File: | gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_c.c |
Location: | line 126, 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_VdwBhamSw_GeomW4W4_VF_c |
51 | * Electrostatics interaction: Ewald |
52 | * VdW interaction: Buckingham |
53 | * Geometry: Water4-Water4 |
54 | * Calculate force/pot: PotentialAndForce |
55 | */ |
56 | void |
57 | nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_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 vdwioffset3; |
79 | real ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3; |
80 | int vdwjidx0; |
81 | real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
82 | int vdwjidx1; |
83 | real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
84 | int vdwjidx2; |
85 | real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
86 | int vdwjidx3; |
87 | real jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3; |
88 | real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00; |
89 | real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11; |
90 | real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12; |
91 | real dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13; |
92 | real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21; |
93 | real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22; |
94 | real dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23; |
95 | real dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31; |
96 | real dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32; |
97 | real dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33; |
98 | real velec,felec,velecsum,facel,crf,krf,krf2; |
99 | real *charge; |
100 | int nvdwtype; |
101 | real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6; |
102 | int *vdwtype; |
103 | real *vdwparam; |
104 | int ewitab; |
105 | real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace; |
106 | real *ewtab; |
107 | real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw; |
108 | |
109 | x = xx[0]; |
110 | f = ff[0]; |
111 | |
112 | nri = nlist->nri; |
113 | iinr = nlist->iinr; |
114 | jindex = nlist->jindex; |
115 | jjnr = nlist->jjnr; |
116 | shiftidx = nlist->shift; |
117 | gid = nlist->gid; |
118 | shiftvec = fr->shift_vec[0]; |
119 | fshift = fr->fshift[0]; |
120 | facel = fr->epsfac; |
121 | charge = mdatoms->chargeA; |
122 | nvdwtype = fr->ntype; |
123 | vdwparam = fr->nbfp; |
124 | vdwtype = mdatoms->typeA; |
125 | |
126 | sh_ewald = fr->ic->sh_ewald; |
Value stored to 'sh_ewald' is never read | |
127 | ewtab = fr->ic->tabq_coul_FDV0; |
128 | ewtabscale = fr->ic->tabq_scale; |
129 | ewtabhalfspace = 0.5/ewtabscale; |
130 | |
131 | /* Setup water-specific parameters */ |
132 | inr = nlist->iinr[0]; |
133 | iq1 = facel*charge[inr+1]; |
134 | iq2 = facel*charge[inr+2]; |
135 | iq3 = facel*charge[inr+3]; |
136 | vdwioffset0 = 3*nvdwtype*vdwtype[inr+0]; |
137 | |
138 | jq1 = charge[inr+1]; |
139 | jq2 = charge[inr+2]; |
140 | jq3 = charge[inr+3]; |
141 | vdwjidx0 = 3*vdwtype[inr+0]; |
142 | c6_00 = vdwparam[vdwioffset0+vdwjidx0]; |
143 | cexp1_00 = vdwparam[vdwioffset0+vdwjidx0+1]; |
144 | cexp2_00 = vdwparam[vdwioffset0+vdwjidx0+2]; |
145 | qq11 = iq1*jq1; |
146 | qq12 = iq1*jq2; |
147 | qq13 = iq1*jq3; |
148 | qq21 = iq2*jq1; |
149 | qq22 = iq2*jq2; |
150 | qq23 = iq2*jq3; |
151 | qq31 = iq3*jq1; |
152 | qq32 = iq3*jq2; |
153 | qq33 = iq3*jq3; |
154 | |
155 | /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */ |
156 | rcutoff = fr->rcoulomb; |
157 | rcutoff2 = rcutoff*rcutoff; |
158 | |
159 | rswitch = fr->rcoulomb_switch; |
160 | /* Setup switch parameters */ |
161 | d = rcutoff-rswitch; |
162 | swV3 = -10.0/(d*d*d); |
163 | swV4 = 15.0/(d*d*d*d); |
164 | swV5 = -6.0/(d*d*d*d*d); |
165 | swF2 = -30.0/(d*d*d); |
166 | swF3 = 60.0/(d*d*d*d); |
167 | swF4 = -30.0/(d*d*d*d*d); |
168 | |
169 | outeriter = 0; |
170 | inneriter = 0; |
171 | |
172 | /* Start outer loop over neighborlists */ |
173 | for(iidx=0; iidx<nri; iidx++) |
174 | { |
175 | /* Load shift vector for this list */ |
176 | i_shift_offset = DIM3*shiftidx[iidx]; |
177 | shX = shiftvec[i_shift_offset+XX0]; |
178 | shY = shiftvec[i_shift_offset+YY1]; |
179 | shZ = shiftvec[i_shift_offset+ZZ2]; |
180 | |
181 | /* Load limits for loop over neighbors */ |
182 | j_index_start = jindex[iidx]; |
183 | j_index_end = jindex[iidx+1]; |
184 | |
185 | /* Get outer coordinate index */ |
186 | inr = iinr[iidx]; |
187 | i_coord_offset = DIM3*inr; |
188 | |
189 | /* Load i particle coords and add shift vector */ |
190 | ix0 = shX + x[i_coord_offset+DIM3*0+XX0]; |
191 | iy0 = shY + x[i_coord_offset+DIM3*0+YY1]; |
192 | iz0 = shZ + x[i_coord_offset+DIM3*0+ZZ2]; |
193 | ix1 = shX + x[i_coord_offset+DIM3*1+XX0]; |
194 | iy1 = shY + x[i_coord_offset+DIM3*1+YY1]; |
195 | iz1 = shZ + x[i_coord_offset+DIM3*1+ZZ2]; |
196 | ix2 = shX + x[i_coord_offset+DIM3*2+XX0]; |
197 | iy2 = shY + x[i_coord_offset+DIM3*2+YY1]; |
198 | iz2 = shZ + x[i_coord_offset+DIM3*2+ZZ2]; |
199 | ix3 = shX + x[i_coord_offset+DIM3*3+XX0]; |
200 | iy3 = shY + x[i_coord_offset+DIM3*3+YY1]; |
201 | iz3 = shZ + x[i_coord_offset+DIM3*3+ZZ2]; |
202 | |
203 | fix0 = 0.0; |
204 | fiy0 = 0.0; |
205 | fiz0 = 0.0; |
206 | fix1 = 0.0; |
207 | fiy1 = 0.0; |
208 | fiz1 = 0.0; |
209 | fix2 = 0.0; |
210 | fiy2 = 0.0; |
211 | fiz2 = 0.0; |
212 | fix3 = 0.0; |
213 | fiy3 = 0.0; |
214 | fiz3 = 0.0; |
215 | |
216 | /* Reset potential sums */ |
217 | velecsum = 0.0; |
218 | vvdwsum = 0.0; |
219 | |
220 | /* Start inner kernel loop */ |
221 | for(jidx=j_index_start; jidx<j_index_end; jidx++) |
222 | { |
223 | /* Get j neighbor index, and coordinate index */ |
224 | jnr = jjnr[jidx]; |
225 | j_coord_offset = DIM3*jnr; |
226 | |
227 | /* load j atom coordinates */ |
228 | jx0 = x[j_coord_offset+DIM3*0+XX0]; |
229 | jy0 = x[j_coord_offset+DIM3*0+YY1]; |
230 | jz0 = x[j_coord_offset+DIM3*0+ZZ2]; |
231 | jx1 = x[j_coord_offset+DIM3*1+XX0]; |
232 | jy1 = x[j_coord_offset+DIM3*1+YY1]; |
233 | jz1 = x[j_coord_offset+DIM3*1+ZZ2]; |
234 | jx2 = x[j_coord_offset+DIM3*2+XX0]; |
235 | jy2 = x[j_coord_offset+DIM3*2+YY1]; |
236 | jz2 = x[j_coord_offset+DIM3*2+ZZ2]; |
237 | jx3 = x[j_coord_offset+DIM3*3+XX0]; |
238 | jy3 = x[j_coord_offset+DIM3*3+YY1]; |
239 | jz3 = x[j_coord_offset+DIM3*3+ZZ2]; |
240 | |
241 | /* Calculate displacement vector */ |
242 | dx00 = ix0 - jx0; |
243 | dy00 = iy0 - jy0; |
244 | dz00 = iz0 - jz0; |
245 | dx11 = ix1 - jx1; |
246 | dy11 = iy1 - jy1; |
247 | dz11 = iz1 - jz1; |
248 | dx12 = ix1 - jx2; |
249 | dy12 = iy1 - jy2; |
250 | dz12 = iz1 - jz2; |
251 | dx13 = ix1 - jx3; |
252 | dy13 = iy1 - jy3; |
253 | dz13 = iz1 - jz3; |
254 | dx21 = ix2 - jx1; |
255 | dy21 = iy2 - jy1; |
256 | dz21 = iz2 - jz1; |
257 | dx22 = ix2 - jx2; |
258 | dy22 = iy2 - jy2; |
259 | dz22 = iz2 - jz2; |
260 | dx23 = ix2 - jx3; |
261 | dy23 = iy2 - jy3; |
262 | dz23 = iz2 - jz3; |
263 | dx31 = ix3 - jx1; |
264 | dy31 = iy3 - jy1; |
265 | dz31 = iz3 - jz1; |
266 | dx32 = ix3 - jx2; |
267 | dy32 = iy3 - jy2; |
268 | dz32 = iz3 - jz2; |
269 | dx33 = ix3 - jx3; |
270 | dy33 = iy3 - jy3; |
271 | dz33 = iz3 - jz3; |
272 | |
273 | /* Calculate squared distance and things based on it */ |
274 | rsq00 = dx00*dx00+dy00*dy00+dz00*dz00; |
275 | rsq11 = dx11*dx11+dy11*dy11+dz11*dz11; |
276 | rsq12 = dx12*dx12+dy12*dy12+dz12*dz12; |
277 | rsq13 = dx13*dx13+dy13*dy13+dz13*dz13; |
278 | rsq21 = dx21*dx21+dy21*dy21+dz21*dz21; |
279 | rsq22 = dx22*dx22+dy22*dy22+dz22*dz22; |
280 | rsq23 = dx23*dx23+dy23*dy23+dz23*dz23; |
281 | rsq31 = dx31*dx31+dy31*dy31+dz31*dz31; |
282 | rsq32 = dx32*dx32+dy32*dy32+dz32*dz32; |
283 | rsq33 = dx33*dx33+dy33*dy33+dz33*dz33; |
284 | |
285 | rinv00 = gmx_invsqrt(rsq00)gmx_software_invsqrt(rsq00); |
286 | rinv11 = gmx_invsqrt(rsq11)gmx_software_invsqrt(rsq11); |
287 | rinv12 = gmx_invsqrt(rsq12)gmx_software_invsqrt(rsq12); |
288 | rinv13 = gmx_invsqrt(rsq13)gmx_software_invsqrt(rsq13); |
289 | rinv21 = gmx_invsqrt(rsq21)gmx_software_invsqrt(rsq21); |
290 | rinv22 = gmx_invsqrt(rsq22)gmx_software_invsqrt(rsq22); |
291 | rinv23 = gmx_invsqrt(rsq23)gmx_software_invsqrt(rsq23); |
292 | rinv31 = gmx_invsqrt(rsq31)gmx_software_invsqrt(rsq31); |
293 | rinv32 = gmx_invsqrt(rsq32)gmx_software_invsqrt(rsq32); |
294 | rinv33 = gmx_invsqrt(rsq33)gmx_software_invsqrt(rsq33); |
295 | |
296 | rinvsq00 = rinv00*rinv00; |
297 | rinvsq11 = rinv11*rinv11; |
298 | rinvsq12 = rinv12*rinv12; |
299 | rinvsq13 = rinv13*rinv13; |
300 | rinvsq21 = rinv21*rinv21; |
301 | rinvsq22 = rinv22*rinv22; |
302 | rinvsq23 = rinv23*rinv23; |
303 | rinvsq31 = rinv31*rinv31; |
304 | rinvsq32 = rinv32*rinv32; |
305 | rinvsq33 = rinv33*rinv33; |
306 | |
307 | /************************** |
308 | * CALCULATE INTERACTIONS * |
309 | **************************/ |
310 | |
311 | if (rsq00<rcutoff2) |
312 | { |
313 | |
314 | r00 = rsq00*rinv00; |
315 | |
316 | /* BUCKINGHAM DISPERSION/REPULSION */ |
317 | rinvsix = rinvsq00*rinvsq00*rinvsq00; |
318 | vvdw6 = c6_00*rinvsix; |
319 | br = cexp2_00*r00; |
320 | vvdwexp = cexp1_00*exp(-br); |
321 | vvdw = vvdwexp - vvdw6*(1.0/6.0); |
322 | fvdw = (br*vvdwexp-vvdw6)*rinvsq00; |
323 | |
324 | d = r00-rswitch; |
325 | d = (d>0.0) ? d : 0.0; |
326 | d2 = d*d; |
327 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
328 | |
329 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
330 | |
331 | /* Evaluate switch function */ |
332 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
333 | fvdw = fvdw*sw - rinv00*vvdw*dsw; |
334 | vvdw *= sw; |
335 | |
336 | /* Update potential sums from outer loop */ |
337 | vvdwsum += vvdw; |
338 | |
339 | fscal = fvdw; |
340 | |
341 | /* Calculate temporary vectorial force */ |
342 | tx = fscal*dx00; |
343 | ty = fscal*dy00; |
344 | tz = fscal*dz00; |
345 | |
346 | /* Update vectorial force */ |
347 | fix0 += tx; |
348 | fiy0 += ty; |
349 | fiz0 += tz; |
350 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
351 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
352 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
353 | |
354 | } |
355 | |
356 | /************************** |
357 | * CALCULATE INTERACTIONS * |
358 | **************************/ |
359 | |
360 | if (rsq11<rcutoff2) |
361 | { |
362 | |
363 | r11 = rsq11*rinv11; |
364 | |
365 | /* EWALD ELECTROSTATICS */ |
366 | |
367 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
368 | ewrt = r11*ewtabscale; |
369 | ewitab = ewrt; |
370 | eweps = ewrt-ewitab; |
371 | ewitab = 4*ewitab; |
372 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
373 | velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
374 | felec = qq11*rinv11*(rinvsq11-felec); |
375 | |
376 | d = r11-rswitch; |
377 | d = (d>0.0) ? d : 0.0; |
378 | d2 = d*d; |
379 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
380 | |
381 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
382 | |
383 | /* Evaluate switch function */ |
384 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
385 | felec = felec*sw - rinv11*velec*dsw; |
386 | velec *= sw; |
387 | |
388 | /* Update potential sums from outer loop */ |
389 | velecsum += velec; |
390 | |
391 | fscal = felec; |
392 | |
393 | /* Calculate temporary vectorial force */ |
394 | tx = fscal*dx11; |
395 | ty = fscal*dy11; |
396 | tz = fscal*dz11; |
397 | |
398 | /* Update vectorial force */ |
399 | fix1 += tx; |
400 | fiy1 += ty; |
401 | fiz1 += tz; |
402 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
403 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
404 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
405 | |
406 | } |
407 | |
408 | /************************** |
409 | * CALCULATE INTERACTIONS * |
410 | **************************/ |
411 | |
412 | if (rsq12<rcutoff2) |
413 | { |
414 | |
415 | r12 = rsq12*rinv12; |
416 | |
417 | /* EWALD ELECTROSTATICS */ |
418 | |
419 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
420 | ewrt = r12*ewtabscale; |
421 | ewitab = ewrt; |
422 | eweps = ewrt-ewitab; |
423 | ewitab = 4*ewitab; |
424 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
425 | velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
426 | felec = qq12*rinv12*(rinvsq12-felec); |
427 | |
428 | d = r12-rswitch; |
429 | d = (d>0.0) ? d : 0.0; |
430 | d2 = d*d; |
431 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
432 | |
433 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
434 | |
435 | /* Evaluate switch function */ |
436 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
437 | felec = felec*sw - rinv12*velec*dsw; |
438 | velec *= sw; |
439 | |
440 | /* Update potential sums from outer loop */ |
441 | velecsum += velec; |
442 | |
443 | fscal = felec; |
444 | |
445 | /* Calculate temporary vectorial force */ |
446 | tx = fscal*dx12; |
447 | ty = fscal*dy12; |
448 | tz = fscal*dz12; |
449 | |
450 | /* Update vectorial force */ |
451 | fix1 += tx; |
452 | fiy1 += ty; |
453 | fiz1 += tz; |
454 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
455 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
456 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
457 | |
458 | } |
459 | |
460 | /************************** |
461 | * CALCULATE INTERACTIONS * |
462 | **************************/ |
463 | |
464 | if (rsq13<rcutoff2) |
465 | { |
466 | |
467 | r13 = rsq13*rinv13; |
468 | |
469 | /* EWALD ELECTROSTATICS */ |
470 | |
471 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
472 | ewrt = r13*ewtabscale; |
473 | ewitab = ewrt; |
474 | eweps = ewrt-ewitab; |
475 | ewitab = 4*ewitab; |
476 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
477 | velec = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
478 | felec = qq13*rinv13*(rinvsq13-felec); |
479 | |
480 | d = r13-rswitch; |
481 | d = (d>0.0) ? d : 0.0; |
482 | d2 = d*d; |
483 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
484 | |
485 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
486 | |
487 | /* Evaluate switch function */ |
488 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
489 | felec = felec*sw - rinv13*velec*dsw; |
490 | velec *= sw; |
491 | |
492 | /* Update potential sums from outer loop */ |
493 | velecsum += velec; |
494 | |
495 | fscal = felec; |
496 | |
497 | /* Calculate temporary vectorial force */ |
498 | tx = fscal*dx13; |
499 | ty = fscal*dy13; |
500 | tz = fscal*dz13; |
501 | |
502 | /* Update vectorial force */ |
503 | fix1 += tx; |
504 | fiy1 += ty; |
505 | fiz1 += tz; |
506 | f[j_coord_offset+DIM3*3+XX0] -= tx; |
507 | f[j_coord_offset+DIM3*3+YY1] -= ty; |
508 | f[j_coord_offset+DIM3*3+ZZ2] -= tz; |
509 | |
510 | } |
511 | |
512 | /************************** |
513 | * CALCULATE INTERACTIONS * |
514 | **************************/ |
515 | |
516 | if (rsq21<rcutoff2) |
517 | { |
518 | |
519 | r21 = rsq21*rinv21; |
520 | |
521 | /* EWALD ELECTROSTATICS */ |
522 | |
523 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
524 | ewrt = r21*ewtabscale; |
525 | ewitab = ewrt; |
526 | eweps = ewrt-ewitab; |
527 | ewitab = 4*ewitab; |
528 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
529 | velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
530 | felec = qq21*rinv21*(rinvsq21-felec); |
531 | |
532 | d = r21-rswitch; |
533 | d = (d>0.0) ? d : 0.0; |
534 | d2 = d*d; |
535 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
536 | |
537 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
538 | |
539 | /* Evaluate switch function */ |
540 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
541 | felec = felec*sw - rinv21*velec*dsw; |
542 | velec *= sw; |
543 | |
544 | /* Update potential sums from outer loop */ |
545 | velecsum += velec; |
546 | |
547 | fscal = felec; |
548 | |
549 | /* Calculate temporary vectorial force */ |
550 | tx = fscal*dx21; |
551 | ty = fscal*dy21; |
552 | tz = fscal*dz21; |
553 | |
554 | /* Update vectorial force */ |
555 | fix2 += tx; |
556 | fiy2 += ty; |
557 | fiz2 += tz; |
558 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
559 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
560 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
561 | |
562 | } |
563 | |
564 | /************************** |
565 | * CALCULATE INTERACTIONS * |
566 | **************************/ |
567 | |
568 | if (rsq22<rcutoff2) |
569 | { |
570 | |
571 | r22 = rsq22*rinv22; |
572 | |
573 | /* EWALD ELECTROSTATICS */ |
574 | |
575 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
576 | ewrt = r22*ewtabscale; |
577 | ewitab = ewrt; |
578 | eweps = ewrt-ewitab; |
579 | ewitab = 4*ewitab; |
580 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
581 | velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
582 | felec = qq22*rinv22*(rinvsq22-felec); |
583 | |
584 | d = r22-rswitch; |
585 | d = (d>0.0) ? d : 0.0; |
586 | d2 = d*d; |
587 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
588 | |
589 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
590 | |
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 - rinv22*velec*dsw; |
594 | velec *= sw; |
595 | |
596 | /* Update potential sums from outer loop */ |
597 | velecsum += velec; |
598 | |
599 | fscal = felec; |
600 | |
601 | /* Calculate temporary vectorial force */ |
602 | tx = fscal*dx22; |
603 | ty = fscal*dy22; |
604 | tz = fscal*dz22; |
605 | |
606 | /* Update vectorial force */ |
607 | fix2 += tx; |
608 | fiy2 += ty; |
609 | fiz2 += tz; |
610 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
611 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
612 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
613 | |
614 | } |
615 | |
616 | /************************** |
617 | * CALCULATE INTERACTIONS * |
618 | **************************/ |
619 | |
620 | if (rsq23<rcutoff2) |
621 | { |
622 | |
623 | r23 = rsq23*rinv23; |
624 | |
625 | /* EWALD ELECTROSTATICS */ |
626 | |
627 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
628 | ewrt = r23*ewtabscale; |
629 | ewitab = ewrt; |
630 | eweps = ewrt-ewitab; |
631 | ewitab = 4*ewitab; |
632 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
633 | velec = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
634 | felec = qq23*rinv23*(rinvsq23-felec); |
635 | |
636 | d = r23-rswitch; |
637 | d = (d>0.0) ? d : 0.0; |
638 | d2 = d*d; |
639 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
640 | |
641 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
642 | |
643 | /* Evaluate switch function */ |
644 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
645 | felec = felec*sw - rinv23*velec*dsw; |
646 | velec *= sw; |
647 | |
648 | /* Update potential sums from outer loop */ |
649 | velecsum += velec; |
650 | |
651 | fscal = felec; |
652 | |
653 | /* Calculate temporary vectorial force */ |
654 | tx = fscal*dx23; |
655 | ty = fscal*dy23; |
656 | tz = fscal*dz23; |
657 | |
658 | /* Update vectorial force */ |
659 | fix2 += tx; |
660 | fiy2 += ty; |
661 | fiz2 += tz; |
662 | f[j_coord_offset+DIM3*3+XX0] -= tx; |
663 | f[j_coord_offset+DIM3*3+YY1] -= ty; |
664 | f[j_coord_offset+DIM3*3+ZZ2] -= tz; |
665 | |
666 | } |
667 | |
668 | /************************** |
669 | * CALCULATE INTERACTIONS * |
670 | **************************/ |
671 | |
672 | if (rsq31<rcutoff2) |
673 | { |
674 | |
675 | r31 = rsq31*rinv31; |
676 | |
677 | /* EWALD ELECTROSTATICS */ |
678 | |
679 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
680 | ewrt = r31*ewtabscale; |
681 | ewitab = ewrt; |
682 | eweps = ewrt-ewitab; |
683 | ewitab = 4*ewitab; |
684 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
685 | velec = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
686 | felec = qq31*rinv31*(rinvsq31-felec); |
687 | |
688 | d = r31-rswitch; |
689 | d = (d>0.0) ? d : 0.0; |
690 | d2 = d*d; |
691 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
692 | |
693 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
694 | |
695 | /* Evaluate switch function */ |
696 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
697 | felec = felec*sw - rinv31*velec*dsw; |
698 | velec *= sw; |
699 | |
700 | /* Update potential sums from outer loop */ |
701 | velecsum += velec; |
702 | |
703 | fscal = felec; |
704 | |
705 | /* Calculate temporary vectorial force */ |
706 | tx = fscal*dx31; |
707 | ty = fscal*dy31; |
708 | tz = fscal*dz31; |
709 | |
710 | /* Update vectorial force */ |
711 | fix3 += tx; |
712 | fiy3 += ty; |
713 | fiz3 += tz; |
714 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
715 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
716 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
717 | |
718 | } |
719 | |
720 | /************************** |
721 | * CALCULATE INTERACTIONS * |
722 | **************************/ |
723 | |
724 | if (rsq32<rcutoff2) |
725 | { |
726 | |
727 | r32 = rsq32*rinv32; |
728 | |
729 | /* EWALD ELECTROSTATICS */ |
730 | |
731 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
732 | ewrt = r32*ewtabscale; |
733 | ewitab = ewrt; |
734 | eweps = ewrt-ewitab; |
735 | ewitab = 4*ewitab; |
736 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
737 | velec = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
738 | felec = qq32*rinv32*(rinvsq32-felec); |
739 | |
740 | d = r32-rswitch; |
741 | d = (d>0.0) ? d : 0.0; |
742 | d2 = d*d; |
743 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
744 | |
745 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
746 | |
747 | /* Evaluate switch function */ |
748 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
749 | felec = felec*sw - rinv32*velec*dsw; |
750 | velec *= sw; |
751 | |
752 | /* Update potential sums from outer loop */ |
753 | velecsum += velec; |
754 | |
755 | fscal = felec; |
756 | |
757 | /* Calculate temporary vectorial force */ |
758 | tx = fscal*dx32; |
759 | ty = fscal*dy32; |
760 | tz = fscal*dz32; |
761 | |
762 | /* Update vectorial force */ |
763 | fix3 += tx; |
764 | fiy3 += ty; |
765 | fiz3 += tz; |
766 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
767 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
768 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
769 | |
770 | } |
771 | |
772 | /************************** |
773 | * CALCULATE INTERACTIONS * |
774 | **************************/ |
775 | |
776 | if (rsq33<rcutoff2) |
777 | { |
778 | |
779 | r33 = rsq33*rinv33; |
780 | |
781 | /* EWALD ELECTROSTATICS */ |
782 | |
783 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
784 | ewrt = r33*ewtabscale; |
785 | ewitab = ewrt; |
786 | eweps = ewrt-ewitab; |
787 | ewitab = 4*ewitab; |
788 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
789 | velec = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
790 | felec = qq33*rinv33*(rinvsq33-felec); |
791 | |
792 | d = r33-rswitch; |
793 | d = (d>0.0) ? d : 0.0; |
794 | d2 = d*d; |
795 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
796 | |
797 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
798 | |
799 | /* Evaluate switch function */ |
800 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
801 | felec = felec*sw - rinv33*velec*dsw; |
802 | velec *= sw; |
803 | |
804 | /* Update potential sums from outer loop */ |
805 | velecsum += velec; |
806 | |
807 | fscal = felec; |
808 | |
809 | /* Calculate temporary vectorial force */ |
810 | tx = fscal*dx33; |
811 | ty = fscal*dy33; |
812 | tz = fscal*dz33; |
813 | |
814 | /* Update vectorial force */ |
815 | fix3 += tx; |
816 | fiy3 += ty; |
817 | fiz3 += tz; |
818 | f[j_coord_offset+DIM3*3+XX0] -= tx; |
819 | f[j_coord_offset+DIM3*3+YY1] -= ty; |
820 | f[j_coord_offset+DIM3*3+ZZ2] -= tz; |
821 | |
822 | } |
823 | |
824 | /* Inner loop uses 601 flops */ |
825 | } |
826 | /* End of innermost loop */ |
827 | |
828 | tx = ty = tz = 0; |
829 | f[i_coord_offset+DIM3*0+XX0] += fix0; |
830 | f[i_coord_offset+DIM3*0+YY1] += fiy0; |
831 | f[i_coord_offset+DIM3*0+ZZ2] += fiz0; |
832 | tx += fix0; |
833 | ty += fiy0; |
834 | tz += fiz0; |
835 | f[i_coord_offset+DIM3*1+XX0] += fix1; |
836 | f[i_coord_offset+DIM3*1+YY1] += fiy1; |
837 | f[i_coord_offset+DIM3*1+ZZ2] += fiz1; |
838 | tx += fix1; |
839 | ty += fiy1; |
840 | tz += fiz1; |
841 | f[i_coord_offset+DIM3*2+XX0] += fix2; |
842 | f[i_coord_offset+DIM3*2+YY1] += fiy2; |
843 | f[i_coord_offset+DIM3*2+ZZ2] += fiz2; |
844 | tx += fix2; |
845 | ty += fiy2; |
846 | tz += fiz2; |
847 | f[i_coord_offset+DIM3*3+XX0] += fix3; |
848 | f[i_coord_offset+DIM3*3+YY1] += fiy3; |
849 | f[i_coord_offset+DIM3*3+ZZ2] += fiz3; |
850 | tx += fix3; |
851 | ty += fiy3; |
852 | tz += fiz3; |
853 | fshift[i_shift_offset+XX0] += tx; |
854 | fshift[i_shift_offset+YY1] += ty; |
855 | fshift[i_shift_offset+ZZ2] += tz; |
856 | |
857 | ggid = gid[iidx]; |
858 | /* Update potential energies */ |
859 | kernel_data->energygrp_elec[ggid] += velecsum; |
860 | kernel_data->energygrp_vdw[ggid] += vvdwsum; |
861 | |
862 | /* Increment number of inner iterations */ |
863 | inneriter += j_index_end - j_index_start; |
864 | |
865 | /* Outer loop uses 41 flops */ |
866 | } |
867 | |
868 | /* Increment number of outer iterations */ |
869 | outeriter += nri; |
870 | |
871 | /* Update outer/inner flops */ |
872 | |
873 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*41 + inneriter*601)(nrnb)->n[eNR_NBKERNEL_ELEC_VDW_W4W4_VF] += outeriter*41 + inneriter*601; |
874 | } |
875 | /* |
876 | * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_F_c |
877 | * Electrostatics interaction: Ewald |
878 | * VdW interaction: Buckingham |
879 | * Geometry: Water4-Water4 |
880 | * Calculate force/pot: Force |
881 | */ |
882 | void |
883 | nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_F_c |
884 | (t_nblist * gmx_restrict__restrict nlist, |
885 | rvec * gmx_restrict__restrict xx, |
886 | rvec * gmx_restrict__restrict ff, |
887 | t_forcerec * gmx_restrict__restrict fr, |
888 | t_mdatoms * gmx_restrict__restrict mdatoms, |
889 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict__restrict kernel_data, |
890 | t_nrnb * gmx_restrict__restrict nrnb) |
891 | { |
892 | int i_shift_offset,i_coord_offset,j_coord_offset; |
893 | int j_index_start,j_index_end; |
894 | int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter; |
895 | real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2; |
896 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
897 | real *shiftvec,*fshift,*x,*f; |
898 | int vdwioffset0; |
899 | real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0; |
900 | int vdwioffset1; |
901 | real ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
902 | int vdwioffset2; |
903 | real ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
904 | int vdwioffset3; |
905 | real ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3; |
906 | int vdwjidx0; |
907 | real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
908 | int vdwjidx1; |
909 | real jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
910 | int vdwjidx2; |
911 | real jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
912 | int vdwjidx3; |
913 | real jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3; |
914 | real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00; |
915 | real dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11; |
916 | real dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12; |
917 | real dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13; |
918 | real dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21; |
919 | real dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22; |
920 | real dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23; |
921 | real dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31; |
922 | real dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32; |
923 | real dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33; |
924 | real velec,felec,velecsum,facel,crf,krf,krf2; |
925 | real *charge; |
926 | int nvdwtype; |
927 | real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6; |
928 | int *vdwtype; |
929 | real *vdwparam; |
930 | int ewitab; |
931 | real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace; |
932 | real *ewtab; |
933 | real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw; |
934 | |
935 | x = xx[0]; |
936 | f = ff[0]; |
937 | |
938 | nri = nlist->nri; |
939 | iinr = nlist->iinr; |
940 | jindex = nlist->jindex; |
941 | jjnr = nlist->jjnr; |
942 | shiftidx = nlist->shift; |
943 | gid = nlist->gid; |
944 | shiftvec = fr->shift_vec[0]; |
945 | fshift = fr->fshift[0]; |
946 | facel = fr->epsfac; |
947 | charge = mdatoms->chargeA; |
948 | nvdwtype = fr->ntype; |
949 | vdwparam = fr->nbfp; |
950 | vdwtype = mdatoms->typeA; |
951 | |
952 | sh_ewald = fr->ic->sh_ewald; |
953 | ewtab = fr->ic->tabq_coul_FDV0; |
954 | ewtabscale = fr->ic->tabq_scale; |
955 | ewtabhalfspace = 0.5/ewtabscale; |
956 | |
957 | /* Setup water-specific parameters */ |
958 | inr = nlist->iinr[0]; |
959 | iq1 = facel*charge[inr+1]; |
960 | iq2 = facel*charge[inr+2]; |
961 | iq3 = facel*charge[inr+3]; |
962 | vdwioffset0 = 3*nvdwtype*vdwtype[inr+0]; |
963 | |
964 | jq1 = charge[inr+1]; |
965 | jq2 = charge[inr+2]; |
966 | jq3 = charge[inr+3]; |
967 | vdwjidx0 = 3*vdwtype[inr+0]; |
968 | c6_00 = vdwparam[vdwioffset0+vdwjidx0]; |
969 | cexp1_00 = vdwparam[vdwioffset0+vdwjidx0+1]; |
970 | cexp2_00 = vdwparam[vdwioffset0+vdwjidx0+2]; |
971 | qq11 = iq1*jq1; |
972 | qq12 = iq1*jq2; |
973 | qq13 = iq1*jq3; |
974 | qq21 = iq2*jq1; |
975 | qq22 = iq2*jq2; |
976 | qq23 = iq2*jq3; |
977 | qq31 = iq3*jq1; |
978 | qq32 = iq3*jq2; |
979 | qq33 = iq3*jq3; |
980 | |
981 | /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */ |
982 | rcutoff = fr->rcoulomb; |
983 | rcutoff2 = rcutoff*rcutoff; |
984 | |
985 | rswitch = fr->rcoulomb_switch; |
986 | /* Setup switch parameters */ |
987 | d = rcutoff-rswitch; |
988 | swV3 = -10.0/(d*d*d); |
989 | swV4 = 15.0/(d*d*d*d); |
990 | swV5 = -6.0/(d*d*d*d*d); |
991 | swF2 = -30.0/(d*d*d); |
992 | swF3 = 60.0/(d*d*d*d); |
993 | swF4 = -30.0/(d*d*d*d*d); |
994 | |
995 | outeriter = 0; |
996 | inneriter = 0; |
997 | |
998 | /* Start outer loop over neighborlists */ |
999 | for(iidx=0; iidx<nri; iidx++) |
1000 | { |
1001 | /* Load shift vector for this list */ |
1002 | i_shift_offset = DIM3*shiftidx[iidx]; |
1003 | shX = shiftvec[i_shift_offset+XX0]; |
1004 | shY = shiftvec[i_shift_offset+YY1]; |
1005 | shZ = shiftvec[i_shift_offset+ZZ2]; |
1006 | |
1007 | /* Load limits for loop over neighbors */ |
1008 | j_index_start = jindex[iidx]; |
1009 | j_index_end = jindex[iidx+1]; |
1010 | |
1011 | /* Get outer coordinate index */ |
1012 | inr = iinr[iidx]; |
1013 | i_coord_offset = DIM3*inr; |
1014 | |
1015 | /* Load i particle coords and add shift vector */ |
1016 | ix0 = shX + x[i_coord_offset+DIM3*0+XX0]; |
1017 | iy0 = shY + x[i_coord_offset+DIM3*0+YY1]; |
1018 | iz0 = shZ + x[i_coord_offset+DIM3*0+ZZ2]; |
1019 | ix1 = shX + x[i_coord_offset+DIM3*1+XX0]; |
1020 | iy1 = shY + x[i_coord_offset+DIM3*1+YY1]; |
1021 | iz1 = shZ + x[i_coord_offset+DIM3*1+ZZ2]; |
1022 | ix2 = shX + x[i_coord_offset+DIM3*2+XX0]; |
1023 | iy2 = shY + x[i_coord_offset+DIM3*2+YY1]; |
1024 | iz2 = shZ + x[i_coord_offset+DIM3*2+ZZ2]; |
1025 | ix3 = shX + x[i_coord_offset+DIM3*3+XX0]; |
1026 | iy3 = shY + x[i_coord_offset+DIM3*3+YY1]; |
1027 | iz3 = shZ + x[i_coord_offset+DIM3*3+ZZ2]; |
1028 | |
1029 | fix0 = 0.0; |
1030 | fiy0 = 0.0; |
1031 | fiz0 = 0.0; |
1032 | fix1 = 0.0; |
1033 | fiy1 = 0.0; |
1034 | fiz1 = 0.0; |
1035 | fix2 = 0.0; |
1036 | fiy2 = 0.0; |
1037 | fiz2 = 0.0; |
1038 | fix3 = 0.0; |
1039 | fiy3 = 0.0; |
1040 | fiz3 = 0.0; |
1041 | |
1042 | /* Start inner kernel loop */ |
1043 | for(jidx=j_index_start; jidx<j_index_end; jidx++) |
1044 | { |
1045 | /* Get j neighbor index, and coordinate index */ |
1046 | jnr = jjnr[jidx]; |
1047 | j_coord_offset = DIM3*jnr; |
1048 | |
1049 | /* load j atom coordinates */ |
1050 | jx0 = x[j_coord_offset+DIM3*0+XX0]; |
1051 | jy0 = x[j_coord_offset+DIM3*0+YY1]; |
1052 | jz0 = x[j_coord_offset+DIM3*0+ZZ2]; |
1053 | jx1 = x[j_coord_offset+DIM3*1+XX0]; |
1054 | jy1 = x[j_coord_offset+DIM3*1+YY1]; |
1055 | jz1 = x[j_coord_offset+DIM3*1+ZZ2]; |
1056 | jx2 = x[j_coord_offset+DIM3*2+XX0]; |
1057 | jy2 = x[j_coord_offset+DIM3*2+YY1]; |
1058 | jz2 = x[j_coord_offset+DIM3*2+ZZ2]; |
1059 | jx3 = x[j_coord_offset+DIM3*3+XX0]; |
1060 | jy3 = x[j_coord_offset+DIM3*3+YY1]; |
1061 | jz3 = x[j_coord_offset+DIM3*3+ZZ2]; |
1062 | |
1063 | /* Calculate displacement vector */ |
1064 | dx00 = ix0 - jx0; |
1065 | dy00 = iy0 - jy0; |
1066 | dz00 = iz0 - jz0; |
1067 | dx11 = ix1 - jx1; |
1068 | dy11 = iy1 - jy1; |
1069 | dz11 = iz1 - jz1; |
1070 | dx12 = ix1 - jx2; |
1071 | dy12 = iy1 - jy2; |
1072 | dz12 = iz1 - jz2; |
1073 | dx13 = ix1 - jx3; |
1074 | dy13 = iy1 - jy3; |
1075 | dz13 = iz1 - jz3; |
1076 | dx21 = ix2 - jx1; |
1077 | dy21 = iy2 - jy1; |
1078 | dz21 = iz2 - jz1; |
1079 | dx22 = ix2 - jx2; |
1080 | dy22 = iy2 - jy2; |
1081 | dz22 = iz2 - jz2; |
1082 | dx23 = ix2 - jx3; |
1083 | dy23 = iy2 - jy3; |
1084 | dz23 = iz2 - jz3; |
1085 | dx31 = ix3 - jx1; |
1086 | dy31 = iy3 - jy1; |
1087 | dz31 = iz3 - jz1; |
1088 | dx32 = ix3 - jx2; |
1089 | dy32 = iy3 - jy2; |
1090 | dz32 = iz3 - jz2; |
1091 | dx33 = ix3 - jx3; |
1092 | dy33 = iy3 - jy3; |
1093 | dz33 = iz3 - jz3; |
1094 | |
1095 | /* Calculate squared distance and things based on it */ |
1096 | rsq00 = dx00*dx00+dy00*dy00+dz00*dz00; |
1097 | rsq11 = dx11*dx11+dy11*dy11+dz11*dz11; |
1098 | rsq12 = dx12*dx12+dy12*dy12+dz12*dz12; |
1099 | rsq13 = dx13*dx13+dy13*dy13+dz13*dz13; |
1100 | rsq21 = dx21*dx21+dy21*dy21+dz21*dz21; |
1101 | rsq22 = dx22*dx22+dy22*dy22+dz22*dz22; |
1102 | rsq23 = dx23*dx23+dy23*dy23+dz23*dz23; |
1103 | rsq31 = dx31*dx31+dy31*dy31+dz31*dz31; |
1104 | rsq32 = dx32*dx32+dy32*dy32+dz32*dz32; |
1105 | rsq33 = dx33*dx33+dy33*dy33+dz33*dz33; |
1106 | |
1107 | rinv00 = gmx_invsqrt(rsq00)gmx_software_invsqrt(rsq00); |
1108 | rinv11 = gmx_invsqrt(rsq11)gmx_software_invsqrt(rsq11); |
1109 | rinv12 = gmx_invsqrt(rsq12)gmx_software_invsqrt(rsq12); |
1110 | rinv13 = gmx_invsqrt(rsq13)gmx_software_invsqrt(rsq13); |
1111 | rinv21 = gmx_invsqrt(rsq21)gmx_software_invsqrt(rsq21); |
1112 | rinv22 = gmx_invsqrt(rsq22)gmx_software_invsqrt(rsq22); |
1113 | rinv23 = gmx_invsqrt(rsq23)gmx_software_invsqrt(rsq23); |
1114 | rinv31 = gmx_invsqrt(rsq31)gmx_software_invsqrt(rsq31); |
1115 | rinv32 = gmx_invsqrt(rsq32)gmx_software_invsqrt(rsq32); |
1116 | rinv33 = gmx_invsqrt(rsq33)gmx_software_invsqrt(rsq33); |
1117 | |
1118 | rinvsq00 = rinv00*rinv00; |
1119 | rinvsq11 = rinv11*rinv11; |
1120 | rinvsq12 = rinv12*rinv12; |
1121 | rinvsq13 = rinv13*rinv13; |
1122 | rinvsq21 = rinv21*rinv21; |
1123 | rinvsq22 = rinv22*rinv22; |
1124 | rinvsq23 = rinv23*rinv23; |
1125 | rinvsq31 = rinv31*rinv31; |
1126 | rinvsq32 = rinv32*rinv32; |
1127 | rinvsq33 = rinv33*rinv33; |
1128 | |
1129 | /************************** |
1130 | * CALCULATE INTERACTIONS * |
1131 | **************************/ |
1132 | |
1133 | if (rsq00<rcutoff2) |
1134 | { |
1135 | |
1136 | r00 = rsq00*rinv00; |
1137 | |
1138 | /* BUCKINGHAM DISPERSION/REPULSION */ |
1139 | rinvsix = rinvsq00*rinvsq00*rinvsq00; |
1140 | vvdw6 = c6_00*rinvsix; |
1141 | br = cexp2_00*r00; |
1142 | vvdwexp = cexp1_00*exp(-br); |
1143 | vvdw = vvdwexp - vvdw6*(1.0/6.0); |
1144 | fvdw = (br*vvdwexp-vvdw6)*rinvsq00; |
1145 | |
1146 | d = r00-rswitch; |
1147 | d = (d>0.0) ? d : 0.0; |
1148 | d2 = d*d; |
1149 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1150 | |
1151 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1152 | |
1153 | /* Evaluate switch function */ |
1154 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1155 | fvdw = fvdw*sw - rinv00*vvdw*dsw; |
1156 | |
1157 | fscal = fvdw; |
1158 | |
1159 | /* Calculate temporary vectorial force */ |
1160 | tx = fscal*dx00; |
1161 | ty = fscal*dy00; |
1162 | tz = fscal*dz00; |
1163 | |
1164 | /* Update vectorial force */ |
1165 | fix0 += tx; |
1166 | fiy0 += ty; |
1167 | fiz0 += tz; |
1168 | f[j_coord_offset+DIM3*0+XX0] -= tx; |
1169 | f[j_coord_offset+DIM3*0+YY1] -= ty; |
1170 | f[j_coord_offset+DIM3*0+ZZ2] -= tz; |
1171 | |
1172 | } |
1173 | |
1174 | /************************** |
1175 | * CALCULATE INTERACTIONS * |
1176 | **************************/ |
1177 | |
1178 | if (rsq11<rcutoff2) |
1179 | { |
1180 | |
1181 | r11 = rsq11*rinv11; |
1182 | |
1183 | /* EWALD ELECTROSTATICS */ |
1184 | |
1185 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1186 | ewrt = r11*ewtabscale; |
1187 | ewitab = ewrt; |
1188 | eweps = ewrt-ewitab; |
1189 | ewitab = 4*ewitab; |
1190 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1191 | velec = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1192 | felec = qq11*rinv11*(rinvsq11-felec); |
1193 | |
1194 | d = r11-rswitch; |
1195 | d = (d>0.0) ? d : 0.0; |
1196 | d2 = d*d; |
1197 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1198 | |
1199 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1200 | |
1201 | /* Evaluate switch function */ |
1202 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1203 | felec = felec*sw - rinv11*velec*dsw; |
1204 | |
1205 | fscal = felec; |
1206 | |
1207 | /* Calculate temporary vectorial force */ |
1208 | tx = fscal*dx11; |
1209 | ty = fscal*dy11; |
1210 | tz = fscal*dz11; |
1211 | |
1212 | /* Update vectorial force */ |
1213 | fix1 += tx; |
1214 | fiy1 += ty; |
1215 | fiz1 += tz; |
1216 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
1217 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
1218 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
1219 | |
1220 | } |
1221 | |
1222 | /************************** |
1223 | * CALCULATE INTERACTIONS * |
1224 | **************************/ |
1225 | |
1226 | if (rsq12<rcutoff2) |
1227 | { |
1228 | |
1229 | r12 = rsq12*rinv12; |
1230 | |
1231 | /* EWALD ELECTROSTATICS */ |
1232 | |
1233 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1234 | ewrt = r12*ewtabscale; |
1235 | ewitab = ewrt; |
1236 | eweps = ewrt-ewitab; |
1237 | ewitab = 4*ewitab; |
1238 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1239 | velec = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1240 | felec = qq12*rinv12*(rinvsq12-felec); |
1241 | |
1242 | d = r12-rswitch; |
1243 | d = (d>0.0) ? d : 0.0; |
1244 | d2 = d*d; |
1245 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1246 | |
1247 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1248 | |
1249 | /* Evaluate switch function */ |
1250 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1251 | felec = felec*sw - rinv12*velec*dsw; |
1252 | |
1253 | fscal = felec; |
1254 | |
1255 | /* Calculate temporary vectorial force */ |
1256 | tx = fscal*dx12; |
1257 | ty = fscal*dy12; |
1258 | tz = fscal*dz12; |
1259 | |
1260 | /* Update vectorial force */ |
1261 | fix1 += tx; |
1262 | fiy1 += ty; |
1263 | fiz1 += tz; |
1264 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
1265 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
1266 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
1267 | |
1268 | } |
1269 | |
1270 | /************************** |
1271 | * CALCULATE INTERACTIONS * |
1272 | **************************/ |
1273 | |
1274 | if (rsq13<rcutoff2) |
1275 | { |
1276 | |
1277 | r13 = rsq13*rinv13; |
1278 | |
1279 | /* EWALD ELECTROSTATICS */ |
1280 | |
1281 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1282 | ewrt = r13*ewtabscale; |
1283 | ewitab = ewrt; |
1284 | eweps = ewrt-ewitab; |
1285 | ewitab = 4*ewitab; |
1286 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1287 | velec = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1288 | felec = qq13*rinv13*(rinvsq13-felec); |
1289 | |
1290 | d = r13-rswitch; |
1291 | d = (d>0.0) ? d : 0.0; |
1292 | d2 = d*d; |
1293 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1294 | |
1295 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1296 | |
1297 | /* Evaluate switch function */ |
1298 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1299 | felec = felec*sw - rinv13*velec*dsw; |
1300 | |
1301 | fscal = felec; |
1302 | |
1303 | /* Calculate temporary vectorial force */ |
1304 | tx = fscal*dx13; |
1305 | ty = fscal*dy13; |
1306 | tz = fscal*dz13; |
1307 | |
1308 | /* Update vectorial force */ |
1309 | fix1 += tx; |
1310 | fiy1 += ty; |
1311 | fiz1 += tz; |
1312 | f[j_coord_offset+DIM3*3+XX0] -= tx; |
1313 | f[j_coord_offset+DIM3*3+YY1] -= ty; |
1314 | f[j_coord_offset+DIM3*3+ZZ2] -= tz; |
1315 | |
1316 | } |
1317 | |
1318 | /************************** |
1319 | * CALCULATE INTERACTIONS * |
1320 | **************************/ |
1321 | |
1322 | if (rsq21<rcutoff2) |
1323 | { |
1324 | |
1325 | r21 = rsq21*rinv21; |
1326 | |
1327 | /* EWALD ELECTROSTATICS */ |
1328 | |
1329 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1330 | ewrt = r21*ewtabscale; |
1331 | ewitab = ewrt; |
1332 | eweps = ewrt-ewitab; |
1333 | ewitab = 4*ewitab; |
1334 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1335 | velec = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1336 | felec = qq21*rinv21*(rinvsq21-felec); |
1337 | |
1338 | d = r21-rswitch; |
1339 | d = (d>0.0) ? d : 0.0; |
1340 | d2 = d*d; |
1341 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1342 | |
1343 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1344 | |
1345 | /* Evaluate switch function */ |
1346 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1347 | felec = felec*sw - rinv21*velec*dsw; |
1348 | |
1349 | fscal = felec; |
1350 | |
1351 | /* Calculate temporary vectorial force */ |
1352 | tx = fscal*dx21; |
1353 | ty = fscal*dy21; |
1354 | tz = fscal*dz21; |
1355 | |
1356 | /* Update vectorial force */ |
1357 | fix2 += tx; |
1358 | fiy2 += ty; |
1359 | fiz2 += tz; |
1360 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
1361 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
1362 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
1363 | |
1364 | } |
1365 | |
1366 | /************************** |
1367 | * CALCULATE INTERACTIONS * |
1368 | **************************/ |
1369 | |
1370 | if (rsq22<rcutoff2) |
1371 | { |
1372 | |
1373 | r22 = rsq22*rinv22; |
1374 | |
1375 | /* EWALD ELECTROSTATICS */ |
1376 | |
1377 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1378 | ewrt = r22*ewtabscale; |
1379 | ewitab = ewrt; |
1380 | eweps = ewrt-ewitab; |
1381 | ewitab = 4*ewitab; |
1382 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1383 | velec = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1384 | felec = qq22*rinv22*(rinvsq22-felec); |
1385 | |
1386 | d = r22-rswitch; |
1387 | d = (d>0.0) ? d : 0.0; |
1388 | d2 = d*d; |
1389 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1390 | |
1391 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1392 | |
1393 | /* Evaluate switch function */ |
1394 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1395 | felec = felec*sw - rinv22*velec*dsw; |
1396 | |
1397 | fscal = felec; |
1398 | |
1399 | /* Calculate temporary vectorial force */ |
1400 | tx = fscal*dx22; |
1401 | ty = fscal*dy22; |
1402 | tz = fscal*dz22; |
1403 | |
1404 | /* Update vectorial force */ |
1405 | fix2 += tx; |
1406 | fiy2 += ty; |
1407 | fiz2 += tz; |
1408 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
1409 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
1410 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
1411 | |
1412 | } |
1413 | |
1414 | /************************** |
1415 | * CALCULATE INTERACTIONS * |
1416 | **************************/ |
1417 | |
1418 | if (rsq23<rcutoff2) |
1419 | { |
1420 | |
1421 | r23 = rsq23*rinv23; |
1422 | |
1423 | /* EWALD ELECTROSTATICS */ |
1424 | |
1425 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1426 | ewrt = r23*ewtabscale; |
1427 | ewitab = ewrt; |
1428 | eweps = ewrt-ewitab; |
1429 | ewitab = 4*ewitab; |
1430 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1431 | velec = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1432 | felec = qq23*rinv23*(rinvsq23-felec); |
1433 | |
1434 | d = r23-rswitch; |
1435 | d = (d>0.0) ? d : 0.0; |
1436 | d2 = d*d; |
1437 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1438 | |
1439 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1440 | |
1441 | /* Evaluate switch function */ |
1442 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1443 | felec = felec*sw - rinv23*velec*dsw; |
1444 | |
1445 | fscal = felec; |
1446 | |
1447 | /* Calculate temporary vectorial force */ |
1448 | tx = fscal*dx23; |
1449 | ty = fscal*dy23; |
1450 | tz = fscal*dz23; |
1451 | |
1452 | /* Update vectorial force */ |
1453 | fix2 += tx; |
1454 | fiy2 += ty; |
1455 | fiz2 += tz; |
1456 | f[j_coord_offset+DIM3*3+XX0] -= tx; |
1457 | f[j_coord_offset+DIM3*3+YY1] -= ty; |
1458 | f[j_coord_offset+DIM3*3+ZZ2] -= tz; |
1459 | |
1460 | } |
1461 | |
1462 | /************************** |
1463 | * CALCULATE INTERACTIONS * |
1464 | **************************/ |
1465 | |
1466 | if (rsq31<rcutoff2) |
1467 | { |
1468 | |
1469 | r31 = rsq31*rinv31; |
1470 | |
1471 | /* EWALD ELECTROSTATICS */ |
1472 | |
1473 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1474 | ewrt = r31*ewtabscale; |
1475 | ewitab = ewrt; |
1476 | eweps = ewrt-ewitab; |
1477 | ewitab = 4*ewitab; |
1478 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1479 | velec = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1480 | felec = qq31*rinv31*(rinvsq31-felec); |
1481 | |
1482 | d = r31-rswitch; |
1483 | d = (d>0.0) ? d : 0.0; |
1484 | d2 = d*d; |
1485 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1486 | |
1487 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1488 | |
1489 | /* Evaluate switch function */ |
1490 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1491 | felec = felec*sw - rinv31*velec*dsw; |
1492 | |
1493 | fscal = felec; |
1494 | |
1495 | /* Calculate temporary vectorial force */ |
1496 | tx = fscal*dx31; |
1497 | ty = fscal*dy31; |
1498 | tz = fscal*dz31; |
1499 | |
1500 | /* Update vectorial force */ |
1501 | fix3 += tx; |
1502 | fiy3 += ty; |
1503 | fiz3 += tz; |
1504 | f[j_coord_offset+DIM3*1+XX0] -= tx; |
1505 | f[j_coord_offset+DIM3*1+YY1] -= ty; |
1506 | f[j_coord_offset+DIM3*1+ZZ2] -= tz; |
1507 | |
1508 | } |
1509 | |
1510 | /************************** |
1511 | * CALCULATE INTERACTIONS * |
1512 | **************************/ |
1513 | |
1514 | if (rsq32<rcutoff2) |
1515 | { |
1516 | |
1517 | r32 = rsq32*rinv32; |
1518 | |
1519 | /* EWALD ELECTROSTATICS */ |
1520 | |
1521 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1522 | ewrt = r32*ewtabscale; |
1523 | ewitab = ewrt; |
1524 | eweps = ewrt-ewitab; |
1525 | ewitab = 4*ewitab; |
1526 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1527 | velec = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1528 | felec = qq32*rinv32*(rinvsq32-felec); |
1529 | |
1530 | d = r32-rswitch; |
1531 | d = (d>0.0) ? d : 0.0; |
1532 | d2 = d*d; |
1533 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1534 | |
1535 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1536 | |
1537 | /* Evaluate switch function */ |
1538 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1539 | felec = felec*sw - rinv32*velec*dsw; |
1540 | |
1541 | fscal = felec; |
1542 | |
1543 | /* Calculate temporary vectorial force */ |
1544 | tx = fscal*dx32; |
1545 | ty = fscal*dy32; |
1546 | tz = fscal*dz32; |
1547 | |
1548 | /* Update vectorial force */ |
1549 | fix3 += tx; |
1550 | fiy3 += ty; |
1551 | fiz3 += tz; |
1552 | f[j_coord_offset+DIM3*2+XX0] -= tx; |
1553 | f[j_coord_offset+DIM3*2+YY1] -= ty; |
1554 | f[j_coord_offset+DIM3*2+ZZ2] -= tz; |
1555 | |
1556 | } |
1557 | |
1558 | /************************** |
1559 | * CALCULATE INTERACTIONS * |
1560 | **************************/ |
1561 | |
1562 | if (rsq33<rcutoff2) |
1563 | { |
1564 | |
1565 | r33 = rsq33*rinv33; |
1566 | |
1567 | /* EWALD ELECTROSTATICS */ |
1568 | |
1569 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1570 | ewrt = r33*ewtabscale; |
1571 | ewitab = ewrt; |
1572 | eweps = ewrt-ewitab; |
1573 | ewitab = 4*ewitab; |
1574 | felec = ewtab[ewitab]+eweps*ewtab[ewitab+1]; |
1575 | velec = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec))); |
1576 | felec = qq33*rinv33*(rinvsq33-felec); |
1577 | |
1578 | d = r33-rswitch; |
1579 | d = (d>0.0) ? d : 0.0; |
1580 | d2 = d*d; |
1581 | sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5)); |
1582 | |
1583 | dsw = d2*(swF2+d*(swF3+d*swF4)); |
1584 | |
1585 | /* Evaluate switch function */ |
1586 | /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */ |
1587 | felec = felec*sw - rinv33*velec*dsw; |
1588 | |
1589 | fscal = felec; |
1590 | |
1591 | /* Calculate temporary vectorial force */ |
1592 | tx = fscal*dx33; |
1593 | ty = fscal*dy33; |
1594 | tz = fscal*dz33; |
1595 | |
1596 | /* Update vectorial force */ |
1597 | fix3 += tx; |
1598 | fiy3 += ty; |
1599 | fiz3 += tz; |
1600 | f[j_coord_offset+DIM3*3+XX0] -= tx; |
1601 | f[j_coord_offset+DIM3*3+YY1] -= ty; |
1602 | f[j_coord_offset+DIM3*3+ZZ2] -= tz; |
1603 | |
1604 | } |
1605 | |
1606 | /* Inner loop uses 581 flops */ |
1607 | } |
1608 | /* End of innermost loop */ |
1609 | |
1610 | tx = ty = tz = 0; |
1611 | f[i_coord_offset+DIM3*0+XX0] += fix0; |
1612 | f[i_coord_offset+DIM3*0+YY1] += fiy0; |
1613 | f[i_coord_offset+DIM3*0+ZZ2] += fiz0; |
1614 | tx += fix0; |
1615 | ty += fiy0; |
1616 | tz += fiz0; |
1617 | f[i_coord_offset+DIM3*1+XX0] += fix1; |
1618 | f[i_coord_offset+DIM3*1+YY1] += fiy1; |
1619 | f[i_coord_offset+DIM3*1+ZZ2] += fiz1; |
1620 | tx += fix1; |
1621 | ty += fiy1; |
1622 | tz += fiz1; |
1623 | f[i_coord_offset+DIM3*2+XX0] += fix2; |
1624 | f[i_coord_offset+DIM3*2+YY1] += fiy2; |
1625 | f[i_coord_offset+DIM3*2+ZZ2] += fiz2; |
1626 | tx += fix2; |
1627 | ty += fiy2; |
1628 | tz += fiz2; |
1629 | f[i_coord_offset+DIM3*3+XX0] += fix3; |
1630 | f[i_coord_offset+DIM3*3+YY1] += fiy3; |
1631 | f[i_coord_offset+DIM3*3+ZZ2] += fiz3; |
1632 | tx += fix3; |
1633 | ty += fiy3; |
1634 | tz += fiz3; |
1635 | fshift[i_shift_offset+XX0] += tx; |
1636 | fshift[i_shift_offset+YY1] += ty; |
1637 | fshift[i_shift_offset+ZZ2] += tz; |
1638 | |
1639 | /* Increment number of inner iterations */ |
1640 | inneriter += j_index_end - j_index_start; |
1641 | |
1642 | /* Outer loop uses 39 flops */ |
1643 | } |
1644 | |
1645 | /* Increment number of outer iterations */ |
1646 | outeriter += nri; |
1647 | |
1648 | /* Update outer/inner flops */ |
1649 | |
1650 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*39 + inneriter*581)(nrnb)->n[eNR_NBKERNEL_ELEC_VDW_W4W4_F] += outeriter*39 + inneriter *581; |
1651 | } |