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_VdwNone_GeomW4W4_VF_c
35 * Electrostatics interaction: Ewald
36 * VdW interaction: None
37 * Geometry: Water4-Water4
38 * Calculate force/pot: PotentialAndForce
41 nb_kernel_ElecEwSw_VdwNone_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 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
59 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
61 real ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
63 real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
65 real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
67 real jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
68 real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
69 real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
70 real dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
71 real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
72 real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
73 real dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
74 real dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
75 real dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
76 real dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
77 real velec,felec,velecsum,facel,crf,krf,krf2;
80 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
82 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
89 jindex = nlist->jindex;
91 shiftidx = nlist->shift;
93 shiftvec = fr->shift_vec[0];
94 fshift = fr->fshift[0];
96 charge = mdatoms->chargeA;
98 sh_ewald = fr->ic->sh_ewald;
99 ewtab = fr->ic->tabq_coul_FDV0;
100 ewtabscale = fr->ic->tabq_scale;
101 ewtabhalfspace = 0.5/ewtabscale;
103 /* Setup water-specific parameters */
104 inr = nlist->iinr[0];
105 iq1 = facel*charge[inr+1];
106 iq2 = facel*charge[inr+2];
107 iq3 = facel*charge[inr+3];
122 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
123 rcutoff = fr->rcoulomb;
124 rcutoff2 = rcutoff*rcutoff;
126 rswitch = fr->rcoulomb_switch;
127 /* Setup switch parameters */
129 swV3 = -10.0/(d*d*d);
130 swV4 = 15.0/(d*d*d*d);
131 swV5 = -6.0/(d*d*d*d*d);
132 swF2 = -30.0/(d*d*d);
133 swF3 = 60.0/(d*d*d*d);
134 swF4 = -30.0/(d*d*d*d*d);
139 /* Start outer loop over neighborlists */
140 for(iidx=0; iidx<nri; iidx++)
142 /* Load shift vector for this list */
143 i_shift_offset = DIM*shiftidx[iidx];
144 shX = shiftvec[i_shift_offset+XX];
145 shY = shiftvec[i_shift_offset+YY];
146 shZ = shiftvec[i_shift_offset+ZZ];
148 /* Load limits for loop over neighbors */
149 j_index_start = jindex[iidx];
150 j_index_end = jindex[iidx+1];
152 /* Get outer coordinate index */
154 i_coord_offset = DIM*inr;
156 /* Load i particle coords and add shift vector */
157 ix1 = shX + x[i_coord_offset+DIM*1+XX];
158 iy1 = shY + x[i_coord_offset+DIM*1+YY];
159 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
160 ix2 = shX + x[i_coord_offset+DIM*2+XX];
161 iy2 = shY + x[i_coord_offset+DIM*2+YY];
162 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
163 ix3 = shX + x[i_coord_offset+DIM*3+XX];
164 iy3 = shY + x[i_coord_offset+DIM*3+YY];
165 iz3 = shZ + x[i_coord_offset+DIM*3+ZZ];
177 /* Reset potential sums */
180 /* Start inner kernel loop */
181 for(jidx=j_index_start; jidx<j_index_end; jidx++)
183 /* Get j neighbor index, and coordinate index */
185 j_coord_offset = DIM*jnr;
187 /* load j atom coordinates */
188 jx1 = x[j_coord_offset+DIM*1+XX];
189 jy1 = x[j_coord_offset+DIM*1+YY];
190 jz1 = x[j_coord_offset+DIM*1+ZZ];
191 jx2 = x[j_coord_offset+DIM*2+XX];
192 jy2 = x[j_coord_offset+DIM*2+YY];
193 jz2 = x[j_coord_offset+DIM*2+ZZ];
194 jx3 = x[j_coord_offset+DIM*3+XX];
195 jy3 = x[j_coord_offset+DIM*3+YY];
196 jz3 = x[j_coord_offset+DIM*3+ZZ];
198 /* Calculate displacement vector */
227 /* Calculate squared distance and things based on it */
228 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
229 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
230 rsq13 = dx13*dx13+dy13*dy13+dz13*dz13;
231 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
232 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
233 rsq23 = dx23*dx23+dy23*dy23+dz23*dz23;
234 rsq31 = dx31*dx31+dy31*dy31+dz31*dz31;
235 rsq32 = dx32*dx32+dy32*dy32+dz32*dz32;
236 rsq33 = dx33*dx33+dy33*dy33+dz33*dz33;
238 rinv11 = gmx_invsqrt(rsq11);
239 rinv12 = gmx_invsqrt(rsq12);
240 rinv13 = gmx_invsqrt(rsq13);
241 rinv21 = gmx_invsqrt(rsq21);
242 rinv22 = gmx_invsqrt(rsq22);
243 rinv23 = gmx_invsqrt(rsq23);
244 rinv31 = gmx_invsqrt(rsq31);
245 rinv32 = gmx_invsqrt(rsq32);
246 rinv33 = gmx_invsqrt(rsq33);
248 rinvsq11 = rinv11*rinv11;
249 rinvsq12 = rinv12*rinv12;
250 rinvsq13 = rinv13*rinv13;
251 rinvsq21 = rinv21*rinv21;
252 rinvsq22 = rinv22*rinv22;
253 rinvsq23 = rinv23*rinv23;
254 rinvsq31 = rinv31*rinv31;
255 rinvsq32 = rinv32*rinv32;
256 rinvsq33 = rinv33*rinv33;
258 /**************************
259 * CALCULATE INTERACTIONS *
260 **************************/
267 /* EWALD ELECTROSTATICS */
269 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
270 ewrt = r11*ewtabscale;
274 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
275 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
276 felec = qq11*rinv11*(rinvsq11-felec);
279 d = (d>0.0) ? d : 0.0;
281 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
283 dsw = d2*(swF2+d*(swF3+d*swF4));
285 /* Evaluate switch function */
286 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
287 felec = felec*sw - rinv11*velec*dsw;
290 /* Update potential sums from outer loop */
295 /* Calculate temporary vectorial force */
300 /* Update vectorial force */
304 f[j_coord_offset+DIM*1+XX] -= tx;
305 f[j_coord_offset+DIM*1+YY] -= ty;
306 f[j_coord_offset+DIM*1+ZZ] -= tz;
310 /**************************
311 * CALCULATE INTERACTIONS *
312 **************************/
319 /* EWALD ELECTROSTATICS */
321 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
322 ewrt = r12*ewtabscale;
326 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
327 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
328 felec = qq12*rinv12*(rinvsq12-felec);
331 d = (d>0.0) ? d : 0.0;
333 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
335 dsw = d2*(swF2+d*(swF3+d*swF4));
337 /* Evaluate switch function */
338 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
339 felec = felec*sw - rinv12*velec*dsw;
342 /* Update potential sums from outer loop */
347 /* Calculate temporary vectorial force */
352 /* Update vectorial force */
356 f[j_coord_offset+DIM*2+XX] -= tx;
357 f[j_coord_offset+DIM*2+YY] -= ty;
358 f[j_coord_offset+DIM*2+ZZ] -= tz;
362 /**************************
363 * CALCULATE INTERACTIONS *
364 **************************/
371 /* EWALD ELECTROSTATICS */
373 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
374 ewrt = r13*ewtabscale;
378 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
379 velec = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
380 felec = qq13*rinv13*(rinvsq13-felec);
383 d = (d>0.0) ? d : 0.0;
385 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
387 dsw = d2*(swF2+d*(swF3+d*swF4));
389 /* Evaluate switch function */
390 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
391 felec = felec*sw - rinv13*velec*dsw;
394 /* Update potential sums from outer loop */
399 /* Calculate temporary vectorial force */
404 /* Update vectorial force */
408 f[j_coord_offset+DIM*3+XX] -= tx;
409 f[j_coord_offset+DIM*3+YY] -= ty;
410 f[j_coord_offset+DIM*3+ZZ] -= tz;
414 /**************************
415 * CALCULATE INTERACTIONS *
416 **************************/
423 /* EWALD ELECTROSTATICS */
425 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
426 ewrt = r21*ewtabscale;
430 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
431 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
432 felec = qq21*rinv21*(rinvsq21-felec);
435 d = (d>0.0) ? d : 0.0;
437 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
439 dsw = d2*(swF2+d*(swF3+d*swF4));
441 /* Evaluate switch function */
442 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
443 felec = felec*sw - rinv21*velec*dsw;
446 /* Update potential sums from outer loop */
451 /* Calculate temporary vectorial force */
456 /* Update vectorial force */
460 f[j_coord_offset+DIM*1+XX] -= tx;
461 f[j_coord_offset+DIM*1+YY] -= ty;
462 f[j_coord_offset+DIM*1+ZZ] -= tz;
466 /**************************
467 * CALCULATE INTERACTIONS *
468 **************************/
475 /* EWALD ELECTROSTATICS */
477 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
478 ewrt = r22*ewtabscale;
482 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
483 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
484 felec = qq22*rinv22*(rinvsq22-felec);
487 d = (d>0.0) ? d : 0.0;
489 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
491 dsw = d2*(swF2+d*(swF3+d*swF4));
493 /* Evaluate switch function */
494 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
495 felec = felec*sw - rinv22*velec*dsw;
498 /* Update potential sums from outer loop */
503 /* Calculate temporary vectorial force */
508 /* Update vectorial force */
512 f[j_coord_offset+DIM*2+XX] -= tx;
513 f[j_coord_offset+DIM*2+YY] -= ty;
514 f[j_coord_offset+DIM*2+ZZ] -= tz;
518 /**************************
519 * CALCULATE INTERACTIONS *
520 **************************/
527 /* EWALD ELECTROSTATICS */
529 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
530 ewrt = r23*ewtabscale;
534 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
535 velec = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
536 felec = qq23*rinv23*(rinvsq23-felec);
539 d = (d>0.0) ? d : 0.0;
541 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
543 dsw = d2*(swF2+d*(swF3+d*swF4));
545 /* Evaluate switch function */
546 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
547 felec = felec*sw - rinv23*velec*dsw;
550 /* Update potential sums from outer loop */
555 /* Calculate temporary vectorial force */
560 /* Update vectorial force */
564 f[j_coord_offset+DIM*3+XX] -= tx;
565 f[j_coord_offset+DIM*3+YY] -= ty;
566 f[j_coord_offset+DIM*3+ZZ] -= tz;
570 /**************************
571 * CALCULATE INTERACTIONS *
572 **************************/
579 /* EWALD ELECTROSTATICS */
581 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
582 ewrt = r31*ewtabscale;
586 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
587 velec = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
588 felec = qq31*rinv31*(rinvsq31-felec);
591 d = (d>0.0) ? d : 0.0;
593 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
595 dsw = d2*(swF2+d*(swF3+d*swF4));
597 /* Evaluate switch function */
598 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
599 felec = felec*sw - rinv31*velec*dsw;
602 /* Update potential sums from outer loop */
607 /* Calculate temporary vectorial force */
612 /* Update vectorial force */
616 f[j_coord_offset+DIM*1+XX] -= tx;
617 f[j_coord_offset+DIM*1+YY] -= ty;
618 f[j_coord_offset+DIM*1+ZZ] -= tz;
622 /**************************
623 * CALCULATE INTERACTIONS *
624 **************************/
631 /* EWALD ELECTROSTATICS */
633 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
634 ewrt = r32*ewtabscale;
638 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
639 velec = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
640 felec = qq32*rinv32*(rinvsq32-felec);
643 d = (d>0.0) ? d : 0.0;
645 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
647 dsw = d2*(swF2+d*(swF3+d*swF4));
649 /* Evaluate switch function */
650 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
651 felec = felec*sw - rinv32*velec*dsw;
654 /* Update potential sums from outer loop */
659 /* Calculate temporary vectorial force */
664 /* Update vectorial force */
668 f[j_coord_offset+DIM*2+XX] -= tx;
669 f[j_coord_offset+DIM*2+YY] -= ty;
670 f[j_coord_offset+DIM*2+ZZ] -= tz;
674 /**************************
675 * CALCULATE INTERACTIONS *
676 **************************/
683 /* EWALD ELECTROSTATICS */
685 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
686 ewrt = r33*ewtabscale;
690 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
691 velec = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
692 felec = qq33*rinv33*(rinvsq33-felec);
695 d = (d>0.0) ? d : 0.0;
697 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
699 dsw = d2*(swF2+d*(swF3+d*swF4));
701 /* Evaluate switch function */
702 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
703 felec = felec*sw - rinv33*velec*dsw;
706 /* Update potential sums from outer loop */
711 /* Calculate temporary vectorial force */
716 /* Update vectorial force */
720 f[j_coord_offset+DIM*3+XX] -= tx;
721 f[j_coord_offset+DIM*3+YY] -= ty;
722 f[j_coord_offset+DIM*3+ZZ] -= tz;
726 /* Inner loop uses 522 flops */
728 /* End of innermost loop */
731 f[i_coord_offset+DIM*1+XX] += fix1;
732 f[i_coord_offset+DIM*1+YY] += fiy1;
733 f[i_coord_offset+DIM*1+ZZ] += fiz1;
737 f[i_coord_offset+DIM*2+XX] += fix2;
738 f[i_coord_offset+DIM*2+YY] += fiy2;
739 f[i_coord_offset+DIM*2+ZZ] += fiz2;
743 f[i_coord_offset+DIM*3+XX] += fix3;
744 f[i_coord_offset+DIM*3+YY] += fiy3;
745 f[i_coord_offset+DIM*3+ZZ] += fiz3;
749 fshift[i_shift_offset+XX] += tx;
750 fshift[i_shift_offset+YY] += ty;
751 fshift[i_shift_offset+ZZ] += tz;
754 /* Update potential energies */
755 kernel_data->energygrp_elec[ggid] += velecsum;
757 /* Increment number of inner iterations */
758 inneriter += j_index_end - j_index_start;
760 /* Outer loop uses 31 flops */
763 /* Increment number of outer iterations */
766 /* Update outer/inner flops */
768 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_VF,outeriter*31 + inneriter*522);
771 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwNone_GeomW4W4_F_c
772 * Electrostatics interaction: Ewald
773 * VdW interaction: None
774 * Geometry: Water4-Water4
775 * Calculate force/pot: Force
778 nb_kernel_ElecEwSw_VdwNone_GeomW4W4_F_c
779 (t_nblist * gmx_restrict nlist,
780 rvec * gmx_restrict xx,
781 rvec * gmx_restrict ff,
782 t_forcerec * gmx_restrict fr,
783 t_mdatoms * gmx_restrict mdatoms,
784 nb_kernel_data_t * gmx_restrict kernel_data,
785 t_nrnb * gmx_restrict nrnb)
787 int i_shift_offset,i_coord_offset,j_coord_offset;
788 int j_index_start,j_index_end;
789 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
790 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
791 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
792 real *shiftvec,*fshift,*x,*f;
794 real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
796 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
798 real ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
800 real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
802 real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
804 real jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
805 real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
806 real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
807 real dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
808 real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
809 real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
810 real dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
811 real dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
812 real dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
813 real dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
814 real velec,felec,velecsum,facel,crf,krf,krf2;
817 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
819 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
826 jindex = nlist->jindex;
828 shiftidx = nlist->shift;
830 shiftvec = fr->shift_vec[0];
831 fshift = fr->fshift[0];
833 charge = mdatoms->chargeA;
835 sh_ewald = fr->ic->sh_ewald;
836 ewtab = fr->ic->tabq_coul_FDV0;
837 ewtabscale = fr->ic->tabq_scale;
838 ewtabhalfspace = 0.5/ewtabscale;
840 /* Setup water-specific parameters */
841 inr = nlist->iinr[0];
842 iq1 = facel*charge[inr+1];
843 iq2 = facel*charge[inr+2];
844 iq3 = facel*charge[inr+3];
859 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
860 rcutoff = fr->rcoulomb;
861 rcutoff2 = rcutoff*rcutoff;
863 rswitch = fr->rcoulomb_switch;
864 /* Setup switch parameters */
866 swV3 = -10.0/(d*d*d);
867 swV4 = 15.0/(d*d*d*d);
868 swV5 = -6.0/(d*d*d*d*d);
869 swF2 = -30.0/(d*d*d);
870 swF3 = 60.0/(d*d*d*d);
871 swF4 = -30.0/(d*d*d*d*d);
876 /* Start outer loop over neighborlists */
877 for(iidx=0; iidx<nri; iidx++)
879 /* Load shift vector for this list */
880 i_shift_offset = DIM*shiftidx[iidx];
881 shX = shiftvec[i_shift_offset+XX];
882 shY = shiftvec[i_shift_offset+YY];
883 shZ = shiftvec[i_shift_offset+ZZ];
885 /* Load limits for loop over neighbors */
886 j_index_start = jindex[iidx];
887 j_index_end = jindex[iidx+1];
889 /* Get outer coordinate index */
891 i_coord_offset = DIM*inr;
893 /* Load i particle coords and add shift vector */
894 ix1 = shX + x[i_coord_offset+DIM*1+XX];
895 iy1 = shY + x[i_coord_offset+DIM*1+YY];
896 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
897 ix2 = shX + x[i_coord_offset+DIM*2+XX];
898 iy2 = shY + x[i_coord_offset+DIM*2+YY];
899 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
900 ix3 = shX + x[i_coord_offset+DIM*3+XX];
901 iy3 = shY + x[i_coord_offset+DIM*3+YY];
902 iz3 = shZ + x[i_coord_offset+DIM*3+ZZ];
914 /* Start inner kernel loop */
915 for(jidx=j_index_start; jidx<j_index_end; jidx++)
917 /* Get j neighbor index, and coordinate index */
919 j_coord_offset = DIM*jnr;
921 /* load j atom coordinates */
922 jx1 = x[j_coord_offset+DIM*1+XX];
923 jy1 = x[j_coord_offset+DIM*1+YY];
924 jz1 = x[j_coord_offset+DIM*1+ZZ];
925 jx2 = x[j_coord_offset+DIM*2+XX];
926 jy2 = x[j_coord_offset+DIM*2+YY];
927 jz2 = x[j_coord_offset+DIM*2+ZZ];
928 jx3 = x[j_coord_offset+DIM*3+XX];
929 jy3 = x[j_coord_offset+DIM*3+YY];
930 jz3 = x[j_coord_offset+DIM*3+ZZ];
932 /* Calculate displacement vector */
961 /* Calculate squared distance and things based on it */
962 rsq11 = dx11*dx11+dy11*dy11+dz11*dz11;
963 rsq12 = dx12*dx12+dy12*dy12+dz12*dz12;
964 rsq13 = dx13*dx13+dy13*dy13+dz13*dz13;
965 rsq21 = dx21*dx21+dy21*dy21+dz21*dz21;
966 rsq22 = dx22*dx22+dy22*dy22+dz22*dz22;
967 rsq23 = dx23*dx23+dy23*dy23+dz23*dz23;
968 rsq31 = dx31*dx31+dy31*dy31+dz31*dz31;
969 rsq32 = dx32*dx32+dy32*dy32+dz32*dz32;
970 rsq33 = dx33*dx33+dy33*dy33+dz33*dz33;
972 rinv11 = gmx_invsqrt(rsq11);
973 rinv12 = gmx_invsqrt(rsq12);
974 rinv13 = gmx_invsqrt(rsq13);
975 rinv21 = gmx_invsqrt(rsq21);
976 rinv22 = gmx_invsqrt(rsq22);
977 rinv23 = gmx_invsqrt(rsq23);
978 rinv31 = gmx_invsqrt(rsq31);
979 rinv32 = gmx_invsqrt(rsq32);
980 rinv33 = gmx_invsqrt(rsq33);
982 rinvsq11 = rinv11*rinv11;
983 rinvsq12 = rinv12*rinv12;
984 rinvsq13 = rinv13*rinv13;
985 rinvsq21 = rinv21*rinv21;
986 rinvsq22 = rinv22*rinv22;
987 rinvsq23 = rinv23*rinv23;
988 rinvsq31 = rinv31*rinv31;
989 rinvsq32 = rinv32*rinv32;
990 rinvsq33 = rinv33*rinv33;
992 /**************************
993 * CALCULATE INTERACTIONS *
994 **************************/
1001 /* EWALD ELECTROSTATICS */
1003 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1004 ewrt = r11*ewtabscale;
1006 eweps = ewrt-ewitab;
1008 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1009 velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1010 felec = qq11*rinv11*(rinvsq11-felec);
1013 d = (d>0.0) ? d : 0.0;
1015 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1017 dsw = d2*(swF2+d*(swF3+d*swF4));
1019 /* Evaluate switch function */
1020 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1021 felec = felec*sw - rinv11*velec*dsw;
1025 /* Calculate temporary vectorial force */
1030 /* Update vectorial force */
1034 f[j_coord_offset+DIM*1+XX] -= tx;
1035 f[j_coord_offset+DIM*1+YY] -= ty;
1036 f[j_coord_offset+DIM*1+ZZ] -= tz;
1040 /**************************
1041 * CALCULATE INTERACTIONS *
1042 **************************/
1049 /* EWALD ELECTROSTATICS */
1051 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1052 ewrt = r12*ewtabscale;
1054 eweps = ewrt-ewitab;
1056 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1057 velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1058 felec = qq12*rinv12*(rinvsq12-felec);
1061 d = (d>0.0) ? d : 0.0;
1063 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1065 dsw = d2*(swF2+d*(swF3+d*swF4));
1067 /* Evaluate switch function */
1068 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1069 felec = felec*sw - rinv12*velec*dsw;
1073 /* Calculate temporary vectorial force */
1078 /* Update vectorial force */
1082 f[j_coord_offset+DIM*2+XX] -= tx;
1083 f[j_coord_offset+DIM*2+YY] -= ty;
1084 f[j_coord_offset+DIM*2+ZZ] -= tz;
1088 /**************************
1089 * CALCULATE INTERACTIONS *
1090 **************************/
1097 /* EWALD ELECTROSTATICS */
1099 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1100 ewrt = r13*ewtabscale;
1102 eweps = ewrt-ewitab;
1104 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1105 velec = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1106 felec = qq13*rinv13*(rinvsq13-felec);
1109 d = (d>0.0) ? d : 0.0;
1111 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1113 dsw = d2*(swF2+d*(swF3+d*swF4));
1115 /* Evaluate switch function */
1116 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1117 felec = felec*sw - rinv13*velec*dsw;
1121 /* Calculate temporary vectorial force */
1126 /* Update vectorial force */
1130 f[j_coord_offset+DIM*3+XX] -= tx;
1131 f[j_coord_offset+DIM*3+YY] -= ty;
1132 f[j_coord_offset+DIM*3+ZZ] -= tz;
1136 /**************************
1137 * CALCULATE INTERACTIONS *
1138 **************************/
1145 /* EWALD ELECTROSTATICS */
1147 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1148 ewrt = r21*ewtabscale;
1150 eweps = ewrt-ewitab;
1152 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1153 velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1154 felec = qq21*rinv21*(rinvsq21-felec);
1157 d = (d>0.0) ? d : 0.0;
1159 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1161 dsw = d2*(swF2+d*(swF3+d*swF4));
1163 /* Evaluate switch function */
1164 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1165 felec = felec*sw - rinv21*velec*dsw;
1169 /* Calculate temporary vectorial force */
1174 /* Update vectorial force */
1178 f[j_coord_offset+DIM*1+XX] -= tx;
1179 f[j_coord_offset+DIM*1+YY] -= ty;
1180 f[j_coord_offset+DIM*1+ZZ] -= tz;
1184 /**************************
1185 * CALCULATE INTERACTIONS *
1186 **************************/
1193 /* EWALD ELECTROSTATICS */
1195 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1196 ewrt = r22*ewtabscale;
1198 eweps = ewrt-ewitab;
1200 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1201 velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1202 felec = qq22*rinv22*(rinvsq22-felec);
1205 d = (d>0.0) ? d : 0.0;
1207 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1209 dsw = d2*(swF2+d*(swF3+d*swF4));
1211 /* Evaluate switch function */
1212 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1213 felec = felec*sw - rinv22*velec*dsw;
1217 /* Calculate temporary vectorial force */
1222 /* Update vectorial force */
1226 f[j_coord_offset+DIM*2+XX] -= tx;
1227 f[j_coord_offset+DIM*2+YY] -= ty;
1228 f[j_coord_offset+DIM*2+ZZ] -= tz;
1232 /**************************
1233 * CALCULATE INTERACTIONS *
1234 **************************/
1241 /* EWALD ELECTROSTATICS */
1243 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1244 ewrt = r23*ewtabscale;
1246 eweps = ewrt-ewitab;
1248 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1249 velec = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1250 felec = qq23*rinv23*(rinvsq23-felec);
1253 d = (d>0.0) ? d : 0.0;
1255 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1257 dsw = d2*(swF2+d*(swF3+d*swF4));
1259 /* Evaluate switch function */
1260 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1261 felec = felec*sw - rinv23*velec*dsw;
1265 /* Calculate temporary vectorial force */
1270 /* Update vectorial force */
1274 f[j_coord_offset+DIM*3+XX] -= tx;
1275 f[j_coord_offset+DIM*3+YY] -= ty;
1276 f[j_coord_offset+DIM*3+ZZ] -= tz;
1280 /**************************
1281 * CALCULATE INTERACTIONS *
1282 **************************/
1289 /* EWALD ELECTROSTATICS */
1291 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1292 ewrt = r31*ewtabscale;
1294 eweps = ewrt-ewitab;
1296 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1297 velec = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1298 felec = qq31*rinv31*(rinvsq31-felec);
1301 d = (d>0.0) ? d : 0.0;
1303 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1305 dsw = d2*(swF2+d*(swF3+d*swF4));
1307 /* Evaluate switch function */
1308 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1309 felec = felec*sw - rinv31*velec*dsw;
1313 /* Calculate temporary vectorial force */
1318 /* Update vectorial force */
1322 f[j_coord_offset+DIM*1+XX] -= tx;
1323 f[j_coord_offset+DIM*1+YY] -= ty;
1324 f[j_coord_offset+DIM*1+ZZ] -= tz;
1328 /**************************
1329 * CALCULATE INTERACTIONS *
1330 **************************/
1337 /* EWALD ELECTROSTATICS */
1339 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1340 ewrt = r32*ewtabscale;
1342 eweps = ewrt-ewitab;
1344 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1345 velec = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1346 felec = qq32*rinv32*(rinvsq32-felec);
1349 d = (d>0.0) ? d : 0.0;
1351 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1353 dsw = d2*(swF2+d*(swF3+d*swF4));
1355 /* Evaluate switch function */
1356 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1357 felec = felec*sw - rinv32*velec*dsw;
1361 /* Calculate temporary vectorial force */
1366 /* Update vectorial force */
1370 f[j_coord_offset+DIM*2+XX] -= tx;
1371 f[j_coord_offset+DIM*2+YY] -= ty;
1372 f[j_coord_offset+DIM*2+ZZ] -= tz;
1376 /**************************
1377 * CALCULATE INTERACTIONS *
1378 **************************/
1385 /* EWALD ELECTROSTATICS */
1387 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1388 ewrt = r33*ewtabscale;
1390 eweps = ewrt-ewitab;
1392 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1393 velec = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1394 felec = qq33*rinv33*(rinvsq33-felec);
1397 d = (d>0.0) ? d : 0.0;
1399 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1401 dsw = d2*(swF2+d*(swF3+d*swF4));
1403 /* Evaluate switch function */
1404 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1405 felec = felec*sw - rinv33*velec*dsw;
1409 /* Calculate temporary vectorial force */
1414 /* Update vectorial force */
1418 f[j_coord_offset+DIM*3+XX] -= tx;
1419 f[j_coord_offset+DIM*3+YY] -= ty;
1420 f[j_coord_offset+DIM*3+ZZ] -= tz;
1424 /* Inner loop uses 504 flops */
1426 /* End of innermost loop */
1429 f[i_coord_offset+DIM*1+XX] += fix1;
1430 f[i_coord_offset+DIM*1+YY] += fiy1;
1431 f[i_coord_offset+DIM*1+ZZ] += fiz1;
1435 f[i_coord_offset+DIM*2+XX] += fix2;
1436 f[i_coord_offset+DIM*2+YY] += fiy2;
1437 f[i_coord_offset+DIM*2+ZZ] += fiz2;
1441 f[i_coord_offset+DIM*3+XX] += fix3;
1442 f[i_coord_offset+DIM*3+YY] += fiy3;
1443 f[i_coord_offset+DIM*3+ZZ] += fiz3;
1447 fshift[i_shift_offset+XX] += tx;
1448 fshift[i_shift_offset+YY] += ty;
1449 fshift[i_shift_offset+ZZ] += tz;
1451 /* Increment number of inner iterations */
1452 inneriter += j_index_end - j_index_start;
1454 /* Outer loop uses 30 flops */
1457 /* Increment number of outer iterations */
1460 /* Update outer/inner flops */
1462 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_F,outeriter*30 + inneriter*504);