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_ElecEw_VdwNone_GeomW4W4_VF_sse2_single
38 * Electrostatics interaction: Ewald
39 * VdW interaction: None
40 * Geometry: Water4-Water4
41 * Calculate force/pot: PotentialAndForce
44 nb_kernel_ElecEw_VdwNone_GeomW4W4_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 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
69 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
71 __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
72 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
73 __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
74 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
75 __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
76 int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D;
77 __m128 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
78 __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
79 __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
80 __m128 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
81 __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
82 __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
83 __m128 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
84 __m128 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
85 __m128 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
86 __m128 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
87 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
90 __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
92 __m128 dummy_mask,cutoff_mask;
93 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
94 __m128 one = _mm_set1_ps(1.0);
95 __m128 two = _mm_set1_ps(2.0);
101 jindex = nlist->jindex;
103 shiftidx = nlist->shift;
105 shiftvec = fr->shift_vec[0];
106 fshift = fr->fshift[0];
107 facel = _mm_set1_ps(fr->epsfac);
108 charge = mdatoms->chargeA;
110 sh_ewald = _mm_set1_ps(fr->ic->sh_ewald);
111 ewtab = fr->ic->tabq_coul_FDV0;
112 ewtabscale = _mm_set1_ps(fr->ic->tabq_scale);
113 ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale);
115 /* Setup water-specific parameters */
116 inr = nlist->iinr[0];
117 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
118 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
119 iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
121 jq1 = _mm_set1_ps(charge[inr+1]);
122 jq2 = _mm_set1_ps(charge[inr+2]);
123 jq3 = _mm_set1_ps(charge[inr+3]);
124 qq11 = _mm_mul_ps(iq1,jq1);
125 qq12 = _mm_mul_ps(iq1,jq2);
126 qq13 = _mm_mul_ps(iq1,jq3);
127 qq21 = _mm_mul_ps(iq2,jq1);
128 qq22 = _mm_mul_ps(iq2,jq2);
129 qq23 = _mm_mul_ps(iq2,jq3);
130 qq31 = _mm_mul_ps(iq3,jq1);
131 qq32 = _mm_mul_ps(iq3,jq2);
132 qq33 = _mm_mul_ps(iq3,jq3);
134 /* Avoid stupid compiler warnings */
135 jnrA = jnrB = jnrC = jnrD = 0;
144 /* Start outer loop over neighborlists */
145 for(iidx=0; iidx<nri; iidx++)
147 /* Load shift vector for this list */
148 i_shift_offset = DIM*shiftidx[iidx];
149 shX = shiftvec[i_shift_offset+XX];
150 shY = shiftvec[i_shift_offset+YY];
151 shZ = shiftvec[i_shift_offset+ZZ];
153 /* Load limits for loop over neighbors */
154 j_index_start = jindex[iidx];
155 j_index_end = jindex[iidx+1];
157 /* Get outer coordinate index */
159 i_coord_offset = DIM*inr;
161 /* Load i particle coords and add shift vector */
162 ix1 = _mm_set1_ps(shX + x[i_coord_offset+DIM*1+XX]);
163 iy1 = _mm_set1_ps(shY + x[i_coord_offset+DIM*1+YY]);
164 iz1 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*1+ZZ]);
165 ix2 = _mm_set1_ps(shX + x[i_coord_offset+DIM*2+XX]);
166 iy2 = _mm_set1_ps(shY + x[i_coord_offset+DIM*2+YY]);
167 iz2 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*2+ZZ]);
168 ix3 = _mm_set1_ps(shX + x[i_coord_offset+DIM*3+XX]);
169 iy3 = _mm_set1_ps(shY + x[i_coord_offset+DIM*3+YY]);
170 iz3 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*3+ZZ]);
172 fix1 = _mm_setzero_ps();
173 fiy1 = _mm_setzero_ps();
174 fiz1 = _mm_setzero_ps();
175 fix2 = _mm_setzero_ps();
176 fiy2 = _mm_setzero_ps();
177 fiz2 = _mm_setzero_ps();
178 fix3 = _mm_setzero_ps();
179 fiy3 = _mm_setzero_ps();
180 fiz3 = _mm_setzero_ps();
182 /* Reset potential sums */
183 velecsum = _mm_setzero_ps();
185 /* Start inner kernel loop */
186 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
189 /* Get j neighbor index, and coordinate index */
195 j_coord_offsetA = DIM*jnrA;
196 j_coord_offsetB = DIM*jnrB;
197 j_coord_offsetC = DIM*jnrC;
198 j_coord_offsetD = DIM*jnrD;
200 /* load j atom coordinates */
201 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
202 x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
203 &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
205 /* Calculate displacement vector */
206 dx11 = _mm_sub_ps(ix1,jx1);
207 dy11 = _mm_sub_ps(iy1,jy1);
208 dz11 = _mm_sub_ps(iz1,jz1);
209 dx12 = _mm_sub_ps(ix1,jx2);
210 dy12 = _mm_sub_ps(iy1,jy2);
211 dz12 = _mm_sub_ps(iz1,jz2);
212 dx13 = _mm_sub_ps(ix1,jx3);
213 dy13 = _mm_sub_ps(iy1,jy3);
214 dz13 = _mm_sub_ps(iz1,jz3);
215 dx21 = _mm_sub_ps(ix2,jx1);
216 dy21 = _mm_sub_ps(iy2,jy1);
217 dz21 = _mm_sub_ps(iz2,jz1);
218 dx22 = _mm_sub_ps(ix2,jx2);
219 dy22 = _mm_sub_ps(iy2,jy2);
220 dz22 = _mm_sub_ps(iz2,jz2);
221 dx23 = _mm_sub_ps(ix2,jx3);
222 dy23 = _mm_sub_ps(iy2,jy3);
223 dz23 = _mm_sub_ps(iz2,jz3);
224 dx31 = _mm_sub_ps(ix3,jx1);
225 dy31 = _mm_sub_ps(iy3,jy1);
226 dz31 = _mm_sub_ps(iz3,jz1);
227 dx32 = _mm_sub_ps(ix3,jx2);
228 dy32 = _mm_sub_ps(iy3,jy2);
229 dz32 = _mm_sub_ps(iz3,jz2);
230 dx33 = _mm_sub_ps(ix3,jx3);
231 dy33 = _mm_sub_ps(iy3,jy3);
232 dz33 = _mm_sub_ps(iz3,jz3);
234 /* Calculate squared distance and things based on it */
235 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
236 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
237 rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13);
238 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
239 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
240 rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23);
241 rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31);
242 rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32);
243 rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33);
245 rinv11 = gmx_mm_invsqrt_ps(rsq11);
246 rinv12 = gmx_mm_invsqrt_ps(rsq12);
247 rinv13 = gmx_mm_invsqrt_ps(rsq13);
248 rinv21 = gmx_mm_invsqrt_ps(rsq21);
249 rinv22 = gmx_mm_invsqrt_ps(rsq22);
250 rinv23 = gmx_mm_invsqrt_ps(rsq23);
251 rinv31 = gmx_mm_invsqrt_ps(rsq31);
252 rinv32 = gmx_mm_invsqrt_ps(rsq32);
253 rinv33 = gmx_mm_invsqrt_ps(rsq33);
255 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
256 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
257 rinvsq13 = _mm_mul_ps(rinv13,rinv13);
258 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
259 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
260 rinvsq23 = _mm_mul_ps(rinv23,rinv23);
261 rinvsq31 = _mm_mul_ps(rinv31,rinv31);
262 rinvsq32 = _mm_mul_ps(rinv32,rinv32);
263 rinvsq33 = _mm_mul_ps(rinv33,rinv33);
265 fjx1 = _mm_setzero_ps();
266 fjy1 = _mm_setzero_ps();
267 fjz1 = _mm_setzero_ps();
268 fjx2 = _mm_setzero_ps();
269 fjy2 = _mm_setzero_ps();
270 fjz2 = _mm_setzero_ps();
271 fjx3 = _mm_setzero_ps();
272 fjy3 = _mm_setzero_ps();
273 fjz3 = _mm_setzero_ps();
275 /**************************
276 * CALCULATE INTERACTIONS *
277 **************************/
279 r11 = _mm_mul_ps(rsq11,rinv11);
281 /* EWALD ELECTROSTATICS */
283 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
284 ewrt = _mm_mul_ps(r11,ewtabscale);
285 ewitab = _mm_cvttps_epi32(ewrt);
286 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
287 ewitab = _mm_slli_epi32(ewitab,2);
288 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
289 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
290 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
291 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
292 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
293 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
294 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
295 velec = _mm_mul_ps(qq11,_mm_sub_ps(rinv11,velec));
296 felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
298 /* Update potential sum for this i atom from the interaction with this j atom. */
299 velecsum = _mm_add_ps(velecsum,velec);
303 /* Calculate temporary vectorial force */
304 tx = _mm_mul_ps(fscal,dx11);
305 ty = _mm_mul_ps(fscal,dy11);
306 tz = _mm_mul_ps(fscal,dz11);
308 /* Update vectorial force */
309 fix1 = _mm_add_ps(fix1,tx);
310 fiy1 = _mm_add_ps(fiy1,ty);
311 fiz1 = _mm_add_ps(fiz1,tz);
313 fjx1 = _mm_add_ps(fjx1,tx);
314 fjy1 = _mm_add_ps(fjy1,ty);
315 fjz1 = _mm_add_ps(fjz1,tz);
317 /**************************
318 * CALCULATE INTERACTIONS *
319 **************************/
321 r12 = _mm_mul_ps(rsq12,rinv12);
323 /* EWALD ELECTROSTATICS */
325 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
326 ewrt = _mm_mul_ps(r12,ewtabscale);
327 ewitab = _mm_cvttps_epi32(ewrt);
328 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
329 ewitab = _mm_slli_epi32(ewitab,2);
330 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
331 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
332 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
333 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
334 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
335 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
336 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
337 velec = _mm_mul_ps(qq12,_mm_sub_ps(rinv12,velec));
338 felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
340 /* Update potential sum for this i atom from the interaction with this j atom. */
341 velecsum = _mm_add_ps(velecsum,velec);
345 /* Calculate temporary vectorial force */
346 tx = _mm_mul_ps(fscal,dx12);
347 ty = _mm_mul_ps(fscal,dy12);
348 tz = _mm_mul_ps(fscal,dz12);
350 /* Update vectorial force */
351 fix1 = _mm_add_ps(fix1,tx);
352 fiy1 = _mm_add_ps(fiy1,ty);
353 fiz1 = _mm_add_ps(fiz1,tz);
355 fjx2 = _mm_add_ps(fjx2,tx);
356 fjy2 = _mm_add_ps(fjy2,ty);
357 fjz2 = _mm_add_ps(fjz2,tz);
359 /**************************
360 * CALCULATE INTERACTIONS *
361 **************************/
363 r13 = _mm_mul_ps(rsq13,rinv13);
365 /* EWALD ELECTROSTATICS */
367 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
368 ewrt = _mm_mul_ps(r13,ewtabscale);
369 ewitab = _mm_cvttps_epi32(ewrt);
370 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
371 ewitab = _mm_slli_epi32(ewitab,2);
372 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
373 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
374 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
375 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
376 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
377 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
378 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
379 velec = _mm_mul_ps(qq13,_mm_sub_ps(rinv13,velec));
380 felec = _mm_mul_ps(_mm_mul_ps(qq13,rinv13),_mm_sub_ps(rinvsq13,felec));
382 /* Update potential sum for this i atom from the interaction with this j atom. */
383 velecsum = _mm_add_ps(velecsum,velec);
387 /* Calculate temporary vectorial force */
388 tx = _mm_mul_ps(fscal,dx13);
389 ty = _mm_mul_ps(fscal,dy13);
390 tz = _mm_mul_ps(fscal,dz13);
392 /* Update vectorial force */
393 fix1 = _mm_add_ps(fix1,tx);
394 fiy1 = _mm_add_ps(fiy1,ty);
395 fiz1 = _mm_add_ps(fiz1,tz);
397 fjx3 = _mm_add_ps(fjx3,tx);
398 fjy3 = _mm_add_ps(fjy3,ty);
399 fjz3 = _mm_add_ps(fjz3,tz);
401 /**************************
402 * CALCULATE INTERACTIONS *
403 **************************/
405 r21 = _mm_mul_ps(rsq21,rinv21);
407 /* EWALD ELECTROSTATICS */
409 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
410 ewrt = _mm_mul_ps(r21,ewtabscale);
411 ewitab = _mm_cvttps_epi32(ewrt);
412 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
413 ewitab = _mm_slli_epi32(ewitab,2);
414 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
415 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
416 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
417 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
418 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
419 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
420 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
421 velec = _mm_mul_ps(qq21,_mm_sub_ps(rinv21,velec));
422 felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
424 /* Update potential sum for this i atom from the interaction with this j atom. */
425 velecsum = _mm_add_ps(velecsum,velec);
429 /* Calculate temporary vectorial force */
430 tx = _mm_mul_ps(fscal,dx21);
431 ty = _mm_mul_ps(fscal,dy21);
432 tz = _mm_mul_ps(fscal,dz21);
434 /* Update vectorial force */
435 fix2 = _mm_add_ps(fix2,tx);
436 fiy2 = _mm_add_ps(fiy2,ty);
437 fiz2 = _mm_add_ps(fiz2,tz);
439 fjx1 = _mm_add_ps(fjx1,tx);
440 fjy1 = _mm_add_ps(fjy1,ty);
441 fjz1 = _mm_add_ps(fjz1,tz);
443 /**************************
444 * CALCULATE INTERACTIONS *
445 **************************/
447 r22 = _mm_mul_ps(rsq22,rinv22);
449 /* EWALD ELECTROSTATICS */
451 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
452 ewrt = _mm_mul_ps(r22,ewtabscale);
453 ewitab = _mm_cvttps_epi32(ewrt);
454 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
455 ewitab = _mm_slli_epi32(ewitab,2);
456 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
457 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
458 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
459 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
460 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
461 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
462 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
463 velec = _mm_mul_ps(qq22,_mm_sub_ps(rinv22,velec));
464 felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
466 /* Update potential sum for this i atom from the interaction with this j atom. */
467 velecsum = _mm_add_ps(velecsum,velec);
471 /* Calculate temporary vectorial force */
472 tx = _mm_mul_ps(fscal,dx22);
473 ty = _mm_mul_ps(fscal,dy22);
474 tz = _mm_mul_ps(fscal,dz22);
476 /* Update vectorial force */
477 fix2 = _mm_add_ps(fix2,tx);
478 fiy2 = _mm_add_ps(fiy2,ty);
479 fiz2 = _mm_add_ps(fiz2,tz);
481 fjx2 = _mm_add_ps(fjx2,tx);
482 fjy2 = _mm_add_ps(fjy2,ty);
483 fjz2 = _mm_add_ps(fjz2,tz);
485 /**************************
486 * CALCULATE INTERACTIONS *
487 **************************/
489 r23 = _mm_mul_ps(rsq23,rinv23);
491 /* EWALD ELECTROSTATICS */
493 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
494 ewrt = _mm_mul_ps(r23,ewtabscale);
495 ewitab = _mm_cvttps_epi32(ewrt);
496 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
497 ewitab = _mm_slli_epi32(ewitab,2);
498 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
499 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
500 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
501 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
502 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
503 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
504 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
505 velec = _mm_mul_ps(qq23,_mm_sub_ps(rinv23,velec));
506 felec = _mm_mul_ps(_mm_mul_ps(qq23,rinv23),_mm_sub_ps(rinvsq23,felec));
508 /* Update potential sum for this i atom from the interaction with this j atom. */
509 velecsum = _mm_add_ps(velecsum,velec);
513 /* Calculate temporary vectorial force */
514 tx = _mm_mul_ps(fscal,dx23);
515 ty = _mm_mul_ps(fscal,dy23);
516 tz = _mm_mul_ps(fscal,dz23);
518 /* Update vectorial force */
519 fix2 = _mm_add_ps(fix2,tx);
520 fiy2 = _mm_add_ps(fiy2,ty);
521 fiz2 = _mm_add_ps(fiz2,tz);
523 fjx3 = _mm_add_ps(fjx3,tx);
524 fjy3 = _mm_add_ps(fjy3,ty);
525 fjz3 = _mm_add_ps(fjz3,tz);
527 /**************************
528 * CALCULATE INTERACTIONS *
529 **************************/
531 r31 = _mm_mul_ps(rsq31,rinv31);
533 /* EWALD ELECTROSTATICS */
535 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
536 ewrt = _mm_mul_ps(r31,ewtabscale);
537 ewitab = _mm_cvttps_epi32(ewrt);
538 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
539 ewitab = _mm_slli_epi32(ewitab,2);
540 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
541 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
542 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
543 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
544 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
545 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
546 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
547 velec = _mm_mul_ps(qq31,_mm_sub_ps(rinv31,velec));
548 felec = _mm_mul_ps(_mm_mul_ps(qq31,rinv31),_mm_sub_ps(rinvsq31,felec));
550 /* Update potential sum for this i atom from the interaction with this j atom. */
551 velecsum = _mm_add_ps(velecsum,velec);
555 /* Calculate temporary vectorial force */
556 tx = _mm_mul_ps(fscal,dx31);
557 ty = _mm_mul_ps(fscal,dy31);
558 tz = _mm_mul_ps(fscal,dz31);
560 /* Update vectorial force */
561 fix3 = _mm_add_ps(fix3,tx);
562 fiy3 = _mm_add_ps(fiy3,ty);
563 fiz3 = _mm_add_ps(fiz3,tz);
565 fjx1 = _mm_add_ps(fjx1,tx);
566 fjy1 = _mm_add_ps(fjy1,ty);
567 fjz1 = _mm_add_ps(fjz1,tz);
569 /**************************
570 * CALCULATE INTERACTIONS *
571 **************************/
573 r32 = _mm_mul_ps(rsq32,rinv32);
575 /* EWALD ELECTROSTATICS */
577 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
578 ewrt = _mm_mul_ps(r32,ewtabscale);
579 ewitab = _mm_cvttps_epi32(ewrt);
580 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
581 ewitab = _mm_slli_epi32(ewitab,2);
582 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
583 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
584 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
585 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
586 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
587 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
588 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
589 velec = _mm_mul_ps(qq32,_mm_sub_ps(rinv32,velec));
590 felec = _mm_mul_ps(_mm_mul_ps(qq32,rinv32),_mm_sub_ps(rinvsq32,felec));
592 /* Update potential sum for this i atom from the interaction with this j atom. */
593 velecsum = _mm_add_ps(velecsum,velec);
597 /* Calculate temporary vectorial force */
598 tx = _mm_mul_ps(fscal,dx32);
599 ty = _mm_mul_ps(fscal,dy32);
600 tz = _mm_mul_ps(fscal,dz32);
602 /* Update vectorial force */
603 fix3 = _mm_add_ps(fix3,tx);
604 fiy3 = _mm_add_ps(fiy3,ty);
605 fiz3 = _mm_add_ps(fiz3,tz);
607 fjx2 = _mm_add_ps(fjx2,tx);
608 fjy2 = _mm_add_ps(fjy2,ty);
609 fjz2 = _mm_add_ps(fjz2,tz);
611 /**************************
612 * CALCULATE INTERACTIONS *
613 **************************/
615 r33 = _mm_mul_ps(rsq33,rinv33);
617 /* EWALD ELECTROSTATICS */
619 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
620 ewrt = _mm_mul_ps(r33,ewtabscale);
621 ewitab = _mm_cvttps_epi32(ewrt);
622 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
623 ewitab = _mm_slli_epi32(ewitab,2);
624 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
625 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
626 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
627 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
628 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
629 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
630 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
631 velec = _mm_mul_ps(qq33,_mm_sub_ps(rinv33,velec));
632 felec = _mm_mul_ps(_mm_mul_ps(qq33,rinv33),_mm_sub_ps(rinvsq33,felec));
634 /* Update potential sum for this i atom from the interaction with this j atom. */
635 velecsum = _mm_add_ps(velecsum,velec);
639 /* Calculate temporary vectorial force */
640 tx = _mm_mul_ps(fscal,dx33);
641 ty = _mm_mul_ps(fscal,dy33);
642 tz = _mm_mul_ps(fscal,dz33);
644 /* Update vectorial force */
645 fix3 = _mm_add_ps(fix3,tx);
646 fiy3 = _mm_add_ps(fiy3,ty);
647 fiz3 = _mm_add_ps(fiz3,tz);
649 fjx3 = _mm_add_ps(fjx3,tx);
650 fjy3 = _mm_add_ps(fjy3,ty);
651 fjz3 = _mm_add_ps(fjz3,tz);
653 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(f+j_coord_offsetA+DIM,f+j_coord_offsetB+DIM,
654 f+j_coord_offsetC+DIM,f+j_coord_offsetD+DIM,
655 fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
657 /* Inner loop uses 369 flops */
663 /* Get j neighbor index, and coordinate index */
669 /* Sign of each element will be negative for non-real atoms.
670 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
671 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
673 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
674 jnrA = (jnrA>=0) ? jnrA : 0;
675 jnrB = (jnrB>=0) ? jnrB : 0;
676 jnrC = (jnrC>=0) ? jnrC : 0;
677 jnrD = (jnrD>=0) ? jnrD : 0;
679 j_coord_offsetA = DIM*jnrA;
680 j_coord_offsetB = DIM*jnrB;
681 j_coord_offsetC = DIM*jnrC;
682 j_coord_offsetD = DIM*jnrD;
684 /* load j atom coordinates */
685 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
686 x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
687 &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
689 /* Calculate displacement vector */
690 dx11 = _mm_sub_ps(ix1,jx1);
691 dy11 = _mm_sub_ps(iy1,jy1);
692 dz11 = _mm_sub_ps(iz1,jz1);
693 dx12 = _mm_sub_ps(ix1,jx2);
694 dy12 = _mm_sub_ps(iy1,jy2);
695 dz12 = _mm_sub_ps(iz1,jz2);
696 dx13 = _mm_sub_ps(ix1,jx3);
697 dy13 = _mm_sub_ps(iy1,jy3);
698 dz13 = _mm_sub_ps(iz1,jz3);
699 dx21 = _mm_sub_ps(ix2,jx1);
700 dy21 = _mm_sub_ps(iy2,jy1);
701 dz21 = _mm_sub_ps(iz2,jz1);
702 dx22 = _mm_sub_ps(ix2,jx2);
703 dy22 = _mm_sub_ps(iy2,jy2);
704 dz22 = _mm_sub_ps(iz2,jz2);
705 dx23 = _mm_sub_ps(ix2,jx3);
706 dy23 = _mm_sub_ps(iy2,jy3);
707 dz23 = _mm_sub_ps(iz2,jz3);
708 dx31 = _mm_sub_ps(ix3,jx1);
709 dy31 = _mm_sub_ps(iy3,jy1);
710 dz31 = _mm_sub_ps(iz3,jz1);
711 dx32 = _mm_sub_ps(ix3,jx2);
712 dy32 = _mm_sub_ps(iy3,jy2);
713 dz32 = _mm_sub_ps(iz3,jz2);
714 dx33 = _mm_sub_ps(ix3,jx3);
715 dy33 = _mm_sub_ps(iy3,jy3);
716 dz33 = _mm_sub_ps(iz3,jz3);
718 /* Calculate squared distance and things based on it */
719 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
720 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
721 rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13);
722 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
723 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
724 rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23);
725 rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31);
726 rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32);
727 rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33);
729 rinv11 = gmx_mm_invsqrt_ps(rsq11);
730 rinv12 = gmx_mm_invsqrt_ps(rsq12);
731 rinv13 = gmx_mm_invsqrt_ps(rsq13);
732 rinv21 = gmx_mm_invsqrt_ps(rsq21);
733 rinv22 = gmx_mm_invsqrt_ps(rsq22);
734 rinv23 = gmx_mm_invsqrt_ps(rsq23);
735 rinv31 = gmx_mm_invsqrt_ps(rsq31);
736 rinv32 = gmx_mm_invsqrt_ps(rsq32);
737 rinv33 = gmx_mm_invsqrt_ps(rsq33);
739 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
740 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
741 rinvsq13 = _mm_mul_ps(rinv13,rinv13);
742 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
743 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
744 rinvsq23 = _mm_mul_ps(rinv23,rinv23);
745 rinvsq31 = _mm_mul_ps(rinv31,rinv31);
746 rinvsq32 = _mm_mul_ps(rinv32,rinv32);
747 rinvsq33 = _mm_mul_ps(rinv33,rinv33);
749 fjx1 = _mm_setzero_ps();
750 fjy1 = _mm_setzero_ps();
751 fjz1 = _mm_setzero_ps();
752 fjx2 = _mm_setzero_ps();
753 fjy2 = _mm_setzero_ps();
754 fjz2 = _mm_setzero_ps();
755 fjx3 = _mm_setzero_ps();
756 fjy3 = _mm_setzero_ps();
757 fjz3 = _mm_setzero_ps();
759 /**************************
760 * CALCULATE INTERACTIONS *
761 **************************/
763 r11 = _mm_mul_ps(rsq11,rinv11);
764 r11 = _mm_andnot_ps(dummy_mask,r11);
766 /* EWALD ELECTROSTATICS */
768 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
769 ewrt = _mm_mul_ps(r11,ewtabscale);
770 ewitab = _mm_cvttps_epi32(ewrt);
771 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
772 ewitab = _mm_slli_epi32(ewitab,2);
773 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
774 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
775 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
776 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
777 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
778 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
779 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
780 velec = _mm_mul_ps(qq11,_mm_sub_ps(rinv11,velec));
781 felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
783 /* Update potential sum for this i atom from the interaction with this j atom. */
784 velec = _mm_andnot_ps(dummy_mask,velec);
785 velecsum = _mm_add_ps(velecsum,velec);
789 fscal = _mm_andnot_ps(dummy_mask,fscal);
791 /* Calculate temporary vectorial force */
792 tx = _mm_mul_ps(fscal,dx11);
793 ty = _mm_mul_ps(fscal,dy11);
794 tz = _mm_mul_ps(fscal,dz11);
796 /* Update vectorial force */
797 fix1 = _mm_add_ps(fix1,tx);
798 fiy1 = _mm_add_ps(fiy1,ty);
799 fiz1 = _mm_add_ps(fiz1,tz);
801 fjx1 = _mm_add_ps(fjx1,tx);
802 fjy1 = _mm_add_ps(fjy1,ty);
803 fjz1 = _mm_add_ps(fjz1,tz);
805 /**************************
806 * CALCULATE INTERACTIONS *
807 **************************/
809 r12 = _mm_mul_ps(rsq12,rinv12);
810 r12 = _mm_andnot_ps(dummy_mask,r12);
812 /* EWALD ELECTROSTATICS */
814 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
815 ewrt = _mm_mul_ps(r12,ewtabscale);
816 ewitab = _mm_cvttps_epi32(ewrt);
817 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
818 ewitab = _mm_slli_epi32(ewitab,2);
819 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
820 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
821 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
822 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
823 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
824 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
825 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
826 velec = _mm_mul_ps(qq12,_mm_sub_ps(rinv12,velec));
827 felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
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);
835 fscal = _mm_andnot_ps(dummy_mask,fscal);
837 /* Calculate temporary vectorial force */
838 tx = _mm_mul_ps(fscal,dx12);
839 ty = _mm_mul_ps(fscal,dy12);
840 tz = _mm_mul_ps(fscal,dz12);
842 /* Update vectorial force */
843 fix1 = _mm_add_ps(fix1,tx);
844 fiy1 = _mm_add_ps(fiy1,ty);
845 fiz1 = _mm_add_ps(fiz1,tz);
847 fjx2 = _mm_add_ps(fjx2,tx);
848 fjy2 = _mm_add_ps(fjy2,ty);
849 fjz2 = _mm_add_ps(fjz2,tz);
851 /**************************
852 * CALCULATE INTERACTIONS *
853 **************************/
855 r13 = _mm_mul_ps(rsq13,rinv13);
856 r13 = _mm_andnot_ps(dummy_mask,r13);
858 /* EWALD ELECTROSTATICS */
860 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
861 ewrt = _mm_mul_ps(r13,ewtabscale);
862 ewitab = _mm_cvttps_epi32(ewrt);
863 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
864 ewitab = _mm_slli_epi32(ewitab,2);
865 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
866 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
867 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
868 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
869 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
870 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
871 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
872 velec = _mm_mul_ps(qq13,_mm_sub_ps(rinv13,velec));
873 felec = _mm_mul_ps(_mm_mul_ps(qq13,rinv13),_mm_sub_ps(rinvsq13,felec));
875 /* Update potential sum for this i atom from the interaction with this j atom. */
876 velec = _mm_andnot_ps(dummy_mask,velec);
877 velecsum = _mm_add_ps(velecsum,velec);
881 fscal = _mm_andnot_ps(dummy_mask,fscal);
883 /* Calculate temporary vectorial force */
884 tx = _mm_mul_ps(fscal,dx13);
885 ty = _mm_mul_ps(fscal,dy13);
886 tz = _mm_mul_ps(fscal,dz13);
888 /* Update vectorial force */
889 fix1 = _mm_add_ps(fix1,tx);
890 fiy1 = _mm_add_ps(fiy1,ty);
891 fiz1 = _mm_add_ps(fiz1,tz);
893 fjx3 = _mm_add_ps(fjx3,tx);
894 fjy3 = _mm_add_ps(fjy3,ty);
895 fjz3 = _mm_add_ps(fjz3,tz);
897 /**************************
898 * CALCULATE INTERACTIONS *
899 **************************/
901 r21 = _mm_mul_ps(rsq21,rinv21);
902 r21 = _mm_andnot_ps(dummy_mask,r21);
904 /* EWALD ELECTROSTATICS */
906 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
907 ewrt = _mm_mul_ps(r21,ewtabscale);
908 ewitab = _mm_cvttps_epi32(ewrt);
909 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
910 ewitab = _mm_slli_epi32(ewitab,2);
911 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
912 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
913 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
914 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
915 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
916 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
917 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
918 velec = _mm_mul_ps(qq21,_mm_sub_ps(rinv21,velec));
919 felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
921 /* Update potential sum for this i atom from the interaction with this j atom. */
922 velec = _mm_andnot_ps(dummy_mask,velec);
923 velecsum = _mm_add_ps(velecsum,velec);
927 fscal = _mm_andnot_ps(dummy_mask,fscal);
929 /* Calculate temporary vectorial force */
930 tx = _mm_mul_ps(fscal,dx21);
931 ty = _mm_mul_ps(fscal,dy21);
932 tz = _mm_mul_ps(fscal,dz21);
934 /* Update vectorial force */
935 fix2 = _mm_add_ps(fix2,tx);
936 fiy2 = _mm_add_ps(fiy2,ty);
937 fiz2 = _mm_add_ps(fiz2,tz);
939 fjx1 = _mm_add_ps(fjx1,tx);
940 fjy1 = _mm_add_ps(fjy1,ty);
941 fjz1 = _mm_add_ps(fjz1,tz);
943 /**************************
944 * CALCULATE INTERACTIONS *
945 **************************/
947 r22 = _mm_mul_ps(rsq22,rinv22);
948 r22 = _mm_andnot_ps(dummy_mask,r22);
950 /* EWALD ELECTROSTATICS */
952 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
953 ewrt = _mm_mul_ps(r22,ewtabscale);
954 ewitab = _mm_cvttps_epi32(ewrt);
955 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
956 ewitab = _mm_slli_epi32(ewitab,2);
957 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
958 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
959 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
960 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
961 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
962 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
963 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
964 velec = _mm_mul_ps(qq22,_mm_sub_ps(rinv22,velec));
965 felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
967 /* Update potential sum for this i atom from the interaction with this j atom. */
968 velec = _mm_andnot_ps(dummy_mask,velec);
969 velecsum = _mm_add_ps(velecsum,velec);
973 fscal = _mm_andnot_ps(dummy_mask,fscal);
975 /* Calculate temporary vectorial force */
976 tx = _mm_mul_ps(fscal,dx22);
977 ty = _mm_mul_ps(fscal,dy22);
978 tz = _mm_mul_ps(fscal,dz22);
980 /* Update vectorial force */
981 fix2 = _mm_add_ps(fix2,tx);
982 fiy2 = _mm_add_ps(fiy2,ty);
983 fiz2 = _mm_add_ps(fiz2,tz);
985 fjx2 = _mm_add_ps(fjx2,tx);
986 fjy2 = _mm_add_ps(fjy2,ty);
987 fjz2 = _mm_add_ps(fjz2,tz);
989 /**************************
990 * CALCULATE INTERACTIONS *
991 **************************/
993 r23 = _mm_mul_ps(rsq23,rinv23);
994 r23 = _mm_andnot_ps(dummy_mask,r23);
996 /* EWALD ELECTROSTATICS */
998 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
999 ewrt = _mm_mul_ps(r23,ewtabscale);
1000 ewitab = _mm_cvttps_epi32(ewrt);
1001 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1002 ewitab = _mm_slli_epi32(ewitab,2);
1003 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1004 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1005 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1006 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1007 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1008 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1009 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1010 velec = _mm_mul_ps(qq23,_mm_sub_ps(rinv23,velec));
1011 felec = _mm_mul_ps(_mm_mul_ps(qq23,rinv23),_mm_sub_ps(rinvsq23,felec));
1013 /* Update potential sum for this i atom from the interaction with this j atom. */
1014 velec = _mm_andnot_ps(dummy_mask,velec);
1015 velecsum = _mm_add_ps(velecsum,velec);
1019 fscal = _mm_andnot_ps(dummy_mask,fscal);
1021 /* Calculate temporary vectorial force */
1022 tx = _mm_mul_ps(fscal,dx23);
1023 ty = _mm_mul_ps(fscal,dy23);
1024 tz = _mm_mul_ps(fscal,dz23);
1026 /* Update vectorial force */
1027 fix2 = _mm_add_ps(fix2,tx);
1028 fiy2 = _mm_add_ps(fiy2,ty);
1029 fiz2 = _mm_add_ps(fiz2,tz);
1031 fjx3 = _mm_add_ps(fjx3,tx);
1032 fjy3 = _mm_add_ps(fjy3,ty);
1033 fjz3 = _mm_add_ps(fjz3,tz);
1035 /**************************
1036 * CALCULATE INTERACTIONS *
1037 **************************/
1039 r31 = _mm_mul_ps(rsq31,rinv31);
1040 r31 = _mm_andnot_ps(dummy_mask,r31);
1042 /* EWALD ELECTROSTATICS */
1044 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1045 ewrt = _mm_mul_ps(r31,ewtabscale);
1046 ewitab = _mm_cvttps_epi32(ewrt);
1047 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1048 ewitab = _mm_slli_epi32(ewitab,2);
1049 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1050 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1051 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1052 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1053 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1054 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1055 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1056 velec = _mm_mul_ps(qq31,_mm_sub_ps(rinv31,velec));
1057 felec = _mm_mul_ps(_mm_mul_ps(qq31,rinv31),_mm_sub_ps(rinvsq31,felec));
1059 /* Update potential sum for this i atom from the interaction with this j atom. */
1060 velec = _mm_andnot_ps(dummy_mask,velec);
1061 velecsum = _mm_add_ps(velecsum,velec);
1065 fscal = _mm_andnot_ps(dummy_mask,fscal);
1067 /* Calculate temporary vectorial force */
1068 tx = _mm_mul_ps(fscal,dx31);
1069 ty = _mm_mul_ps(fscal,dy31);
1070 tz = _mm_mul_ps(fscal,dz31);
1072 /* Update vectorial force */
1073 fix3 = _mm_add_ps(fix3,tx);
1074 fiy3 = _mm_add_ps(fiy3,ty);
1075 fiz3 = _mm_add_ps(fiz3,tz);
1077 fjx1 = _mm_add_ps(fjx1,tx);
1078 fjy1 = _mm_add_ps(fjy1,ty);
1079 fjz1 = _mm_add_ps(fjz1,tz);
1081 /**************************
1082 * CALCULATE INTERACTIONS *
1083 **************************/
1085 r32 = _mm_mul_ps(rsq32,rinv32);
1086 r32 = _mm_andnot_ps(dummy_mask,r32);
1088 /* EWALD ELECTROSTATICS */
1090 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1091 ewrt = _mm_mul_ps(r32,ewtabscale);
1092 ewitab = _mm_cvttps_epi32(ewrt);
1093 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1094 ewitab = _mm_slli_epi32(ewitab,2);
1095 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1096 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1097 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1098 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1099 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1100 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1101 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1102 velec = _mm_mul_ps(qq32,_mm_sub_ps(rinv32,velec));
1103 felec = _mm_mul_ps(_mm_mul_ps(qq32,rinv32),_mm_sub_ps(rinvsq32,felec));
1105 /* Update potential sum for this i atom from the interaction with this j atom. */
1106 velec = _mm_andnot_ps(dummy_mask,velec);
1107 velecsum = _mm_add_ps(velecsum,velec);
1111 fscal = _mm_andnot_ps(dummy_mask,fscal);
1113 /* Calculate temporary vectorial force */
1114 tx = _mm_mul_ps(fscal,dx32);
1115 ty = _mm_mul_ps(fscal,dy32);
1116 tz = _mm_mul_ps(fscal,dz32);
1118 /* Update vectorial force */
1119 fix3 = _mm_add_ps(fix3,tx);
1120 fiy3 = _mm_add_ps(fiy3,ty);
1121 fiz3 = _mm_add_ps(fiz3,tz);
1123 fjx2 = _mm_add_ps(fjx2,tx);
1124 fjy2 = _mm_add_ps(fjy2,ty);
1125 fjz2 = _mm_add_ps(fjz2,tz);
1127 /**************************
1128 * CALCULATE INTERACTIONS *
1129 **************************/
1131 r33 = _mm_mul_ps(rsq33,rinv33);
1132 r33 = _mm_andnot_ps(dummy_mask,r33);
1134 /* EWALD ELECTROSTATICS */
1136 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1137 ewrt = _mm_mul_ps(r33,ewtabscale);
1138 ewitab = _mm_cvttps_epi32(ewrt);
1139 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1140 ewitab = _mm_slli_epi32(ewitab,2);
1141 ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1142 ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1143 ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1144 ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1145 _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1146 felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1147 velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1148 velec = _mm_mul_ps(qq33,_mm_sub_ps(rinv33,velec));
1149 felec = _mm_mul_ps(_mm_mul_ps(qq33,rinv33),_mm_sub_ps(rinvsq33,felec));
1151 /* Update potential sum for this i atom from the interaction with this j atom. */
1152 velec = _mm_andnot_ps(dummy_mask,velec);
1153 velecsum = _mm_add_ps(velecsum,velec);
1157 fscal = _mm_andnot_ps(dummy_mask,fscal);
1159 /* Calculate temporary vectorial force */
1160 tx = _mm_mul_ps(fscal,dx33);
1161 ty = _mm_mul_ps(fscal,dy33);
1162 tz = _mm_mul_ps(fscal,dz33);
1164 /* Update vectorial force */
1165 fix3 = _mm_add_ps(fix3,tx);
1166 fiy3 = _mm_add_ps(fiy3,ty);
1167 fiz3 = _mm_add_ps(fiz3,tz);
1169 fjx3 = _mm_add_ps(fjx3,tx);
1170 fjy3 = _mm_add_ps(fjy3,ty);
1171 fjz3 = _mm_add_ps(fjz3,tz);
1173 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(f+j_coord_offsetA+DIM,f+j_coord_offsetB+DIM,
1174 f+j_coord_offsetC+DIM,f+j_coord_offsetD+DIM,
1175 fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1177 /* Inner loop uses 378 flops */
1180 /* End of innermost loop */
1182 gmx_mm_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1183 f+i_coord_offset+DIM,fshift+i_shift_offset);
1186 /* Update potential energies */
1187 gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1189 /* Increment number of inner iterations */
1190 inneriter += j_index_end - j_index_start;
1192 /* Outer loop uses 28 flops */
1195 /* Increment number of outer iterations */
1198 /* Update outer/inner flops */
1200 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_VF,outeriter*28 + inneriter*378);
1203 * Gromacs nonbonded kernel: nb_kernel_ElecEw_VdwNone_GeomW4W4_F_sse2_single
1204 * Electrostatics interaction: Ewald
1205 * VdW interaction: None
1206 * Geometry: Water4-Water4
1207 * Calculate force/pot: Force
1210 nb_kernel_ElecEw_VdwNone_GeomW4W4_F_sse2_single
1211 (t_nblist * gmx_restrict nlist,
1212 rvec * gmx_restrict xx,
1213 rvec * gmx_restrict ff,
1214 t_forcerec * gmx_restrict fr,
1215 t_mdatoms * gmx_restrict mdatoms,
1216 nb_kernel_data_t * gmx_restrict kernel_data,
1217 t_nrnb * gmx_restrict nrnb)
1219 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1220 * just 0 for non-waters.
1221 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
1222 * jnr indices corresponding to data put in the four positions in the SIMD register.
1224 int i_shift_offset,i_coord_offset,outeriter,inneriter;
1225 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1226 int jnrA,jnrB,jnrC,jnrD;
1227 int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1228 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
1229 real shX,shY,shZ,rcutoff_scalar;
1230 real *shiftvec,*fshift,*x,*f;
1231 __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1233 __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1235 __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1237 __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
1238 int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
1239 __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1240 int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
1241 __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1242 int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D;
1243 __m128 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
1244 __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1245 __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1246 __m128 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
1247 __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1248 __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1249 __m128 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
1250 __m128 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
1251 __m128 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
1252 __m128 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
1253 __m128 velec,felec,velecsum,facel,crf,krf,krf2;
1256 __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1258 __m128 dummy_mask,cutoff_mask;
1259 __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
1260 __m128 one = _mm_set1_ps(1.0);
1261 __m128 two = _mm_set1_ps(2.0);
1267 jindex = nlist->jindex;
1269 shiftidx = nlist->shift;
1271 shiftvec = fr->shift_vec[0];
1272 fshift = fr->fshift[0];
1273 facel = _mm_set1_ps(fr->epsfac);
1274 charge = mdatoms->chargeA;
1276 sh_ewald = _mm_set1_ps(fr->ic->sh_ewald);
1277 ewtab = fr->ic->tabq_coul_F;
1278 ewtabscale = _mm_set1_ps(fr->ic->tabq_scale);
1279 ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale);
1281 /* Setup water-specific parameters */
1282 inr = nlist->iinr[0];
1283 iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
1284 iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
1285 iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
1287 jq1 = _mm_set1_ps(charge[inr+1]);
1288 jq2 = _mm_set1_ps(charge[inr+2]);
1289 jq3 = _mm_set1_ps(charge[inr+3]);
1290 qq11 = _mm_mul_ps(iq1,jq1);
1291 qq12 = _mm_mul_ps(iq1,jq2);
1292 qq13 = _mm_mul_ps(iq1,jq3);
1293 qq21 = _mm_mul_ps(iq2,jq1);
1294 qq22 = _mm_mul_ps(iq2,jq2);
1295 qq23 = _mm_mul_ps(iq2,jq3);
1296 qq31 = _mm_mul_ps(iq3,jq1);
1297 qq32 = _mm_mul_ps(iq3,jq2);
1298 qq33 = _mm_mul_ps(iq3,jq3);
1300 /* Avoid stupid compiler warnings */
1301 jnrA = jnrB = jnrC = jnrD = 0;
1302 j_coord_offsetA = 0;
1303 j_coord_offsetB = 0;
1304 j_coord_offsetC = 0;
1305 j_coord_offsetD = 0;
1310 /* Start outer loop over neighborlists */
1311 for(iidx=0; iidx<nri; iidx++)
1313 /* Load shift vector for this list */
1314 i_shift_offset = DIM*shiftidx[iidx];
1315 shX = shiftvec[i_shift_offset+XX];
1316 shY = shiftvec[i_shift_offset+YY];
1317 shZ = shiftvec[i_shift_offset+ZZ];
1319 /* Load limits for loop over neighbors */
1320 j_index_start = jindex[iidx];
1321 j_index_end = jindex[iidx+1];
1323 /* Get outer coordinate index */
1325 i_coord_offset = DIM*inr;
1327 /* Load i particle coords and add shift vector */
1328 ix1 = _mm_set1_ps(shX + x[i_coord_offset+DIM*1+XX]);
1329 iy1 = _mm_set1_ps(shY + x[i_coord_offset+DIM*1+YY]);
1330 iz1 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*1+ZZ]);
1331 ix2 = _mm_set1_ps(shX + x[i_coord_offset+DIM*2+XX]);
1332 iy2 = _mm_set1_ps(shY + x[i_coord_offset+DIM*2+YY]);
1333 iz2 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*2+ZZ]);
1334 ix3 = _mm_set1_ps(shX + x[i_coord_offset+DIM*3+XX]);
1335 iy3 = _mm_set1_ps(shY + x[i_coord_offset+DIM*3+YY]);
1336 iz3 = _mm_set1_ps(shZ + x[i_coord_offset+DIM*3+ZZ]);
1338 fix1 = _mm_setzero_ps();
1339 fiy1 = _mm_setzero_ps();
1340 fiz1 = _mm_setzero_ps();
1341 fix2 = _mm_setzero_ps();
1342 fiy2 = _mm_setzero_ps();
1343 fiz2 = _mm_setzero_ps();
1344 fix3 = _mm_setzero_ps();
1345 fiy3 = _mm_setzero_ps();
1346 fiz3 = _mm_setzero_ps();
1348 /* Start inner kernel loop */
1349 for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1352 /* Get j neighbor index, and coordinate index */
1354 jnrB = jjnr[jidx+1];
1355 jnrC = jjnr[jidx+2];
1356 jnrD = jjnr[jidx+3];
1358 j_coord_offsetA = DIM*jnrA;
1359 j_coord_offsetB = DIM*jnrB;
1360 j_coord_offsetC = DIM*jnrC;
1361 j_coord_offsetD = DIM*jnrD;
1363 /* load j atom coordinates */
1364 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
1365 x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
1366 &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
1368 /* Calculate displacement vector */
1369 dx11 = _mm_sub_ps(ix1,jx1);
1370 dy11 = _mm_sub_ps(iy1,jy1);
1371 dz11 = _mm_sub_ps(iz1,jz1);
1372 dx12 = _mm_sub_ps(ix1,jx2);
1373 dy12 = _mm_sub_ps(iy1,jy2);
1374 dz12 = _mm_sub_ps(iz1,jz2);
1375 dx13 = _mm_sub_ps(ix1,jx3);
1376 dy13 = _mm_sub_ps(iy1,jy3);
1377 dz13 = _mm_sub_ps(iz1,jz3);
1378 dx21 = _mm_sub_ps(ix2,jx1);
1379 dy21 = _mm_sub_ps(iy2,jy1);
1380 dz21 = _mm_sub_ps(iz2,jz1);
1381 dx22 = _mm_sub_ps(ix2,jx2);
1382 dy22 = _mm_sub_ps(iy2,jy2);
1383 dz22 = _mm_sub_ps(iz2,jz2);
1384 dx23 = _mm_sub_ps(ix2,jx3);
1385 dy23 = _mm_sub_ps(iy2,jy3);
1386 dz23 = _mm_sub_ps(iz2,jz3);
1387 dx31 = _mm_sub_ps(ix3,jx1);
1388 dy31 = _mm_sub_ps(iy3,jy1);
1389 dz31 = _mm_sub_ps(iz3,jz1);
1390 dx32 = _mm_sub_ps(ix3,jx2);
1391 dy32 = _mm_sub_ps(iy3,jy2);
1392 dz32 = _mm_sub_ps(iz3,jz2);
1393 dx33 = _mm_sub_ps(ix3,jx3);
1394 dy33 = _mm_sub_ps(iy3,jy3);
1395 dz33 = _mm_sub_ps(iz3,jz3);
1397 /* Calculate squared distance and things based on it */
1398 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1399 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1400 rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13);
1401 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1402 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1403 rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23);
1404 rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31);
1405 rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32);
1406 rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33);
1408 rinv11 = gmx_mm_invsqrt_ps(rsq11);
1409 rinv12 = gmx_mm_invsqrt_ps(rsq12);
1410 rinv13 = gmx_mm_invsqrt_ps(rsq13);
1411 rinv21 = gmx_mm_invsqrt_ps(rsq21);
1412 rinv22 = gmx_mm_invsqrt_ps(rsq22);
1413 rinv23 = gmx_mm_invsqrt_ps(rsq23);
1414 rinv31 = gmx_mm_invsqrt_ps(rsq31);
1415 rinv32 = gmx_mm_invsqrt_ps(rsq32);
1416 rinv33 = gmx_mm_invsqrt_ps(rsq33);
1418 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
1419 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
1420 rinvsq13 = _mm_mul_ps(rinv13,rinv13);
1421 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
1422 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
1423 rinvsq23 = _mm_mul_ps(rinv23,rinv23);
1424 rinvsq31 = _mm_mul_ps(rinv31,rinv31);
1425 rinvsq32 = _mm_mul_ps(rinv32,rinv32);
1426 rinvsq33 = _mm_mul_ps(rinv33,rinv33);
1428 fjx1 = _mm_setzero_ps();
1429 fjy1 = _mm_setzero_ps();
1430 fjz1 = _mm_setzero_ps();
1431 fjx2 = _mm_setzero_ps();
1432 fjy2 = _mm_setzero_ps();
1433 fjz2 = _mm_setzero_ps();
1434 fjx3 = _mm_setzero_ps();
1435 fjy3 = _mm_setzero_ps();
1436 fjz3 = _mm_setzero_ps();
1438 /**************************
1439 * CALCULATE INTERACTIONS *
1440 **************************/
1442 r11 = _mm_mul_ps(rsq11,rinv11);
1444 /* EWALD ELECTROSTATICS */
1446 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1447 ewrt = _mm_mul_ps(r11,ewtabscale);
1448 ewitab = _mm_cvttps_epi32(ewrt);
1449 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1450 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1451 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1453 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1454 felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
1458 /* Calculate temporary vectorial force */
1459 tx = _mm_mul_ps(fscal,dx11);
1460 ty = _mm_mul_ps(fscal,dy11);
1461 tz = _mm_mul_ps(fscal,dz11);
1463 /* Update vectorial force */
1464 fix1 = _mm_add_ps(fix1,tx);
1465 fiy1 = _mm_add_ps(fiy1,ty);
1466 fiz1 = _mm_add_ps(fiz1,tz);
1468 fjx1 = _mm_add_ps(fjx1,tx);
1469 fjy1 = _mm_add_ps(fjy1,ty);
1470 fjz1 = _mm_add_ps(fjz1,tz);
1472 /**************************
1473 * CALCULATE INTERACTIONS *
1474 **************************/
1476 r12 = _mm_mul_ps(rsq12,rinv12);
1478 /* EWALD ELECTROSTATICS */
1480 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1481 ewrt = _mm_mul_ps(r12,ewtabscale);
1482 ewitab = _mm_cvttps_epi32(ewrt);
1483 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1484 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1485 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1487 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1488 felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
1492 /* Calculate temporary vectorial force */
1493 tx = _mm_mul_ps(fscal,dx12);
1494 ty = _mm_mul_ps(fscal,dy12);
1495 tz = _mm_mul_ps(fscal,dz12);
1497 /* Update vectorial force */
1498 fix1 = _mm_add_ps(fix1,tx);
1499 fiy1 = _mm_add_ps(fiy1,ty);
1500 fiz1 = _mm_add_ps(fiz1,tz);
1502 fjx2 = _mm_add_ps(fjx2,tx);
1503 fjy2 = _mm_add_ps(fjy2,ty);
1504 fjz2 = _mm_add_ps(fjz2,tz);
1506 /**************************
1507 * CALCULATE INTERACTIONS *
1508 **************************/
1510 r13 = _mm_mul_ps(rsq13,rinv13);
1512 /* EWALD ELECTROSTATICS */
1514 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1515 ewrt = _mm_mul_ps(r13,ewtabscale);
1516 ewitab = _mm_cvttps_epi32(ewrt);
1517 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1518 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1519 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1521 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1522 felec = _mm_mul_ps(_mm_mul_ps(qq13,rinv13),_mm_sub_ps(rinvsq13,felec));
1526 /* Calculate temporary vectorial force */
1527 tx = _mm_mul_ps(fscal,dx13);
1528 ty = _mm_mul_ps(fscal,dy13);
1529 tz = _mm_mul_ps(fscal,dz13);
1531 /* Update vectorial force */
1532 fix1 = _mm_add_ps(fix1,tx);
1533 fiy1 = _mm_add_ps(fiy1,ty);
1534 fiz1 = _mm_add_ps(fiz1,tz);
1536 fjx3 = _mm_add_ps(fjx3,tx);
1537 fjy3 = _mm_add_ps(fjy3,ty);
1538 fjz3 = _mm_add_ps(fjz3,tz);
1540 /**************************
1541 * CALCULATE INTERACTIONS *
1542 **************************/
1544 r21 = _mm_mul_ps(rsq21,rinv21);
1546 /* EWALD ELECTROSTATICS */
1548 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1549 ewrt = _mm_mul_ps(r21,ewtabscale);
1550 ewitab = _mm_cvttps_epi32(ewrt);
1551 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1552 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1553 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1555 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1556 felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
1560 /* Calculate temporary vectorial force */
1561 tx = _mm_mul_ps(fscal,dx21);
1562 ty = _mm_mul_ps(fscal,dy21);
1563 tz = _mm_mul_ps(fscal,dz21);
1565 /* Update vectorial force */
1566 fix2 = _mm_add_ps(fix2,tx);
1567 fiy2 = _mm_add_ps(fiy2,ty);
1568 fiz2 = _mm_add_ps(fiz2,tz);
1570 fjx1 = _mm_add_ps(fjx1,tx);
1571 fjy1 = _mm_add_ps(fjy1,ty);
1572 fjz1 = _mm_add_ps(fjz1,tz);
1574 /**************************
1575 * CALCULATE INTERACTIONS *
1576 **************************/
1578 r22 = _mm_mul_ps(rsq22,rinv22);
1580 /* EWALD ELECTROSTATICS */
1582 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1583 ewrt = _mm_mul_ps(r22,ewtabscale);
1584 ewitab = _mm_cvttps_epi32(ewrt);
1585 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1586 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1587 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1589 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1590 felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
1594 /* Calculate temporary vectorial force */
1595 tx = _mm_mul_ps(fscal,dx22);
1596 ty = _mm_mul_ps(fscal,dy22);
1597 tz = _mm_mul_ps(fscal,dz22);
1599 /* Update vectorial force */
1600 fix2 = _mm_add_ps(fix2,tx);
1601 fiy2 = _mm_add_ps(fiy2,ty);
1602 fiz2 = _mm_add_ps(fiz2,tz);
1604 fjx2 = _mm_add_ps(fjx2,tx);
1605 fjy2 = _mm_add_ps(fjy2,ty);
1606 fjz2 = _mm_add_ps(fjz2,tz);
1608 /**************************
1609 * CALCULATE INTERACTIONS *
1610 **************************/
1612 r23 = _mm_mul_ps(rsq23,rinv23);
1614 /* EWALD ELECTROSTATICS */
1616 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1617 ewrt = _mm_mul_ps(r23,ewtabscale);
1618 ewitab = _mm_cvttps_epi32(ewrt);
1619 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1620 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1621 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1623 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1624 felec = _mm_mul_ps(_mm_mul_ps(qq23,rinv23),_mm_sub_ps(rinvsq23,felec));
1628 /* Calculate temporary vectorial force */
1629 tx = _mm_mul_ps(fscal,dx23);
1630 ty = _mm_mul_ps(fscal,dy23);
1631 tz = _mm_mul_ps(fscal,dz23);
1633 /* Update vectorial force */
1634 fix2 = _mm_add_ps(fix2,tx);
1635 fiy2 = _mm_add_ps(fiy2,ty);
1636 fiz2 = _mm_add_ps(fiz2,tz);
1638 fjx3 = _mm_add_ps(fjx3,tx);
1639 fjy3 = _mm_add_ps(fjy3,ty);
1640 fjz3 = _mm_add_ps(fjz3,tz);
1642 /**************************
1643 * CALCULATE INTERACTIONS *
1644 **************************/
1646 r31 = _mm_mul_ps(rsq31,rinv31);
1648 /* EWALD ELECTROSTATICS */
1650 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1651 ewrt = _mm_mul_ps(r31,ewtabscale);
1652 ewitab = _mm_cvttps_epi32(ewrt);
1653 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1654 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1655 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1657 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1658 felec = _mm_mul_ps(_mm_mul_ps(qq31,rinv31),_mm_sub_ps(rinvsq31,felec));
1662 /* Calculate temporary vectorial force */
1663 tx = _mm_mul_ps(fscal,dx31);
1664 ty = _mm_mul_ps(fscal,dy31);
1665 tz = _mm_mul_ps(fscal,dz31);
1667 /* Update vectorial force */
1668 fix3 = _mm_add_ps(fix3,tx);
1669 fiy3 = _mm_add_ps(fiy3,ty);
1670 fiz3 = _mm_add_ps(fiz3,tz);
1672 fjx1 = _mm_add_ps(fjx1,tx);
1673 fjy1 = _mm_add_ps(fjy1,ty);
1674 fjz1 = _mm_add_ps(fjz1,tz);
1676 /**************************
1677 * CALCULATE INTERACTIONS *
1678 **************************/
1680 r32 = _mm_mul_ps(rsq32,rinv32);
1682 /* EWALD ELECTROSTATICS */
1684 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1685 ewrt = _mm_mul_ps(r32,ewtabscale);
1686 ewitab = _mm_cvttps_epi32(ewrt);
1687 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1688 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1689 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1691 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1692 felec = _mm_mul_ps(_mm_mul_ps(qq32,rinv32),_mm_sub_ps(rinvsq32,felec));
1696 /* Calculate temporary vectorial force */
1697 tx = _mm_mul_ps(fscal,dx32);
1698 ty = _mm_mul_ps(fscal,dy32);
1699 tz = _mm_mul_ps(fscal,dz32);
1701 /* Update vectorial force */
1702 fix3 = _mm_add_ps(fix3,tx);
1703 fiy3 = _mm_add_ps(fiy3,ty);
1704 fiz3 = _mm_add_ps(fiz3,tz);
1706 fjx2 = _mm_add_ps(fjx2,tx);
1707 fjy2 = _mm_add_ps(fjy2,ty);
1708 fjz2 = _mm_add_ps(fjz2,tz);
1710 /**************************
1711 * CALCULATE INTERACTIONS *
1712 **************************/
1714 r33 = _mm_mul_ps(rsq33,rinv33);
1716 /* EWALD ELECTROSTATICS */
1718 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1719 ewrt = _mm_mul_ps(r33,ewtabscale);
1720 ewitab = _mm_cvttps_epi32(ewrt);
1721 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1722 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1723 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1725 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1726 felec = _mm_mul_ps(_mm_mul_ps(qq33,rinv33),_mm_sub_ps(rinvsq33,felec));
1730 /* Calculate temporary vectorial force */
1731 tx = _mm_mul_ps(fscal,dx33);
1732 ty = _mm_mul_ps(fscal,dy33);
1733 tz = _mm_mul_ps(fscal,dz33);
1735 /* Update vectorial force */
1736 fix3 = _mm_add_ps(fix3,tx);
1737 fiy3 = _mm_add_ps(fiy3,ty);
1738 fiz3 = _mm_add_ps(fiz3,tz);
1740 fjx3 = _mm_add_ps(fjx3,tx);
1741 fjy3 = _mm_add_ps(fjy3,ty);
1742 fjz3 = _mm_add_ps(fjz3,tz);
1744 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(f+j_coord_offsetA+DIM,f+j_coord_offsetB+DIM,
1745 f+j_coord_offsetC+DIM,f+j_coord_offsetD+DIM,
1746 fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1748 /* Inner loop uses 324 flops */
1751 if(jidx<j_index_end)
1754 /* Get j neighbor index, and coordinate index */
1756 jnrB = jjnr[jidx+1];
1757 jnrC = jjnr[jidx+2];
1758 jnrD = jjnr[jidx+3];
1760 /* Sign of each element will be negative for non-real atoms.
1761 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1762 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1764 dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1765 jnrA = (jnrA>=0) ? jnrA : 0;
1766 jnrB = (jnrB>=0) ? jnrB : 0;
1767 jnrC = (jnrC>=0) ? jnrC : 0;
1768 jnrD = (jnrD>=0) ? jnrD : 0;
1770 j_coord_offsetA = DIM*jnrA;
1771 j_coord_offsetB = DIM*jnrB;
1772 j_coord_offsetC = DIM*jnrC;
1773 j_coord_offsetD = DIM*jnrD;
1775 /* load j atom coordinates */
1776 gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
1777 x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
1778 &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
1780 /* Calculate displacement vector */
1781 dx11 = _mm_sub_ps(ix1,jx1);
1782 dy11 = _mm_sub_ps(iy1,jy1);
1783 dz11 = _mm_sub_ps(iz1,jz1);
1784 dx12 = _mm_sub_ps(ix1,jx2);
1785 dy12 = _mm_sub_ps(iy1,jy2);
1786 dz12 = _mm_sub_ps(iz1,jz2);
1787 dx13 = _mm_sub_ps(ix1,jx3);
1788 dy13 = _mm_sub_ps(iy1,jy3);
1789 dz13 = _mm_sub_ps(iz1,jz3);
1790 dx21 = _mm_sub_ps(ix2,jx1);
1791 dy21 = _mm_sub_ps(iy2,jy1);
1792 dz21 = _mm_sub_ps(iz2,jz1);
1793 dx22 = _mm_sub_ps(ix2,jx2);
1794 dy22 = _mm_sub_ps(iy2,jy2);
1795 dz22 = _mm_sub_ps(iz2,jz2);
1796 dx23 = _mm_sub_ps(ix2,jx3);
1797 dy23 = _mm_sub_ps(iy2,jy3);
1798 dz23 = _mm_sub_ps(iz2,jz3);
1799 dx31 = _mm_sub_ps(ix3,jx1);
1800 dy31 = _mm_sub_ps(iy3,jy1);
1801 dz31 = _mm_sub_ps(iz3,jz1);
1802 dx32 = _mm_sub_ps(ix3,jx2);
1803 dy32 = _mm_sub_ps(iy3,jy2);
1804 dz32 = _mm_sub_ps(iz3,jz2);
1805 dx33 = _mm_sub_ps(ix3,jx3);
1806 dy33 = _mm_sub_ps(iy3,jy3);
1807 dz33 = _mm_sub_ps(iz3,jz3);
1809 /* Calculate squared distance and things based on it */
1810 rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1811 rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1812 rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13);
1813 rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1814 rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1815 rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23);
1816 rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31);
1817 rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32);
1818 rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33);
1820 rinv11 = gmx_mm_invsqrt_ps(rsq11);
1821 rinv12 = gmx_mm_invsqrt_ps(rsq12);
1822 rinv13 = gmx_mm_invsqrt_ps(rsq13);
1823 rinv21 = gmx_mm_invsqrt_ps(rsq21);
1824 rinv22 = gmx_mm_invsqrt_ps(rsq22);
1825 rinv23 = gmx_mm_invsqrt_ps(rsq23);
1826 rinv31 = gmx_mm_invsqrt_ps(rsq31);
1827 rinv32 = gmx_mm_invsqrt_ps(rsq32);
1828 rinv33 = gmx_mm_invsqrt_ps(rsq33);
1830 rinvsq11 = _mm_mul_ps(rinv11,rinv11);
1831 rinvsq12 = _mm_mul_ps(rinv12,rinv12);
1832 rinvsq13 = _mm_mul_ps(rinv13,rinv13);
1833 rinvsq21 = _mm_mul_ps(rinv21,rinv21);
1834 rinvsq22 = _mm_mul_ps(rinv22,rinv22);
1835 rinvsq23 = _mm_mul_ps(rinv23,rinv23);
1836 rinvsq31 = _mm_mul_ps(rinv31,rinv31);
1837 rinvsq32 = _mm_mul_ps(rinv32,rinv32);
1838 rinvsq33 = _mm_mul_ps(rinv33,rinv33);
1840 fjx1 = _mm_setzero_ps();
1841 fjy1 = _mm_setzero_ps();
1842 fjz1 = _mm_setzero_ps();
1843 fjx2 = _mm_setzero_ps();
1844 fjy2 = _mm_setzero_ps();
1845 fjz2 = _mm_setzero_ps();
1846 fjx3 = _mm_setzero_ps();
1847 fjy3 = _mm_setzero_ps();
1848 fjz3 = _mm_setzero_ps();
1850 /**************************
1851 * CALCULATE INTERACTIONS *
1852 **************************/
1854 r11 = _mm_mul_ps(rsq11,rinv11);
1855 r11 = _mm_andnot_ps(dummy_mask,r11);
1857 /* EWALD ELECTROSTATICS */
1859 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1860 ewrt = _mm_mul_ps(r11,ewtabscale);
1861 ewitab = _mm_cvttps_epi32(ewrt);
1862 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1863 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1864 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1866 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1867 felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
1871 fscal = _mm_andnot_ps(dummy_mask,fscal);
1873 /* Calculate temporary vectorial force */
1874 tx = _mm_mul_ps(fscal,dx11);
1875 ty = _mm_mul_ps(fscal,dy11);
1876 tz = _mm_mul_ps(fscal,dz11);
1878 /* Update vectorial force */
1879 fix1 = _mm_add_ps(fix1,tx);
1880 fiy1 = _mm_add_ps(fiy1,ty);
1881 fiz1 = _mm_add_ps(fiz1,tz);
1883 fjx1 = _mm_add_ps(fjx1,tx);
1884 fjy1 = _mm_add_ps(fjy1,ty);
1885 fjz1 = _mm_add_ps(fjz1,tz);
1887 /**************************
1888 * CALCULATE INTERACTIONS *
1889 **************************/
1891 r12 = _mm_mul_ps(rsq12,rinv12);
1892 r12 = _mm_andnot_ps(dummy_mask,r12);
1894 /* EWALD ELECTROSTATICS */
1896 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1897 ewrt = _mm_mul_ps(r12,ewtabscale);
1898 ewitab = _mm_cvttps_epi32(ewrt);
1899 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1900 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1901 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1903 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1904 felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
1908 fscal = _mm_andnot_ps(dummy_mask,fscal);
1910 /* Calculate temporary vectorial force */
1911 tx = _mm_mul_ps(fscal,dx12);
1912 ty = _mm_mul_ps(fscal,dy12);
1913 tz = _mm_mul_ps(fscal,dz12);
1915 /* Update vectorial force */
1916 fix1 = _mm_add_ps(fix1,tx);
1917 fiy1 = _mm_add_ps(fiy1,ty);
1918 fiz1 = _mm_add_ps(fiz1,tz);
1920 fjx2 = _mm_add_ps(fjx2,tx);
1921 fjy2 = _mm_add_ps(fjy2,ty);
1922 fjz2 = _mm_add_ps(fjz2,tz);
1924 /**************************
1925 * CALCULATE INTERACTIONS *
1926 **************************/
1928 r13 = _mm_mul_ps(rsq13,rinv13);
1929 r13 = _mm_andnot_ps(dummy_mask,r13);
1931 /* EWALD ELECTROSTATICS */
1933 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1934 ewrt = _mm_mul_ps(r13,ewtabscale);
1935 ewitab = _mm_cvttps_epi32(ewrt);
1936 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1937 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1938 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1940 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1941 felec = _mm_mul_ps(_mm_mul_ps(qq13,rinv13),_mm_sub_ps(rinvsq13,felec));
1945 fscal = _mm_andnot_ps(dummy_mask,fscal);
1947 /* Calculate temporary vectorial force */
1948 tx = _mm_mul_ps(fscal,dx13);
1949 ty = _mm_mul_ps(fscal,dy13);
1950 tz = _mm_mul_ps(fscal,dz13);
1952 /* Update vectorial force */
1953 fix1 = _mm_add_ps(fix1,tx);
1954 fiy1 = _mm_add_ps(fiy1,ty);
1955 fiz1 = _mm_add_ps(fiz1,tz);
1957 fjx3 = _mm_add_ps(fjx3,tx);
1958 fjy3 = _mm_add_ps(fjy3,ty);
1959 fjz3 = _mm_add_ps(fjz3,tz);
1961 /**************************
1962 * CALCULATE INTERACTIONS *
1963 **************************/
1965 r21 = _mm_mul_ps(rsq21,rinv21);
1966 r21 = _mm_andnot_ps(dummy_mask,r21);
1968 /* EWALD ELECTROSTATICS */
1970 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1971 ewrt = _mm_mul_ps(r21,ewtabscale);
1972 ewitab = _mm_cvttps_epi32(ewrt);
1973 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
1974 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
1975 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
1977 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1978 felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
1982 fscal = _mm_andnot_ps(dummy_mask,fscal);
1984 /* Calculate temporary vectorial force */
1985 tx = _mm_mul_ps(fscal,dx21);
1986 ty = _mm_mul_ps(fscal,dy21);
1987 tz = _mm_mul_ps(fscal,dz21);
1989 /* Update vectorial force */
1990 fix2 = _mm_add_ps(fix2,tx);
1991 fiy2 = _mm_add_ps(fiy2,ty);
1992 fiz2 = _mm_add_ps(fiz2,tz);
1994 fjx1 = _mm_add_ps(fjx1,tx);
1995 fjy1 = _mm_add_ps(fjy1,ty);
1996 fjz1 = _mm_add_ps(fjz1,tz);
1998 /**************************
1999 * CALCULATE INTERACTIONS *
2000 **************************/
2002 r22 = _mm_mul_ps(rsq22,rinv22);
2003 r22 = _mm_andnot_ps(dummy_mask,r22);
2005 /* EWALD ELECTROSTATICS */
2007 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2008 ewrt = _mm_mul_ps(r22,ewtabscale);
2009 ewitab = _mm_cvttps_epi32(ewrt);
2010 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
2011 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
2012 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
2014 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2015 felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
2019 fscal = _mm_andnot_ps(dummy_mask,fscal);
2021 /* Calculate temporary vectorial force */
2022 tx = _mm_mul_ps(fscal,dx22);
2023 ty = _mm_mul_ps(fscal,dy22);
2024 tz = _mm_mul_ps(fscal,dz22);
2026 /* Update vectorial force */
2027 fix2 = _mm_add_ps(fix2,tx);
2028 fiy2 = _mm_add_ps(fiy2,ty);
2029 fiz2 = _mm_add_ps(fiz2,tz);
2031 fjx2 = _mm_add_ps(fjx2,tx);
2032 fjy2 = _mm_add_ps(fjy2,ty);
2033 fjz2 = _mm_add_ps(fjz2,tz);
2035 /**************************
2036 * CALCULATE INTERACTIONS *
2037 **************************/
2039 r23 = _mm_mul_ps(rsq23,rinv23);
2040 r23 = _mm_andnot_ps(dummy_mask,r23);
2042 /* EWALD ELECTROSTATICS */
2044 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2045 ewrt = _mm_mul_ps(r23,ewtabscale);
2046 ewitab = _mm_cvttps_epi32(ewrt);
2047 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
2048 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
2049 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
2051 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2052 felec = _mm_mul_ps(_mm_mul_ps(qq23,rinv23),_mm_sub_ps(rinvsq23,felec));
2056 fscal = _mm_andnot_ps(dummy_mask,fscal);
2058 /* Calculate temporary vectorial force */
2059 tx = _mm_mul_ps(fscal,dx23);
2060 ty = _mm_mul_ps(fscal,dy23);
2061 tz = _mm_mul_ps(fscal,dz23);
2063 /* Update vectorial force */
2064 fix2 = _mm_add_ps(fix2,tx);
2065 fiy2 = _mm_add_ps(fiy2,ty);
2066 fiz2 = _mm_add_ps(fiz2,tz);
2068 fjx3 = _mm_add_ps(fjx3,tx);
2069 fjy3 = _mm_add_ps(fjy3,ty);
2070 fjz3 = _mm_add_ps(fjz3,tz);
2072 /**************************
2073 * CALCULATE INTERACTIONS *
2074 **************************/
2076 r31 = _mm_mul_ps(rsq31,rinv31);
2077 r31 = _mm_andnot_ps(dummy_mask,r31);
2079 /* EWALD ELECTROSTATICS */
2081 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2082 ewrt = _mm_mul_ps(r31,ewtabscale);
2083 ewitab = _mm_cvttps_epi32(ewrt);
2084 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
2085 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
2086 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
2088 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2089 felec = _mm_mul_ps(_mm_mul_ps(qq31,rinv31),_mm_sub_ps(rinvsq31,felec));
2093 fscal = _mm_andnot_ps(dummy_mask,fscal);
2095 /* Calculate temporary vectorial force */
2096 tx = _mm_mul_ps(fscal,dx31);
2097 ty = _mm_mul_ps(fscal,dy31);
2098 tz = _mm_mul_ps(fscal,dz31);
2100 /* Update vectorial force */
2101 fix3 = _mm_add_ps(fix3,tx);
2102 fiy3 = _mm_add_ps(fiy3,ty);
2103 fiz3 = _mm_add_ps(fiz3,tz);
2105 fjx1 = _mm_add_ps(fjx1,tx);
2106 fjy1 = _mm_add_ps(fjy1,ty);
2107 fjz1 = _mm_add_ps(fjz1,tz);
2109 /**************************
2110 * CALCULATE INTERACTIONS *
2111 **************************/
2113 r32 = _mm_mul_ps(rsq32,rinv32);
2114 r32 = _mm_andnot_ps(dummy_mask,r32);
2116 /* EWALD ELECTROSTATICS */
2118 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2119 ewrt = _mm_mul_ps(r32,ewtabscale);
2120 ewitab = _mm_cvttps_epi32(ewrt);
2121 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
2122 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
2123 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
2125 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2126 felec = _mm_mul_ps(_mm_mul_ps(qq32,rinv32),_mm_sub_ps(rinvsq32,felec));
2130 fscal = _mm_andnot_ps(dummy_mask,fscal);
2132 /* Calculate temporary vectorial force */
2133 tx = _mm_mul_ps(fscal,dx32);
2134 ty = _mm_mul_ps(fscal,dy32);
2135 tz = _mm_mul_ps(fscal,dz32);
2137 /* Update vectorial force */
2138 fix3 = _mm_add_ps(fix3,tx);
2139 fiy3 = _mm_add_ps(fiy3,ty);
2140 fiz3 = _mm_add_ps(fiz3,tz);
2142 fjx2 = _mm_add_ps(fjx2,tx);
2143 fjy2 = _mm_add_ps(fjy2,ty);
2144 fjz2 = _mm_add_ps(fjz2,tz);
2146 /**************************
2147 * CALCULATE INTERACTIONS *
2148 **************************/
2150 r33 = _mm_mul_ps(rsq33,rinv33);
2151 r33 = _mm_andnot_ps(dummy_mask,r33);
2153 /* EWALD ELECTROSTATICS */
2155 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2156 ewrt = _mm_mul_ps(r33,ewtabscale);
2157 ewitab = _mm_cvttps_epi32(ewrt);
2158 eweps = _mm_sub_ps(ewrt,_mm_cvtepi32_ps(ewitab));
2159 gmx_mm_load_4pair_swizzle_ps(ewtab+gmx_mm_extract_epi32(ewitab,0),ewtab+gmx_mm_extract_epi32(ewitab,1),
2160 ewtab+gmx_mm_extract_epi32(ewitab,2),ewtab+gmx_mm_extract_epi32(ewitab,3),
2162 felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2163 felec = _mm_mul_ps(_mm_mul_ps(qq33,rinv33),_mm_sub_ps(rinvsq33,felec));
2167 fscal = _mm_andnot_ps(dummy_mask,fscal);
2169 /* Calculate temporary vectorial force */
2170 tx = _mm_mul_ps(fscal,dx33);
2171 ty = _mm_mul_ps(fscal,dy33);
2172 tz = _mm_mul_ps(fscal,dz33);
2174 /* Update vectorial force */
2175 fix3 = _mm_add_ps(fix3,tx);
2176 fiy3 = _mm_add_ps(fiy3,ty);
2177 fiz3 = _mm_add_ps(fiz3,tz);
2179 fjx3 = _mm_add_ps(fjx3,tx);
2180 fjy3 = _mm_add_ps(fjy3,ty);
2181 fjz3 = _mm_add_ps(fjz3,tz);
2183 gmx_mm_decrement_3rvec_4ptr_swizzle_ps(f+j_coord_offsetA+DIM,f+j_coord_offsetB+DIM,
2184 f+j_coord_offsetC+DIM,f+j_coord_offsetD+DIM,
2185 fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
2187 /* Inner loop uses 333 flops */
2190 /* End of innermost loop */
2192 gmx_mm_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
2193 f+i_coord_offset+DIM,fshift+i_shift_offset);
2195 /* Increment number of inner iterations */
2196 inneriter += j_index_end - j_index_start;
2198 /* Outer loop uses 27 flops */
2201 /* Increment number of outer iterations */
2204 /* Update outer/inner flops */
2206 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_F,outeriter*27 + inneriter*333);