscale,rinv00)));
821 /* LENNARD-JONES DISPERSION/REPULSION */
823 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
824 vvdw6 = _mm_mul_ps(c6_00,rinvsix);
825 vvdw12 = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
826 vvdw = _mm_sub_ps( _mm_mul_ps(vvdw12,one_twelfth) , _mm_mul_ps(vvdw6,one_sixth) );
827 fvdw = _mm_mul_ps(_mm_sub_ps(vvdw12,vvdw6),rinvsq00);
829 /* Update potential sum for this i atom from the interaction with this j atom. */
830 velec = _mm_andnot_ps(dummy_mask,velec);
831 velecsum = _mm_add_ps(velecsum,velec);
832 vvdw = _mm_andnot_ps(dummy_mask,vvdw);
833 vvdwsum = _mm_add_ps(vvdwsum,vvdw);
835 fscal = _mm_add_ps(felec,fvdw);
837 fscal = _mm_andnot_ps(dummy_mask,fscal);
839 /* Calculate temporary vectorial force */
840 tx = _mm_mul_ps(fscal,dx00);
841 ty = _mm_mul_ps(fscal,dy00);
842 tz = _mm_mul_ps(fscal,dz00);
844 /* Update vectorial force */
845 fix0 = _mm_add_ps(fix0,tx);
846 fiy0 = _mm_add_ps(fiy0,ty);
847 fiz0 = _mm_add_ps(fiz0,tz);
849 fjx0 = _mm_add_ps(fjx0,tx);
850 fjy0 = _mm_add_ps(fjy0,ty);
851 fjz0 = _mm_add_ps(fjz0,tz);
853 /**************************
854 * CALCULATE INTERACTIONS *
855 **************************/
857 r01 = _mm_mul_ps(rsq01,rinv01);
858 r01 = _mm_andnot_ps(dummy_mask,r01);
860 /* Calculate table index by multiplying r with table scale and truncate to integer */
861 rt = _mm_mul_ps(r01,vftabscale);
862 vfitab = _mm_cvttps_epi32(rt);
863 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
864 vfitab = _mm_slli_epi32(vfitab,2);
866 /* CUBIC SPLINE TABLE ELECTROSTATICS */
867 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
868 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
869 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
870 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
871 _MM_TRANSPOSE4_PS(Y,F,G,H);
872 Heps = _mm_mul_ps(vfeps,H);
873 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
874 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
875 velec = _mm_mul_ps(qq01,VV);
876 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
877 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq01,FF),_mm_mul_ps(vftabscale,rinv01)));
879 /* Update potential sum for this i atom from the interaction with this j atom. */
880 velec = _mm_andnot_ps(dummy_mask,velec);
881 velecsum = _mm_add_ps(velecsum,velec);
885 fscal = _mm_andnot_ps(dummy_mask,fscal);
887 /* Calculate temporary vectorial force */
888 tx = _mm_mul_ps(fscal,dx01);
889 ty = _mm_mul_ps(fscal,dy01);
890 tz = _mm_mul_ps(fscal,dz01);
892 /* Update vectorial force */
893 fix0 = _mm_add_ps(fix0,tx);
894 fiy0 = _mm_add_ps(fiy0,ty);
895 fiz0 = _mm_add_ps(fiz0,tz);
897 fjx1 = _mm_add_ps(fjx1,tx);
898 fjy1 = _mm_add_ps(fjy1,ty);
899 fjz1 = _mm_add_ps(fjz1,tz);
901 /**************************
902 * CALCULATE INTERACTIONS *
903 **************************/
905 r02 = _mm_mul_ps(rsq02,rinv02);
906 r02 = _mm_andnot_ps(dummy_mask,r02);
908 /* Calculate table index by multiplying r with table scale and truncate to integer */
909 rt = _mm_mul_ps(r02,vftabscale);
910 vfitab = _mm_cvttps_epi32(rt);
911 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
912 vfitab = _mm_slli_epi32(vfitab,2);
914 /* CUBIC SPLINE TABLE ELECTROSTATICS */
915 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
916 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
917 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
918 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
919 _MM_TRANSPOSE4_PS(Y,F,G,H);
920 Heps = _mm_mul_ps(vfeps,H);
921 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
922 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
923 velec = _mm_mul_ps(qq02,VV);
924 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
925 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq02,FF),_mm_mul_ps(vftabscale,rinv02)));
927 /* Update potential sum for this i atom from the interaction with this j atom. */
928 velec = _mm_andnot_ps(dummy_mask,velec);
929 velecsum = _mm_add_ps(velecsum,velec);
933 fscal = _mm_andnot_ps(dummy_mask,fscal);
935 /* Calculate temporary vectorial force */
936 tx = _mm_mul_ps(fscal,dx02);
937 ty = _mm_mul_ps(fscal,dy02);
938 tz = _mm_mul_ps(fscal,dz02);
940 /* Update vectorial force */
941 fix0 = _mm_add_ps(fix0,tx);
942 fiy0 = _mm_add_ps(fiy0,ty);
943 fiz0 = _mm_add_ps(fiz0,tz);
945 fjx2 = _mm_add_ps(fjx2,tx);
946 fjy2 = _mm_add_ps(fjy2,ty);
947 fjz2 = _mm_add_ps(fjz2,tz);
949 /**************************
950 * CALCULATE INTERACTIONS *
951 **************************/
953 r10 = _mm_mul_ps(rsq10,rinv10);
954 r10 = _mm_andnot_ps(dummy_mask,r10);
956 /* Calculate table index by multiplying r with table scale and truncate to integer */
957 rt = _mm_mul_ps(r10,vftabscale);
958 vfitab = _mm_cvttps_epi32(rt);
959 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
960 vfitab = _mm_slli_epi32(vfitab,2);
962 /* CUBIC SPLINE TABLE ELECTROSTATICS */
963 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
964 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
965 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
966 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
967 _MM_TRANSPOSE4_PS(Y,F,G,H);
968 Heps = _mm_mul_ps(vfeps,H);
969 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
970 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
971 velec = _mm_mul_ps(qq10,VV);
972 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
973 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq10,FF),_mm_mul_ps(vftabscale,rinv10)));
975 /* Update potential sum for this i atom from the interaction with this j atom. */
976 velec = _mm_andnot_ps(dummy_mask,velec);
977 velecsum = _mm_add_ps(velecsum,velec);
981 fscal = _mm_andnot_ps(dummy_mask,fscal);
983 /* Calculate temporary vectorial force */
984 tx = _mm_mul_ps(fscal,dx10);
985 ty = _mm_mul_ps(fscal,dy10);
986 tz = _mm_mul_ps(fscal,dz10);
988 /* Update vectorial force */
989 fix1 = _mm_add_ps(fix1,tx);
990 fiy1 = _mm_add_ps(fiy1,ty);
991 fiz1 = _mm_add_ps(fiz1,tz);
993 fjx0 = _mm_add_ps(fjx0,tx);
994 fjy0 = _mm_add_ps(fjy0,ty);
995 fjz0 = _mm_add_ps(fjz0,tz);
997 /**************************
998 * CALCULATE INTERACTIONS *
999 **************************/
1001 r11 = _mm_mul_ps(rsq11,rinv11);
1002 r11 = _mm_andnot_ps(dummy_mask,r11);
1004 /* Calculate table index by multiplying r with table scale and truncate to integer */
1005 rt = _mm_mul_ps(r11,vftabscale);
1006 vfitab = _mm_cvttps_epi32(rt);
1007 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1008 vfitab = _mm_slli_epi32(vfitab,2);
1010 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1011 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1012 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1013 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1014 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1015 _MM_TRANSPOSE4_PS(Y,F,G,H);
1016 Heps = _mm_mul_ps(vfeps,H);
1017 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1018 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
1019 velec = _mm_mul_ps(qq11,VV);
1020 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1021 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11)));
1023 /* Update potential sum for this i atom from the interaction with this j atom. */
1024 velec = _mm_andnot_ps(dummy_mask,velec);
1025 velecsum = _mm_add_ps(velecsum,velec);
1029 fscal = _mm_andnot_ps(dummy_mask,fscal);
1031 /* Calculate temporary vectorial force */
1032 tx = _mm_mul_ps(fscal,dx11);
1033 ty = _mm_mul_ps(fscal,dy11);
1034 tz = _mm_mul_ps(fscal,dz11);
1036 /* Update vectorial force */
1037 fix1 = _mm_add_ps(fix1,tx);
1038 fiy1 = _mm_add_ps(fiy1,ty);
1039 fiz1 = _mm_add_ps(fiz1,tz);
1041 fjx1 = _mm_add_ps(fjx1,tx);
1042 fjy1 = _mm_add_ps(fjy1,ty);
1043 fjz1 = _mm_add_ps(fjz1,tz);
1045 /**************************
1046 * CALCULATE INTERACTIONS *
1047 **************************/
1049 r12 = _mm_mul_ps(rsq12,rinv12);
1050 r12 = _mm_andnot_ps(dummy_mask,r12);
1052 /* Calculate table index by multiplying r with table scale and truncate to integer */
1053 rt = _mm_mul_ps(r12,vftabscale);
1054 vfitab = _mm_cvttps_epi32(rt);
1055 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1056 vfitab = _mm_slli_epi32(vfitab,2);
1058 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1059 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1060 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1061 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1062 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1063 _MM_TRANSPOSE4_PS(Y,F,G,H);
1064 Heps = _mm_mul_ps(vfeps,H);
1065 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1066 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
1067 velec = _mm_mul_ps(qq12,VV);
1068 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1069 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12)));
1071 /* Update potential sum for this i atom from the interaction with this j atom. */
1072 velec = _mm_andnot_ps(dummy_mask,velec);
1073 velecsum = _mm_add_ps(velecsum,velec);
1077 fscal = _mm_andnot_ps(dummy_mask,fscal);
1079 /* Calculate temporary vectorial force */
1080 tx = _mm_mul_ps(fscal,dx12);
1081 ty = _mm_mul_ps(fscal,dy12);
1082 tz = _mm_mul_ps(fscal,dz12);
1084 /* Update vectorial force */
1085 fix1 = _mm_add_ps(fix1,tx);
1086 fiy1 = _mm_add_ps(fiy1,ty);
1087 fiz1 = _mm_add_ps(fiz1,tz);
1089 fjx2 = _mm_add_ps(fjx2,tx);
1090 fjy2 = _mm_add_ps(fjy2,ty);
1091 fjz2 = _mm_add_ps(fjz2,tz);
1093 /**************************
1094 * CALCULATE INTERACTIONS *
1095 **************************/
1097 r20 = _mm_mul_ps(rsq20,rinv20);
1098 r20 = _mm_andnot_ps(dummy_mask,r20);
1100 /* Calculate table index by multiplying r with table scale and truncate to integer */
1101 rt = _mm_mul_ps(r20,vftabscale);
1102 vfitab = _mm_cvttps_epi32(rt);
1103 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1104 vfitab = _mm_slli_epi32(vfitab,2);
1106 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1107 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1108 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1109 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1110 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1111 _MM_TRANSPOSE4_PS(Y,F,G,H);
1112 Heps = _mm_mul_ps(vfeps,H);
1113 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1114 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
1115 velec = _mm_mul_ps(qq20,VV);
1116 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1117 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq20,FF),_mm_mul_ps(vftabscale,rinv20)));
1119 /* Update potential sum for this i atom from the interaction with this j atom. */
1120 velec = _mm_andnot_ps(dummy_mask,velec);
1121 velecsum = _mm_add_ps(velecsum,velec);
1125 fscal = _mm_andnot_ps(dummy_mask,fscal);
1127 /* Calculate temporary vectorial force */
1128 tx = _mm_mul_ps(fscal,dx20);
1129 ty = _mm_mul_ps(fscal,dy20);
1130 tz = _mm_mul_ps(fscal,dz20);
1132 /* Update vectorial force */
1133 fix2 = _mm_add_ps(fix2,tx);
1134 fiy2 = _mm_add_ps(fiy2,ty);
1135 fiz2 = _mm_add_ps(fiz2,tz);
1137 fjx0 = _mm_add_ps(fjx0,tx);
1138 fjy0 = _mm_add_ps(fjy0,ty);
1139 fjz0 = _mm_add_ps(fjz0,tz);
1141 /**************************
1142 * CALCULATE INTERACTIONS *
1143 **************************/
1145 r21 = _mm_mul_ps(rsq21,rinv21);
1146 r21 = _mm_andnot_ps(dummy_mask,r21);
1148 /* Calculate table index by multiplying r with table scale and truncate to integer */
1149 rt = _mm_mul_ps(r21,vftabscale);
1150 vfitab = _mm_cvttps_epi32(rt);
1151 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1152 vfitab = _mm_slli_epi32(vfitab,2);
1154 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1155 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1156 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1157 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1158 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1159 _MM_TRANSPOSE4_PS(Y,F,G,H);
1160 Heps = _mm_mul_ps(vfeps,H);
1161 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1162 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
1163 velec = _mm_mul_ps(qq21,VV);
1164 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1165 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21)));
1167 /* Update potential sum for this i atom from the interaction with this j atom. */
1168 velec = _mm_andnot_ps(dummy_mask,velec);
1169 velecsum = _mm_add_ps(velecsum,velec);
1173 fscal = _mm_andnot_ps(dummy_mask,fscal);
1175 /* Calculate temporary vectorial force */
1176 tx = _mm_mul_ps(fscal,dx21);
1177 ty = _mm_mul_ps(fscal,dy21);
1178 tz = _mm_mul_ps(fscal,dz21);
1180 /* Update vectorial force */
1181 fix2 = _mm_add_ps(fix2,tx);
1182 fiy2 = _mm_add_ps(fiy2,ty);
1183 fiz2 = _mm_add_ps(fiz2,tz);
1185 fjx1 = _mm_add_ps(fjx1,tx);
1186 fjy1 = _mm_add_ps(fjy1,ty);
1187 fjz1 = _mm_add_ps(fjz1,tz);
1189 /**************************
1190 * CALCULATE INTERACTIONS *
1191 **************************/
1193 r22 = _mm_mul_ps(rsq22,rinv22);
1194 r22 = _mm_andnot_ps(dummy_mask,r22);
1196 /* Calculate table index by multiplying r with table scale and truncate to integer */
1197 rt = _mm_mul_ps(r22,vftabscale);
1198 vfitab = _mm_cvttps_epi32(rt);
1199 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1200 vfitab = _mm_slli_epi32(vfitab,2);
1202 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1203 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1204 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1205 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1206 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1207 _MM_TRANSPOSE4_PS(Y,F,G,H);
1208 Heps = _mm_mul_ps(vfeps,H);
1209 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1210 VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp));
1211 velec = _mm_mul_ps(qq22,VV);
1212 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1213 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22)));
1215 /* Update potential sum for this i atom from the interaction with this j atom. */
1216 velec = _mm_andnot_ps(dummy_mask,velec);
1217 velecsum = _mm_add_ps(velecsum,velec);
1221 fscal = _mm_andnot_ps(dummy_mask,fscal);
1223 /* Calculate temporary vectorial force */
1224 tx = _mm_mul_ps(fscal,dx22);
1225 ty = _mm_mul_ps(fscal,dy22);
1226 tz = _mm_mul_ps(fscal,dz22);
1228 /* Update vectorial force */
1229 fix2 = _mm_add_ps(fix2,tx);
1230 fiy2 = _mm_add_ps(fiy2,ty);
1231 fiz2 = _mm_add_ps(fiz2,tz);
1233 fjx2 = _mm_add_ps(fjx2,tx);
1234 fjy2 = _mm_add_ps(fjy2,ty);
1235 fjz2 = _mm_add_ps(fjz2,tz);
1237 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1238 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1239 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1240 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1242 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1243 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1245 /* Inner loop uses 409 flops */
1248 /* End of innermost loop */
1250 gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1251 f+i_coord_offset,fshift+i_shift_offset);
1254 /* Update potential energies */
1255 gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1256 gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
1258 /* Increment number of inner iterations */
1259 inneriter += j_index_end - j_index_start;
1261 /* Outer loop uses 20 flops */
1264 /* Increment number of outer iterations */
1267 /* Update outer/inner flops */
1269 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*409);
1272 * Gromacs nonbonded kernel: nb_kernel_ElecCSTab_VdwLJ_GeomW3W3_F_sse2_single
1273 * Electrostatics interaction: CubicSplineTable
1274 * VdW interaction: LennardJones
1275 * Geometry: Water3-Water3
1276 * Calculate force/pot: Force
1279 nb_kernel_ElecCSTab_VdwLJ_GeomW3W3_F_sse2_single
1280 (t_nblist * gmx_restrict nlist,
1281 rvec * gmx_restrict xx,
1282 rvec * gmx_restrict ff,
1283 struct t_forcerec * gmx_restrict fr,
1284 t_mdatoms * gmx_restrict mdatoms,
1285 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1286 t_nrnb * gmx_restrict nrnb)
1288 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1289 * just 0 for non-waters.
1290 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
1291 * jnr indices corresponding to data put in the four positions in the SIMD register.
1293 int i_shift_offset,i_coord_offset,outeriter,inneriter;
1294 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1295 int jnrA,jnrB,jnrC,jnrD;
1296 int jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1297 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1298 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
1299 real rcutoff_scalar;
1300 real *shiftvec,*fshift,*x,*f;
1301 real *fjptrA,*fjptrB,*fjptrC,*fjptrD;
1302 real scratch[4*DIM];
1303 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1305 __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1307 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1309 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1310 int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
1311 __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1312 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
1313 __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1314 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
1315 __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1316 __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1317 __m128 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
1318 __m128 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
1319 __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
1320 __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1321 __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1322 __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
1323 __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1324 __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1325 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
1328 __m128 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1331 __m128 one_sixth = _mm_set1_ps(1.0/6.0);
1332 __m128 one_twelfth = _mm_set1_ps(1.0/12.0);
1334 __m128i ifour = _mm_set1_epi32(4);
1335 __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
1337 __m128 dummy_mask,cutoff_mask;
1338 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
1339 __m128 one = _mm_set1_ps(1.0);
1340 __m128 two = _mm_set1_ps(2.0);
1346 jindex = nlist->jindex;
1348 shiftidx = nlist->shift;
1350 shiftvec = fr->shift_vec[0];
1351 fshift = fr->fshift[0];
1352 facel = _mm_set1_ps(fr->ic->epsfac);
1353 charge = mdatoms->chargeA;
1354 nvdwtype = fr->ntype;
1355 vdwparam = fr->nbfp;
1356 vdwtype = mdatoms->typeA;
1358 vftab = kernel_data->table_elec->data;
1359 vftabscale = _mm_set1_ps(kernel_data->table_elec->scale);
1361 /* Setup water-specific parameters */
1362 inr = nlist->iinr[0];
1363 iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
1364 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
1365 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
1366 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
1368 jq0 = _mm_set1_ps(charge[inr+0]);
1369 jq1 = _mm_set1_ps(charge[inr+1]);
1370 jq2 = _mm_set1_ps(charge[inr+2]);
1371 vdwjidx0A = 2*vdwtype[inr+0];
1372 qq00 = _mm_mul_ps(iq0,jq0);
1373 c6_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
1374 c12_00 = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
1375 qq01 = _mm_mul_ps(iq0,jq1);
1376 qq02 = _mm_mul_ps(iq0,jq2);
1377 qq10 = _mm_mul_ps(iq1,jq0);
1378 qq11 = _mm_mul_ps(iq1,jq1);
1379 qq12 = _mm_mul_ps(iq1,jq2);
1380 qq20 = _mm_mul_ps(iq2,jq0);
1381 qq21 = _mm_mul_ps(iq2,jq1);
1382 qq22 = _mm_mul_ps(iq2,jq2);
1384 /* Avoid stupid compiler warnings */
1385 jnrA = jnrB = jnrC = jnrD = 0;
1386 j_coord_offsetA = 0;
1387 j_coord_offsetB = 0;
1388 j_coord_offsetC = 0;
1389 j_coord_offsetD = 0;
1394 for(iidx=0;iidx<4*DIM;iidx++)
1396 scratch[iidx] = 0.0;
1399 /* Start outer loop over neighborlists */
1400 for(iidx=0; iidx<nri; iidx++)
1402 /* Load shift vector for this list */
1403 i_shift_offset = DIM*shiftidx[iidx];
1405 /* Load limits for loop over neighbors */
1406 j_index_start = jindex[iidx];
1407 j_index_end = jindex[iidx+1];
1409 /* Get outer coordinate index */
1411 i_coord_offset = DIM*inr;
1413 /* Load i particle coords and add shift vector */
1414 gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1415 &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1417 fix0 = _mm_setzero_ps();
1418 fiy0 = _mm_setzero_ps();
1419 fiz0 = _mm_setzero_ps();
1420 fix1 = _mm_setzero_ps();
1421 fiy1 = _mm_setzero_ps();
1422 fiz1 = _mm_setzero_ps();
1423 fix2 = _mm_setzero_ps();
1424 fiy2 = _mm_setzero_ps();
1425 fiz2 = _mm_setzero_ps();
1427 /* Start inner kernel loop */
1428 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1431 /* Get j neighbor index, and coordinate index */
1433 jnrB = jjnr[jidx+1];
1434 jnrC = jjnr[jidx+2];
1435 jnrD = jjnr[jidx+3];
1436 j_coord_offsetA = DIM*jnrA;
1437 j_coord_offsetB = DIM*jnrB;
1438 j_coord_offsetC = DIM*jnrC;
1439 j_coord_offsetD = DIM*jnrD;
1441 /* load j atom coordinates */
1442 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1443 x+j_coord_offsetC,x+j_coord_offsetD,
1444 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1446 /* Calculate displacement vector */
1447 dx00 = _mm_sub_ps(ix0,jx0);
1448 dy00 = _mm_sub_ps(iy0,jy0);
1449 dz00 = _mm_sub_ps(iz0,jz0);
1450 dx01 = _mm_sub_ps(ix0,jx1);
1451 dy01 = _mm_sub_ps(iy0,jy1);
1452 dz01 = _mm_sub_ps(iz0,jz1);
1453 dx02 = _mm_sub_ps(ix0,jx2);
1454 dy02 = _mm_sub_ps(iy0,jy2);
1455 dz02 = _mm_sub_ps(iz0,jz2);
1456 dx10 = _mm_sub_ps(ix1,jx0);
1457 dy10 = _mm_sub_ps(iy1,jy0);
1458 dz10 = _mm_sub_ps(iz1,jz0);
1459 dx11 = _mm_sub_ps(ix1,jx1);
1460 dy11 = _mm_sub_ps(iy1,jy1);
1461 dz11 = _mm_sub_ps(iz1,jz1);
1462 dx12 = _mm_sub_ps(ix1,jx2);
1463 dy12 = _mm_sub_ps(iy1,jy2);
1464 dz12 = _mm_sub_ps(iz1,jz2);
1465 dx20 = _mm_sub_ps(ix2,jx0);
1466 dy20 = _mm_sub_ps(iy2,jy0);
1467 dz20 = _mm_sub_ps(iz2,jz0);
1468 dx21 = _mm_sub_ps(ix2,jx1);
1469 dy21 = _mm_sub_ps(iy2,jy1);
1470 dz21 = _mm_sub_ps(iz2,jz1);
1471 dx22 = _mm_sub_ps(ix2,jx2);
1472 dy22 = _mm_sub_ps(iy2,jy2);
1473 dz22 = _mm_sub_ps(iz2,jz2);
1475 /* Calculate squared distance and things based on it */
1476 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1477 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
1478 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
1479 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1480 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1481 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1482 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1483 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1484 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1486 rinv00 = sse2_invsqrt_f(rsq00);
1487 rinv01 = sse2_invsqrt_f(rsq01);
1488 rinv02 = sse2_invsqrt_f(rsq02);
1489 rinv10 = sse2_invsqrt_f(rsq10);
1490 rinv11 = sse2_invsqrt_f(rsq11);
1491 rinv12 = sse2_invsqrt_f(rsq12);
1492 rinv20 = sse2_invsqrt_f(rsq20);
1493 rinv21 = sse2_invsqrt_f(rsq21);
1494 rinv22 = sse2_invsqrt_f(rsq22);
1496 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
1498 fjx0 = _mm_setzero_ps();
1499 fjy0 = _mm_setzero_ps();
1500 fjz0 = _mm_setzero_ps();
1501 fjx1 = _mm_setzero_ps();
1502 fjy1 = _mm_setzero_ps();
1503 fjz1 = _mm_setzero_ps();
1504 fjx2 = _mm_setzero_ps();
1505 fjy2 = _mm_setzero_ps();
1506 fjz2 = _mm_setzero_ps();
1508 /**************************
1509 * CALCULATE INTERACTIONS *
1510 **************************/
1512 r00 = _mm_mul_ps(rsq00,rinv00);
1514 /* Calculate table index by multiplying r with table scale and truncate to integer */
1515 rt = _mm_mul_ps(r00,vftabscale);
1516 vfitab = _mm_cvttps_epi32(rt);
1517 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1518 vfitab = _mm_slli_epi32(vfitab,2);
1520 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1521 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1522 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1523 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1524 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1525 _MM_TRANSPOSE4_PS(Y,F,G,H);
1526 Heps = _mm_mul_ps(vfeps,H);
1527 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1528 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1529 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq00,FF),_mm_mul_ps(vftabscale,rinv00)));
1531 /* LENNARD-JONES DISPERSION/REPULSION */
1533 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1534 fvdw = _mm_mul_ps(_mm_sub_ps(_mm_mul_ps(c12_00,rinvsix),c6_00),_mm_mul_ps(rinvsix,rinvsq00));
1536 fscal = _mm_add_ps(felec,fvdw);
1538 /* Calculate temporary vectorial force */
1539 tx = _mm_mul_ps(fscal,dx00);
1540 ty = _mm_mul_ps(fscal,dy00);
1541 tz = _mm_mul_ps(fscal,dz00);
1543 /* Update vectorial force */
1544 fix0 = _mm_add_ps(fix0,tx);
1545 fiy0 = _mm_add_ps(fiy0,ty);
1546 fiz0 = _mm_add_ps(fiz0,tz);
1548 fjx0 = _mm_add_ps(fjx0,tx);
1549 fjy0 = _mm_add_ps(fjy0,ty);
1550 fjz0 = _mm_add_ps(fjz0,tz);
1552 /**************************
1553 * CALCULATE INTERACTIONS *
1554 **************************/
1556 r01 = _mm_mul_ps(rsq01,rinv01);
1558 /* Calculate table index by multiplying r with table scale and truncate to integer */
1559 rt = _mm_mul_ps(r01,vftabscale);
1560 vfitab = _mm_cvttps_epi32(rt);
1561 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1562 vfitab = _mm_slli_epi32(vfitab,2);
1564 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1565 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1566 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1567 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1568 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1569 _MM_TRANSPOSE4_PS(Y,F,G,H);
1570 Heps = _mm_mul_ps(vfeps,H);
1571 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1572 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1573 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq01,FF),_mm_mul_ps(vftabscale,rinv01)));
1577 /* Calculate temporary vectorial force */
1578 tx = _mm_mul_ps(fscal,dx01);
1579 ty = _mm_mul_ps(fscal,dy01);
1580 tz = _mm_mul_ps(fscal,dz01);
1582 /* Update vectorial force */
1583 fix0 = _mm_add_ps(fix0,tx);
1584 fiy0 = _mm_add_ps(fiy0,ty);
1585 fiz0 = _mm_add_ps(fiz0,tz);
1587 fjx1 = _mm_add_ps(fjx1,tx);
1588 fjy1 = _mm_add_ps(fjy1,ty);
1589 fjz1 = _mm_add_ps(fjz1,tz);
1591 /**************************
1592 * CALCULATE INTERACTIONS *
1593 **************************/
1595 r02 = _mm_mul_ps(rsq02,rinv02);
1597 /* Calculate table index by multiplying r with table scale and truncate to integer */
1598 rt = _mm_mul_ps(r02,vftabscale);
1599 vfitab = _mm_cvttps_epi32(rt);
1600 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1601 vfitab = _mm_slli_epi32(vfitab,2);
1603 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1604 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1605 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1606 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1607 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1608 _MM_TRANSPOSE4_PS(Y,F,G,H);
1609 Heps = _mm_mul_ps(vfeps,H);
1610 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1611 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1612 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq02,FF),_mm_mul_ps(vftabscale,rinv02)));
1616 /* Calculate temporary vectorial force */
1617 tx = _mm_mul_ps(fscal,dx02);
1618 ty = _mm_mul_ps(fscal,dy02);
1619 tz = _mm_mul_ps(fscal,dz02);
1621 /* Update vectorial force */
1622 fix0 = _mm_add_ps(fix0,tx);
1623 fiy0 = _mm_add_ps(fiy0,ty);
1624 fiz0 = _mm_add_ps(fiz0,tz);
1626 fjx2 = _mm_add_ps(fjx2,tx);
1627 fjy2 = _mm_add_ps(fjy2,ty);
1628 fjz2 = _mm_add_ps(fjz2,tz);
1630 /**************************
1631 * CALCULATE INTERACTIONS *
1632 **************************/
1634 r10 = _mm_mul_ps(rsq10,rinv10);
1636 /* Calculate table index by multiplying r with table scale and truncate to integer */
1637 rt = _mm_mul_ps(r10,vftabscale);
1638 vfitab = _mm_cvttps_epi32(rt);
1639 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1640 vfitab = _mm_slli_epi32(vfitab,2);
1642 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1643 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1644 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1645 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1646 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1647 _MM_TRANSPOSE4_PS(Y,F,G,H);
1648 Heps = _mm_mul_ps(vfeps,H);
1649 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1650 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1651 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq10,FF),_mm_mul_ps(vftabscale,rinv10)));
1655 /* Calculate temporary vectorial force */
1656 tx = _mm_mul_ps(fscal,dx10);
1657 ty = _mm_mul_ps(fscal,dy10);
1658 tz = _mm_mul_ps(fscal,dz10);
1660 /* Update vectorial force */
1661 fix1 = _mm_add_ps(fix1,tx);
1662 fiy1 = _mm_add_ps(fiy1,ty);
1663 fiz1 = _mm_add_ps(fiz1,tz);
1665 fjx0 = _mm_add_ps(fjx0,tx);
1666 fjy0 = _mm_add_ps(fjy0,ty);
1667 fjz0 = _mm_add_ps(fjz0,tz);
1669 /**************************
1670 * CALCULATE INTERACTIONS *
1671 **************************/
1673 r11 = _mm_mul_ps(rsq11,rinv11);
1675 /* Calculate table index by multiplying r with table scale and truncate to integer */
1676 rt = _mm_mul_ps(r11,vftabscale);
1677 vfitab = _mm_cvttps_epi32(rt);
1678 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1679 vfitab = _mm_slli_epi32(vfitab,2);
1681 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1682 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1683 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1684 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1685 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1686 _MM_TRANSPOSE4_PS(Y,F,G,H);
1687 Heps = _mm_mul_ps(vfeps,H);
1688 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1689 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1690 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11)));
1694 /* Calculate temporary vectorial force */
1695 tx = _mm_mul_ps(fscal,dx11);
1696 ty = _mm_mul_ps(fscal,dy11);
1697 tz = _mm_mul_ps(fscal,dz11);
1699 /* Update vectorial force */
1700 fix1 = _mm_add_ps(fix1,tx);
1701 fiy1 = _mm_add_ps(fiy1,ty);
1702 fiz1 = _mm_add_ps(fiz1,tz);
1704 fjx1 = _mm_add_ps(fjx1,tx);
1705 fjy1 = _mm_add_ps(fjy1,ty);
1706 fjz1 = _mm_add_ps(fjz1,tz);
1708 /**************************
1709 * CALCULATE INTERACTIONS *
1710 **************************/
1712 r12 = _mm_mul_ps(rsq12,rinv12);
1714 /* Calculate table index by multiplying r with table scale and truncate to integer */
1715 rt = _mm_mul_ps(r12,vftabscale);
1716 vfitab = _mm_cvttps_epi32(rt);
1717 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1718 vfitab = _mm_slli_epi32(vfitab,2);
1720 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1721 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1722 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1723 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1724 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1725 _MM_TRANSPOSE4_PS(Y,F,G,H);
1726 Heps = _mm_mul_ps(vfeps,H);
1727 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1728 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1729 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12)));
1733 /* Calculate temporary vectorial force */
1734 tx = _mm_mul_ps(fscal,dx12);
1735 ty = _mm_mul_ps(fscal,dy12);
1736 tz = _mm_mul_ps(fscal,dz12);
1738 /* Update vectorial force */
1739 fix1 = _mm_add_ps(fix1,tx);
1740 fiy1 = _mm_add_ps(fiy1,ty);
1741 fiz1 = _mm_add_ps(fiz1,tz);
1743 fjx2 = _mm_add_ps(fjx2,tx);
1744 fjy2 = _mm_add_ps(fjy2,ty);
1745 fjz2 = _mm_add_ps(fjz2,tz);
1747 /**************************
1748 * CALCULATE INTERACTIONS *
1749 **************************/
1751 r20 = _mm_mul_ps(rsq20,rinv20);
1753 /* Calculate table index by multiplying r with table scale and truncate to integer */
1754 rt = _mm_mul_ps(r20,vftabscale);
1755 vfitab = _mm_cvttps_epi32(rt);
1756 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1757 vfitab = _mm_slli_epi32(vfitab,2);
1759 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1760 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1761 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1762 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1763 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1764 _MM_TRANSPOSE4_PS(Y,F,G,H);
1765 Heps = _mm_mul_ps(vfeps,H);
1766 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1767 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1768 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq20,FF),_mm_mul_ps(vftabscale,rinv20)));
1772 /* Calculate temporary vectorial force */
1773 tx = _mm_mul_ps(fscal,dx20);
1774 ty = _mm_mul_ps(fscal,dy20);
1775 tz = _mm_mul_ps(fscal,dz20);
1777 /* Update vectorial force */
1778 fix2 = _mm_add_ps(fix2,tx);
1779 fiy2 = _mm_add_ps(fiy2,ty);
1780 fiz2 = _mm_add_ps(fiz2,tz);
1782 fjx0 = _mm_add_ps(fjx0,tx);
1783 fjy0 = _mm_add_ps(fjy0,ty);
1784 fjz0 = _mm_add_ps(fjz0,tz);
1786 /**************************
1787 * CALCULATE INTERACTIONS *
1788 **************************/
1790 r21 = _mm_mul_ps(rsq21,rinv21);
1792 /* Calculate table index by multiplying r with table scale and truncate to integer */
1793 rt = _mm_mul_ps(r21,vftabscale);
1794 vfitab = _mm_cvttps_epi32(rt);
1795 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1796 vfitab = _mm_slli_epi32(vfitab,2);
1798 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1799 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1800 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1801 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1802 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1803 _MM_TRANSPOSE4_PS(Y,F,G,H);
1804 Heps = _mm_mul_ps(vfeps,H);
1805 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1806 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1807 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21)));
1811 /* Calculate temporary vectorial force */
1812 tx = _mm_mul_ps(fscal,dx21);
1813 ty = _mm_mul_ps(fscal,dy21);
1814 tz = _mm_mul_ps(fscal,dz21);
1816 /* Update vectorial force */
1817 fix2 = _mm_add_ps(fix2,tx);
1818 fiy2 = _mm_add_ps(fiy2,ty);
1819 fiz2 = _mm_add_ps(fiz2,tz);
1821 fjx1 = _mm_add_ps(fjx1,tx);
1822 fjy1 = _mm_add_ps(fjy1,ty);
1823 fjz1 = _mm_add_ps(fjz1,tz);
1825 /**************************
1826 * CALCULATE INTERACTIONS *
1827 **************************/
1829 r22 = _mm_mul_ps(rsq22,rinv22);
1831 /* Calculate table index by multiplying r with table scale and truncate to integer */
1832 rt = _mm_mul_ps(r22,vftabscale);
1833 vfitab = _mm_cvttps_epi32(rt);
1834 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1835 vfitab = _mm_slli_epi32(vfitab,2);
1837 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1838 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1839 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1840 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1841 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1842 _MM_TRANSPOSE4_PS(Y,F,G,H);
1843 Heps = _mm_mul_ps(vfeps,H);
1844 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1845 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1846 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22)));
1850 /* Calculate temporary vectorial force */
1851 tx = _mm_mul_ps(fscal,dx22);
1852 ty = _mm_mul_ps(fscal,dy22);
1853 tz = _mm_mul_ps(fscal,dz22);
1855 /* Update vectorial force */
1856 fix2 = _mm_add_ps(fix2,tx);
1857 fiy2 = _mm_add_ps(fiy2,ty);
1858 fiz2 = _mm_add_ps(fiz2,tz);
1860 fjx2 = _mm_add_ps(fjx2,tx);
1861 fjy2 = _mm_add_ps(fjy2,ty);
1862 fjz2 = _mm_add_ps(fjz2,tz);
1864 fjptrA = f+j_coord_offsetA;
1865 fjptrB = f+j_coord_offsetB;
1866 fjptrC = f+j_coord_offsetC;
1867 fjptrD = f+j_coord_offsetD;
1869 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1870 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1872 /* Inner loop uses 359 flops */
1875 if(jidx<j_index_end)
1878 /* Get j neighbor index, and coordinate index */
1879 jnrlistA = jjnr[jidx];
1880 jnrlistB = jjnr[jidx+1];
1881 jnrlistC = jjnr[jidx+2];
1882 jnrlistD = jjnr[jidx+3];
1883 /* Sign of each element will be negative for non-real atoms.
1884 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1885 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1887 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1888 jnrA = (jnrlistA>=0) ? jnrlistA : 0;
1889 jnrB = (jnrlistB>=0) ? jnrlistB : 0;
1890 jnrC = (jnrlistC>=0) ? jnrlistC : 0;
1891 jnrD = (jnrlistD>=0) ? jnrlistD : 0;
1892 j_coord_offsetA = DIM*jnrA;
1893 j_coord_offsetB = DIM*jnrB;
1894 j_coord_offsetC = DIM*jnrC;
1895 j_coord_offsetD = DIM*jnrD;
1897 /* load j atom coordinates */
1898 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1899 x+j_coord_offsetC,x+j_coord_offsetD,
1900 &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1902 /* Calculate displacement vector */
1903 dx00 = _mm_sub_ps(ix0,jx0);
1904 dy00 = _mm_sub_ps(iy0,jy0);
1905 dz00 = _mm_sub_ps(iz0,jz0);
1906 dx01 = _mm_sub_ps(ix0,jx1);
1907 dy01 = _mm_sub_ps(iy0,jy1);
1908 dz01 = _mm_sub_ps(iz0,jz1);
1909 dx02 = _mm_sub_ps(ix0,jx2);
1910 dy02 = _mm_sub_ps(iy0,jy2);
1911 dz02 = _mm_sub_ps(iz0,jz2);
1912 dx10 = _mm_sub_ps(ix1,jx0);
1913 dy10 = _mm_sub_ps(iy1,jy0);
1914 dz10 = _mm_sub_ps(iz1,jz0);
1915 dx11 = _mm_sub_ps(ix1,jx1);
1916 dy11 = _mm_sub_ps(iy1,jy1);
1917 dz11 = _mm_sub_ps(iz1,jz1);
1918 dx12 = _mm_sub_ps(ix1,jx2);
1919 dy12 = _mm_sub_ps(iy1,jy2);
1920 dz12 = _mm_sub_ps(iz1,jz2);
1921 dx20 = _mm_sub_ps(ix2,jx0);
1922 dy20 = _mm_sub_ps(iy2,jy0);
1923 dz20 = _mm_sub_ps(iz2,jz0);
1924 dx21 = _mm_sub_ps(ix2,jx1);
1925 dy21 = _mm_sub_ps(iy2,jy1);
1926 dz21 = _mm_sub_ps(iz2,jz1);
1927 dx22 = _mm_sub_ps(ix2,jx2);
1928 dy22 = _mm_sub_ps(iy2,jy2);
1929 dz22 = _mm_sub_ps(iz2,jz2);
1931 /* Calculate squared distance and things based on it */
1932 rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1933 rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
1934 rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
1935 rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1936 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1937 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1938 rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1939 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1940 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1942 rinv00 = sse2_invsqrt_f(rsq00);
1943 rinv01 = sse2_invsqrt_f(rsq01);
1944 rinv02 = sse2_invsqrt_f(rsq02);
1945 rinv10 = sse2_invsqrt_f(rsq10);
1946 rinv11 = sse2_invsqrt_f(rsq11);
1947 rinv12 = sse2_invsqrt_f(rsq12);
1948 rinv20 = sse2_invsqrt_f(rsq20);
1949 rinv21 = sse2_invsqrt_f(rsq21);
1950 rinv22 = sse2_invsqrt_f(rsq22);
1952 rinvsq00 = _mm_mul_ps(rinv00,rinv00);
1954 fjx0 = _mm_setzero_ps();
1955 fjy0 = _mm_setzero_ps();
1956 fjz0 = _mm_setzero_ps();
1957 fjx1 = _mm_setzero_ps();
1958 fjy1 = _mm_setzero_ps();
1959 fjz1 = _mm_setzero_ps();
1960 fjx2 = _mm_setzero_ps();
1961 fjy2 = _mm_setzero_ps();
1962 fjz2 = _mm_setzero_ps();
1964 /**************************
1965 * CALCULATE INTERACTIONS *
1966 **************************/
1968 r00 = _mm_mul_ps(rsq00,rinv00);
1969 r00 = _mm_andnot_ps(dummy_mask,r00);
1971 /* Calculate table index by multiplying r with table scale and truncate to integer */
1972 rt = _mm_mul_ps(r00,vftabscale);
1973 vfitab = _mm_cvttps_epi32(rt);
1974 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
1975 vfitab = _mm_slli_epi32(vfitab,2);
1977 /* CUBIC SPLINE TABLE ELECTROSTATICS */
1978 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
1979 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
1980 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
1981 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
1982 _MM_TRANSPOSE4_PS(Y,F,G,H);
1983 Heps = _mm_mul_ps(vfeps,H);
1984 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
1985 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
1986 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq00,FF),_mm_mul_ps(vftabscale,rinv00)));
1988 /* LENNARD-JONES DISPERSION/REPULSION */
1990 rinvsix = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1991 fvdw = _mm_mul_ps(_mm_sub_ps(_mm_mul_ps(c12_00,rinvsix),c6_00),_mm_mul_ps(rinvsix,rinvsq00));
1993 fscal = _mm_add_ps(felec,fvdw);
1995 fscal = _mm_andnot_ps(dummy_mask,fscal);
1997 /* Calculate temporary vectorial force */
1998 tx = _mm_mul_ps(fscal,dx00);
1999 ty = _mm_mul_ps(fscal,dy00);
2000 tz = _mm_mul_ps(fscal,dz00);
2002 /* Update vectorial force */
2003 fix0 = _mm_add_ps(fix0,tx);
2004 fiy0 = _mm_add_ps(fiy0,ty);
2005 fiz0 = _mm_add_ps(fiz0,tz);
2007 fjx0 = _mm_add_ps(fjx0,tx);
2008 fjy0 = _mm_add_ps(fjy0,ty);
2009 fjz0 = _mm_add_ps(fjz0,tz);
2011 /**************************
2012 * CALCULATE INTERACTIONS *
2013 **************************/
2015 r01 = _mm_mul_ps(rsq01,rinv01);
2016 r01 = _mm_andnot_ps(dummy_mask,r01);
2018 /* Calculate table index by multiplying r with table scale and truncate to integer */
2019 rt = _mm_mul_ps(r01,vftabscale);
2020 vfitab = _mm_cvttps_epi32(rt);
2021 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2022 vfitab = _mm_slli_epi32(vfitab,2);
2024 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2025 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2026 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2027 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2028 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2029 _MM_TRANSPOSE4_PS(Y,F,G,H);
2030 Heps = _mm_mul_ps(vfeps,H);
2031 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2032 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2033 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq01,FF),_mm_mul_ps(vftabscale,rinv01)));
2037 fscal = _mm_andnot_ps(dummy_mask,fscal);
2039 /* Calculate temporary vectorial force */
2040 tx = _mm_mul_ps(fscal,dx01);
2041 ty = _mm_mul_ps(fscal,dy01);
2042 tz = _mm_mul_ps(fscal,dz01);
2044 /* Update vectorial force */
2045 fix0 = _mm_add_ps(fix0,tx);
2046 fiy0 = _mm_add_ps(fiy0,ty);
2047 fiz0 = _mm_add_ps(fiz0,tz);
2049 fjx1 = _mm_add_ps(fjx1,tx);
2050 fjy1 = _mm_add_ps(fjy1,ty);
2051 fjz1 = _mm_add_ps(fjz1,tz);
2053 /**************************
2054 * CALCULATE INTERACTIONS *
2055 **************************/
2057 r02 = _mm_mul_ps(rsq02,rinv02);
2058 r02 = _mm_andnot_ps(dummy_mask,r02);
2060 /* Calculate table index by multiplying r with table scale and truncate to integer */
2061 rt = _mm_mul_ps(r02,vftabscale);
2062 vfitab = _mm_cvttps_epi32(rt);
2063 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2064 vfitab = _mm_slli_epi32(vfitab,2);
2066 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2067 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2068 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2069 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2070 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2071 _MM_TRANSPOSE4_PS(Y,F,G,H);
2072 Heps = _mm_mul_ps(vfeps,H);
2073 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2074 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2075 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq02,FF),_mm_mul_ps(vftabscale,rinv02)));
2079 fscal = _mm_andnot_ps(dummy_mask,fscal);
2081 /* Calculate temporary vectorial force */
2082 tx = _mm_mul_ps(fscal,dx02);
2083 ty = _mm_mul_ps(fscal,dy02);
2084 tz = _mm_mul_ps(fscal,dz02);
2086 /* Update vectorial force */
2087 fix0 = _mm_add_ps(fix0,tx);
2088 fiy0 = _mm_add_ps(fiy0,ty);
2089 fiz0 = _mm_add_ps(fiz0,tz);
2091 fjx2 = _mm_add_ps(fjx2,tx);
2092 fjy2 = _mm_add_ps(fjy2,ty);
2093 fjz2 = _mm_add_ps(fjz2,tz);
2095 /**************************
2096 * CALCULATE INTERACTIONS *
2097 **************************/
2099 r10 = _mm_mul_ps(rsq10,rinv10);
2100 r10 = _mm_andnot_ps(dummy_mask,r10);
2102 /* Calculate table index by multiplying r with table scale and truncate to integer */
2103 rt = _mm_mul_ps(r10,vftabscale);
2104 vfitab = _mm_cvttps_epi32(rt);
2105 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2106 vfitab = _mm_slli_epi32(vfitab,2);
2108 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2109 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2110 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2111 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2112 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2113 _MM_TRANSPOSE4_PS(Y,F,G,H);
2114 Heps = _mm_mul_ps(vfeps,H);
2115 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2116 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2117 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq10,FF),_mm_mul_ps(vftabscale,rinv10)));
2121 fscal = _mm_andnot_ps(dummy_mask,fscal);
2123 /* Calculate temporary vectorial force */
2124 tx = _mm_mul_ps(fscal,dx10);
2125 ty = _mm_mul_ps(fscal,dy10);
2126 tz = _mm_mul_ps(fscal,dz10);
2128 /* Update vectorial force */
2129 fix1 = _mm_add_ps(fix1,tx);
2130 fiy1 = _mm_add_ps(fiy1,ty);
2131 fiz1 = _mm_add_ps(fiz1,tz);
2133 fjx0 = _mm_add_ps(fjx0,tx);
2134 fjy0 = _mm_add_ps(fjy0,ty);
2135 fjz0 = _mm_add_ps(fjz0,tz);
2137 /**************************
2138 * CALCULATE INTERACTIONS *
2139 **************************/
2141 r11 = _mm_mul_ps(rsq11,rinv11);
2142 r11 = _mm_andnot_ps(dummy_mask,r11);
2144 /* Calculate table index by multiplying r with table scale and truncate to integer */
2145 rt = _mm_mul_ps(r11,vftabscale);
2146 vfitab = _mm_cvttps_epi32(rt);
2147 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2148 vfitab = _mm_slli_epi32(vfitab,2);
2150 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2151 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2152 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2153 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2154 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2155 _MM_TRANSPOSE4_PS(Y,F,G,H);
2156 Heps = _mm_mul_ps(vfeps,H);
2157 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2158 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2159 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11)));
2163 fscal = _mm_andnot_ps(dummy_mask,fscal);
2165 /* Calculate temporary vectorial force */
2166 tx = _mm_mul_ps(fscal,dx11);
2167 ty = _mm_mul_ps(fscal,dy11);
2168 tz = _mm_mul_ps(fscal,dz11);
2170 /* Update vectorial force */
2171 fix1 = _mm_add_ps(fix1,tx);
2172 fiy1 = _mm_add_ps(fiy1,ty);
2173 fiz1 = _mm_add_ps(fiz1,tz);
2175 fjx1 = _mm_add_ps(fjx1,tx);
2176 fjy1 = _mm_add_ps(fjy1,ty);
2177 fjz1 = _mm_add_ps(fjz1,tz);
2179 /**************************
2180 * CALCULATE INTERACTIONS *
2181 **************************/
2183 r12 = _mm_mul_ps(rsq12,rinv12);
2184 r12 = _mm_andnot_ps(dummy_mask,r12);
2186 /* Calculate table index by multiplying r with table scale and truncate to integer */
2187 rt = _mm_mul_ps(r12,vftabscale);
2188 vfitab = _mm_cvttps_epi32(rt);
2189 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2190 vfitab = _mm_slli_epi32(vfitab,2);
2192 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2193 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2194 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2195 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2196 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2197 _MM_TRANSPOSE4_PS(Y,F,G,H);
2198 Heps = _mm_mul_ps(vfeps,H);
2199 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2200 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2201 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12)));
2205 fscal = _mm_andnot_ps(dummy_mask,fscal);
2207 /* Calculate temporary vectorial force */
2208 tx = _mm_mul_ps(fscal,dx12);
2209 ty = _mm_mul_ps(fscal,dy12);
2210 tz = _mm_mul_ps(fscal,dz12);
2212 /* Update vectorial force */
2213 fix1 = _mm_add_ps(fix1,tx);
2214 fiy1 = _mm_add_ps(fiy1,ty);
2215 fiz1 = _mm_add_ps(fiz1,tz);
2217 fjx2 = _mm_add_ps(fjx2,tx);
2218 fjy2 = _mm_add_ps(fjy2,ty);
2219 fjz2 = _mm_add_ps(fjz2,tz);
2221 /**************************
2222 * CALCULATE INTERACTIONS *
2223 **************************/
2225 r20 = _mm_mul_ps(rsq20,rinv20);
2226 r20 = _mm_andnot_ps(dummy_mask,r20);
2228 /* Calculate table index by multiplying r with table scale and truncate to integer */
2229 rt = _mm_mul_ps(r20,vftabscale);
2230 vfitab = _mm_cvttps_epi32(rt);
2231 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2232 vfitab = _mm_slli_epi32(vfitab,2);
2234 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2235 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2236 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2237 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2238 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2239 _MM_TRANSPOSE4_PS(Y,F,G,H);
2240 Heps = _mm_mul_ps(vfeps,H);
2241 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2242 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2243 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq20,FF),_mm_mul_ps(vftabscale,rinv20)));
2247 fscal = _mm_andnot_ps(dummy_mask,fscal);
2249 /* Calculate temporary vectorial force */
2250 tx = _mm_mul_ps(fscal,dx20);
2251 ty = _mm_mul_ps(fscal,dy20);
2252 tz = _mm_mul_ps(fscal,dz20);
2254 /* Update vectorial force */
2255 fix2 = _mm_add_ps(fix2,tx);
2256 fiy2 = _mm_add_ps(fiy2,ty);
2257 fiz2 = _mm_add_ps(fiz2,tz);
2259 fjx0 = _mm_add_ps(fjx0,tx);
2260 fjy0 = _mm_add_ps(fjy0,ty);
2261 fjz0 = _mm_add_ps(fjz0,tz);
2263 /**************************
2264 * CALCULATE INTERACTIONS *
2265 **************************/
2267 r21 = _mm_mul_ps(rsq21,rinv21);
2268 r21 = _mm_andnot_ps(dummy_mask,r21);
2270 /* Calculate table index by multiplying r with table scale and truncate to integer */
2271 rt = _mm_mul_ps(r21,vftabscale);
2272 vfitab = _mm_cvttps_epi32(rt);
2273 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2274 vfitab = _mm_slli_epi32(vfitab,2);
2276 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2277 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2278 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2279 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2280 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2281 _MM_TRANSPOSE4_PS(Y,F,G,H);
2282 Heps = _mm_mul_ps(vfeps,H);
2283 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2284 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2285 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21)));
2289 fscal = _mm_andnot_ps(dummy_mask,fscal);
2291 /* Calculate temporary vectorial force */
2292 tx = _mm_mul_ps(fscal,dx21);
2293 ty = _mm_mul_ps(fscal,dy21);
2294 tz = _mm_mul_ps(fscal,dz21);
2296 /* Update vectorial force */
2297 fix2 = _mm_add_ps(fix2,tx);
2298 fiy2 = _mm_add_ps(fiy2,ty);
2299 fiz2 = _mm_add_ps(fiz2,tz);
2301 fjx1 = _mm_add_ps(fjx1,tx);
2302 fjy1 = _mm_add_ps(fjy1,ty);
2303 fjz1 = _mm_add_ps(fjz1,tz);
2305 /**************************
2306 * CALCULATE INTERACTIONS *
2307 **************************/
2309 r22 = _mm_mul_ps(rsq22,rinv22);
2310 r22 = _mm_andnot_ps(dummy_mask,r22);
2312 /* Calculate table index by multiplying r with table scale and truncate to integer */
2313 rt = _mm_mul_ps(r22,vftabscale);
2314 vfitab = _mm_cvttps_epi32(rt);
2315 vfeps = _mm_sub_ps(rt,_mm_cvtepi32_ps(vfitab));
2316 vfitab = _mm_slli_epi32(vfitab,2);
2318 /* CUBIC SPLINE TABLE ELECTROSTATICS */
2319 Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0) );
2320 F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1) );
2321 G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2) );
2322 H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3) );
2323 _MM_TRANSPOSE4_PS(Y,F,G,H);
2324 Heps = _mm_mul_ps(vfeps,H);
2325 Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps)));
2326 FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps))));
2327 felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22)));
2331 fscal = _mm_andnot_ps(dummy_mask,fscal);
2333 /* Calculate temporary vectorial force */
2334 tx = _mm_mul_ps(fscal,dx22);
2335 ty = _mm_mul_ps(fscal,dy22);
2336 tz = _mm_mul_ps(fscal,dz22);
2338 /* Update vectorial force */
2339 fix2 = _mm_add_ps(fix2,tx);
2340 fiy2 = _mm_add_ps(fiy2,ty);
2341 fiz2 = _mm_add_ps(fiz2,tz);
2343 fjx2 = _mm_add_ps(fjx2,tx);
2344 fjy2 = _mm_add_ps(fjy2,ty);
2345 fjz2 = _mm_add_ps(fjz2,tz);
2347 fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2348 fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
2349 fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
2350 fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
2352 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
2353 fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2355 /* Inner loop uses 368 flops */
2358 /* End of innermost loop */
2360 gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
2361 f+i_coord_offset,fshift+i_shift_offset);
2363 /* Increment number of inner iterations */
2364 inneriter += j_index_end - j_index_start;
2366 /* Outer loop uses 18 flops */
2369 /* Increment number of outer iterations */
2372 /* Update outer/inner flops */
2374 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*368);