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_GeomW4P1_VF_c
35 * Electrostatics interaction: Ewald
36 * VdW interaction: None
37 * Geometry: Water4-Particle
38 * Calculate force/pot: PotentialAndForce
41 nb_kernel_ElecEwSw_VdwNone_GeomW4P1_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 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
64 real dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
65 real dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
66 real dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
67 real velec,felec,velecsum,facel,crf,krf,krf2;
70 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
72 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
79 jindex = nlist->jindex;
81 shiftidx = nlist->shift;
83 shiftvec = fr->shift_vec[0];
84 fshift = fr->fshift[0];
86 charge = mdatoms->chargeA;
88 sh_ewald = fr->ic->sh_ewald;
89 ewtab = fr->ic->tabq_coul_FDV0;
90 ewtabscale = fr->ic->tabq_scale;
91 ewtabhalfspace = 0.5/ewtabscale;
93 /* Setup water-specific parameters */
95 iq1 = facel*charge[inr+1];
96 iq2 = facel*charge[inr+2];
97 iq3 = facel*charge[inr+3];
99 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
100 rcutoff = fr->rcoulomb;
101 rcutoff2 = rcutoff*rcutoff;
103 rswitch = fr->rcoulomb_switch;
104 /* Setup switch parameters */
106 swV3 = -10.0/(d*d*d);
107 swV4 = 15.0/(d*d*d*d);
108 swV5 = -6.0/(d*d*d*d*d);
109 swF2 = -30.0/(d*d*d);
110 swF3 = 60.0/(d*d*d*d);
111 swF4 = -30.0/(d*d*d*d*d);
116 /* Start outer loop over neighborlists */
117 for(iidx=0; iidx<nri; iidx++)
119 /* Load shift vector for this list */
120 i_shift_offset = DIM*shiftidx[iidx];
121 shX = shiftvec[i_shift_offset+XX];
122 shY = shiftvec[i_shift_offset+YY];
123 shZ = shiftvec[i_shift_offset+ZZ];
125 /* Load limits for loop over neighbors */
126 j_index_start = jindex[iidx];
127 j_index_end = jindex[iidx+1];
129 /* Get outer coordinate index */
131 i_coord_offset = DIM*inr;
133 /* Load i particle coords and add shift vector */
134 ix1 = shX + x[i_coord_offset+DIM*1+XX];
135 iy1 = shY + x[i_coord_offset+DIM*1+YY];
136 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
137 ix2 = shX + x[i_coord_offset+DIM*2+XX];
138 iy2 = shY + x[i_coord_offset+DIM*2+YY];
139 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
140 ix3 = shX + x[i_coord_offset+DIM*3+XX];
141 iy3 = shY + x[i_coord_offset+DIM*3+YY];
142 iz3 = shZ + x[i_coord_offset+DIM*3+ZZ];
154 /* Reset potential sums */
157 /* Start inner kernel loop */
158 for(jidx=j_index_start; jidx<j_index_end; jidx++)
160 /* Get j neighbor index, and coordinate index */
162 j_coord_offset = DIM*jnr;
164 /* load j atom coordinates */
165 jx0 = x[j_coord_offset+DIM*0+XX];
166 jy0 = x[j_coord_offset+DIM*0+YY];
167 jz0 = x[j_coord_offset+DIM*0+ZZ];
169 /* Calculate displacement vector */
180 /* Calculate squared distance and things based on it */
181 rsq10 = dx10*dx10+dy10*dy10+dz10*dz10;
182 rsq20 = dx20*dx20+dy20*dy20+dz20*dz20;
183 rsq30 = dx30*dx30+dy30*dy30+dz30*dz30;
185 rinv10 = gmx_invsqrt(rsq10);
186 rinv20 = gmx_invsqrt(rsq20);
187 rinv30 = gmx_invsqrt(rsq30);
189 rinvsq10 = rinv10*rinv10;
190 rinvsq20 = rinv20*rinv20;
191 rinvsq30 = rinv30*rinv30;
193 /* Load parameters for j particles */
196 /**************************
197 * CALCULATE INTERACTIONS *
198 **************************/
207 /* EWALD ELECTROSTATICS */
209 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
210 ewrt = r10*ewtabscale;
214 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
215 velec = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
216 felec = qq10*rinv10*(rinvsq10-felec);
219 d = (d>0.0) ? d : 0.0;
221 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
223 dsw = d2*(swF2+d*(swF3+d*swF4));
225 /* Evaluate switch function */
226 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
227 felec = felec*sw - rinv10*velec*dsw;
230 /* Update potential sums from outer loop */
235 /* Calculate temporary vectorial force */
240 /* Update vectorial force */
244 f[j_coord_offset+DIM*0+XX] -= tx;
245 f[j_coord_offset+DIM*0+YY] -= ty;
246 f[j_coord_offset+DIM*0+ZZ] -= tz;
250 /**************************
251 * CALCULATE INTERACTIONS *
252 **************************/
261 /* EWALD ELECTROSTATICS */
263 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
264 ewrt = r20*ewtabscale;
268 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
269 velec = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
270 felec = qq20*rinv20*(rinvsq20-felec);
273 d = (d>0.0) ? d : 0.0;
275 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
277 dsw = d2*(swF2+d*(swF3+d*swF4));
279 /* Evaluate switch function */
280 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
281 felec = felec*sw - rinv20*velec*dsw;
284 /* Update potential sums from outer loop */
289 /* Calculate temporary vectorial force */
294 /* Update vectorial force */
298 f[j_coord_offset+DIM*0+XX] -= tx;
299 f[j_coord_offset+DIM*0+YY] -= ty;
300 f[j_coord_offset+DIM*0+ZZ] -= tz;
304 /**************************
305 * CALCULATE INTERACTIONS *
306 **************************/
315 /* EWALD ELECTROSTATICS */
317 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
318 ewrt = r30*ewtabscale;
322 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
323 velec = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
324 felec = qq30*rinv30*(rinvsq30-felec);
327 d = (d>0.0) ? d : 0.0;
329 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
331 dsw = d2*(swF2+d*(swF3+d*swF4));
333 /* Evaluate switch function */
334 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
335 felec = felec*sw - rinv30*velec*dsw;
338 /* Update potential sums from outer loop */
343 /* Calculate temporary vectorial force */
348 /* Update vectorial force */
352 f[j_coord_offset+DIM*0+XX] -= tx;
353 f[j_coord_offset+DIM*0+YY] -= ty;
354 f[j_coord_offset+DIM*0+ZZ] -= tz;
358 /* Inner loop uses 177 flops */
360 /* End of innermost loop */
363 f[i_coord_offset+DIM*1+XX] += fix1;
364 f[i_coord_offset+DIM*1+YY] += fiy1;
365 f[i_coord_offset+DIM*1+ZZ] += fiz1;
369 f[i_coord_offset+DIM*2+XX] += fix2;
370 f[i_coord_offset+DIM*2+YY] += fiy2;
371 f[i_coord_offset+DIM*2+ZZ] += fiz2;
375 f[i_coord_offset+DIM*3+XX] += fix3;
376 f[i_coord_offset+DIM*3+YY] += fiy3;
377 f[i_coord_offset+DIM*3+ZZ] += fiz3;
381 fshift[i_shift_offset+XX] += tx;
382 fshift[i_shift_offset+YY] += ty;
383 fshift[i_shift_offset+ZZ] += tz;
386 /* Update potential energies */
387 kernel_data->energygrp_elec[ggid] += velecsum;
389 /* Increment number of inner iterations */
390 inneriter += j_index_end - j_index_start;
392 /* Outer loop uses 31 flops */
395 /* Increment number of outer iterations */
398 /* Update outer/inner flops */
400 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4_VF,outeriter*31 + inneriter*177);
403 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwNone_GeomW4P1_F_c
404 * Electrostatics interaction: Ewald
405 * VdW interaction: None
406 * Geometry: Water4-Particle
407 * Calculate force/pot: Force
410 nb_kernel_ElecEwSw_VdwNone_GeomW4P1_F_c
411 (t_nblist * gmx_restrict nlist,
412 rvec * gmx_restrict xx,
413 rvec * gmx_restrict ff,
414 t_forcerec * gmx_restrict fr,
415 t_mdatoms * gmx_restrict mdatoms,
416 nb_kernel_data_t * gmx_restrict kernel_data,
417 t_nrnb * gmx_restrict nrnb)
419 int i_shift_offset,i_coord_offset,j_coord_offset;
420 int j_index_start,j_index_end;
421 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
422 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
423 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
424 real *shiftvec,*fshift,*x,*f;
426 real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
428 real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
430 real ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
432 real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
433 real dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
434 real dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
435 real dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
436 real velec,felec,velecsum,facel,crf,krf,krf2;
439 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
441 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
448 jindex = nlist->jindex;
450 shiftidx = nlist->shift;
452 shiftvec = fr->shift_vec[0];
453 fshift = fr->fshift[0];
455 charge = mdatoms->chargeA;
457 sh_ewald = fr->ic->sh_ewald;
458 ewtab = fr->ic->tabq_coul_FDV0;
459 ewtabscale = fr->ic->tabq_scale;
460 ewtabhalfspace = 0.5/ewtabscale;
462 /* Setup water-specific parameters */
463 inr = nlist->iinr[0];
464 iq1 = facel*charge[inr+1];
465 iq2 = facel*charge[inr+2];
466 iq3 = facel*charge[inr+3];
468 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
469 rcutoff = fr->rcoulomb;
470 rcutoff2 = rcutoff*rcutoff;
472 rswitch = fr->rcoulomb_switch;
473 /* Setup switch parameters */
475 swV3 = -10.0/(d*d*d);
476 swV4 = 15.0/(d*d*d*d);
477 swV5 = -6.0/(d*d*d*d*d);
478 swF2 = -30.0/(d*d*d);
479 swF3 = 60.0/(d*d*d*d);
480 swF4 = -30.0/(d*d*d*d*d);
485 /* Start outer loop over neighborlists */
486 for(iidx=0; iidx<nri; iidx++)
488 /* Load shift vector for this list */
489 i_shift_offset = DIM*shiftidx[iidx];
490 shX = shiftvec[i_shift_offset+XX];
491 shY = shiftvec[i_shift_offset+YY];
492 shZ = shiftvec[i_shift_offset+ZZ];
494 /* Load limits for loop over neighbors */
495 j_index_start = jindex[iidx];
496 j_index_end = jindex[iidx+1];
498 /* Get outer coordinate index */
500 i_coord_offset = DIM*inr;
502 /* Load i particle coords and add shift vector */
503 ix1 = shX + x[i_coord_offset+DIM*1+XX];
504 iy1 = shY + x[i_coord_offset+DIM*1+YY];
505 iz1 = shZ + x[i_coord_offset+DIM*1+ZZ];
506 ix2 = shX + x[i_coord_offset+DIM*2+XX];
507 iy2 = shY + x[i_coord_offset+DIM*2+YY];
508 iz2 = shZ + x[i_coord_offset+DIM*2+ZZ];
509 ix3 = shX + x[i_coord_offset+DIM*3+XX];
510 iy3 = shY + x[i_coord_offset+DIM*3+YY];
511 iz3 = shZ + x[i_coord_offset+DIM*3+ZZ];
523 /* Start inner kernel loop */
524 for(jidx=j_index_start; jidx<j_index_end; jidx++)
526 /* Get j neighbor index, and coordinate index */
528 j_coord_offset = DIM*jnr;
530 /* load j atom coordinates */
531 jx0 = x[j_coord_offset+DIM*0+XX];
532 jy0 = x[j_coord_offset+DIM*0+YY];
533 jz0 = x[j_coord_offset+DIM*0+ZZ];
535 /* Calculate displacement vector */
546 /* Calculate squared distance and things based on it */
547 rsq10 = dx10*dx10+dy10*dy10+dz10*dz10;
548 rsq20 = dx20*dx20+dy20*dy20+dz20*dz20;
549 rsq30 = dx30*dx30+dy30*dy30+dz30*dz30;
551 rinv10 = gmx_invsqrt(rsq10);
552 rinv20 = gmx_invsqrt(rsq20);
553 rinv30 = gmx_invsqrt(rsq30);
555 rinvsq10 = rinv10*rinv10;
556 rinvsq20 = rinv20*rinv20;
557 rinvsq30 = rinv30*rinv30;
559 /* Load parameters for j particles */
562 /**************************
563 * CALCULATE INTERACTIONS *
564 **************************/
573 /* EWALD ELECTROSTATICS */
575 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
576 ewrt = r10*ewtabscale;
580 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
581 velec = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
582 felec = qq10*rinv10*(rinvsq10-felec);
585 d = (d>0.0) ? d : 0.0;
587 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
589 dsw = d2*(swF2+d*(swF3+d*swF4));
591 /* Evaluate switch function */
592 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
593 felec = felec*sw - rinv10*velec*dsw;
597 /* Calculate temporary vectorial force */
602 /* Update vectorial force */
606 f[j_coord_offset+DIM*0+XX] -= tx;
607 f[j_coord_offset+DIM*0+YY] -= ty;
608 f[j_coord_offset+DIM*0+ZZ] -= tz;
612 /**************************
613 * CALCULATE INTERACTIONS *
614 **************************/
623 /* EWALD ELECTROSTATICS */
625 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
626 ewrt = r20*ewtabscale;
630 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
631 velec = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
632 felec = qq20*rinv20*(rinvsq20-felec);
635 d = (d>0.0) ? d : 0.0;
637 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
639 dsw = d2*(swF2+d*(swF3+d*swF4));
641 /* Evaluate switch function */
642 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
643 felec = felec*sw - rinv20*velec*dsw;
647 /* Calculate temporary vectorial force */
652 /* Update vectorial force */
656 f[j_coord_offset+DIM*0+XX] -= tx;
657 f[j_coord_offset+DIM*0+YY] -= ty;
658 f[j_coord_offset+DIM*0+ZZ] -= tz;
662 /**************************
663 * CALCULATE INTERACTIONS *
664 **************************/
673 /* EWALD ELECTROSTATICS */
675 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
676 ewrt = r30*ewtabscale;
680 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
681 velec = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
682 felec = qq30*rinv30*(rinvsq30-felec);
685 d = (d>0.0) ? d : 0.0;
687 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
689 dsw = d2*(swF2+d*(swF3+d*swF4));
691 /* Evaluate switch function */
692 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
693 felec = felec*sw - rinv30*velec*dsw;
697 /* Calculate temporary vectorial force */
702 /* Update vectorial force */
706 f[j_coord_offset+DIM*0+XX] -= tx;
707 f[j_coord_offset+DIM*0+YY] -= ty;
708 f[j_coord_offset+DIM*0+ZZ] -= tz;
712 /* Inner loop uses 171 flops */
714 /* End of innermost loop */
717 f[i_coord_offset+DIM*1+XX] += fix1;
718 f[i_coord_offset+DIM*1+YY] += fiy1;
719 f[i_coord_offset+DIM*1+ZZ] += fiz1;
723 f[i_coord_offset+DIM*2+XX] += fix2;
724 f[i_coord_offset+DIM*2+YY] += fiy2;
725 f[i_coord_offset+DIM*2+ZZ] += fiz2;
729 f[i_coord_offset+DIM*3+XX] += fix3;
730 f[i_coord_offset+DIM*3+YY] += fiy3;
731 f[i_coord_offset+DIM*3+ZZ] += fiz3;
735 fshift[i_shift_offset+XX] += tx;
736 fshift[i_shift_offset+YY] += ty;
737 fshift[i_shift_offset+ZZ] += tz;
739 /* Increment number of inner iterations */
740 inneriter += j_index_end - j_index_start;
742 /* Outer loop uses 30 flops */
745 /* Increment number of outer iterations */
748 /* Update outer/inner flops */
750 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4_F,outeriter*30 + inneriter*171);