2 * Note: this file was generated by the Gromacs c 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"
34 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW4W4_VF_c
35 * Electrostatics interaction: Ewald
36 * VdW interaction: LennardJones
37 * Geometry: Water4-Water4
38 * Calculate force/pot: PotentialAndForce
41 nb_kernel_ElecEwSw_VdwLJSw_GeomW4W4_VF_c
42 (t_nblist * gmx_restrict nlist,
43 rvec * gmx_restrict xx,
44 rvec * gmx_restrict ff,
45 t_forcerec * gmx_restrict fr,
46 t_mdatoms * gmx_restrict mdatoms,
47 nb_kernel_data_t * gmx_restrict kernel_data,
48 t_nrnb * gmx_restrict nrnb)
50 int i_shift_offset,i_coord_offset,j_coord_offset;
51 int j_index_start,j_index_end;
52 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
53 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
54 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
55 real *shiftvec,*fshift,*x,*f;
57 real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
59 real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
61 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
63 real ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
65 real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
67 real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
69 real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
71 real jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
72 real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
73 real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
74 real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
75 real dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
76 real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
77 real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
78 real dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
79 real dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
80 real dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
81 real dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
82 real velec,felec,velecsum,facel,crf,krf,krf2;
85 real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
89 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
91 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
98 jindex = nlist->jindex;
100 shiftidx = nlist->shift;
102 shiftvec = fr->shift_vec[0];
103 fshift = fr->fshift[0];
105 charge = mdatoms->chargeA;
106 nvdwtype = fr->ntype;
108 vdwtype = mdatoms->typeA;
110 sh_ewald = fr->ic->sh_ewald;
111 ewtab = fr->ic->tabq_coul_FDV0;
112 ewtabscale = fr->ic->tabq_scale;
113 ewtabhalfspace = 0.5/ewtabscale;
115 /* Setup water-specific parameters */
116 inr = nlist->iinr[0];
117 iq1 = facel*charge[inr+1];
118 iq2 = facel*charge[inr+2];
119 iq3 = facel*charge[inr+3];
120 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
125 vdwjidx0 = 2*vdwtype[inr+0];
126 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
127 c12_00 = vdwparam[vdwioffset0+vdwjidx0+1];
138 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
139 rcutoff = fr->rcoulomb;
140 rcutoff2 = rcutoff*rcutoff;
142 rswitch = fr->rcoulomb_switch;
143 /* Setup switch parameters */
145 swV3 = -10.0/(d*d*d);
146 swV4 = 15.0/(d*d*d*d);
147 swV5 = -6.0/(d*d*d*d*d);
148 swF2 = -30.0/(d*d*d);
149 swF3 = 60.0/(d*d*d*d);
150 swF4 = -30.0/(d*d*d*d*d);
155 /* Start outer loop over neighborlists */
156 for(iidx=0; iidx<nri; iidx++)
158 /* Load shift vector for this list */
159 i_shift_offset = DIM*shiftidx[iidx];
160 shX = shiftvec[i_shift_offset+XX];
161 shY = shiftvec[i_shift_offset+YY];
162 shZ = shiftvec[i_shift_offset+ZZ];
164 /* Load limits for loop over neighbors */
165 j_index_start = jindex[iidx];
166 j_index_end = jindex[iidx+1];
168 /* Get outer coordinate index */
170 i_coord_offset = DIM*inr;
172 /* Load i particle coords and add shift vector */
173 ix0 = shX + x[i_coord_offset+DIM*0+XX];
174 iy0 = shY + x[i_coord_offset+DIM*0+YY];
175 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
176 ix1 = shX + x[i_coord_offset+DIM*1+XX];
177 iy1 = shY + x[i_coord_offset+DIM*1+YY];
178 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
179 ix2 = shX + x[i_coord_offset+DIM*2+XX];
180 iy2 = shY + x[i_coord_offset+DIM*2+YY];
181 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
182 ix3 = shX + x[i_coord_offset+DIM*3+XX];
183 iy3 = shY + x[i_coord_offset+DIM*3+YY];
184 iz3 = shZ + x[i_coord_offset+DIM*3+ZZ];
199 /* Reset potential sums */
203 /* Start inner kernel loop */
204 for(jidx=j_index_start; jidx<j_index_end; jidx++)
206 /* Get j neighbor index, and coordinate index */
208 j_coord_offset = DIM*jnr;
210 /* load j atom coordinates */
211 jx0 = x[j_coord_offset+DIM*0+XX];
212 jy0 = x[j_coord_offset+DIM*0+YY];
213 jz0 = x[j_coord_offset+DIM*0+ZZ];
214 jx1 = x[j_coord_offset+DIM*1+XX];
215 jy1 = x[j_coord_offset+DIM*1+YY];
216 jz1 = x[j_coord_offset+DIM*1+ZZ];
217 jx2 = x[j_coord_offset+DIM*2+XX];
218 jy2 = x[j_coord_offset+DIM*2+YY];
219 jz2 = x[j_coord_offset+DIM*2+ZZ];
220 jx3 = x[j_coord_offset+DIM*3+XX];
221 jy3 = x[j_coord_offset+DIM*3+YY];
222 jz3 = x[j_coord_offset+DIM*3+ZZ];
224 /* Calculate displacement vector */
256 /* Calculate squared distance and things based on it */
257 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
258 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
259 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
260 rsq13 = dx13*dx13+dy13*dy13+dz13*dz13;
261 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
262 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
263 rsq23 = dx23*dx23+dy23*dy23+dz23*dz23;
264 rsq31 = dx31*dx31+dy31*dy31+dz31*dz31;
265 rsq32 = dx32*dx32+dy32*dy32+dz32*dz32;
266 rsq33 = dx33*dx33+dy33*dy33+dz33*dz33;
268 rinv00 = gmx_invsqrt(rsq00);
269 rinv11 = gmx_invsqrt(rsq11);
270 rinv12 = gmx_invsqrt(rsq12);
271 rinv13 = gmx_invsqrt(rsq13);
272 rinv21 = gmx_invsqrt(rsq21);
273 rinv22 = gmx_invsqrt(rsq22);
274 rinv23 = gmx_invsqrt(rsq23);
275 rinv31 = gmx_invsqrt(rsq31);
276 rinv32 = gmx_invsqrt(rsq32);
277 rinv33 = gmx_invsqrt(rsq33);
279 rinvsq00 = rinv00*rinv00;
280 rinvsq11 = rinv11*rinv11;
281 rinvsq12 = rinv12*rinv12;
282 rinvsq13 = rinv13*rinv13;
283 rinvsq21 = rinv21*rinv21;
284 rinvsq22 = rinv22*rinv22;
285 rinvsq23 = rinv23*rinv23;
286 rinvsq31 = rinv31*rinv31;
287 rinvsq32 = rinv32*rinv32;
288 rinvsq33 = rinv33*rinv33;
290 /**************************
291 * CALCULATE INTERACTIONS *
292 **************************/
299 /* LENNARD-JONES DISPERSION/REPULSION */
301 rinvsix = rinvsq00*rinvsq00*rinvsq00;
302 vvdw6 = c6_00*rinvsix;
303 vvdw12 = c12_00*rinvsix*rinvsix;
304 vvdw = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
305 fvdw = (vvdw12-vvdw6)*rinvsq00;
308 d = (d>0.0) ? d : 0.0;
310 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
312 dsw = d2*(swF2+d*(swF3+d*swF4));
314 /* Evaluate switch function */
315 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
316 fvdw = fvdw*sw - rinv00*vvdw*dsw;
319 /* Update potential sums from outer loop */
324 /* Calculate temporary vectorial force */
329 /* Update vectorial force */
333 f[j_coord_offset+DIM*0+XX] -= tx;
334 f[j_coord_offset+DIM*0+YY] -= ty;
335 f[j_coord_offset+DIM*0+ZZ] -= tz;
339 /**************************
340 * CALCULATE INTERACTIONS *
341 **************************/
348 /* EWALD ELECTROSTATICS */
350 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
351 ewrt = r11*ewtabscale;
355 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
356 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
357 felec = qq11*rinv11*(rinvsq11-felec);
360 d = (d>0.0) ? d : 0.0;
362 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
364 dsw = d2*(swF2+d*(swF3+d*swF4));
366 /* Evaluate switch function */
367 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
368 felec = felec*sw - rinv11*velec*dsw;
371 /* Update potential sums from outer loop */
376 /* Calculate temporary vectorial force */
381 /* Update vectorial force */
385 f[j_coord_offset+DIM*1+XX] -= tx;
386 f[j_coord_offset+DIM*1+YY] -= ty;
387 f[j_coord_offset+DIM*1+ZZ] -= tz;
391 /**************************
392 * CALCULATE INTERACTIONS *
393 **************************/
400 /* EWALD ELECTROSTATICS */
402 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
403 ewrt = r12*ewtabscale;
407 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
408 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
409 felec = qq12*rinv12*(rinvsq12-felec);
412 d = (d>0.0) ? d : 0.0;
414 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
416 dsw = d2*(swF2+d*(swF3+d*swF4));
418 /* Evaluate switch function */
419 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
420 felec = felec*sw - rinv12*velec*dsw;
423 /* Update potential sums from outer loop */
428 /* Calculate temporary vectorial force */
433 /* Update vectorial force */
437 f[j_coord_offset+DIM*2+XX] -= tx;
438 f[j_coord_offset+DIM*2+YY] -= ty;
439 f[j_coord_offset+DIM*2+ZZ] -= tz;
443 /**************************
444 * CALCULATE INTERACTIONS *
445 **************************/
452 /* EWALD ELECTROSTATICS */
454 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
455 ewrt = r13*ewtabscale;
459 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
460 velec = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
461 felec = qq13*rinv13*(rinvsq13-felec);
464 d = (d>0.0) ? d : 0.0;
466 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
468 dsw = d2*(swF2+d*(swF3+d*swF4));
470 /* Evaluate switch function */
471 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
472 felec = felec*sw - rinv13*velec*dsw;
475 /* Update potential sums from outer loop */
480 /* Calculate temporary vectorial force */
485 /* Update vectorial force */
489 f[j_coord_offset+DIM*3+XX] -= tx;
490 f[j_coord_offset+DIM*3+YY] -= ty;
491 f[j_coord_offset+DIM*3+ZZ] -= tz;
495 /**************************
496 * CALCULATE INTERACTIONS *
497 **************************/
504 /* EWALD ELECTROSTATICS */
506 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
507 ewrt = r21*ewtabscale;
511 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
512 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
513 felec = qq21*rinv21*(rinvsq21-felec);
516 d = (d>0.0) ? d : 0.0;
518 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
520 dsw = d2*(swF2+d*(swF3+d*swF4));
522 /* Evaluate switch function */
523 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
524 felec = felec*sw - rinv21*velec*dsw;
527 /* Update potential sums from outer loop */
532 /* Calculate temporary vectorial force */
537 /* Update vectorial force */
541 f[j_coord_offset+DIM*1+XX] -= tx;
542 f[j_coord_offset+DIM*1+YY] -= ty;
543 f[j_coord_offset+DIM*1+ZZ] -= tz;
547 /**************************
548 * CALCULATE INTERACTIONS *
549 **************************/
556 /* EWALD ELECTROSTATICS */
558 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
559 ewrt = r22*ewtabscale;
563 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
564 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
565 felec = qq22*rinv22*(rinvsq22-felec);
568 d = (d>0.0) ? d : 0.0;
570 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
572 dsw = d2*(swF2+d*(swF3+d*swF4));
574 /* Evaluate switch function */
575 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
576 felec = felec*sw - rinv22*velec*dsw;
579 /* Update potential sums from outer loop */
584 /* Calculate temporary vectorial force */
589 /* Update vectorial force */
593 f[j_coord_offset+DIM*2+XX] -= tx;
594 f[j_coord_offset+DIM*2+YY] -= ty;
595 f[j_coord_offset+DIM*2+ZZ] -= tz;
599 /**************************
600 * CALCULATE INTERACTIONS *
601 **************************/
608 /* EWALD ELECTROSTATICS */
610 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
611 ewrt = r23*ewtabscale;
615 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
616 velec = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
617 felec = qq23*rinv23*(rinvsq23-felec);
620 d = (d>0.0) ? d : 0.0;
622 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
624 dsw = d2*(swF2+d*(swF3+d*swF4));
626 /* Evaluate switch function */
627 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
628 felec = felec*sw - rinv23*velec*dsw;
631 /* Update potential sums from outer loop */
636 /* Calculate temporary vectorial force */
641 /* Update vectorial force */
645 f[j_coord_offset+DIM*3+XX] -= tx;
646 f[j_coord_offset+DIM*3+YY] -= ty;
647 f[j_coord_offset+DIM*3+ZZ] -= tz;
651 /**************************
652 * CALCULATE INTERACTIONS *
653 **************************/
660 /* EWALD ELECTROSTATICS */
662 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
663 ewrt = r31*ewtabscale;
667 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
668 velec = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
669 felec = qq31*rinv31*(rinvsq31-felec);
672 d = (d>0.0) ? d : 0.0;
674 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
676 dsw = d2*(swF2+d*(swF3+d*swF4));
678 /* Evaluate switch function */
679 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
680 felec = felec*sw - rinv31*velec*dsw;
683 /* Update potential sums from outer loop */
688 /* Calculate temporary vectorial force */
693 /* Update vectorial force */
697 f[j_coord_offset+DIM*1+XX] -= tx;
698 f[j_coord_offset+DIM*1+YY] -= ty;
699 f[j_coord_offset+DIM*1+ZZ] -= tz;
703 /**************************
704 * CALCULATE INTERACTIONS *
705 **************************/
712 /* EWALD ELECTROSTATICS */
714 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
715 ewrt = r32*ewtabscale;
719 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
720 velec = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
721 felec = qq32*rinv32*(rinvsq32-felec);
724 d = (d>0.0) ? d : 0.0;
726 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
728 dsw = d2*(swF2+d*(swF3+d*swF4));
730 /* Evaluate switch function */
731 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
732 felec = felec*sw - rinv32*velec*dsw;
735 /* Update potential sums from outer loop */
740 /* Calculate temporary vectorial force */
745 /* Update vectorial force */
749 f[j_coord_offset+DIM*2+XX] -= tx;
750 f[j_coord_offset+DIM*2+YY] -= ty;
751 f[j_coord_offset+DIM*2+ZZ] -= tz;
755 /**************************
756 * CALCULATE INTERACTIONS *
757 **************************/
764 /* EWALD ELECTROSTATICS */
766 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
767 ewrt = r33*ewtabscale;
771 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
772 velec = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
773 felec = qq33*rinv33*(rinvsq33-felec);
776 d = (d>0.0) ? d : 0.0;
778 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
780 dsw = d2*(swF2+d*(swF3+d*swF4));
782 /* Evaluate switch function */
783 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
784 felec = felec*sw - rinv33*velec*dsw;
787 /* Update potential sums from outer loop */
792 /* Calculate temporary vectorial force */
797 /* Update vectorial force */
801 f[j_coord_offset+DIM*3+XX] -= tx;
802 f[j_coord_offset+DIM*3+YY] -= ty;
803 f[j_coord_offset+DIM*3+ZZ] -= tz;
807 /* Inner loop uses 575 flops */
809 /* End of innermost loop */
812 f[i_coord_offset+DIM*0+XX] += fix0;
813 f[i_coord_offset+DIM*0+YY] += fiy0;
814 f[i_coord_offset+DIM*0+ZZ] += fiz0;
818 f[i_coord_offset+DIM*1+XX] += fix1;
819 f[i_coord_offset+DIM*1+YY] += fiy1;
820 f[i_coord_offset+DIM*1+ZZ] += fiz1;
824 f[i_coord_offset+DIM*2+XX] += fix2;
825 f[i_coord_offset+DIM*2+YY] += fiy2;
826 f[i_coord_offset+DIM*2+ZZ] += fiz2;
830 f[i_coord_offset+DIM*3+XX] += fix3;
831 f[i_coord_offset+DIM*3+YY] += fiy3;
832 f[i_coord_offset+DIM*3+ZZ] += fiz3;
836 fshift[i_shift_offset+XX] += tx;
837 fshift[i_shift_offset+YY] += ty;
838 fshift[i_shift_offset+ZZ] += tz;
841 /* Update potential energies */
842 kernel_data->energygrp_elec[ggid] += velecsum;
843 kernel_data->energygrp_vdw[ggid] += vvdwsum;
845 /* Increment number of inner iterations */
846 inneriter += j_index_end - j_index_start;
848 /* Outer loop uses 41 flops */
851 /* Increment number of outer iterations */
854 /* Update outer/inner flops */
856 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*41 + inneriter*575);
859 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW4W4_F_c
860 * Electrostatics interaction: Ewald
861 * VdW interaction: LennardJones
862 * Geometry: Water4-Water4
863 * Calculate force/pot: Force
866 nb_kernel_ElecEwSw_VdwLJSw_GeomW4W4_F_c
867 (t_nblist * gmx_restrict nlist,
868 rvec * gmx_restrict xx,
869 rvec * gmx_restrict ff,
870 t_forcerec * gmx_restrict fr,
871 t_mdatoms * gmx_restrict mdatoms,
872 nb_kernel_data_t * gmx_restrict kernel_data,
873 t_nrnb * gmx_restrict nrnb)
875 int i_shift_offset,i_coord_offset,j_coord_offset;
876 int j_index_start,j_index_end;
877 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
878 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
879 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
880 real *shiftvec,*fshift,*x,*f;
882 real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
884 real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
886 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
888 real ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
890 real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
892 real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
894 real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
896 real jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
897 real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
898 real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
899 real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
900 real dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
901 real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
902 real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
903 real dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
904 real dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
905 real dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
906 real dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
907 real velec,felec,velecsum,facel,crf,krf,krf2;
910 real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
914 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
916 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
923 jindex = nlist->jindex;
925 shiftidx = nlist->shift;
927 shiftvec = fr->shift_vec[0];
928 fshift = fr->fshift[0];
930 charge = mdatoms->chargeA;
931 nvdwtype = fr->ntype;
933 vdwtype = mdatoms->typeA;
935 sh_ewald = fr->ic->sh_ewald;
936 ewtab = fr->ic->tabq_coul_FDV0;
937 ewtabscale = fr->ic->tabq_scale;
938 ewtabhalfspace = 0.5/ewtabscale;
940 /* Setup water-specific parameters */
941 inr = nlist->iinr[0];
942 iq1 = facel*charge[inr+1];
943 iq2 = facel*charge[inr+2];
944 iq3 = facel*charge[inr+3];
945 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
950 vdwjidx0 = 2*vdwtype[inr+0];
951 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
952 c12_00 = vdwparam[vdwioffset0+vdwjidx0+1];
963 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
964 rcutoff = fr->rcoulomb;
965 rcutoff2 = rcutoff*rcutoff;
967 rswitch = fr->rcoulomb_switch;
968 /* Setup switch parameters */
970 swV3 = -10.0/(d*d*d);
971 swV4 = 15.0/(d*d*d*d);
972 swV5 = -6.0/(d*d*d*d*d);
973 swF2 = -30.0/(d*d*d);
974 swF3 = 60.0/(d*d*d*d);
975 swF4 = -30.0/(d*d*d*d*d);
980 /* Start outer loop over neighborlists */
981 for(iidx=0; iidx<nri; iidx++)
983 /* Load shift vector for this list */
984 i_shift_offset = DIM*shiftidx[iidx];
985 shX = shiftvec[i_shift_offset+XX];
986 shY = shiftvec[i_shift_offset+YY];
987 shZ = shiftvec[i_shift_offset+ZZ];
989 /* Load limits for loop over neighbors */
990 j_index_start = jindex[iidx];
991 j_index_end = jindex[iidx+1];
993 /* Get outer coordinate index */
995 i_coord_offset = DIM*inr;
997 /* Load i particle coords and add shift vector */
998 ix0 = shX + x[i_coord_offset+DIM*0+XX];
999 iy0 = shY + x[i_coord_offset+DIM*0+YY];
1000 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
1001 ix1 = shX + x[i_coord_offset+DIM*1+XX];
1002 iy1 = shY + x[i_coord_offset+DIM*1+YY];
1003 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
1004 ix2 = shX + x[i_coord_offset+DIM*2+XX];
1005 iy2 = shY + x[i_coord_offset+DIM*2+YY];
1006 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
1007 ix3 = shX + x[i_coord_offset+DIM*3+XX];
1008 iy3 = shY + x[i_coord_offset+DIM*3+YY];
1009 iz3 = shZ + x[i_coord_offset+DIM*3+ZZ];
1024 /* Start inner kernel loop */
1025 for(jidx=j_index_start; jidx<j_index_end; jidx++)
1027 /* Get j neighbor index, and coordinate index */
1029 j_coord_offset = DIM*jnr;
1031 /* load j atom coordinates */
1032 jx0 = x[j_coord_offset+DIM*0+XX];
1033 jy0 = x[j_coord_offset+DIM*0+YY];
1034 jz0 = x[j_coord_offset+DIM*0+ZZ];
1035 jx1 = x[j_coord_offset+DIM*1+XX];
1036 jy1 = x[j_coord_offset+DIM*1+YY];
1037 jz1 = x[j_coord_offset+DIM*1+ZZ];
1038 jx2 = x[j_coord_offset+DIM*2+XX];
1039 jy2 = x[j_coord_offset+DIM*2+YY];
1040 jz2 = x[j_coord_offset+DIM*2+ZZ];
1041 jx3 = x[j_coord_offset+DIM*3+XX];
1042 jy3 = x[j_coord_offset+DIM*3+YY];
1043 jz3 = x[j_coord_offset+DIM*3+ZZ];
1045 /* Calculate displacement vector */
1077 /* Calculate squared distance and things based on it */
1078 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
1079 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
1080 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
1081 rsq13 = dx13*dx13+dy13*dy13+dz13*dz13;
1082 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
1083 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
1084 rsq23 = dx23*dx23+dy23*dy23+dz23*dz23;
1085 rsq31 = dx31*dx31+dy31*dy31+dz31*dz31;
1086 rsq32 = dx32*dx32+dy32*dy32+dz32*dz32;
1087 rsq33 = dx33*dx33+dy33*dy33+dz33*dz33;
1089 rinv00 = gmx_invsqrt(rsq00);
1090 rinv11 = gmx_invsqrt(rsq11);
1091 rinv12 = gmx_invsqrt(rsq12);
1092 rinv13 = gmx_invsqrt(rsq13);
1093 rinv21 = gmx_invsqrt(rsq21);
1094 rinv22 = gmx_invsqrt(rsq22);
1095 rinv23 = gmx_invsqrt(rsq23);
1096 rinv31 = gmx_invsqrt(rsq31);
1097 rinv32 = gmx_invsqrt(rsq32);
1098 rinv33 = gmx_invsqrt(rsq33);
1100 rinvsq00 = rinv00*rinv00;
1101 rinvsq11 = rinv11*rinv11;
1102 rinvsq12 = rinv12*rinv12;
1103 rinvsq13 = rinv13*rinv13;
1104 rinvsq21 = rinv21*rinv21;
1105 rinvsq22 = rinv22*rinv22;
1106 rinvsq23 = rinv23*rinv23;
1107 rinvsq31 = rinv31*rinv31;
1108 rinvsq32 = rinv32*rinv32;
1109 rinvsq33 = rinv33*rinv33;
1111 /**************************
1112 * CALCULATE INTERACTIONS *
1113 **************************/
1120 /* LENNARD-JONES DISPERSION/REPULSION */
1122 rinvsix = rinvsq00*rinvsq00*rinvsq00;
1123 vvdw6 = c6_00*rinvsix;
1124 vvdw12 = c12_00*rinvsix*rinvsix;
1125 vvdw = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
1126 fvdw = (vvdw12-vvdw6)*rinvsq00;
1129 d = (d>0.0) ? d : 0.0;
1131 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1133 dsw = d2*(swF2+d*(swF3+d*swF4));
1135 /* Evaluate switch function */
1136 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1137 fvdw = fvdw*sw - rinv00*vvdw*dsw;
1141 /* Calculate temporary vectorial force */
1146 /* Update vectorial force */
1150 f[j_coord_offset+DIM*0+XX] -= tx;
1151 f[j_coord_offset+DIM*0+YY] -= ty;
1152 f[j_coord_offset+DIM*0+ZZ] -= tz;
1156 /**************************
1157 * CALCULATE INTERACTIONS *
1158 **************************/
1165 /* EWALD ELECTROSTATICS */
1167 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1168 ewrt = r11*ewtabscale;
1170 eweps = ewrt-ewitab;
1172 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1173 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1174 felec = qq11*rinv11*(rinvsq11-felec);
1177 d = (d>0.0) ? d : 0.0;
1179 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1181 dsw = d2*(swF2+d*(swF3+d*swF4));
1183 /* Evaluate switch function */
1184 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1185 felec = felec*sw - rinv11*velec*dsw;
1189 /* Calculate temporary vectorial force */
1194 /* Update vectorial force */
1198 f[j_coord_offset+DIM*1+XX] -= tx;
1199 f[j_coord_offset+DIM*1+YY] -= ty;
1200 f[j_coord_offset+DIM*1+ZZ] -= tz;
1204 /**************************
1205 * CALCULATE INTERACTIONS *
1206 **************************/
1213 /* EWALD ELECTROSTATICS */
1215 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1216 ewrt = r12*ewtabscale;
1218 eweps = ewrt-ewitab;
1220 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1221 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1222 felec = qq12*rinv12*(rinvsq12-felec);
1225 d = (d>0.0) ? d : 0.0;
1227 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1229 dsw = d2*(swF2+d*(swF3+d*swF4));
1231 /* Evaluate switch function */
1232 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1233 felec = felec*sw - rinv12*velec*dsw;
1237 /* Calculate temporary vectorial force */
1242 /* Update vectorial force */
1246 f[j_coord_offset+DIM*2+XX] -= tx;
1247 f[j_coord_offset+DIM*2+YY] -= ty;
1248 f[j_coord_offset+DIM*2+ZZ] -= tz;
1252 /**************************
1253 * CALCULATE INTERACTIONS *
1254 **************************/
1261 /* EWALD ELECTROSTATICS */
1263 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1264 ewrt = r13*ewtabscale;
1266 eweps = ewrt-ewitab;
1268 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1269 velec = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1270 felec = qq13*rinv13*(rinvsq13-felec);
1273 d = (d>0.0) ? d : 0.0;
1275 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1277 dsw = d2*(swF2+d*(swF3+d*swF4));
1279 /* Evaluate switch function */
1280 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1281 felec = felec*sw - rinv13*velec*dsw;
1285 /* Calculate temporary vectorial force */
1290 /* Update vectorial force */
1294 f[j_coord_offset+DIM*3+XX] -= tx;
1295 f[j_coord_offset+DIM*3+YY] -= ty;
1296 f[j_coord_offset+DIM*3+ZZ] -= tz;
1300 /**************************
1301 * CALCULATE INTERACTIONS *
1302 **************************/
1309 /* EWALD ELECTROSTATICS */
1311 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1312 ewrt = r21*ewtabscale;
1314 eweps = ewrt-ewitab;
1316 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1317 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1318 felec = qq21*rinv21*(rinvsq21-felec);
1321 d = (d>0.0) ? d : 0.0;
1323 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1325 dsw = d2*(swF2+d*(swF3+d*swF4));
1327 /* Evaluate switch function */
1328 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1329 felec = felec*sw - rinv21*velec*dsw;
1333 /* Calculate temporary vectorial force */
1338 /* Update vectorial force */
1342 f[j_coord_offset+DIM*1+XX] -= tx;
1343 f[j_coord_offset+DIM*1+YY] -= ty;
1344 f[j_coord_offset+DIM*1+ZZ] -= tz;
1348 /**************************
1349 * CALCULATE INTERACTIONS *
1350 **************************/
1357 /* EWALD ELECTROSTATICS */
1359 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1360 ewrt = r22*ewtabscale;
1362 eweps = ewrt-ewitab;
1364 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1365 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1366 felec = qq22*rinv22*(rinvsq22-felec);
1369 d = (d>0.0) ? d : 0.0;
1371 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1373 dsw = d2*(swF2+d*(swF3+d*swF4));
1375 /* Evaluate switch function */
1376 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1377 felec = felec*sw - rinv22*velec*dsw;
1381 /* Calculate temporary vectorial force */
1386 /* Update vectorial force */
1390 f[j_coord_offset+DIM*2+XX] -= tx;
1391 f[j_coord_offset+DIM*2+YY] -= ty;
1392 f[j_coord_offset+DIM*2+ZZ] -= tz;
1396 /**************************
1397 * CALCULATE INTERACTIONS *
1398 **************************/
1405 /* EWALD ELECTROSTATICS */
1407 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1408 ewrt = r23*ewtabscale;
1410 eweps = ewrt-ewitab;
1412 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1413 velec = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1414 felec = qq23*rinv23*(rinvsq23-felec);
1417 d = (d>0.0) ? d : 0.0;
1419 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1421 dsw = d2*(swF2+d*(swF3+d*swF4));
1423 /* Evaluate switch function */
1424 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1425 felec = felec*sw - rinv23*velec*dsw;
1429 /* Calculate temporary vectorial force */
1434 /* Update vectorial force */
1438 f[j_coord_offset+DIM*3+XX] -= tx;
1439 f[j_coord_offset+DIM*3+YY] -= ty;
1440 f[j_coord_offset+DIM*3+ZZ] -= tz;
1444 /**************************
1445 * CALCULATE INTERACTIONS *
1446 **************************/
1453 /* EWALD ELECTROSTATICS */
1455 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1456 ewrt = r31*ewtabscale;
1458 eweps = ewrt-ewitab;
1460 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1461 velec = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1462 felec = qq31*rinv31*(rinvsq31-felec);
1465 d = (d>0.0) ? d : 0.0;
1467 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1469 dsw = d2*(swF2+d*(swF3+d*swF4));
1471 /* Evaluate switch function */
1472 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1473 felec = felec*sw - rinv31*velec*dsw;
1477 /* Calculate temporary vectorial force */
1482 /* Update vectorial force */
1486 f[j_coord_offset+DIM*1+XX] -= tx;
1487 f[j_coord_offset+DIM*1+YY] -= ty;
1488 f[j_coord_offset+DIM*1+ZZ] -= tz;
1492 /**************************
1493 * CALCULATE INTERACTIONS *
1494 **************************/
1501 /* EWALD ELECTROSTATICS */
1503 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1504 ewrt = r32*ewtabscale;
1506 eweps = ewrt-ewitab;
1508 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1509 velec = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1510 felec = qq32*rinv32*(rinvsq32-felec);
1513 d = (d>0.0) ? d : 0.0;
1515 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1517 dsw = d2*(swF2+d*(swF3+d*swF4));
1519 /* Evaluate switch function */
1520 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1521 felec = felec*sw - rinv32*velec*dsw;
1525 /* Calculate temporary vectorial force */
1530 /* Update vectorial force */
1534 f[j_coord_offset+DIM*2+XX] -= tx;
1535 f[j_coord_offset+DIM*2+YY] -= ty;
1536 f[j_coord_offset+DIM*2+ZZ] -= tz;
1540 /**************************
1541 * CALCULATE INTERACTIONS *
1542 **************************/
1549 /* EWALD ELECTROSTATICS */
1551 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1552 ewrt = r33*ewtabscale;
1554 eweps = ewrt-ewitab;
1556 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1557 velec = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1558 felec = qq33*rinv33*(rinvsq33-felec);
1561 d = (d>0.0) ? d : 0.0;
1563 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1565 dsw = d2*(swF2+d*(swF3+d*swF4));
1567 /* Evaluate switch function */
1568 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1569 felec = felec*sw - rinv33*velec*dsw;
1573 /* Calculate temporary vectorial force */
1578 /* Update vectorial force */
1582 f[j_coord_offset+DIM*3+XX] -= tx;
1583 f[j_coord_offset+DIM*3+YY] -= ty;
1584 f[j_coord_offset+DIM*3+ZZ] -= tz;
1588 /* Inner loop uses 555 flops */
1590 /* End of innermost loop */
1593 f[i_coord_offset+DIM*0+XX] += fix0;
1594 f[i_coord_offset+DIM*0+YY] += fiy0;
1595 f[i_coord_offset+DIM*0+ZZ] += fiz0;
1599 f[i_coord_offset+DIM*1+XX] += fix1;
1600 f[i_coord_offset+DIM*1+YY] += fiy1;
1601 f[i_coord_offset+DIM*1+ZZ] += fiz1;
1605 f[i_coord_offset+DIM*2+XX] += fix2;
1606 f[i_coord_offset+DIM*2+YY] += fiy2;
1607 f[i_coord_offset+DIM*2+ZZ] += fiz2;
1611 f[i_coord_offset+DIM*3+XX] += fix3;
1612 f[i_coord_offset+DIM*3+YY] += fiy3;
1613 f[i_coord_offset+DIM*3+ZZ] += fiz3;
1617 fshift[i_shift_offset+XX] += tx;
1618 fshift[i_shift_offset+YY] += ty;
1619 fshift[i_shift_offset+ZZ] += tz;
1621 /* Increment number of inner iterations */
1622 inneriter += j_index_end - j_index_start;
1624 /* Outer loop uses 39 flops */
1627 /* Increment number of outer iterations */
1630 /* Update outer/inner flops */
1632 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*39 + inneriter*555);