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 "gromacs/legacyheaders/types/simple.h"
44 #include "gromacs/math/vec.h"
45 #include "gromacs/legacyheaders/nrnb.h"
48 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwBhamSw_GeomW3W3_VF_c
49 * Electrostatics interaction: Ewald
50 * VdW interaction: Buckingham
51 * Geometry: Water3-Water3
52 * Calculate force/pot: PotentialAndForce
55 nb_kernel_ElecEwSw_VdwBhamSw_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 = 3*nvdwtype*vdwtype[inr+0];
134 vdwjidx0 = 3*vdwtype[inr+0];
136 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
137 cexp1_00 = vdwparam[vdwioffset0+vdwjidx0+1];
138 cexp2_00 = vdwparam[vdwioffset0+vdwjidx0+2];
148 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
149 rcutoff = fr->rcoulomb;
150 rcutoff2 = rcutoff*rcutoff;
152 rswitch = fr->rcoulomb_switch;
153 /* Setup switch parameters */
155 swV3 = -10.0/(d*d*d);
156 swV4 = 15.0/(d*d*d*d);
157 swV5 = -6.0/(d*d*d*d*d);
158 swF2 = -30.0/(d*d*d);
159 swF3 = 60.0/(d*d*d*d);
160 swF4 = -30.0/(d*d*d*d*d);
165 /* Start outer loop over neighborlists */
166 for(iidx=0; iidx<nri; iidx++)
168 /* Load shift vector for this list */
169 i_shift_offset = DIM*shiftidx[iidx];
170 shX = shiftvec[i_shift_offset+XX];
171 shY = shiftvec[i_shift_offset+YY];
172 shZ = shiftvec[i_shift_offset+ZZ];
174 /* Load limits for loop over neighbors */
175 j_index_start = jindex[iidx];
176 j_index_end = jindex[iidx+1];
178 /* Get outer coordinate index */
180 i_coord_offset = DIM*inr;
182 /* Load i particle coords and add shift vector */
183 ix0 = shX + x[i_coord_offset+DIM*0+XX];
184 iy0 = shY + x[i_coord_offset+DIM*0+YY];
185 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
186 ix1 = shX + x[i_coord_offset+DIM*1+XX];
187 iy1 = shY + x[i_coord_offset+DIM*1+YY];
188 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
189 ix2 = shX + x[i_coord_offset+DIM*2+XX];
190 iy2 = shY + x[i_coord_offset+DIM*2+YY];
191 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
203 /* Reset potential sums */
207 /* Start inner kernel loop */
208 for(jidx=j_index_start; jidx<j_index_end; jidx++)
210 /* Get j neighbor index, and coordinate index */
212 j_coord_offset = DIM*jnr;
214 /* load j atom coordinates */
215 jx0 = x[j_coord_offset+DIM*0+XX];
216 jy0 = x[j_coord_offset+DIM*0+YY];
217 jz0 = x[j_coord_offset+DIM*0+ZZ];
218 jx1 = x[j_coord_offset+DIM*1+XX];
219 jy1 = x[j_coord_offset+DIM*1+YY];
220 jz1 = x[j_coord_offset+DIM*1+ZZ];
221 jx2 = x[j_coord_offset+DIM*2+XX];
222 jy2 = x[j_coord_offset+DIM*2+YY];
223 jz2 = x[j_coord_offset+DIM*2+ZZ];
225 /* Calculate displacement vector */
254 /* Calculate squared distance and things based on it */
255 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
256 rsq01 = dx01*dx01+dy01*dy01+dz01*dz01;
257 rsq02 = dx02*dx02+dy02*dy02+dz02*dz02;
258 rsq10 = dx10*dx10+dy10*dy10+dz10*dz10;
259 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
260 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
261 rsq20 = dx20*dx20+dy20*dy20+dz20*dz20;
262 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
263 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
265 rinv00 = gmx_invsqrt(rsq00);
266 rinv01 = gmx_invsqrt(rsq01);
267 rinv02 = gmx_invsqrt(rsq02);
268 rinv10 = gmx_invsqrt(rsq10);
269 rinv11 = gmx_invsqrt(rsq11);
270 rinv12 = gmx_invsqrt(rsq12);
271 rinv20 = gmx_invsqrt(rsq20);
272 rinv21 = gmx_invsqrt(rsq21);
273 rinv22 = gmx_invsqrt(rsq22);
275 rinvsq00 = rinv00*rinv00;
276 rinvsq01 = rinv01*rinv01;
277 rinvsq02 = rinv02*rinv02;
278 rinvsq10 = rinv10*rinv10;
279 rinvsq11 = rinv11*rinv11;
280 rinvsq12 = rinv12*rinv12;
281 rinvsq20 = rinv20*rinv20;
282 rinvsq21 = rinv21*rinv21;
283 rinvsq22 = rinv22*rinv22;
285 /**************************
286 * CALCULATE INTERACTIONS *
287 **************************/
294 /* EWALD ELECTROSTATICS */
296 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
297 ewrt = r00*ewtabscale;
301 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
302 velec = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
303 felec = qq00*rinv00*(rinvsq00-felec);
305 /* BUCKINGHAM DISPERSION/REPULSION */
306 rinvsix = rinvsq00*rinvsq00*rinvsq00;
307 vvdw6 = c6_00*rinvsix;
309 vvdwexp = cexp1_00*exp(-br);
310 vvdw = vvdwexp - vvdw6*(1.0/6.0);
311 fvdw = (br*vvdwexp-vvdw6)*rinvsq00;
314 d = (d>0.0) ? d : 0.0;
316 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
318 dsw = d2*(swF2+d*(swF3+d*swF4));
320 /* Evaluate switch function */
321 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
322 felec = felec*sw - rinv00*velec*dsw;
323 fvdw = fvdw*sw - rinv00*vvdw*dsw;
327 /* Update potential sums from outer loop */
333 /* Calculate temporary vectorial force */
338 /* Update vectorial force */
342 f[j_coord_offset+DIM*0+XX] -= tx;
343 f[j_coord_offset+DIM*0+YY] -= ty;
344 f[j_coord_offset+DIM*0+ZZ] -= tz;
348 /**************************
349 * CALCULATE INTERACTIONS *
350 **************************/
357 /* EWALD ELECTROSTATICS */
359 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
360 ewrt = r01*ewtabscale;
364 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
365 velec = qq01*(rinv01-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
366 felec = qq01*rinv01*(rinvsq01-felec);
369 d = (d>0.0) ? d : 0.0;
371 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
373 dsw = d2*(swF2+d*(swF3+d*swF4));
375 /* Evaluate switch function */
376 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
377 felec = felec*sw - rinv01*velec*dsw;
380 /* Update potential sums from outer loop */
385 /* Calculate temporary vectorial force */
390 /* Update vectorial force */
394 f[j_coord_offset+DIM*1+XX] -= tx;
395 f[j_coord_offset+DIM*1+YY] -= ty;
396 f[j_coord_offset+DIM*1+ZZ] -= tz;
400 /**************************
401 * CALCULATE INTERACTIONS *
402 **************************/
409 /* EWALD ELECTROSTATICS */
411 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
412 ewrt = r02*ewtabscale;
416 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
417 velec = qq02*(rinv02-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
418 felec = qq02*rinv02*(rinvsq02-felec);
421 d = (d>0.0) ? d : 0.0;
423 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
425 dsw = d2*(swF2+d*(swF3+d*swF4));
427 /* Evaluate switch function */
428 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
429 felec = felec*sw - rinv02*velec*dsw;
432 /* Update potential sums from outer loop */
437 /* Calculate temporary vectorial force */
442 /* Update vectorial force */
446 f[j_coord_offset+DIM*2+XX] -= tx;
447 f[j_coord_offset+DIM*2+YY] -= ty;
448 f[j_coord_offset+DIM*2+ZZ] -= tz;
452 /**************************
453 * CALCULATE INTERACTIONS *
454 **************************/
461 /* EWALD ELECTROSTATICS */
463 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
464 ewrt = r10*ewtabscale;
468 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
469 velec = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
470 felec = qq10*rinv10*(rinvsq10-felec);
473 d = (d>0.0) ? d : 0.0;
475 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
477 dsw = d2*(swF2+d*(swF3+d*swF4));
479 /* Evaluate switch function */
480 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
481 felec = felec*sw - rinv10*velec*dsw;
484 /* Update potential sums from outer loop */
489 /* Calculate temporary vectorial force */
494 /* Update vectorial force */
498 f[j_coord_offset+DIM*0+XX] -= tx;
499 f[j_coord_offset+DIM*0+YY] -= ty;
500 f[j_coord_offset+DIM*0+ZZ] -= tz;
504 /**************************
505 * CALCULATE INTERACTIONS *
506 **************************/
513 /* EWALD ELECTROSTATICS */
515 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
516 ewrt = r11*ewtabscale;
520 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
521 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
522 felec = qq11*rinv11*(rinvsq11-felec);
525 d = (d>0.0) ? d : 0.0;
527 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
529 dsw = d2*(swF2+d*(swF3+d*swF4));
531 /* Evaluate switch function */
532 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
533 felec = felec*sw - rinv11*velec*dsw;
536 /* Update potential sums from outer loop */
541 /* Calculate temporary vectorial force */
546 /* Update vectorial force */
550 f[j_coord_offset+DIM*1+XX] -= tx;
551 f[j_coord_offset+DIM*1+YY] -= ty;
552 f[j_coord_offset+DIM*1+ZZ] -= tz;
556 /**************************
557 * CALCULATE INTERACTIONS *
558 **************************/
565 /* EWALD ELECTROSTATICS */
567 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
568 ewrt = r12*ewtabscale;
572 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
573 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
574 felec = qq12*rinv12*(rinvsq12-felec);
577 d = (d>0.0) ? d : 0.0;
579 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
581 dsw = d2*(swF2+d*(swF3+d*swF4));
583 /* Evaluate switch function */
584 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
585 felec = felec*sw - rinv12*velec*dsw;
588 /* Update potential sums from outer loop */
593 /* Calculate temporary vectorial force */
598 /* Update vectorial force */
602 f[j_coord_offset+DIM*2+XX] -= tx;
603 f[j_coord_offset+DIM*2+YY] -= ty;
604 f[j_coord_offset+DIM*2+ZZ] -= tz;
608 /**************************
609 * CALCULATE INTERACTIONS *
610 **************************/
617 /* EWALD ELECTROSTATICS */
619 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
620 ewrt = r20*ewtabscale;
624 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
625 velec = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
626 felec = qq20*rinv20*(rinvsq20-felec);
629 d = (d>0.0) ? d : 0.0;
631 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
633 dsw = d2*(swF2+d*(swF3+d*swF4));
635 /* Evaluate switch function */
636 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
637 felec = felec*sw - rinv20*velec*dsw;
640 /* Update potential sums from outer loop */
645 /* Calculate temporary vectorial force */
650 /* Update vectorial force */
654 f[j_coord_offset+DIM*0+XX] -= tx;
655 f[j_coord_offset+DIM*0+YY] -= ty;
656 f[j_coord_offset+DIM*0+ZZ] -= tz;
660 /**************************
661 * CALCULATE INTERACTIONS *
662 **************************/
669 /* EWALD ELECTROSTATICS */
671 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
672 ewrt = r21*ewtabscale;
676 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
677 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
678 felec = qq21*rinv21*(rinvsq21-felec);
681 d = (d>0.0) ? d : 0.0;
683 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
685 dsw = d2*(swF2+d*(swF3+d*swF4));
687 /* Evaluate switch function */
688 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
689 felec = felec*sw - rinv21*velec*dsw;
692 /* Update potential sums from outer loop */
697 /* Calculate temporary vectorial force */
702 /* Update vectorial force */
706 f[j_coord_offset+DIM*1+XX] -= tx;
707 f[j_coord_offset+DIM*1+YY] -= ty;
708 f[j_coord_offset+DIM*1+ZZ] -= tz;
712 /**************************
713 * CALCULATE INTERACTIONS *
714 **************************/
721 /* EWALD ELECTROSTATICS */
723 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
724 ewrt = r22*ewtabscale;
728 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
729 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
730 felec = qq22*rinv22*(rinvsq22-felec);
733 d = (d>0.0) ? d : 0.0;
735 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
737 dsw = d2*(swF2+d*(swF3+d*swF4));
739 /* Evaluate switch function */
740 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
741 felec = felec*sw - rinv22*velec*dsw;
744 /* Update potential sums from outer loop */
749 /* Calculate temporary vectorial force */
754 /* Update vectorial force */
758 f[j_coord_offset+DIM*2+XX] -= tx;
759 f[j_coord_offset+DIM*2+YY] -= ty;
760 f[j_coord_offset+DIM*2+ZZ] -= tz;
764 /* Inner loop uses 564 flops */
766 /* End of innermost loop */
769 f[i_coord_offset+DIM*0+XX] += fix0;
770 f[i_coord_offset+DIM*0+YY] += fiy0;
771 f[i_coord_offset+DIM*0+ZZ] += fiz0;
775 f[i_coord_offset+DIM*1+XX] += fix1;
776 f[i_coord_offset+DIM*1+YY] += fiy1;
777 f[i_coord_offset+DIM*1+ZZ] += fiz1;
781 f[i_coord_offset+DIM*2+XX] += fix2;
782 f[i_coord_offset+DIM*2+YY] += fiy2;
783 f[i_coord_offset+DIM*2+ZZ] += fiz2;
787 fshift[i_shift_offset+XX] += tx;
788 fshift[i_shift_offset+YY] += ty;
789 fshift[i_shift_offset+ZZ] += tz;
792 /* Update potential energies */
793 kernel_data->energygrp_elec[ggid] += velecsum;
794 kernel_data->energygrp_vdw[ggid] += vvdwsum;
796 /* Increment number of inner iterations */
797 inneriter += j_index_end - j_index_start;
799 /* Outer loop uses 32 flops */
802 /* Increment number of outer iterations */
805 /* Update outer/inner flops */
807 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*32 + inneriter*564);
810 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwBhamSw_GeomW3W3_F_c
811 * Electrostatics interaction: Ewald
812 * VdW interaction: Buckingham
813 * Geometry: Water3-Water3
814 * Calculate force/pot: Force
817 nb_kernel_ElecEwSw_VdwBhamSw_GeomW3W3_F_c
818 (t_nblist * gmx_restrict nlist,
819 rvec * gmx_restrict xx,
820 rvec * gmx_restrict ff,
821 t_forcerec * gmx_restrict fr,
822 t_mdatoms * gmx_restrict mdatoms,
823 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
824 t_nrnb * gmx_restrict nrnb)
826 int i_shift_offset,i_coord_offset,j_coord_offset;
827 int j_index_start,j_index_end;
828 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
829 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
830 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
831 real *shiftvec,*fshift,*x,*f;
833 real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
835 real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
837 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
839 real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
841 real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
843 real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
844 real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
845 real dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01;
846 real dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02;
847 real dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
848 real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
849 real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
850 real dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
851 real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
852 real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
853 real velec,felec,velecsum,facel,crf,krf,krf2;
856 real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
860 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
862 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
869 jindex = nlist->jindex;
871 shiftidx = nlist->shift;
873 shiftvec = fr->shift_vec[0];
874 fshift = fr->fshift[0];
876 charge = mdatoms->chargeA;
877 nvdwtype = fr->ntype;
879 vdwtype = mdatoms->typeA;
881 sh_ewald = fr->ic->sh_ewald;
882 ewtab = fr->ic->tabq_coul_FDV0;
883 ewtabscale = fr->ic->tabq_scale;
884 ewtabhalfspace = 0.5/ewtabscale;
886 /* Setup water-specific parameters */
887 inr = nlist->iinr[0];
888 iq0 = facel*charge[inr+0];
889 iq1 = facel*charge[inr+1];
890 iq2 = facel*charge[inr+2];
891 vdwioffset0 = 3*nvdwtype*vdwtype[inr+0];
896 vdwjidx0 = 3*vdwtype[inr+0];
898 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
899 cexp1_00 = vdwparam[vdwioffset0+vdwjidx0+1];
900 cexp2_00 = vdwparam[vdwioffset0+vdwjidx0+2];
910 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
911 rcutoff = fr->rcoulomb;
912 rcutoff2 = rcutoff*rcutoff;
914 rswitch = fr->rcoulomb_switch;
915 /* Setup switch parameters */
917 swV3 = -10.0/(d*d*d);
918 swV4 = 15.0/(d*d*d*d);
919 swV5 = -6.0/(d*d*d*d*d);
920 swF2 = -30.0/(d*d*d);
921 swF3 = 60.0/(d*d*d*d);
922 swF4 = -30.0/(d*d*d*d*d);
927 /* Start outer loop over neighborlists */
928 for(iidx=0; iidx<nri; iidx++)
930 /* Load shift vector for this list */
931 i_shift_offset = DIM*shiftidx[iidx];
932 shX = shiftvec[i_shift_offset+XX];
933 shY = shiftvec[i_shift_offset+YY];
934 shZ = shiftvec[i_shift_offset+ZZ];
936 /* Load limits for loop over neighbors */
937 j_index_start = jindex[iidx];
938 j_index_end = jindex[iidx+1];
940 /* Get outer coordinate index */
942 i_coord_offset = DIM*inr;
944 /* Load i particle coords and add shift vector */
945 ix0 = shX + x[i_coord_offset+DIM*0+XX];
946 iy0 = shY + x[i_coord_offset+DIM*0+YY];
947 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
948 ix1 = shX + x[i_coord_offset+DIM*1+XX];
949 iy1 = shY + x[i_coord_offset+DIM*1+YY];
950 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
951 ix2 = shX + x[i_coord_offset+DIM*2+XX];
952 iy2 = shY + x[i_coord_offset+DIM*2+YY];
953 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
965 /* Start inner kernel loop */
966 for(jidx=j_index_start; jidx<j_index_end; jidx++)
968 /* Get j neighbor index, and coordinate index */
970 j_coord_offset = DIM*jnr;
972 /* load j atom coordinates */
973 jx0 = x[j_coord_offset+DIM*0+XX];
974 jy0 = x[j_coord_offset+DIM*0+YY];
975 jz0 = x[j_coord_offset+DIM*0+ZZ];
976 jx1 = x[j_coord_offset+DIM*1+XX];
977 jy1 = x[j_coord_offset+DIM*1+YY];
978 jz1 = x[j_coord_offset+DIM*1+ZZ];
979 jx2 = x[j_coord_offset+DIM*2+XX];
980 jy2 = x[j_coord_offset+DIM*2+YY];
981 jz2 = x[j_coord_offset+DIM*2+ZZ];
983 /* Calculate displacement vector */
1012 /* Calculate squared distance and things based on it */
1013 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
1014 rsq01 = dx01*dx01+dy01*dy01+dz01*dz01;
1015 rsq02 = dx02*dx02+dy02*dy02+dz02*dz02;
1016 rsq10 = dx10*dx10+dy10*dy10+dz10*dz10;
1017 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
1018 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
1019 rsq20 = dx20*dx20+dy20*dy20+dz20*dz20;
1020 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
1021 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
1023 rinv00 = gmx_invsqrt(rsq00);
1024 rinv01 = gmx_invsqrt(rsq01);
1025 rinv02 = gmx_invsqrt(rsq02);
1026 rinv10 = gmx_invsqrt(rsq10);
1027 rinv11 = gmx_invsqrt(rsq11);
1028 rinv12 = gmx_invsqrt(rsq12);
1029 rinv20 = gmx_invsqrt(rsq20);
1030 rinv21 = gmx_invsqrt(rsq21);
1031 rinv22 = gmx_invsqrt(rsq22);
1033 rinvsq00 = rinv00*rinv00;
1034 rinvsq01 = rinv01*rinv01;
1035 rinvsq02 = rinv02*rinv02;
1036 rinvsq10 = rinv10*rinv10;
1037 rinvsq11 = rinv11*rinv11;
1038 rinvsq12 = rinv12*rinv12;
1039 rinvsq20 = rinv20*rinv20;
1040 rinvsq21 = rinv21*rinv21;
1041 rinvsq22 = rinv22*rinv22;
1043 /**************************
1044 * CALCULATE INTERACTIONS *
1045 **************************/
1052 /* EWALD ELECTROSTATICS */
1054 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1055 ewrt = r00*ewtabscale;
1057 eweps = ewrt-ewitab;
1059 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1060 velec = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1061 felec = qq00*rinv00*(rinvsq00-felec);
1063 /* BUCKINGHAM DISPERSION/REPULSION */
1064 rinvsix = rinvsq00*rinvsq00*rinvsq00;
1065 vvdw6 = c6_00*rinvsix;
1067 vvdwexp = cexp1_00*exp(-br);
1068 vvdw = vvdwexp - vvdw6*(1.0/6.0);
1069 fvdw = (br*vvdwexp-vvdw6)*rinvsq00;
1072 d = (d>0.0) ? d : 0.0;
1074 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1076 dsw = d2*(swF2+d*(swF3+d*swF4));
1078 /* Evaluate switch function */
1079 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1080 felec = felec*sw - rinv00*velec*dsw;
1081 fvdw = fvdw*sw - rinv00*vvdw*dsw;
1085 /* Calculate temporary vectorial force */
1090 /* Update vectorial force */
1094 f[j_coord_offset+DIM*0+XX] -= tx;
1095 f[j_coord_offset+DIM*0+YY] -= ty;
1096 f[j_coord_offset+DIM*0+ZZ] -= tz;
1100 /**************************
1101 * CALCULATE INTERACTIONS *
1102 **************************/
1109 /* EWALD ELECTROSTATICS */
1111 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1112 ewrt = r01*ewtabscale;
1114 eweps = ewrt-ewitab;
1116 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1117 velec = qq01*(rinv01-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1118 felec = qq01*rinv01*(rinvsq01-felec);
1121 d = (d>0.0) ? d : 0.0;
1123 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1125 dsw = d2*(swF2+d*(swF3+d*swF4));
1127 /* Evaluate switch function */
1128 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1129 felec = felec*sw - rinv01*velec*dsw;
1133 /* Calculate temporary vectorial force */
1138 /* Update vectorial force */
1142 f[j_coord_offset+DIM*1+XX] -= tx;
1143 f[j_coord_offset+DIM*1+YY] -= ty;
1144 f[j_coord_offset+DIM*1+ZZ] -= tz;
1148 /**************************
1149 * CALCULATE INTERACTIONS *
1150 **************************/
1157 /* EWALD ELECTROSTATICS */
1159 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1160 ewrt = r02*ewtabscale;
1162 eweps = ewrt-ewitab;
1164 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1165 velec = qq02*(rinv02-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1166 felec = qq02*rinv02*(rinvsq02-felec);
1169 d = (d>0.0) ? d : 0.0;
1171 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1173 dsw = d2*(swF2+d*(swF3+d*swF4));
1175 /* Evaluate switch function */
1176 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1177 felec = felec*sw - rinv02*velec*dsw;
1181 /* Calculate temporary vectorial force */
1186 /* Update vectorial force */
1190 f[j_coord_offset+DIM*2+XX] -= tx;
1191 f[j_coord_offset+DIM*2+YY] -= ty;
1192 f[j_coord_offset+DIM*2+ZZ] -= tz;
1196 /**************************
1197 * CALCULATE INTERACTIONS *
1198 **************************/
1205 /* EWALD ELECTROSTATICS */
1207 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1208 ewrt = r10*ewtabscale;
1210 eweps = ewrt-ewitab;
1212 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1213 velec = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1214 felec = qq10*rinv10*(rinvsq10-felec);
1217 d = (d>0.0) ? d : 0.0;
1219 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1221 dsw = d2*(swF2+d*(swF3+d*swF4));
1223 /* Evaluate switch function */
1224 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1225 felec = felec*sw - rinv10*velec*dsw;
1229 /* Calculate temporary vectorial force */
1234 /* Update vectorial force */
1238 f[j_coord_offset+DIM*0+XX] -= tx;
1239 f[j_coord_offset+DIM*0+YY] -= ty;
1240 f[j_coord_offset+DIM*0+ZZ] -= tz;
1244 /**************************
1245 * CALCULATE INTERACTIONS *
1246 **************************/
1253 /* EWALD ELECTROSTATICS */
1255 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1256 ewrt = r11*ewtabscale;
1258 eweps = ewrt-ewitab;
1260 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1261 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1262 felec = qq11*rinv11*(rinvsq11-felec);
1265 d = (d>0.0) ? d : 0.0;
1267 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1269 dsw = d2*(swF2+d*(swF3+d*swF4));
1271 /* Evaluate switch function */
1272 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1273 felec = felec*sw - rinv11*velec*dsw;
1277 /* Calculate temporary vectorial force */
1282 /* Update vectorial force */
1286 f[j_coord_offset+DIM*1+XX] -= tx;
1287 f[j_coord_offset+DIM*1+YY] -= ty;
1288 f[j_coord_offset+DIM*1+ZZ] -= tz;
1292 /**************************
1293 * CALCULATE INTERACTIONS *
1294 **************************/
1301 /* EWALD ELECTROSTATICS */
1303 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1304 ewrt = r12*ewtabscale;
1306 eweps = ewrt-ewitab;
1308 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1309 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1310 felec = qq12*rinv12*(rinvsq12-felec);
1313 d = (d>0.0) ? d : 0.0;
1315 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1317 dsw = d2*(swF2+d*(swF3+d*swF4));
1319 /* Evaluate switch function */
1320 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1321 felec = felec*sw - rinv12*velec*dsw;
1325 /* Calculate temporary vectorial force */
1330 /* Update vectorial force */
1334 f[j_coord_offset+DIM*2+XX] -= tx;
1335 f[j_coord_offset+DIM*2+YY] -= ty;
1336 f[j_coord_offset+DIM*2+ZZ] -= tz;
1340 /**************************
1341 * CALCULATE INTERACTIONS *
1342 **************************/
1349 /* EWALD ELECTROSTATICS */
1351 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1352 ewrt = r20*ewtabscale;
1354 eweps = ewrt-ewitab;
1356 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1357 velec = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1358 felec = qq20*rinv20*(rinvsq20-felec);
1361 d = (d>0.0) ? d : 0.0;
1363 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1365 dsw = d2*(swF2+d*(swF3+d*swF4));
1367 /* Evaluate switch function */
1368 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1369 felec = felec*sw - rinv20*velec*dsw;
1373 /* Calculate temporary vectorial force */
1378 /* Update vectorial force */
1382 f[j_coord_offset+DIM*0+XX] -= tx;
1383 f[j_coord_offset+DIM*0+YY] -= ty;
1384 f[j_coord_offset+DIM*0+ZZ] -= tz;
1388 /**************************
1389 * CALCULATE INTERACTIONS *
1390 **************************/
1397 /* EWALD ELECTROSTATICS */
1399 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1400 ewrt = r21*ewtabscale;
1402 eweps = ewrt-ewitab;
1404 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1405 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1406 felec = qq21*rinv21*(rinvsq21-felec);
1409 d = (d>0.0) ? d : 0.0;
1411 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1413 dsw = d2*(swF2+d*(swF3+d*swF4));
1415 /* Evaluate switch function */
1416 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1417 felec = felec*sw - rinv21*velec*dsw;
1421 /* Calculate temporary vectorial force */
1426 /* Update vectorial force */
1430 f[j_coord_offset+DIM*1+XX] -= tx;
1431 f[j_coord_offset+DIM*1+YY] -= ty;
1432 f[j_coord_offset+DIM*1+ZZ] -= tz;
1436 /**************************
1437 * CALCULATE INTERACTIONS *
1438 **************************/
1445 /* EWALD ELECTROSTATICS */
1447 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1448 ewrt = r22*ewtabscale;
1450 eweps = ewrt-ewitab;
1452 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1453 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1454 felec = qq22*rinv22*(rinvsq22-felec);
1457 d = (d>0.0) ? d : 0.0;
1459 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1461 dsw = d2*(swF2+d*(swF3+d*swF4));
1463 /* Evaluate switch function */
1464 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1465 felec = felec*sw - rinv22*velec*dsw;
1469 /* Calculate temporary vectorial force */
1474 /* Update vectorial force */
1478 f[j_coord_offset+DIM*2+XX] -= tx;
1479 f[j_coord_offset+DIM*2+YY] -= ty;
1480 f[j_coord_offset+DIM*2+ZZ] -= tz;
1484 /* Inner loop uses 544 flops */
1486 /* End of innermost loop */
1489 f[i_coord_offset+DIM*0+XX] += fix0;
1490 f[i_coord_offset+DIM*0+YY] += fiy0;
1491 f[i_coord_offset+DIM*0+ZZ] += fiz0;
1495 f[i_coord_offset+DIM*1+XX] += fix1;
1496 f[i_coord_offset+DIM*1+YY] += fiy1;
1497 f[i_coord_offset+DIM*1+ZZ] += fiz1;
1501 f[i_coord_offset+DIM*2+XX] += fix2;
1502 f[i_coord_offset+DIM*2+YY] += fiy2;
1503 f[i_coord_offset+DIM*2+ZZ] += fiz2;
1507 fshift[i_shift_offset+XX] += tx;
1508 fshift[i_shift_offset+YY] += ty;
1509 fshift[i_shift_offset+ZZ] += tz;
1511 /* Increment number of inner iterations */
1512 inneriter += j_index_end - j_index_start;
1514 /* Outer loop uses 30 flops */
1517 /* Increment number of outer iterations */
1520 /* Update outer/inner flops */
1522 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*30 + inneriter*544);