2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
36 * Note: this file was generated by the GROMACS c kernel generator.
42 #include "../nb_kernel.h"
43 #include "types/simple.h"
44 #include "gromacs/math/vec.h"
48 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_VF_c
49 * Electrostatics interaction: Ewald
50 * VdW interaction: LennardJones
51 * Geometry: Water3-Water3
52 * Calculate force/pot: PotentialAndForce
55 nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_VF_c
56 (t_nblist * gmx_restrict nlist,
57 rvec * gmx_restrict xx,
58 rvec * gmx_restrict ff,
59 t_forcerec * gmx_restrict fr,
60 t_mdatoms * gmx_restrict mdatoms,
61 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
62 t_nrnb * gmx_restrict nrnb)
64 int i_shift_offset,i_coord_offset,j_coord_offset;
65 int j_index_start,j_index_end;
66 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
67 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
68 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
69 real *shiftvec,*fshift,*x,*f;
71 real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
73 real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
75 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
77 real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
79 real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
81 real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
82 real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
83 real dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01;
84 real dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02;
85 real dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
86 real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
87 real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
88 real dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
89 real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
90 real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
91 real velec,felec,velecsum,facel,crf,krf,krf2;
94 real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
98 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
100 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
107 jindex = nlist->jindex;
109 shiftidx = nlist->shift;
111 shiftvec = fr->shift_vec[0];
112 fshift = fr->fshift[0];
114 charge = mdatoms->chargeA;
115 nvdwtype = fr->ntype;
117 vdwtype = mdatoms->typeA;
119 sh_ewald = fr->ic->sh_ewald;
120 ewtab = fr->ic->tabq_coul_FDV0;
121 ewtabscale = fr->ic->tabq_scale;
122 ewtabhalfspace = 0.5/ewtabscale;
124 /* Setup water-specific parameters */
125 inr = nlist->iinr[0];
126 iq0 = facel*charge[inr+0];
127 iq1 = facel*charge[inr+1];
128 iq2 = facel*charge[inr+2];
129 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
134 vdwjidx0 = 2*vdwtype[inr+0];
136 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
137 c12_00 = vdwparam[vdwioffset0+vdwjidx0+1];
147 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
148 rcutoff = fr->rcoulomb;
149 rcutoff2 = rcutoff*rcutoff;
151 rswitch = fr->rcoulomb_switch;
152 /* Setup switch parameters */
154 swV3 = -10.0/(d*d*d);
155 swV4 = 15.0/(d*d*d*d);
156 swV5 = -6.0/(d*d*d*d*d);
157 swF2 = -30.0/(d*d*d);
158 swF3 = 60.0/(d*d*d*d);
159 swF4 = -30.0/(d*d*d*d*d);
164 /* Start outer loop over neighborlists */
165 for(iidx=0; iidx<nri; iidx++)
167 /* Load shift vector for this list */
168 i_shift_offset = DIM*shiftidx[iidx];
169 shX = shiftvec[i_shift_offset+XX];
170 shY = shiftvec[i_shift_offset+YY];
171 shZ = shiftvec[i_shift_offset+ZZ];
173 /* Load limits for loop over neighbors */
174 j_index_start = jindex[iidx];
175 j_index_end = jindex[iidx+1];
177 /* Get outer coordinate index */
179 i_coord_offset = DIM*inr;
181 /* Load i particle coords and add shift vector */
182 ix0 = shX + x[i_coord_offset+DIM*0+XX];
183 iy0 = shY + x[i_coord_offset+DIM*0+YY];
184 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
185 ix1 = shX + x[i_coord_offset+DIM*1+XX];
186 iy1 = shY + x[i_coord_offset+DIM*1+YY];
187 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
188 ix2 = shX + x[i_coord_offset+DIM*2+XX];
189 iy2 = shY + x[i_coord_offset+DIM*2+YY];
190 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
202 /* Reset potential sums */
206 /* Start inner kernel loop */
207 for(jidx=j_index_start; jidx<j_index_end; jidx++)
209 /* Get j neighbor index, and coordinate index */
211 j_coord_offset = DIM*jnr;
213 /* load j atom coordinates */
214 jx0 = x[j_coord_offset+DIM*0+XX];
215 jy0 = x[j_coord_offset+DIM*0+YY];
216 jz0 = x[j_coord_offset+DIM*0+ZZ];
217 jx1 = x[j_coord_offset+DIM*1+XX];
218 jy1 = x[j_coord_offset+DIM*1+YY];
219 jz1 = x[j_coord_offset+DIM*1+ZZ];
220 jx2 = x[j_coord_offset+DIM*2+XX];
221 jy2 = x[j_coord_offset+DIM*2+YY];
222 jz2 = x[j_coord_offset+DIM*2+ZZ];
224 /* Calculate displacement vector */
253 /* Calculate squared distance and things based on it */
254 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
255 rsq01 = dx01*dx01+dy01*dy01+dz01*dz01;
256 rsq02 = dx02*dx02+dy02*dy02+dz02*dz02;
257 rsq10 = dx10*dx10+dy10*dy10+dz10*dz10;
258 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
259 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
260 rsq20 = dx20*dx20+dy20*dy20+dz20*dz20;
261 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
262 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
264 rinv00 = gmx_invsqrt(rsq00);
265 rinv01 = gmx_invsqrt(rsq01);
266 rinv02 = gmx_invsqrt(rsq02);
267 rinv10 = gmx_invsqrt(rsq10);
268 rinv11 = gmx_invsqrt(rsq11);
269 rinv12 = gmx_invsqrt(rsq12);
270 rinv20 = gmx_invsqrt(rsq20);
271 rinv21 = gmx_invsqrt(rsq21);
272 rinv22 = gmx_invsqrt(rsq22);
274 rinvsq00 = rinv00*rinv00;
275 rinvsq01 = rinv01*rinv01;
276 rinvsq02 = rinv02*rinv02;
277 rinvsq10 = rinv10*rinv10;
278 rinvsq11 = rinv11*rinv11;
279 rinvsq12 = rinv12*rinv12;
280 rinvsq20 = rinv20*rinv20;
281 rinvsq21 = rinv21*rinv21;
282 rinvsq22 = rinv22*rinv22;
284 /**************************
285 * CALCULATE INTERACTIONS *
286 **************************/
293 /* EWALD ELECTROSTATICS */
295 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
296 ewrt = r00*ewtabscale;
300 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
301 velec = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
302 felec = qq00*rinv00*(rinvsq00-felec);
304 /* LENNARD-JONES DISPERSION/REPULSION */
306 rinvsix = rinvsq00*rinvsq00*rinvsq00;
307 vvdw6 = c6_00*rinvsix;
308 vvdw12 = c12_00*rinvsix*rinvsix;
309 vvdw = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
310 fvdw = (vvdw12-vvdw6)*rinvsq00;
313 d = (d>0.0) ? d : 0.0;
315 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
317 dsw = d2*(swF2+d*(swF3+d*swF4));
319 /* Evaluate switch function */
320 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
321 felec = felec*sw - rinv00*velec*dsw;
322 fvdw = fvdw*sw - rinv00*vvdw*dsw;
326 /* Update potential sums from outer loop */
332 /* Calculate temporary vectorial force */
337 /* Update vectorial force */
341 f[j_coord_offset+DIM*0+XX] -= tx;
342 f[j_coord_offset+DIM*0+YY] -= ty;
343 f[j_coord_offset+DIM*0+ZZ] -= tz;
347 /**************************
348 * CALCULATE INTERACTIONS *
349 **************************/
356 /* EWALD ELECTROSTATICS */
358 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
359 ewrt = r01*ewtabscale;
363 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
364 velec = qq01*(rinv01-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
365 felec = qq01*rinv01*(rinvsq01-felec);
368 d = (d>0.0) ? d : 0.0;
370 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
372 dsw = d2*(swF2+d*(swF3+d*swF4));
374 /* Evaluate switch function */
375 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
376 felec = felec*sw - rinv01*velec*dsw;
379 /* Update potential sums from outer loop */
384 /* Calculate temporary vectorial force */
389 /* Update vectorial force */
393 f[j_coord_offset+DIM*1+XX] -= tx;
394 f[j_coord_offset+DIM*1+YY] -= ty;
395 f[j_coord_offset+DIM*1+ZZ] -= tz;
399 /**************************
400 * CALCULATE INTERACTIONS *
401 **************************/
408 /* EWALD ELECTROSTATICS */
410 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
411 ewrt = r02*ewtabscale;
415 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
416 velec = qq02*(rinv02-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
417 felec = qq02*rinv02*(rinvsq02-felec);
420 d = (d>0.0) ? d : 0.0;
422 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
424 dsw = d2*(swF2+d*(swF3+d*swF4));
426 /* Evaluate switch function */
427 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
428 felec = felec*sw - rinv02*velec*dsw;
431 /* Update potential sums from outer loop */
436 /* Calculate temporary vectorial force */
441 /* Update vectorial force */
445 f[j_coord_offset+DIM*2+XX] -= tx;
446 f[j_coord_offset+DIM*2+YY] -= ty;
447 f[j_coord_offset+DIM*2+ZZ] -= tz;
451 /**************************
452 * CALCULATE INTERACTIONS *
453 **************************/
460 /* EWALD ELECTROSTATICS */
462 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
463 ewrt = r10*ewtabscale;
467 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
468 velec = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
469 felec = qq10*rinv10*(rinvsq10-felec);
472 d = (d>0.0) ? d : 0.0;
474 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
476 dsw = d2*(swF2+d*(swF3+d*swF4));
478 /* Evaluate switch function */
479 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
480 felec = felec*sw - rinv10*velec*dsw;
483 /* Update potential sums from outer loop */
488 /* Calculate temporary vectorial force */
493 /* Update vectorial force */
497 f[j_coord_offset+DIM*0+XX] -= tx;
498 f[j_coord_offset+DIM*0+YY] -= ty;
499 f[j_coord_offset+DIM*0+ZZ] -= tz;
503 /**************************
504 * CALCULATE INTERACTIONS *
505 **************************/
512 /* EWALD ELECTROSTATICS */
514 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
515 ewrt = r11*ewtabscale;
519 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
520 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
521 felec = qq11*rinv11*(rinvsq11-felec);
524 d = (d>0.0) ? d : 0.0;
526 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
528 dsw = d2*(swF2+d*(swF3+d*swF4));
530 /* Evaluate switch function */
531 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
532 felec = felec*sw - rinv11*velec*dsw;
535 /* Update potential sums from outer loop */
540 /* Calculate temporary vectorial force */
545 /* Update vectorial force */
549 f[j_coord_offset+DIM*1+XX] -= tx;
550 f[j_coord_offset+DIM*1+YY] -= ty;
551 f[j_coord_offset+DIM*1+ZZ] -= tz;
555 /**************************
556 * CALCULATE INTERACTIONS *
557 **************************/
564 /* EWALD ELECTROSTATICS */
566 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
567 ewrt = r12*ewtabscale;
571 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
572 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
573 felec = qq12*rinv12*(rinvsq12-felec);
576 d = (d>0.0) ? d : 0.0;
578 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
580 dsw = d2*(swF2+d*(swF3+d*swF4));
582 /* Evaluate switch function */
583 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
584 felec = felec*sw - rinv12*velec*dsw;
587 /* Update potential sums from outer loop */
592 /* Calculate temporary vectorial force */
597 /* Update vectorial force */
601 f[j_coord_offset+DIM*2+XX] -= tx;
602 f[j_coord_offset+DIM*2+YY] -= ty;
603 f[j_coord_offset+DIM*2+ZZ] -= tz;
607 /**************************
608 * CALCULATE INTERACTIONS *
609 **************************/
616 /* EWALD ELECTROSTATICS */
618 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
619 ewrt = r20*ewtabscale;
623 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
624 velec = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
625 felec = qq20*rinv20*(rinvsq20-felec);
628 d = (d>0.0) ? d : 0.0;
630 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
632 dsw = d2*(swF2+d*(swF3+d*swF4));
634 /* Evaluate switch function */
635 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
636 felec = felec*sw - rinv20*velec*dsw;
639 /* Update potential sums from outer loop */
644 /* Calculate temporary vectorial force */
649 /* Update vectorial force */
653 f[j_coord_offset+DIM*0+XX] -= tx;
654 f[j_coord_offset+DIM*0+YY] -= ty;
655 f[j_coord_offset+DIM*0+ZZ] -= tz;
659 /**************************
660 * CALCULATE INTERACTIONS *
661 **************************/
668 /* EWALD ELECTROSTATICS */
670 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
671 ewrt = r21*ewtabscale;
675 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
676 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
677 felec = qq21*rinv21*(rinvsq21-felec);
680 d = (d>0.0) ? d : 0.0;
682 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
684 dsw = d2*(swF2+d*(swF3+d*swF4));
686 /* Evaluate switch function */
687 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
688 felec = felec*sw - rinv21*velec*dsw;
691 /* Update potential sums from outer loop */
696 /* Calculate temporary vectorial force */
701 /* Update vectorial force */
705 f[j_coord_offset+DIM*1+XX] -= tx;
706 f[j_coord_offset+DIM*1+YY] -= ty;
707 f[j_coord_offset+DIM*1+ZZ] -= tz;
711 /**************************
712 * CALCULATE INTERACTIONS *
713 **************************/
720 /* EWALD ELECTROSTATICS */
722 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
723 ewrt = r22*ewtabscale;
727 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
728 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
729 felec = qq22*rinv22*(rinvsq22-felec);
732 d = (d>0.0) ? d : 0.0;
734 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
736 dsw = d2*(swF2+d*(swF3+d*swF4));
738 /* Evaluate switch function */
739 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
740 felec = felec*sw - rinv22*velec*dsw;
743 /* Update potential sums from outer loop */
748 /* Calculate temporary vectorial force */
753 /* Update vectorial force */
757 f[j_coord_offset+DIM*2+XX] -= tx;
758 f[j_coord_offset+DIM*2+YY] -= ty;
759 f[j_coord_offset+DIM*2+ZZ] -= tz;
763 /* Inner loop uses 538 flops */
765 /* End of innermost loop */
768 f[i_coord_offset+DIM*0+XX] += fix0;
769 f[i_coord_offset+DIM*0+YY] += fiy0;
770 f[i_coord_offset+DIM*0+ZZ] += fiz0;
774 f[i_coord_offset+DIM*1+XX] += fix1;
775 f[i_coord_offset+DIM*1+YY] += fiy1;
776 f[i_coord_offset+DIM*1+ZZ] += fiz1;
780 f[i_coord_offset+DIM*2+XX] += fix2;
781 f[i_coord_offset+DIM*2+YY] += fiy2;
782 f[i_coord_offset+DIM*2+ZZ] += fiz2;
786 fshift[i_shift_offset+XX] += tx;
787 fshift[i_shift_offset+YY] += ty;
788 fshift[i_shift_offset+ZZ] += tz;
791 /* Update potential energies */
792 kernel_data->energygrp_elec[ggid] += velecsum;
793 kernel_data->energygrp_vdw[ggid] += vvdwsum;
795 /* Increment number of inner iterations */
796 inneriter += j_index_end - j_index_start;
798 /* Outer loop uses 32 flops */
801 /* Increment number of outer iterations */
804 /* Update outer/inner flops */
806 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*32 + inneriter*538);
809 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_F_c
810 * Electrostatics interaction: Ewald
811 * VdW interaction: LennardJones
812 * Geometry: Water3-Water3
813 * Calculate force/pot: Force
816 nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_F_c
817 (t_nblist * gmx_restrict nlist,
818 rvec * gmx_restrict xx,
819 rvec * gmx_restrict ff,
820 t_forcerec * gmx_restrict fr,
821 t_mdatoms * gmx_restrict mdatoms,
822 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
823 t_nrnb * gmx_restrict nrnb)
825 int i_shift_offset,i_coord_offset,j_coord_offset;
826 int j_index_start,j_index_end;
827 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
828 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
829 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
830 real *shiftvec,*fshift,*x,*f;
832 real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
834 real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
836 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
838 real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
840 real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
842 real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
843 real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
844 real dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01;
845 real dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02;
846 real dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
847 real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
848 real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
849 real dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
850 real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
851 real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
852 real velec,felec,velecsum,facel,crf,krf,krf2;
855 real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
859 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
861 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
868 jindex = nlist->jindex;
870 shiftidx = nlist->shift;
872 shiftvec = fr->shift_vec[0];
873 fshift = fr->fshift[0];
875 charge = mdatoms->chargeA;
876 nvdwtype = fr->ntype;
878 vdwtype = mdatoms->typeA;
880 sh_ewald = fr->ic->sh_ewald;
881 ewtab = fr->ic->tabq_coul_FDV0;
882 ewtabscale = fr->ic->tabq_scale;
883 ewtabhalfspace = 0.5/ewtabscale;
885 /* Setup water-specific parameters */
886 inr = nlist->iinr[0];
887 iq0 = facel*charge[inr+0];
888 iq1 = facel*charge[inr+1];
889 iq2 = facel*charge[inr+2];
890 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
895 vdwjidx0 = 2*vdwtype[inr+0];
897 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
898 c12_00 = vdwparam[vdwioffset0+vdwjidx0+1];
908 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
909 rcutoff = fr->rcoulomb;
910 rcutoff2 = rcutoff*rcutoff;
912 rswitch = fr->rcoulomb_switch;
913 /* Setup switch parameters */
915 swV3 = -10.0/(d*d*d);
916 swV4 = 15.0/(d*d*d*d);
917 swV5 = -6.0/(d*d*d*d*d);
918 swF2 = -30.0/(d*d*d);
919 swF3 = 60.0/(d*d*d*d);
920 swF4 = -30.0/(d*d*d*d*d);
925 /* Start outer loop over neighborlists */
926 for(iidx=0; iidx<nri; iidx++)
928 /* Load shift vector for this list */
929 i_shift_offset = DIM*shiftidx[iidx];
930 shX = shiftvec[i_shift_offset+XX];
931 shY = shiftvec[i_shift_offset+YY];
932 shZ = shiftvec[i_shift_offset+ZZ];
934 /* Load limits for loop over neighbors */
935 j_index_start = jindex[iidx];
936 j_index_end = jindex[iidx+1];
938 /* Get outer coordinate index */
940 i_coord_offset = DIM*inr;
942 /* Load i particle coords and add shift vector */
943 ix0 = shX + x[i_coord_offset+DIM*0+XX];
944 iy0 = shY + x[i_coord_offset+DIM*0+YY];
945 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
946 ix1 = shX + x[i_coord_offset+DIM*1+XX];
947 iy1 = shY + x[i_coord_offset+DIM*1+YY];
948 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
949 ix2 = shX + x[i_coord_offset+DIM*2+XX];
950 iy2 = shY + x[i_coord_offset+DIM*2+YY];
951 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
963 /* Start inner kernel loop */
964 for(jidx=j_index_start; jidx<j_index_end; jidx++)
966 /* Get j neighbor index, and coordinate index */
968 j_coord_offset = DIM*jnr;
970 /* load j atom coordinates */
971 jx0 = x[j_coord_offset+DIM*0+XX];
972 jy0 = x[j_coord_offset+DIM*0+YY];
973 jz0 = x[j_coord_offset+DIM*0+ZZ];
974 jx1 = x[j_coord_offset+DIM*1+XX];
975 jy1 = x[j_coord_offset+DIM*1+YY];
976 jz1 = x[j_coord_offset+DIM*1+ZZ];
977 jx2 = x[j_coord_offset+DIM*2+XX];
978 jy2 = x[j_coord_offset+DIM*2+YY];
979 jz2 = x[j_coord_offset+DIM*2+ZZ];
981 /* Calculate displacement vector */
1010 /* Calculate squared distance and things based on it */
1011 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
1012 rsq01 = dx01*dx01+dy01*dy01+dz01*dz01;
1013 rsq02 = dx02*dx02+dy02*dy02+dz02*dz02;
1014 rsq10 = dx10*dx10+dy10*dy10+dz10*dz10;
1015 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
1016 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
1017 rsq20 = dx20*dx20+dy20*dy20+dz20*dz20;
1018 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
1019 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
1021 rinv00 = gmx_invsqrt(rsq00);
1022 rinv01 = gmx_invsqrt(rsq01);
1023 rinv02 = gmx_invsqrt(rsq02);
1024 rinv10 = gmx_invsqrt(rsq10);
1025 rinv11 = gmx_invsqrt(rsq11);
1026 rinv12 = gmx_invsqrt(rsq12);
1027 rinv20 = gmx_invsqrt(rsq20);
1028 rinv21 = gmx_invsqrt(rsq21);
1029 rinv22 = gmx_invsqrt(rsq22);
1031 rinvsq00 = rinv00*rinv00;
1032 rinvsq01 = rinv01*rinv01;
1033 rinvsq02 = rinv02*rinv02;
1034 rinvsq10 = rinv10*rinv10;
1035 rinvsq11 = rinv11*rinv11;
1036 rinvsq12 = rinv12*rinv12;
1037 rinvsq20 = rinv20*rinv20;
1038 rinvsq21 = rinv21*rinv21;
1039 rinvsq22 = rinv22*rinv22;
1041 /**************************
1042 * CALCULATE INTERACTIONS *
1043 **************************/
1050 /* EWALD ELECTROSTATICS */
1052 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1053 ewrt = r00*ewtabscale;
1055 eweps = ewrt-ewitab;
1057 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1058 velec = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1059 felec = qq00*rinv00*(rinvsq00-felec);
1061 /* LENNARD-JONES DISPERSION/REPULSION */
1063 rinvsix = rinvsq00*rinvsq00*rinvsq00;
1064 vvdw6 = c6_00*rinvsix;
1065 vvdw12 = c12_00*rinvsix*rinvsix;
1066 vvdw = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
1067 fvdw = (vvdw12-vvdw6)*rinvsq00;
1070 d = (d>0.0) ? d : 0.0;
1072 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1074 dsw = d2*(swF2+d*(swF3+d*swF4));
1076 /* Evaluate switch function */
1077 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1078 felec = felec*sw - rinv00*velec*dsw;
1079 fvdw = fvdw*sw - rinv00*vvdw*dsw;
1083 /* Calculate temporary vectorial force */
1088 /* Update vectorial force */
1092 f[j_coord_offset+DIM*0+XX] -= tx;
1093 f[j_coord_offset+DIM*0+YY] -= ty;
1094 f[j_coord_offset+DIM*0+ZZ] -= tz;
1098 /**************************
1099 * CALCULATE INTERACTIONS *
1100 **************************/
1107 /* EWALD ELECTROSTATICS */
1109 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1110 ewrt = r01*ewtabscale;
1112 eweps = ewrt-ewitab;
1114 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1115 velec = qq01*(rinv01-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1116 felec = qq01*rinv01*(rinvsq01-felec);
1119 d = (d>0.0) ? d : 0.0;
1121 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1123 dsw = d2*(swF2+d*(swF3+d*swF4));
1125 /* Evaluate switch function */
1126 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1127 felec = felec*sw - rinv01*velec*dsw;
1131 /* Calculate temporary vectorial force */
1136 /* Update vectorial force */
1140 f[j_coord_offset+DIM*1+XX] -= tx;
1141 f[j_coord_offset+DIM*1+YY] -= ty;
1142 f[j_coord_offset+DIM*1+ZZ] -= tz;
1146 /**************************
1147 * CALCULATE INTERACTIONS *
1148 **************************/
1155 /* EWALD ELECTROSTATICS */
1157 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1158 ewrt = r02*ewtabscale;
1160 eweps = ewrt-ewitab;
1162 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1163 velec = qq02*(rinv02-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1164 felec = qq02*rinv02*(rinvsq02-felec);
1167 d = (d>0.0) ? d : 0.0;
1169 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1171 dsw = d2*(swF2+d*(swF3+d*swF4));
1173 /* Evaluate switch function */
1174 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1175 felec = felec*sw - rinv02*velec*dsw;
1179 /* Calculate temporary vectorial force */
1184 /* Update vectorial force */
1188 f[j_coord_offset+DIM*2+XX] -= tx;
1189 f[j_coord_offset+DIM*2+YY] -= ty;
1190 f[j_coord_offset+DIM*2+ZZ] -= tz;
1194 /**************************
1195 * CALCULATE INTERACTIONS *
1196 **************************/
1203 /* EWALD ELECTROSTATICS */
1205 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1206 ewrt = r10*ewtabscale;
1208 eweps = ewrt-ewitab;
1210 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1211 velec = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1212 felec = qq10*rinv10*(rinvsq10-felec);
1215 d = (d>0.0) ? d : 0.0;
1217 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1219 dsw = d2*(swF2+d*(swF3+d*swF4));
1221 /* Evaluate switch function */
1222 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1223 felec = felec*sw - rinv10*velec*dsw;
1227 /* Calculate temporary vectorial force */
1232 /* Update vectorial force */
1236 f[j_coord_offset+DIM*0+XX] -= tx;
1237 f[j_coord_offset+DIM*0+YY] -= ty;
1238 f[j_coord_offset+DIM*0+ZZ] -= tz;
1242 /**************************
1243 * CALCULATE INTERACTIONS *
1244 **************************/
1251 /* EWALD ELECTROSTATICS */
1253 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1254 ewrt = r11*ewtabscale;
1256 eweps = ewrt-ewitab;
1258 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1259 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1260 felec = qq11*rinv11*(rinvsq11-felec);
1263 d = (d>0.0) ? d : 0.0;
1265 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1267 dsw = d2*(swF2+d*(swF3+d*swF4));
1269 /* Evaluate switch function */
1270 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1271 felec = felec*sw - rinv11*velec*dsw;
1275 /* Calculate temporary vectorial force */
1280 /* Update vectorial force */
1284 f[j_coord_offset+DIM*1+XX] -= tx;
1285 f[j_coord_offset+DIM*1+YY] -= ty;
1286 f[j_coord_offset+DIM*1+ZZ] -= tz;
1290 /**************************
1291 * CALCULATE INTERACTIONS *
1292 **************************/
1299 /* EWALD ELECTROSTATICS */
1301 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1302 ewrt = r12*ewtabscale;
1304 eweps = ewrt-ewitab;
1306 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1307 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1308 felec = qq12*rinv12*(rinvsq12-felec);
1311 d = (d>0.0) ? d : 0.0;
1313 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1315 dsw = d2*(swF2+d*(swF3+d*swF4));
1317 /* Evaluate switch function */
1318 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1319 felec = felec*sw - rinv12*velec*dsw;
1323 /* Calculate temporary vectorial force */
1328 /* Update vectorial force */
1332 f[j_coord_offset+DIM*2+XX] -= tx;
1333 f[j_coord_offset+DIM*2+YY] -= ty;
1334 f[j_coord_offset+DIM*2+ZZ] -= tz;
1338 /**************************
1339 * CALCULATE INTERACTIONS *
1340 **************************/
1347 /* EWALD ELECTROSTATICS */
1349 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1350 ewrt = r20*ewtabscale;
1352 eweps = ewrt-ewitab;
1354 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1355 velec = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1356 felec = qq20*rinv20*(rinvsq20-felec);
1359 d = (d>0.0) ? d : 0.0;
1361 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1363 dsw = d2*(swF2+d*(swF3+d*swF4));
1365 /* Evaluate switch function */
1366 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1367 felec = felec*sw - rinv20*velec*dsw;
1371 /* Calculate temporary vectorial force */
1376 /* Update vectorial force */
1380 f[j_coord_offset+DIM*0+XX] -= tx;
1381 f[j_coord_offset+DIM*0+YY] -= ty;
1382 f[j_coord_offset+DIM*0+ZZ] -= tz;
1386 /**************************
1387 * CALCULATE INTERACTIONS *
1388 **************************/
1395 /* EWALD ELECTROSTATICS */
1397 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1398 ewrt = r21*ewtabscale;
1400 eweps = ewrt-ewitab;
1402 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1403 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1404 felec = qq21*rinv21*(rinvsq21-felec);
1407 d = (d>0.0) ? d : 0.0;
1409 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1411 dsw = d2*(swF2+d*(swF3+d*swF4));
1413 /* Evaluate switch function */
1414 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1415 felec = felec*sw - rinv21*velec*dsw;
1419 /* Calculate temporary vectorial force */
1424 /* Update vectorial force */
1428 f[j_coord_offset+DIM*1+XX] -= tx;
1429 f[j_coord_offset+DIM*1+YY] -= ty;
1430 f[j_coord_offset+DIM*1+ZZ] -= tz;
1434 /**************************
1435 * CALCULATE INTERACTIONS *
1436 **************************/
1443 /* EWALD ELECTROSTATICS */
1445 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1446 ewrt = r22*ewtabscale;
1448 eweps = ewrt-ewitab;
1450 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1451 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1452 felec = qq22*rinv22*(rinvsq22-felec);
1455 d = (d>0.0) ? d : 0.0;
1457 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1459 dsw = d2*(swF2+d*(swF3+d*swF4));
1461 /* Evaluate switch function */
1462 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1463 felec = felec*sw - rinv22*velec*dsw;
1467 /* Calculate temporary vectorial force */
1472 /* Update vectorial force */
1476 f[j_coord_offset+DIM*2+XX] -= tx;
1477 f[j_coord_offset+DIM*2+YY] -= ty;
1478 f[j_coord_offset+DIM*2+ZZ] -= tz;
1482 /* Inner loop uses 518 flops */
1484 /* End of innermost loop */
1487 f[i_coord_offset+DIM*0+XX] += fix0;
1488 f[i_coord_offset+DIM*0+YY] += fiy0;
1489 f[i_coord_offset+DIM*0+ZZ] += fiz0;
1493 f[i_coord_offset+DIM*1+XX] += fix1;
1494 f[i_coord_offset+DIM*1+YY] += fiy1;
1495 f[i_coord_offset+DIM*1+ZZ] += fiz1;
1499 f[i_coord_offset+DIM*2+XX] += fix2;
1500 f[i_coord_offset+DIM*2+YY] += fiy2;
1501 f[i_coord_offset+DIM*2+ZZ] += fiz2;
1505 fshift[i_shift_offset+XX] += tx;
1506 fshift[i_shift_offset+YY] += ty;
1507 fshift[i_shift_offset+ZZ] += tz;
1509 /* Increment number of inner iterations */
1510 inneriter += j_index_end - j_index_start;
1512 /* Outer loop uses 30 flops */
1515 /* Increment number of outer iterations */
1518 /* Update outer/inner flops */
1520 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*30 + inneriter*518);