File: | gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_c.c |
Location: | line 882, column 5 |
Description: | Value stored to 'sh_ewald' is never read |
1 | /* |
2 | * This file is part of the GROMACS molecular simulation package. |
3 | * |
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. |
8 | * |
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. |
13 | * |
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. |
18 | * |
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. |
23 | * |
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. |
31 | * |
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. |
34 | */ |
35 | /* |
36 | * Note: this file was generated by the GROMACS c kernel generator. |
37 | */ |
38 | #ifdef HAVE_CONFIG_H1 |
39 | #include <config.h> |
40 | #endif |
41 | |
42 | #include <math.h> |
43 | |
44 | #include "../nb_kernel.h" |
45 | #include "types/simple.h" |
46 | #include "gromacs/math/vec.h" |
47 | #include "nrnb.h" |
48 | |
49 | /* |
50 | * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_VF_c |
51 | * Electrostatics interaction: Ewald |
52 | * VdW interaction: LennardJones |
53 | * Geometry: Water3-Water3 |
54 | * Calculate force/pot: PotentialAndForce |
55 | */ |
56 | void |
57 | nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_VF_c |
58 | (t_nblist * gmx_restrict__restrict nlist, |
59 | rvec * gmx_restrict__restrict xx, |
60 | rvec * gmx_restrict__restrict ff, |
61 | t_forcerec * gmx_restrict__restrict fr, |
62 | t_mdatoms * gmx_restrict__restrict mdatoms, |
63 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict__restrict kernel_data, |
64 | t_nrnb * gmx_restrict__restrict nrnb) |
65 | { |
66 | int i_shift_offset,i_coord_offset,j_coord_offset; |
67 | int j_index_start,j_index_end; |
68 | int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter; |
69 | real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2; |
70 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
71 | real *shiftvec,*fshift,*x,*f; |
72 | int vdwioffset0; |
73 | real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0; |
74 | int vdwioffset1; |
75 | real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
76 | int vdwioffset2; |
77 | real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
78 | int vdwjidx0; |
79 | real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
80 | int vdwjidx1; |
81 | real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
82 | int vdwjidx2; |
83 | real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
84 | real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00; |
85 | real dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01; |
86 | real dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02; |
87 | real dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10; |
88 | real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11; |
89 | real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12; |
90 | real dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20; |
91 | real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21; |
92 | real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22; |
93 | real velec,felec,velecsum,facel,crf,krf,krf2; |
94 | real *charge; |
95 | int nvdwtype; |
96 | real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6; |
97 | int *vdwtype; |
98 | real *vdwparam; |
99 | int ewitab; |
100 | real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace; |
101 | real *ewtab; |
102 | real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw; |
103 | |
104 | x = xx[0]; |
105 | f = ff[0]; |
106 | |
107 | nri = nlist->nri; |
108 | iinr = nlist->iinr; |
109 | jindex = nlist->jindex; |
110 | jjnr = nlist->jjnr; |
111 | shiftidx = nlist->shift; |
112 | gid = nlist->gid; |
113 | shiftvec = fr->shift_vec[0]; |
114 | fshift = fr->fshift[0]; |
115 | facel = fr->epsfac; |
116 | charge = mdatoms->chargeA; |
117 | nvdwtype = fr->ntype; |
118 | vdwparam = fr->nbfp; |
119 | vdwtype = mdatoms->typeA; |
120 | |
121 | sh_ewald = fr->ic->sh_ewald; |
122 | ewtab = fr->ic->tabq_coul_FDV0; |
123 | ewtabscale = fr->ic->tabq_scale; |
124 | ewtabhalfspace = 0.5/ewtabscale; |
125 | |
126 | /* Setup water-specific parameters */ |
127 | inr = nlist->iinr[0]; |
128 | iq0 = facel*charge[inr+0]; |
129 | iq1 = facel*charge[inr+1]; |
130 | iq2 = facel*charge[inr+2]; |
131 | vdwioffset0 = 2*nvdwtype*vdwtype[inr+0]; |
132 | |
133 | jq0 = charge[inr+0]; |
134 | jq1 = charge[inr+1]; |
135 | jq2 = charge[inr+2]; |
136 | vdwjidx0 = 2*vdwtype[inr+0]; |
137 | qq00 = iq0*jq0; |
138 | c6_00 = vdwparam[vdwioffset0+vdwjidx0]; |
139 | c12_00 = vdwparam[vdwioffset0+vdwjidx0+1]; |
140 | qq01 = iq0*jq1; |
141 | qq02 = iq0*jq2; |
142 | qq10 = iq1*jq0; |
143 | qq11 = iq1*jq1; |
144 | qq12 = iq1*jq2; |
145 | qq20 = iq2*jq0; |
146 | qq21 = iq2*jq1; |
147 | qq22 = iq2*jq2; |
148 | |
149 | /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */ |
150 | rcutoff = fr->rcoulomb; |
151 | rcutoff2 = rcutoff*rcutoff; |
152 | |
153 | rswitch = fr->rcoulomb_switch; |
154 | /* Setup switch parameters */ |
155 | d = rcutoff-rswitch; |
156 | swV3 = -10.0/(d*d*d); |
157 | swV4 = 15.0/(d*d*d*d); |
158 | swV5 = -6.0/(d*d*d*d*d); |
159 | swF2 = -30.0/(d*d*d); |
160 | swF3 = 60.0/(d*d*d*d); |
161 | swF4 = -30.0/(d*d*d*d*d); |
162 | |
163 | outeriter = 0; |
164 | inneriter = 0; |
165 | |
166 | /* Start outer loop over neighborlists */ |
167 | for(iidx=0; iidx<nri; iidx++) |
168 | { |
169 | /* Load shift vector for this list */ |
170 | i_shift_offset = DIM3*shiftidx[iidx]; |
171 | shX = shiftvec[i_shift_offset+XX0]; |
172 | shY = shiftvec[i_shift_offset+YY1]; |
173 | shZ = shiftvec[i_shift_offset+ZZ2]; |
174 | |
175 | /* Load limits for loop over neighbors */ |
176 | j_index_start = jindex[iidx]; |
177 | j_index_end = jindex[iidx+1]; |
178 | |
179 | /* Get outer coordinate index */ |
180 | inr = iinr[iidx]; |
181 | i_coord_offset = DIM3*inr; |
182 | |
183 | /* Load i particle coords and add shift vector */ |
184 | ix0 = shX + x[i_coord_offset+DIM3*0+XX0]; |
185 | iy0 = shY + x[i_coord_offset+DIM3*0+YY1]; |
186 | iz0 = shZ + x[i_coord_offset+DIM3*0+ZZ2]; |
187 | ix1 = shX + x[i_coord_offset+DIM3*1+XX0]; |
188 | iy1 = shY + x[i_coord_offset+DIM3*1+YY1]; |
189 | iz1 = shZ + x[i_coord_offset+DIM3*1+ZZ2]; |
190 | ix2 = shX + x[i_coord_offset+DIM3*2+XX0]; |
191 | iy2 = shY + x[i_coord_offset+DIM3*2+YY1]; |
192 | iz2 = shZ + x[i_coord_offset+DIM3*2+ZZ2]; |
193 | |
194 | fix0 = 0.0; |
195 | fiy0 = 0.0; |
196 | fiz0 = 0.0; |
197 | fix1 = 0.0; |
198 | fiy1 = 0.0; |
199 | fiz1 = 0.0; |
200 | fix2 = 0.0; |
201 | fiy2 = 0.0; |
202 | fiz2 = 0.0; |
203 | |
204 | /* Reset potential sums */ |
205 | velecsum = 0.0; |
206 | vvdwsum = 0.0; |
207 | |
208 | /* Start inner kernel loop */ |
209 | for(jidx=j_index_start; jidx<j_index_end; jidx++) |
210 | { |
211 | /* Get j neighbor index, and coordinate index */ |
212 | jnr = jjnr[jidx]; |
213 | j_coord_offset = DIM3*jnr; |
214 | |
215 | /* load j atom coordinates */ |
216 | jx0 = x[j_coord_offset+DIM3*0+XX0]; |
217 | jy0 = x[j_coord_offset+DIM3*0+YY1]; |
218 | jz0 = x[j_coord_offset+DIM3*0+ZZ2]; |
219 | jx1 = x[j_coord_offset+DIM3*1+XX0]; |
220 | jy1 = x[j_coord_offset+DIM3*1+YY1]; |
221 | jz1 = x[j_coord_offset+DIM3*1+ZZ2]; |
222 | jx2 = x[j_coord_offset+DIM3*2+XX0]; |
223 | jy2 = x[j_coord_offset+DIM3*2+YY1]; |
224 | jz2 = x[j_coord_offset+DIM3*2+ZZ2]; |
225 | |
226 | /* Calculate displacement vector */ |
227 | dx00 = ix0 - jx0; |
228 | dy00 = iy0 - jy0; |
229 | dz00 = iz0 - jz0; |
230 | dx01 = ix0 - jx1; |
231 | dy01 = iy0 - jy1; |
232 | dz01 = iz0 - jz1; |
233 | dx02 = ix0 - jx2; |
234 | dy02 = iy0 - jy2; |
235 | dz02 = iz0 - jz2; |
236 | dx10 = ix1 - jx0; |
237 | dy10 = iy1 - jy0; |
238 | dz10 = iz1 - jz0; |
239 | dx11 = ix1 - jx1; |
240 | dy11 = iy1 - jy1; |
241 | dz11 = iz1 - jz1; |
242 | dx12 = ix1 - jx2; |
243 | dy12 = iy1 - jy2; |
244 | dz12 = iz1 - jz2; |
245 | dx20 = ix2 - jx0; |
246 | dy20 = iy2 - jy0; |
247 | dz20 = iz2 - jz0; |
248 | dx21 = ix2 - jx1; |
249 | dy21 = iy2 - jy1; |
250 | dz21 = iz2 - jz1; |
251 | dx22 = ix2 - jx2; |
252 | dy22 = iy2 - jy2; |
253 | dz22 = iz2 - jz2; |
254 | |
255 | /* Calculate squared distance and things based on it */ |
256 | rsq00 = dx00*dx00+dy00*dy00+dz00*dz00; |
257 | rsq01 = dx01*dx01+dy01*dy01+dz01*dz01; |
258 | rsq02 = dx02*dx02+dy02*dy02+dz02*dz02; |
259 | rsq10 = dx10*dx10+dy10*dy10+dz10*dz10; |
260 | rsq11 = dx11*dx11+dy11*dy11+dz11*dz11; |
261 | rsq12 = dx12*dx12+dy12*dy12+dz12*dz12; |
262 | rsq20 = dx20*dx20+dy20*dy20+dz20*dz20; |
263 | rsq21 = dx21*dx21+dy21*dy21+dz21*dz21; |
264 | rsq22 = dx22*dx22+dy22*dy22+dz22*dz22; |
265 | |
266 | rinv00 = gmx_invsqrt(rsq00)gmx_software_invsqrt(rsq00); |
267 | rinv01 = gmx_invsqrt(rsq01)gmx_software_invsqrt(rsq01); |
268 | rinv02 = gmx_invsqrt(rsq02)gmx_software_invsqrt(rsq02); |
269 | rinv10 = gmx_invsqrt(rsq10)gmx_software_invsqrt(rsq10); |
270 | rinv11 = gmx_invsqrt(rsq11)gmx_software_invsqrt(rsq11); |
271 | rinv12 = gmx_invsqrt(rsq12)gmx_software_invsqrt(rsq12); |
272 | rinv20 = gmx_invsqrt(rsq20)gmx_software_invsqrt(rsq20); |
273 | rinv21 = gmx_invsqrt(rsq21)gmx_software_invsqrt(rsq21); |
274 | rinv22 = gmx_invsqrt(rsq22)gmx_software_invsqrt(rsq22); |
275 | |
276 | rinvsq00 = rinv00*rinv00; |
277 | rinvsq01 = rinv01*rinv01; |
278 | rinvsq02 = rinv02*rinv02; |
279 | rinvsq10 = rinv10*rinv10; |
280 | rinvsq11 = rinv11*rinv11; |
281 | rinvsq12 = rinv12*rinv12; |
282 | rinvsq20 = rinv20*rinv20; |
283 | rinvsq21 = rinv21*rinv21; |
284 | rinvsq22 = rinv22*rinv22; |
285 | |
286 | /************************** |
287 | * CALCULATE INTERACTIONS * |
288 | **************************/ |
289 | |
290 | if (rsq00<rcutoff2) |
291 | { |
292 | |
293 | r00 = rsq00*rinv00; |
294 | |
295 | /* EWALD ELECTROSTATICS */ |
296 | |
297 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
298 | ewrt = r00*ewtabscale; |
299 | ewitab = ewrt; |
300 | eweps = ewrt-ewitab; |
301 | ewitab = 4*ewitab; |
302 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
303 | velec = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
304 | felec = qq00*rinv00*(rinvsq00-felec); |
305 | |
306 | /* LENNARD-JONES DISPERSION/REPULSION */ |
307 | |
308 | rinvsix = rinvsq00*rinvsq00*rinvsq00; |
309 | vvdw6 = c6_00*rinvsix; |
310 | vvdw12 = c12_00*rinvsix*rinvsix; |
311 | vvdw = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0); |
312 | fvdw = (vvdw12-vvdw6)*rinvsq00; |
313 | |
314 | d = r00-rswitch; |
315 | d = (d>0.0) ? d : 0.0; |
316 | d2 = d*d; |
317 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
318 | |
319 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
320 | |
321 | /* Evaluate switch function */ |
322 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
323 | felec = felec*sw - rinv00*velec*dsw; |
324 | fvdw = fvdw*sw - rinv00*vvdw*dsw; |
325 | velec *= sw; |
326 | vvdw *= sw; |
327 | |
328 | /* Update potential sums from outer loop */ |
329 | velecsum += velec; |
330 | vvdwsum += vvdw; |
331 | |
332 | fscal = felec+fvdw; |
333 | |
334 | /* Calculate temporary vectorial force */ |
335 | tx = fscal*dx00; |
336 | ty = fscal*dy00; |
337 | tz = fscal*dz00; |
338 | |
339 | /* Update vectorial force */ |
340 | fix0 += tx; |
341 | fiy0 += ty; |
342 | fiz0 += tz; |
343 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
344 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
345 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
346 | |
347 | } |
348 | |
349 | /************************** |
350 | * CALCULATE INTERACTIONS * |
351 | **************************/ |
352 | |
353 | if (rsq01<rcutoff2) |
354 | { |
355 | |
356 | r01 = rsq01*rinv01; |
357 | |
358 | /* EWALD ELECTROSTATICS */ |
359 | |
360 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
361 | ewrt = r01*ewtabscale; |
362 | ewitab = ewrt; |
363 | eweps = ewrt-ewitab; |
364 | ewitab = 4*ewitab; |
365 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
366 | velec = qq01*(rinv01-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
367 | felec = qq01*rinv01*(rinvsq01-felec); |
368 | |
369 | d = r01-rswitch; |
370 | d = (d>0.0) ? d : 0.0; |
371 | d2 = d*d; |
372 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
373 | |
374 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
375 | |
376 | /* Evaluate switch function */ |
377 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
378 | felec = felec*sw - rinv01*velec*dsw; |
379 | velec *= sw; |
380 | |
381 | /* Update potential sums from outer loop */ |
382 | velecsum += velec; |
383 | |
384 | fscal = felec; |
385 | |
386 | /* Calculate temporary vectorial force */ |
387 | tx = fscal*dx01; |
388 | ty = fscal*dy01; |
389 | tz = fscal*dz01; |
390 | |
391 | /* Update vectorial force */ |
392 | fix0 += tx; |
393 | fiy0 += ty; |
394 | fiz0 += tz; |
395 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
396 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
397 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
398 | |
399 | } |
400 | |
401 | /************************** |
402 | * CALCULATE INTERACTIONS * |
403 | **************************/ |
404 | |
405 | if (rsq02<rcutoff2) |
406 | { |
407 | |
408 | r02 = rsq02*rinv02; |
409 | |
410 | /* EWALD ELECTROSTATICS */ |
411 | |
412 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
413 | ewrt = r02*ewtabscale; |
414 | ewitab = ewrt; |
415 | eweps = ewrt-ewitab; |
416 | ewitab = 4*ewitab; |
417 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
418 | velec = qq02*(rinv02-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
419 | felec = qq02*rinv02*(rinvsq02-felec); |
420 | |
421 | d = r02-rswitch; |
422 | d = (d>0.0) ? d : 0.0; |
423 | d2 = d*d; |
424 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
425 | |
426 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
427 | |
428 | /* Evaluate switch function */ |
429 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
430 | felec = felec*sw - rinv02*velec*dsw; |
431 | velec *= sw; |
432 | |
433 | /* Update potential sums from outer loop */ |
434 | velecsum += velec; |
435 | |
436 | fscal = felec; |
437 | |
438 | /* Calculate temporary vectorial force */ |
439 | tx = fscal*dx02; |
440 | ty = fscal*dy02; |
441 | tz = fscal*dz02; |
442 | |
443 | /* Update vectorial force */ |
444 | fix0 += tx; |
445 | fiy0 += ty; |
446 | fiz0 += tz; |
447 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
448 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
449 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
450 | |
451 | } |
452 | |
453 | /************************** |
454 | * CALCULATE INTERACTIONS * |
455 | **************************/ |
456 | |
457 | if (rsq10<rcutoff2) |
458 | { |
459 | |
460 | r10 = rsq10*rinv10; |
461 | |
462 | /* EWALD ELECTROSTATICS */ |
463 | |
464 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
465 | ewrt = r10*ewtabscale; |
466 | ewitab = ewrt; |
467 | eweps = ewrt-ewitab; |
468 | ewitab = 4*ewitab; |
469 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
470 | velec = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
471 | felec = qq10*rinv10*(rinvsq10-felec); |
472 | |
473 | d = r10-rswitch; |
474 | d = (d>0.0) ? d : 0.0; |
475 | d2 = d*d; |
476 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
477 | |
478 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
479 | |
480 | /* Evaluate switch function */ |
481 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
482 | felec = felec*sw - rinv10*velec*dsw; |
483 | velec *= sw; |
484 | |
485 | /* Update potential sums from outer loop */ |
486 | velecsum += velec; |
487 | |
488 | fscal = felec; |
489 | |
490 | /* Calculate temporary vectorial force */ |
491 | tx = fscal*dx10; |
492 | ty = fscal*dy10; |
493 | tz = fscal*dz10; |
494 | |
495 | /* Update vectorial force */ |
496 | fix1 += tx; |
497 | fiy1 += ty; |
498 | fiz1 += tz; |
499 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
500 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
501 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
502 | |
503 | } |
504 | |
505 | /************************** |
506 | * CALCULATE INTERACTIONS * |
507 | **************************/ |
508 | |
509 | if (rsq11<rcutoff2) |
510 | { |
511 | |
512 | r11 = rsq11*rinv11; |
513 | |
514 | /* EWALD ELECTROSTATICS */ |
515 | |
516 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
517 | ewrt = r11*ewtabscale; |
518 | ewitab = ewrt; |
519 | eweps = ewrt-ewitab; |
520 | ewitab = 4*ewitab; |
521 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
522 | velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
523 | felec = qq11*rinv11*(rinvsq11-felec); |
524 | |
525 | d = r11-rswitch; |
526 | d = (d>0.0) ? d : 0.0; |
527 | d2 = d*d; |
528 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
529 | |
530 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
531 | |
532 | /* Evaluate switch function */ |
533 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
534 | felec = felec*sw - rinv11*velec*dsw; |
535 | velec *= sw; |
536 | |
537 | /* Update potential sums from outer loop */ |
538 | velecsum += velec; |
539 | |
540 | fscal = felec; |
541 | |
542 | /* Calculate temporary vectorial force */ |
543 | tx = fscal*dx11; |
544 | ty = fscal*dy11; |
545 | tz = fscal*dz11; |
546 | |
547 | /* Update vectorial force */ |
548 | fix1 += tx; |
549 | fiy1 += ty; |
550 | fiz1 += tz; |
551 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
552 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
553 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
554 | |
555 | } |
556 | |
557 | /************************** |
558 | * CALCULATE INTERACTIONS * |
559 | **************************/ |
560 | |
561 | if (rsq12<rcutoff2) |
562 | { |
563 | |
564 | r12 = rsq12*rinv12; |
565 | |
566 | /* EWALD ELECTROSTATICS */ |
567 | |
568 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
569 | ewrt = r12*ewtabscale; |
570 | ewitab = ewrt; |
571 | eweps = ewrt-ewitab; |
572 | ewitab = 4*ewitab; |
573 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
574 | velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
575 | felec = qq12*rinv12*(rinvsq12-felec); |
576 | |
577 | d = r12-rswitch; |
578 | d = (d>0.0) ? d : 0.0; |
579 | d2 = d*d; |
580 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
581 | |
582 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
583 | |
584 | /* Evaluate switch function */ |
585 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
586 | felec = felec*sw - rinv12*velec*dsw; |
587 | velec *= sw; |
588 | |
589 | /* Update potential sums from outer loop */ |
590 | velecsum += velec; |
591 | |
592 | fscal = felec; |
593 | |
594 | /* Calculate temporary vectorial force */ |
595 | tx = fscal*dx12; |
596 | ty = fscal*dy12; |
597 | tz = fscal*dz12; |
598 | |
599 | /* Update vectorial force */ |
600 | fix1 += tx; |
601 | fiy1 += ty; |
602 | fiz1 += tz; |
603 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
604 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
605 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
606 | |
607 | } |
608 | |
609 | /************************** |
610 | * CALCULATE INTERACTIONS * |
611 | **************************/ |
612 | |
613 | if (rsq20<rcutoff2) |
614 | { |
615 | |
616 | r20 = rsq20*rinv20; |
617 | |
618 | /* EWALD ELECTROSTATICS */ |
619 | |
620 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
621 | ewrt = r20*ewtabscale; |
622 | ewitab = ewrt; |
623 | eweps = ewrt-ewitab; |
624 | ewitab = 4*ewitab; |
625 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
626 | velec = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
627 | felec = qq20*rinv20*(rinvsq20-felec); |
628 | |
629 | d = r20-rswitch; |
630 | d = (d>0.0) ? d : 0.0; |
631 | d2 = d*d; |
632 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
633 | |
634 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
635 | |
636 | /* Evaluate switch function */ |
637 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
638 | felec = felec*sw - rinv20*velec*dsw; |
639 | velec *= sw; |
640 | |
641 | /* Update potential sums from outer loop */ |
642 | velecsum += velec; |
643 | |
644 | fscal = felec; |
645 | |
646 | /* Calculate temporary vectorial force */ |
647 | tx = fscal*dx20; |
648 | ty = fscal*dy20; |
649 | tz = fscal*dz20; |
650 | |
651 | /* Update vectorial force */ |
652 | fix2 += tx; |
653 | fiy2 += ty; |
654 | fiz2 += tz; |
655 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
656 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
657 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
658 | |
659 | } |
660 | |
661 | /************************** |
662 | * CALCULATE INTERACTIONS * |
663 | **************************/ |
664 | |
665 | if (rsq21<rcutoff2) |
666 | { |
667 | |
668 | r21 = rsq21*rinv21; |
669 | |
670 | /* EWALD ELECTROSTATICS */ |
671 | |
672 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
673 | ewrt = r21*ewtabscale; |
674 | ewitab = ewrt; |
675 | eweps = ewrt-ewitab; |
676 | ewitab = 4*ewitab; |
677 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
678 | velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
679 | felec = qq21*rinv21*(rinvsq21-felec); |
680 | |
681 | d = r21-rswitch; |
682 | d = (d>0.0) ? d : 0.0; |
683 | d2 = d*d; |
684 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
685 | |
686 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
687 | |
688 | /* Evaluate switch function */ |
689 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
690 | felec = felec*sw - rinv21*velec*dsw; |
691 | velec *= sw; |
692 | |
693 | /* Update potential sums from outer loop */ |
694 | velecsum += velec; |
695 | |
696 | fscal = felec; |
697 | |
698 | /* Calculate temporary vectorial force */ |
699 | tx = fscal*dx21; |
700 | ty = fscal*dy21; |
701 | tz = fscal*dz21; |
702 | |
703 | /* Update vectorial force */ |
704 | fix2 += tx; |
705 | fiy2 += ty; |
706 | fiz2 += tz; |
707 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
708 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
709 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
710 | |
711 | } |
712 | |
713 | /************************** |
714 | * CALCULATE INTERACTIONS * |
715 | **************************/ |
716 | |
717 | if (rsq22<rcutoff2) |
718 | { |
719 | |
720 | r22 = rsq22*rinv22; |
721 | |
722 | /* EWALD ELECTROSTATICS */ |
723 | |
724 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
725 | ewrt = r22*ewtabscale; |
726 | ewitab = ewrt; |
727 | eweps = ewrt-ewitab; |
728 | ewitab = 4*ewitab; |
729 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
730 | velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
731 | felec = qq22*rinv22*(rinvsq22-felec); |
732 | |
733 | d = r22-rswitch; |
734 | d = (d>0.0) ? d : 0.0; |
735 | d2 = d*d; |
736 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
737 | |
738 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
739 | |
740 | /* Evaluate switch function */ |
741 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
742 | felec = felec*sw - rinv22*velec*dsw; |
743 | velec *= sw; |
744 | |
745 | /* Update potential sums from outer loop */ |
746 | velecsum += velec; |
747 | |
748 | fscal = felec; |
749 | |
750 | /* Calculate temporary vectorial force */ |
751 | tx = fscal*dx22; |
752 | ty = fscal*dy22; |
753 | tz = fscal*dz22; |
754 | |
755 | /* Update vectorial force */ |
756 | fix2 += tx; |
757 | fiy2 += ty; |
758 | fiz2 += tz; |
759 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
760 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
761 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
762 | |
763 | } |
764 | |
765 | /* Inner loop uses 538 flops */ |
766 | } |
767 | /* End of innermost loop */ |
768 | |
769 | tx = ty = tz = 0; |
770 | f[i_coord_offset+DIM3*0+XX0] += fix0; |
771 | f[i_coord_offset+DIM3*0+YY1] += fiy0; |
772 | f[i_coord_offset+DIM3*0+ZZ2] += fiz0; |
773 | tx += fix0; |
774 | ty += fiy0; |
775 | tz += fiz0; |
776 | f[i_coord_offset+DIM3*1+XX0] += fix1; |
777 | f[i_coord_offset+DIM3*1+YY1] += fiy1; |
778 | f[i_coord_offset+DIM3*1+ZZ2] += fiz1; |
779 | tx += fix1; |
780 | ty += fiy1; |
781 | tz += fiz1; |
782 | f[i_coord_offset+DIM3*2+XX0] += fix2; |
783 | f[i_coord_offset+DIM3*2+YY1] += fiy2; |
784 | f[i_coord_offset+DIM3*2+ZZ2] += fiz2; |
785 | tx += fix2; |
786 | ty += fiy2; |
787 | tz += fiz2; |
788 | fshift[i_shift_offset+XX0] += tx; |
789 | fshift[i_shift_offset+YY1] += ty; |
790 | fshift[i_shift_offset+ZZ2] += tz; |
791 | |
792 | ggid = gid[iidx]; |
793 | /* Update potential energies */ |
794 | kernel_data->energygrp_elec[ggid] += velecsum; |
795 | kernel_data->energygrp_vdw[ggid] += vvdwsum; |
796 | |
797 | /* Increment number of inner iterations */ |
798 | inneriter += j_index_end - j_index_start; |
799 | |
800 | /* Outer loop uses 32 flops */ |
801 | } |
802 | |
803 | /* Increment number of outer iterations */ |
804 | outeriter += nri; |
805 | |
806 | /* Update outer/inner flops */ |
807 | |
808 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*32 + inneriter*538)(nrnb)->n[eNR_NBKERNEL_ELEC_VDW_W3W3_VF] += outeriter*32 + inneriter*538; |
809 | } |
810 | /* |
811 | * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_F_c |
812 | * Electrostatics interaction: Ewald |
813 | * VdW interaction: LennardJones |
814 | * Geometry: Water3-Water3 |
815 | * Calculate force/pot: Force |
816 | */ |
817 | void |
818 | nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_F_c |
819 | (t_nblist * gmx_restrict__restrict nlist, |
820 | rvec * gmx_restrict__restrict xx, |
821 | rvec * gmx_restrict__restrict ff, |
822 | t_forcerec * gmx_restrict__restrict fr, |
823 | t_mdatoms * gmx_restrict__restrict mdatoms, |
824 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict__restrict kernel_data, |
825 | t_nrnb * gmx_restrict__restrict nrnb) |
826 | { |
827 | int i_shift_offset,i_coord_offset,j_coord_offset; |
828 | int j_index_start,j_index_end; |
829 | int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter; |
830 | real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2; |
831 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
832 | real *shiftvec,*fshift,*x,*f; |
833 | int vdwioffset0; |
834 | real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0; |
835 | int vdwioffset1; |
836 | real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
837 | int vdwioffset2; |
838 | real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
839 | int vdwjidx0; |
840 | real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
841 | int vdwjidx1; |
842 | real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
843 | int vdwjidx2; |
844 | real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
845 | real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00; |
846 | real dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01; |
847 | real dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02; |
848 | real dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10; |
849 | real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11; |
850 | real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12; |
851 | real dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20; |
852 | real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21; |
853 | real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22; |
854 | real velec,felec,velecsum,facel,crf,krf,krf2; |
855 | real *charge; |
856 | int nvdwtype; |
857 | real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6; |
858 | int *vdwtype; |
859 | real *vdwparam; |
860 | int ewitab; |
861 | real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace; |
862 | real *ewtab; |
863 | real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw; |
864 | |
865 | x = xx[0]; |
866 | f = ff[0]; |
867 | |
868 | nri = nlist->nri; |
869 | iinr = nlist->iinr; |
870 | jindex = nlist->jindex; |
871 | jjnr = nlist->jjnr; |
872 | shiftidx = nlist->shift; |
873 | gid = nlist->gid; |
874 | shiftvec = fr->shift_vec[0]; |
875 | fshift = fr->fshift[0]; |
876 | facel = fr->epsfac; |
877 | charge = mdatoms->chargeA; |
878 | nvdwtype = fr->ntype; |
879 | vdwparam = fr->nbfp; |
880 | vdwtype = mdatoms->typeA; |
881 | |
882 | sh_ewald = fr->ic->sh_ewald; |
Value stored to 'sh_ewald' is never read | |
883 | ewtab = fr->ic->tabq_coul_FDV0; |
884 | ewtabscale = fr->ic->tabq_scale; |
885 | ewtabhalfspace = 0.5/ewtabscale; |
886 | |
887 | /* Setup water-specific parameters */ |
888 | inr = nlist->iinr[0]; |
889 | iq0 = facel*charge[inr+0]; |
890 | iq1 = facel*charge[inr+1]; |
891 | iq2 = facel*charge[inr+2]; |
892 | vdwioffset0 = 2*nvdwtype*vdwtype[inr+0]; |
893 | |
894 | jq0 = charge[inr+0]; |
895 | jq1 = charge[inr+1]; |
896 | jq2 = charge[inr+2]; |
897 | vdwjidx0 = 2*vdwtype[inr+0]; |
898 | qq00 = iq0*jq0; |
899 | c6_00 = vdwparam[vdwioffset0+vdwjidx0]; |
900 | c12_00 = vdwparam[vdwioffset0+vdwjidx0+1]; |
901 | qq01 = iq0*jq1; |
902 | qq02 = iq0*jq2; |
903 | qq10 = iq1*jq0; |
904 | qq11 = iq1*jq1; |
905 | qq12 = iq1*jq2; |
906 | qq20 = iq2*jq0; |
907 | qq21 = iq2*jq1; |
908 | qq22 = iq2*jq2; |
909 | |
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; |
913 | |
914 | rswitch = fr->rcoulomb_switch; |
915 | /* Setup switch parameters */ |
916 | d = rcutoff-rswitch; |
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); |
923 | |
924 | outeriter = 0; |
925 | inneriter = 0; |
926 | |
927 | /* Start outer loop over neighborlists */ |
928 | for(iidx=0; iidx<nri; iidx++) |
929 | { |
930 | /* Load shift vector for this list */ |
931 | i_shift_offset = DIM3*shiftidx[iidx]; |
932 | shX = shiftvec[i_shift_offset+XX0]; |
933 | shY = shiftvec[i_shift_offset+YY1]; |
934 | shZ = shiftvec[i_shift_offset+ZZ2]; |
935 | |
936 | /* Load limits for loop over neighbors */ |
937 | j_index_start = jindex[iidx]; |
938 | j_index_end = jindex[iidx+1]; |
939 | |
940 | /* Get outer coordinate index */ |
941 | inr = iinr[iidx]; |
942 | i_coord_offset = DIM3*inr; |
943 | |
944 | /* Load i particle coords and add shift vector */ |
945 | ix0 = shX + x[i_coord_offset+DIM3*0+XX0]; |
946 | iy0 = shY + x[i_coord_offset+DIM3*0+YY1]; |
947 | iz0 = shZ + x[i_coord_offset+DIM3*0+ZZ2]; |
948 | ix1 = shX + x[i_coord_offset+DIM3*1+XX0]; |
949 | iy1 = shY + x[i_coord_offset+DIM3*1+YY1]; |
950 | iz1 = shZ + x[i_coord_offset+DIM3*1+ZZ2]; |
951 | ix2 = shX + x[i_coord_offset+DIM3*2+XX0]; |
952 | iy2 = shY + x[i_coord_offset+DIM3*2+YY1]; |
953 | iz2 = shZ + x[i_coord_offset+DIM3*2+ZZ2]; |
954 | |
955 | fix0 = 0.0; |
956 | fiy0 = 0.0; |
957 | fiz0 = 0.0; |
958 | fix1 = 0.0; |
959 | fiy1 = 0.0; |
960 | fiz1 = 0.0; |
961 | fix2 = 0.0; |
962 | fiy2 = 0.0; |
963 | fiz2 = 0.0; |
964 | |
965 | /* Start inner kernel loop */ |
966 | for(jidx=j_index_start; jidx<j_index_end; jidx++) |
967 | { |
968 | /* Get j neighbor index, and coordinate index */ |
969 | jnr = jjnr[jidx]; |
970 | j_coord_offset = DIM3*jnr; |
971 | |
972 | /* load j atom coordinates */ |
973 | jx0 = x[j_coord_offset+DIM3*0+XX0]; |
974 | jy0 = x[j_coord_offset+DIM3*0+YY1]; |
975 | jz0 = x[j_coord_offset+DIM3*0+ZZ2]; |
976 | jx1 = x[j_coord_offset+DIM3*1+XX0]; |
977 | jy1 = x[j_coord_offset+DIM3*1+YY1]; |
978 | jz1 = x[j_coord_offset+DIM3*1+ZZ2]; |
979 | jx2 = x[j_coord_offset+DIM3*2+XX0]; |
980 | jy2 = x[j_coord_offset+DIM3*2+YY1]; |
981 | jz2 = x[j_coord_offset+DIM3*2+ZZ2]; |
982 | |
983 | /* Calculate displacement vector */ |
984 | dx00 = ix0 - jx0; |
985 | dy00 = iy0 - jy0; |
986 | dz00 = iz0 - jz0; |
987 | dx01 = ix0 - jx1; |
988 | dy01 = iy0 - jy1; |
989 | dz01 = iz0 - jz1; |
990 | dx02 = ix0 - jx2; |
991 | dy02 = iy0 - jy2; |
992 | dz02 = iz0 - jz2; |
993 | dx10 = ix1 - jx0; |
994 | dy10 = iy1 - jy0; |
995 | dz10 = iz1 - jz0; |
996 | dx11 = ix1 - jx1; |
997 | dy11 = iy1 - jy1; |
998 | dz11 = iz1 - jz1; |
999 | dx12 = ix1 - jx2; |
1000 | dy12 = iy1 - jy2; |
1001 | dz12 = iz1 - jz2; |
1002 | dx20 = ix2 - jx0; |
1003 | dy20 = iy2 - jy0; |
1004 | dz20 = iz2 - jz0; |
1005 | dx21 = ix2 - jx1; |
1006 | dy21 = iy2 - jy1; |
1007 | dz21 = iz2 - jz1; |
1008 | dx22 = ix2 - jx2; |
1009 | dy22 = iy2 - jy2; |
1010 | dz22 = iz2 - jz2; |
1011 | |
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; |
1022 | |
1023 | rinv00 = gmx_invsqrt(rsq00)gmx_software_invsqrt(rsq00); |
1024 | rinv01 = gmx_invsqrt(rsq01)gmx_software_invsqrt(rsq01); |
1025 | rinv02 = gmx_invsqrt(rsq02)gmx_software_invsqrt(rsq02); |
1026 | rinv10 = gmx_invsqrt(rsq10)gmx_software_invsqrt(rsq10); |
1027 | rinv11 = gmx_invsqrt(rsq11)gmx_software_invsqrt(rsq11); |
1028 | rinv12 = gmx_invsqrt(rsq12)gmx_software_invsqrt(rsq12); |
1029 | rinv20 = gmx_invsqrt(rsq20)gmx_software_invsqrt(rsq20); |
1030 | rinv21 = gmx_invsqrt(rsq21)gmx_software_invsqrt(rsq21); |
1031 | rinv22 = gmx_invsqrt(rsq22)gmx_software_invsqrt(rsq22); |
1032 | |
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; |
1042 | |
1043 | /************************** |
1044 | * CALCULATE INTERACTIONS * |
1045 | **************************/ |
1046 | |
1047 | if (rsq00<rcutoff2) |
1048 | { |
1049 | |
1050 | r00 = rsq00*rinv00; |
1051 | |
1052 | /* EWALD ELECTROSTATICS */ |
1053 | |
1054 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1055 | ewrt = r00*ewtabscale; |
1056 | ewitab = ewrt; |
1057 | eweps = ewrt-ewitab; |
1058 | ewitab = 4*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); |
1062 | |
1063 | /* LENNARD-JONES DISPERSION/REPULSION */ |
1064 | |
1065 | rinvsix = rinvsq00*rinvsq00*rinvsq00; |
1066 | vvdw6 = c6_00*rinvsix; |
1067 | vvdw12 = c12_00*rinvsix*rinvsix; |
1068 | vvdw = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0); |
1069 | fvdw = (vvdw12-vvdw6)*rinvsq00; |
1070 | |
1071 | d = r00-rswitch; |
1072 | d = (d>0.0) ? d : 0.0; |
1073 | d2 = d*d; |
1074 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1075 | |
1076 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1077 | |
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; |
1082 | |
1083 | fscal = felec+fvdw; |
1084 | |
1085 | /* Calculate temporary vectorial force */ |
1086 | tx = fscal*dx00; |
1087 | ty = fscal*dy00; |
1088 | tz = fscal*dz00; |
1089 | |
1090 | /* Update vectorial force */ |
1091 | fix0 += tx; |
1092 | fiy0 += ty; |
1093 | fiz0 += tz; |
1094 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
1095 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
1096 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
1097 | |
1098 | } |
1099 | |
1100 | /************************** |
1101 | * CALCULATE INTERACTIONS * |
1102 | **************************/ |
1103 | |
1104 | if (rsq01<rcutoff2) |
1105 | { |
1106 | |
1107 | r01 = rsq01*rinv01; |
1108 | |
1109 | /* EWALD ELECTROSTATICS */ |
1110 | |
1111 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1112 | ewrt = r01*ewtabscale; |
1113 | ewitab = ewrt; |
1114 | eweps = ewrt-ewitab; |
1115 | ewitab = 4*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); |
1119 | |
1120 | d = r01-rswitch; |
1121 | d = (d>0.0) ? d : 0.0; |
1122 | d2 = d*d; |
1123 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1124 | |
1125 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1126 | |
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; |
1130 | |
1131 | fscal = felec; |
1132 | |
1133 | /* Calculate temporary vectorial force */ |
1134 | tx = fscal*dx01; |
1135 | ty = fscal*dy01; |
1136 | tz = fscal*dz01; |
1137 | |
1138 | /* Update vectorial force */ |
1139 | fix0 += tx; |
1140 | fiy0 += ty; |
1141 | fiz0 += tz; |
1142 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
1143 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
1144 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
1145 | |
1146 | } |
1147 | |
1148 | /************************** |
1149 | * CALCULATE INTERACTIONS * |
1150 | **************************/ |
1151 | |
1152 | if (rsq02<rcutoff2) |
1153 | { |
1154 | |
1155 | r02 = rsq02*rinv02; |
1156 | |
1157 | /* EWALD ELECTROSTATICS */ |
1158 | |
1159 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1160 | ewrt = r02*ewtabscale; |
1161 | ewitab = ewrt; |
1162 | eweps = ewrt-ewitab; |
1163 | ewitab = 4*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); |
1167 | |
1168 | d = r02-rswitch; |
1169 | d = (d>0.0) ? d : 0.0; |
1170 | d2 = d*d; |
1171 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1172 | |
1173 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1174 | |
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; |
1178 | |
1179 | fscal = felec; |
1180 | |
1181 | /* Calculate temporary vectorial force */ |
1182 | tx = fscal*dx02; |
1183 | ty = fscal*dy02; |
1184 | tz = fscal*dz02; |
1185 | |
1186 | /* Update vectorial force */ |
1187 | fix0 += tx; |
1188 | fiy0 += ty; |
1189 | fiz0 += tz; |
1190 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
1191 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
1192 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
1193 | |
1194 | } |
1195 | |
1196 | /************************** |
1197 | * CALCULATE INTERACTIONS * |
1198 | **************************/ |
1199 | |
1200 | if (rsq10<rcutoff2) |
1201 | { |
1202 | |
1203 | r10 = rsq10*rinv10; |
1204 | |
1205 | /* EWALD ELECTROSTATICS */ |
1206 | |
1207 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1208 | ewrt = r10*ewtabscale; |
1209 | ewitab = ewrt; |
1210 | eweps = ewrt-ewitab; |
1211 | ewitab = 4*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); |
1215 | |
1216 | d = r10-rswitch; |
1217 | d = (d>0.0) ? d : 0.0; |
1218 | d2 = d*d; |
1219 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1220 | |
1221 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1222 | |
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; |
1226 | |
1227 | fscal = felec; |
1228 | |
1229 | /* Calculate temporary vectorial force */ |
1230 | tx = fscal*dx10; |
1231 | ty = fscal*dy10; |
1232 | tz = fscal*dz10; |
1233 | |
1234 | /* Update vectorial force */ |
1235 | fix1 += tx; |
1236 | fiy1 += ty; |
1237 | fiz1 += tz; |
1238 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
1239 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
1240 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
1241 | |
1242 | } |
1243 | |
1244 | /************************** |
1245 | * CALCULATE INTERACTIONS * |
1246 | **************************/ |
1247 | |
1248 | if (rsq11<rcutoff2) |
1249 | { |
1250 | |
1251 | r11 = rsq11*rinv11; |
1252 | |
1253 | /* EWALD ELECTROSTATICS */ |
1254 | |
1255 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1256 | ewrt = r11*ewtabscale; |
1257 | ewitab = ewrt; |
1258 | eweps = ewrt-ewitab; |
1259 | ewitab = 4*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); |
1263 | |
1264 | d = r11-rswitch; |
1265 | d = (d>0.0) ? d : 0.0; |
1266 | d2 = d*d; |
1267 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1268 | |
1269 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1270 | |
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; |
1274 | |
1275 | fscal = felec; |
1276 | |
1277 | /* Calculate temporary vectorial force */ |
1278 | tx = fscal*dx11; |
1279 | ty = fscal*dy11; |
1280 | tz = fscal*dz11; |
1281 | |
1282 | /* Update vectorial force */ |
1283 | fix1 += tx; |
1284 | fiy1 += ty; |
1285 | fiz1 += tz; |
1286 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
1287 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
1288 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
1289 | |
1290 | } |
1291 | |
1292 | /************************** |
1293 | * CALCULATE INTERACTIONS * |
1294 | **************************/ |
1295 | |
1296 | if (rsq12<rcutoff2) |
1297 | { |
1298 | |
1299 | r12 = rsq12*rinv12; |
1300 | |
1301 | /* EWALD ELECTROSTATICS */ |
1302 | |
1303 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1304 | ewrt = r12*ewtabscale; |
1305 | ewitab = ewrt; |
1306 | eweps = ewrt-ewitab; |
1307 | ewitab = 4*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); |
1311 | |
1312 | d = r12-rswitch; |
1313 | d = (d>0.0) ? d : 0.0; |
1314 | d2 = d*d; |
1315 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1316 | |
1317 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1318 | |
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; |
1322 | |
1323 | fscal = felec; |
1324 | |
1325 | /* Calculate temporary vectorial force */ |
1326 | tx = fscal*dx12; |
1327 | ty = fscal*dy12; |
1328 | tz = fscal*dz12; |
1329 | |
1330 | /* Update vectorial force */ |
1331 | fix1 += tx; |
1332 | fiy1 += ty; |
1333 | fiz1 += tz; |
1334 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
1335 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
1336 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
1337 | |
1338 | } |
1339 | |
1340 | /************************** |
1341 | * CALCULATE INTERACTIONS * |
1342 | **************************/ |
1343 | |
1344 | if (rsq20<rcutoff2) |
1345 | { |
1346 | |
1347 | r20 = rsq20*rinv20; |
1348 | |
1349 | /* EWALD ELECTROSTATICS */ |
1350 | |
1351 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1352 | ewrt = r20*ewtabscale; |
1353 | ewitab = ewrt; |
1354 | eweps = ewrt-ewitab; |
1355 | ewitab = 4*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); |
1359 | |
1360 | d = r20-rswitch; |
1361 | d = (d>0.0) ? d : 0.0; |
1362 | d2 = d*d; |
1363 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1364 | |
1365 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1366 | |
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; |
1370 | |
1371 | fscal = felec; |
1372 | |
1373 | /* Calculate temporary vectorial force */ |
1374 | tx = fscal*dx20; |
1375 | ty = fscal*dy20; |
1376 | tz = fscal*dz20; |
1377 | |
1378 | /* Update vectorial force */ |
1379 | fix2 += tx; |
1380 | fiy2 += ty; |
1381 | fiz2 += tz; |
1382 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
1383 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
1384 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
1385 | |
1386 | } |
1387 | |
1388 | /************************** |
1389 | * CALCULATE INTERACTIONS * |
1390 | **************************/ |
1391 | |
1392 | if (rsq21<rcutoff2) |
1393 | { |
1394 | |
1395 | r21 = rsq21*rinv21; |
1396 | |
1397 | /* EWALD ELECTROSTATICS */ |
1398 | |
1399 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1400 | ewrt = r21*ewtabscale; |
1401 | ewitab = ewrt; |
1402 | eweps = ewrt-ewitab; |
1403 | ewitab = 4*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); |
1407 | |
1408 | d = r21-rswitch; |
1409 | d = (d>0.0) ? d : 0.0; |
1410 | d2 = d*d; |
1411 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1412 | |
1413 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1414 | |
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; |
1418 | |
1419 | fscal = felec; |
1420 | |
1421 | /* Calculate temporary vectorial force */ |
1422 | tx = fscal*dx21; |
1423 | ty = fscal*dy21; |
1424 | tz = fscal*dz21; |
1425 | |
1426 | /* Update vectorial force */ |
1427 | fix2 += tx; |
1428 | fiy2 += ty; |
1429 | fiz2 += tz; |
1430 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
1431 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
1432 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
1433 | |
1434 | } |
1435 | |
1436 | /************************** |
1437 | * CALCULATE INTERACTIONS * |
1438 | **************************/ |
1439 | |
1440 | if (rsq22<rcutoff2) |
1441 | { |
1442 | |
1443 | r22 = rsq22*rinv22; |
1444 | |
1445 | /* EWALD ELECTROSTATICS */ |
1446 | |
1447 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1448 | ewrt = r22*ewtabscale; |
1449 | ewitab = ewrt; |
1450 | eweps = ewrt-ewitab; |
1451 | ewitab = 4*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); |
1455 | |
1456 | d = r22-rswitch; |
1457 | d = (d>0.0) ? d : 0.0; |
1458 | d2 = d*d; |
1459 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1460 | |
1461 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1462 | |
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; |
1466 | |
1467 | fscal = felec; |
1468 | |
1469 | /* Calculate temporary vectorial force */ |
1470 | tx = fscal*dx22; |
1471 | ty = fscal*dy22; |
1472 | tz = fscal*dz22; |
1473 | |
1474 | /* Update vectorial force */ |
1475 | fix2 += tx; |
1476 | fiy2 += ty; |
1477 | fiz2 += tz; |
1478 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
1479 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
1480 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
1481 | |
1482 | } |
1483 | |
1484 | /* Inner loop uses 518 flops */ |
1485 | } |
1486 | /* End of innermost loop */ |
1487 | |
1488 | tx = ty = tz = 0; |
1489 | f[i_coord_offset+DIM3*0+XX0] += fix0; |
1490 | f[i_coord_offset+DIM3*0+YY1] += fiy0; |
1491 | f[i_coord_offset+DIM3*0+ZZ2] += fiz0; |
1492 | tx += fix0; |
1493 | ty += fiy0; |
1494 | tz += fiz0; |
1495 | f[i_coord_offset+DIM3*1+XX0] += fix1; |
1496 | f[i_coord_offset+DIM3*1+YY1] += fiy1; |
1497 | f[i_coord_offset+DIM3*1+ZZ2] += fiz1; |
1498 | tx += fix1; |
1499 | ty += fiy1; |
1500 | tz += fiz1; |
1501 | f[i_coord_offset+DIM3*2+XX0] += fix2; |
1502 | f[i_coord_offset+DIM3*2+YY1] += fiy2; |
1503 | f[i_coord_offset+DIM3*2+ZZ2] += fiz2; |
1504 | tx += fix2; |
1505 | ty += fiy2; |
1506 | tz += fiz2; |
1507 | fshift[i_shift_offset+XX0] += tx; |
1508 | fshift[i_shift_offset+YY1] += ty; |
1509 | fshift[i_shift_offset+ZZ2] += tz; |
1510 | |
1511 | /* Increment number of inner iterations */ |
1512 | inneriter += j_index_end - j_index_start; |
1513 | |
1514 | /* Outer loop uses 30 flops */ |
1515 | } |
1516 | |
1517 | /* Increment number of outer iterations */ |
1518 | outeriter += nri; |
1519 | |
1520 | /* Update outer/inner flops */ |
1521 | |
1522 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*30 + inneriter*518)(nrnb)->n[eNR_NBKERNEL_ELEC_VDW_W3W3_F] += outeriter*30 + inneriter *518; |
1523 | } |