2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2009, The GROMACS Development Team
6 * Copyright (c) 2012, by the GROMACS development team, led by
7 * David van der Spoel, Berk Hess, Erik Lindahl, and including many
8 * others, as listed in the AUTHORS file in the top-level source
9 * directory and at http://www.gromacs.org.
11 * GROMACS is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public License
13 * as published by the Free Software Foundation; either version 2.1
14 * of the License, or (at your option) any later version.
16 * GROMACS is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with GROMACS; if not, see
23 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
24 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
26 * If you want to redistribute modifications to GROMACS, please
27 * consider that scientific software is very special. Version
28 * control is crucial - bugs must be traceable. We will be happy to
29 * consider code for inclusion in the official distribution, but
30 * derived work must not be called official GROMACS. Details are found
31 * in the README & COPYING files - if they are missing, get the
32 * official version at http://www.gromacs.org.
34 * To help us fund GROMACS development, we humbly ask that you cite
35 * the research papers on the package. Check out http://www.gromacs.org.
38 /* This is the innermost loop contents for the 4 x N atom SIMD kernel.
39 * This flavor of the kernel calculates interactions of 4 i-atoms
40 * with N j-atoms stored in N wide SIMD registers.
44 /* When calculating RF or Ewald interactions we calculate the electrostatic
45 * forces on excluded atom pairs here in the non-bonded loops.
46 * But when energies and/or virial is required we calculate them
47 * separately to as then it is easier to separate the energy and virial
50 #if defined CHECK_EXCLS && defined CALC_COULOMB
54 /* Without exclusions and energies we only need to mask the cut-off,
55 * this can be faster when we have defined gmx_blendv_pr, i.e. an instruction
56 * that selects from two SIMD registers based on the contents of a third.
58 #if !(defined CHECK_EXCLS || defined CALC_ENERGIES) && defined GMX_HAVE_SIMD_BLENDV && !defined COUNT_PAIRS
59 /* With RF and tabulated Coulomb we replace cmp+and with sub+blendv.
60 * With gcc this is slower, except for RF on Sandy Bridge.
61 * Tested with gcc 4.6.2, 4.6.3 and 4.7.1.
63 #if (defined CALC_COUL_RF || defined CALC_COUL_TAB) && (!defined __GNUC__ || (defined CALC_COUL_RF && defined GMX_X86_AVX_256))
64 #define NBNXN_CUTOFF_USE_BLENDV
66 /* With analytical Ewald we replace cmp+and+and with sub+blendv+blendv.
67 * This is only faster with icc on Sandy Bridge (PS kernel slower than gcc 4.7).
70 #if defined CALC_COUL_EWALD && defined __INTEL_COMPILER && defined GMX_X86_AVX_256
71 #define NBNXN_CUTOFF_USE_BLENDV
76 int cj, aj, ajx, ajy, ajz;
79 /* Energy group indices for two atoms packed into one int */
80 int egp_jj[UNROLLJ/2];
84 /* Interaction (non-exclusion) mask of all 1's or 0's */
91 gmx_mm_pr jx_S, jy_S, jz_S;
92 gmx_mm_pr dx_S0, dy_S0, dz_S0;
93 gmx_mm_pr dx_S1, dy_S1, dz_S1;
94 gmx_mm_pr dx_S2, dy_S2, dz_S2;
95 gmx_mm_pr dx_S3, dy_S3, dz_S3;
96 gmx_mm_pr tx_S0, ty_S0, tz_S0;
97 gmx_mm_pr tx_S1, ty_S1, tz_S1;
98 gmx_mm_pr tx_S2, ty_S2, tz_S2;
99 gmx_mm_pr tx_S3, ty_S3, tz_S3;
100 gmx_mm_pr rsq_S0, rinv_S0, rinvsq_S0;
101 gmx_mm_pr rsq_S1, rinv_S1, rinvsq_S1;
102 gmx_mm_pr rsq_S2, rinv_S2, rinvsq_S2;
103 gmx_mm_pr rsq_S3, rinv_S3, rinvsq_S3;
104 #ifndef NBNXN_CUTOFF_USE_BLENDV
105 /* wco: within cut-off, mask of all 1's or 0's */
111 #ifdef VDW_CUTOFF_CHECK
112 gmx_mm_pr wco_vdw_S0;
113 gmx_mm_pr wco_vdw_S1;
115 gmx_mm_pr wco_vdw_S2;
116 gmx_mm_pr wco_vdw_S3;
121 /* 1/r masked with the interaction mask */
122 gmx_mm_pr rinv_ex_S0;
123 gmx_mm_pr rinv_ex_S1;
124 gmx_mm_pr rinv_ex_S2;
125 gmx_mm_pr rinv_ex_S3;
133 /* The force (PME mesh force) we need to subtract from 1/r^2 */
139 #ifdef CALC_COUL_EWALD
140 gmx_mm_pr brsq_S0, brsq_S1, brsq_S2, brsq_S3;
141 gmx_mm_pr ewcorr_S0, ewcorr_S1, ewcorr_S2, ewcorr_S3;
144 /* frcoul = (1/r - fsub)*r */
150 /* For tables: r, rs=r/sp, rf=floor(rs), frac=rs-rf */
151 gmx_mm_pr r_S0, rs_S0, rf_S0, frac_S0;
152 gmx_mm_pr r_S1, rs_S1, rf_S1, frac_S1;
153 gmx_mm_pr r_S2, rs_S2, rf_S2, frac_S2;
154 gmx_mm_pr r_S3, rs_S3, rf_S3, frac_S3;
155 /* Table index: rs truncated to an int */
156 gmx_epi32 ti_S0, ti_S1, ti_S2, ti_S3;
157 /* Linear force table values */
158 gmx_mm_pr ctab0_S0, ctab1_S0;
159 gmx_mm_pr ctab0_S1, ctab1_S1;
160 gmx_mm_pr ctab0_S2, ctab1_S2;
161 gmx_mm_pr ctab0_S3, ctab1_S3;
163 /* Quadratic energy table value */
170 #if defined CALC_ENERGIES && (defined CALC_COUL_EWALD || defined CALC_COUL_TAB)
171 /* The potential (PME mesh) we need to subtract from 1/r */
178 /* Electrostatic potential */
185 /* The force times 1/r */
193 /* LJ sigma_j/2 and sqrt(epsilon_j) */
194 gmx_mm_pr hsig_j_S, seps_j_S;
195 /* LJ sigma_ij and epsilon_ij */
196 gmx_mm_pr sig_S0, eps_S0;
197 gmx_mm_pr sig_S1, eps_S1;
199 gmx_mm_pr sig_S2, eps_S2;
200 gmx_mm_pr sig_S3, eps_S3;
203 gmx_mm_pr sig2_S0, sig6_S0;
204 gmx_mm_pr sig2_S1, sig6_S1;
206 gmx_mm_pr sig2_S2, sig6_S2;
207 gmx_mm_pr sig2_S3, sig6_S3;
209 #endif /* LJ_COMB_LB */
213 gmx_mm_pr c6s_j_S, c12s_j_S;
216 #if defined LJ_COMB_GEOM || defined LJ_COMB_LB
217 /* Index for loading LJ parameters, complicated when interleaving */
222 /* LJ C6 and C12 parameters, used with geometric comb. rule */
223 gmx_mm_pr c6_S0, c12_S0;
224 gmx_mm_pr c6_S1, c12_S1;
226 gmx_mm_pr c6_S2, c12_S2;
227 gmx_mm_pr c6_S3, c12_S3;
231 /* Intermediate variables for LJ calculation */
233 gmx_mm_pr rinvsix_S0;
234 gmx_mm_pr rinvsix_S1;
236 gmx_mm_pr rinvsix_S2;
237 gmx_mm_pr rinvsix_S3;
241 gmx_mm_pr sir_S0, sir2_S0, sir6_S0;
242 gmx_mm_pr sir_S1, sir2_S1, sir6_S1;
244 gmx_mm_pr sir_S2, sir2_S2, sir6_S2;
245 gmx_mm_pr sir_S3, sir2_S3, sir6_S3;
249 gmx_mm_pr FrLJ6_S0, FrLJ12_S0;
250 gmx_mm_pr FrLJ6_S1, FrLJ12_S1;
252 gmx_mm_pr FrLJ6_S2, FrLJ12_S2;
253 gmx_mm_pr FrLJ6_S3, FrLJ12_S3;
256 gmx_mm_pr VLJ6_S0, VLJ12_S0, VLJ_S0;
257 gmx_mm_pr VLJ6_S1, VLJ12_S1, VLJ_S1;
259 gmx_mm_pr VLJ6_S2, VLJ12_S2, VLJ_S2;
260 gmx_mm_pr VLJ6_S3, VLJ12_S3, VLJ_S3;
265 /* j-cluster index */
268 /* Atom indices (of the first atom in the cluster) */
270 #if defined CALC_LJ && (defined LJ_COMB_GEOM || defined LJ_COMB_LB)
271 #if UNROLLJ == STRIDE
274 aj2 = (cj>>1)*2*STRIDE + (cj & 1)*UNROLLJ;
277 #if UNROLLJ == STRIDE
280 ajx = (cj>>1)*DIM*STRIDE + (cj & 1)*UNROLLJ;
286 #ifdef gmx_checkbitmask_epi32
288 /* Integer mask set and operations, cast result to real */
289 gmx_epi32 mask_pr_S = gmx_set1_epi32(l_cj[cjind].excl);
291 int_S0 = gmx_castsi_pr(gmx_checkbitmask_epi32(mask_pr_S, mask_S0));
292 int_S1 = gmx_castsi_pr(gmx_checkbitmask_epi32(mask_pr_S, mask_S1));
293 int_S2 = gmx_castsi_pr(gmx_checkbitmask_epi32(mask_pr_S, mask_S2));
294 int_S3 = gmx_castsi_pr(gmx_checkbitmask_epi32(mask_pr_S, mask_S3));
298 /* Integer mask set, cast to real and real mask operations */
299 gmx_mm_pr mask_pr_S = gmx_castsi_pr(gmx_set1_epi32(l_cj[cjind].excl));
301 int_S0 = gmx_checkbitmask_pr(mask_pr_S, mask_S0);
302 int_S1 = gmx_checkbitmask_pr(mask_pr_S, mask_S1);
303 int_S2 = gmx_checkbitmask_pr(mask_pr_S, mask_S2);
304 int_S3 = gmx_checkbitmask_pr(mask_pr_S, mask_S3);
309 /* load j atom coordinates */
310 jx_S = gmx_load_pr(x+ajx);
311 jy_S = gmx_load_pr(x+ajy);
312 jz_S = gmx_load_pr(x+ajz);
314 /* Calculate distance */
315 dx_S0 = gmx_sub_pr(ix_S0, jx_S);
316 dy_S0 = gmx_sub_pr(iy_S0, jy_S);
317 dz_S0 = gmx_sub_pr(iz_S0, jz_S);
318 dx_S1 = gmx_sub_pr(ix_S1, jx_S);
319 dy_S1 = gmx_sub_pr(iy_S1, jy_S);
320 dz_S1 = gmx_sub_pr(iz_S1, jz_S);
321 dx_S2 = gmx_sub_pr(ix_S2, jx_S);
322 dy_S2 = gmx_sub_pr(iy_S2, jy_S);
323 dz_S2 = gmx_sub_pr(iz_S2, jz_S);
324 dx_S3 = gmx_sub_pr(ix_S3, jx_S);
325 dy_S3 = gmx_sub_pr(iy_S3, jy_S);
326 dz_S3 = gmx_sub_pr(iz_S3, jz_S);
328 /* rsq = dx*dx+dy*dy+dz*dz */
329 rsq_S0 = gmx_calc_rsq_pr(dx_S0, dy_S0, dz_S0);
330 rsq_S1 = gmx_calc_rsq_pr(dx_S1, dy_S1, dz_S1);
331 rsq_S2 = gmx_calc_rsq_pr(dx_S2, dy_S2, dz_S2);
332 rsq_S3 = gmx_calc_rsq_pr(dx_S3, dy_S3, dz_S3);
334 #ifndef NBNXN_CUTOFF_USE_BLENDV
335 wco_S0 = gmx_cmplt_pr(rsq_S0, rc2_S);
336 wco_S1 = gmx_cmplt_pr(rsq_S1, rc2_S);
337 wco_S2 = gmx_cmplt_pr(rsq_S2, rc2_S);
338 wco_S3 = gmx_cmplt_pr(rsq_S3, rc2_S);
343 /* Only remove the (sub-)diagonal to avoid double counting */
344 #if UNROLLJ == UNROLLI
347 wco_S0 = gmx_and_pr(wco_S0, diag_S0);
348 wco_S1 = gmx_and_pr(wco_S1, diag_S1);
349 wco_S2 = gmx_and_pr(wco_S2, diag_S2);
350 wco_S3 = gmx_and_pr(wco_S3, diag_S3);
353 #if UNROLLJ < UNROLLI
356 wco_S0 = gmx_and_pr(wco_S0, diag0_S0);
357 wco_S1 = gmx_and_pr(wco_S1, diag0_S1);
358 wco_S2 = gmx_and_pr(wco_S2, diag0_S2);
359 wco_S3 = gmx_and_pr(wco_S3, diag0_S3);
361 if (cj == ci_sh*2 + 1)
363 wco_S0 = gmx_and_pr(wco_S0, diag1_S0);
364 wco_S1 = gmx_and_pr(wco_S1, diag1_S1);
365 wco_S2 = gmx_and_pr(wco_S2, diag1_S2);
366 wco_S3 = gmx_and_pr(wco_S3, diag1_S3);
371 wco_S0 = gmx_and_pr(wco_S0, diag0_S0);
372 wco_S1 = gmx_and_pr(wco_S1, diag0_S1);
373 wco_S2 = gmx_and_pr(wco_S2, diag0_S2);
374 wco_S3 = gmx_and_pr(wco_S3, diag0_S3);
376 else if (cj*2 + 1 == ci_sh)
378 wco_S0 = gmx_and_pr(wco_S0, diag1_S0);
379 wco_S1 = gmx_and_pr(wco_S1, diag1_S1);
380 wco_S2 = gmx_and_pr(wco_S2, diag1_S2);
381 wco_S3 = gmx_and_pr(wco_S3, diag1_S3);
385 #else /* EXCL_FORCES */
386 /* No exclusion forces: remove all excluded atom pairs from the list */
387 wco_S0 = gmx_and_pr(wco_S0, int_S0);
388 wco_S1 = gmx_and_pr(wco_S1, int_S1);
389 wco_S2 = gmx_and_pr(wco_S2, int_S2);
390 wco_S3 = gmx_and_pr(wco_S3, int_S3);
397 real tmpa[2*GMX_SIMD_WIDTH_HERE], *tmp;
398 tmp = gmx_simd_align_real(tmpa);
399 for (i = 0; i < UNROLLI; i++)
401 gmx_store_pr(tmp, i == 0 ? wco_S0 : (i == 1 ? wco_S1 : (i == 2 ? wco_S2 : wco_S3)));
402 for (j = 0; j < UNROLLJ; j++)
414 /* For excluded pairs add a small number to avoid r^-6 = NaN */
415 rsq_S0 = gmx_add_pr(rsq_S0, gmx_andnot_pr(int_S0, avoid_sing_S));
416 rsq_S1 = gmx_add_pr(rsq_S1, gmx_andnot_pr(int_S1, avoid_sing_S));
417 rsq_S2 = gmx_add_pr(rsq_S2, gmx_andnot_pr(int_S2, avoid_sing_S));
418 rsq_S3 = gmx_add_pr(rsq_S3, gmx_andnot_pr(int_S3, avoid_sing_S));
423 rinv_S0 = gmx_invsqrt_pr(rsq_S0);
424 rinv_S1 = gmx_invsqrt_pr(rsq_S1);
425 rinv_S2 = gmx_invsqrt_pr(rsq_S2);
426 rinv_S3 = gmx_invsqrt_pr(rsq_S3);
428 GMX_MM_INVSQRT2_PD(rsq_S0, rsq_S1, rinv_S0, rinv_S1);
429 GMX_MM_INVSQRT2_PD(rsq_S2, rsq_S3, rinv_S2, rinv_S3);
433 /* Load parameters for j atom */
434 jq_S = gmx_load_pr(q+aj);
435 qq_S0 = gmx_mul_pr(iq_S0, jq_S);
436 qq_S1 = gmx_mul_pr(iq_S1, jq_S);
437 qq_S2 = gmx_mul_pr(iq_S2, jq_S);
438 qq_S3 = gmx_mul_pr(iq_S3, jq_S);
443 #if !defined LJ_COMB_GEOM && !defined LJ_COMB_LB && !defined FIX_LJ_C
444 load_lj_pair_params(nbfp0, type, aj, c6_S0, c12_S0);
445 load_lj_pair_params(nbfp1, type, aj, c6_S1, c12_S1);
447 load_lj_pair_params(nbfp2, type, aj, c6_S2, c12_S2);
448 load_lj_pair_params(nbfp3, type, aj, c6_S3, c12_S3);
450 #endif /* not defined any LJ rule */
453 c6s_j_S = gmx_load_pr(ljc+aj2+0);
454 c12s_j_S = gmx_load_pr(ljc+aj2+STRIDE);
455 c6_S0 = gmx_mul_pr(c6s_S0, c6s_j_S );
456 c6_S1 = gmx_mul_pr(c6s_S1, c6s_j_S );
458 c6_S2 = gmx_mul_pr(c6s_S2, c6s_j_S );
459 c6_S3 = gmx_mul_pr(c6s_S3, c6s_j_S );
461 c12_S0 = gmx_mul_pr(c12s_S0, c12s_j_S);
462 c12_S1 = gmx_mul_pr(c12s_S1, c12s_j_S);
464 c12_S2 = gmx_mul_pr(c12s_S2, c12s_j_S);
465 c12_S3 = gmx_mul_pr(c12s_S3, c12s_j_S);
467 #endif /* LJ_COMB_GEOM */
470 hsig_j_S = gmx_load_pr(ljc+aj2+0);
471 seps_j_S = gmx_load_pr(ljc+aj2+STRIDE);
473 sig_S0 = gmx_add_pr(hsig_i_S0, hsig_j_S);
474 sig_S1 = gmx_add_pr(hsig_i_S1, hsig_j_S);
475 eps_S0 = gmx_mul_pr(seps_i_S0, seps_j_S);
476 eps_S1 = gmx_mul_pr(seps_i_S1, seps_j_S);
478 sig_S2 = gmx_add_pr(hsig_i_S2, hsig_j_S);
479 sig_S3 = gmx_add_pr(hsig_i_S3, hsig_j_S);
480 eps_S2 = gmx_mul_pr(seps_i_S2, seps_j_S);
481 eps_S3 = gmx_mul_pr(seps_i_S3, seps_j_S);
483 #endif /* LJ_COMB_LB */
487 #ifndef NBNXN_CUTOFF_USE_BLENDV
488 rinv_S0 = gmx_blendzero_pr(rinv_S0, wco_S0);
489 rinv_S1 = gmx_blendzero_pr(rinv_S1, wco_S1);
490 rinv_S2 = gmx_blendzero_pr(rinv_S2, wco_S2);
491 rinv_S3 = gmx_blendzero_pr(rinv_S3, wco_S3);
493 /* We only need to mask for the cut-off: blendv is faster */
494 rinv_S0 = gmx_blendv_pr(rinv_S0, zero_S, gmx_sub_pr(rc2_S, rsq_S0));
495 rinv_S1 = gmx_blendv_pr(rinv_S1, zero_S, gmx_sub_pr(rc2_S, rsq_S1));
496 rinv_S2 = gmx_blendv_pr(rinv_S2, zero_S, gmx_sub_pr(rc2_S, rsq_S2));
497 rinv_S3 = gmx_blendv_pr(rinv_S3, zero_S, gmx_sub_pr(rc2_S, rsq_S3));
500 rinvsq_S0 = gmx_mul_pr(rinv_S0, rinv_S0);
501 rinvsq_S1 = gmx_mul_pr(rinv_S1, rinv_S1);
502 rinvsq_S2 = gmx_mul_pr(rinv_S2, rinv_S2);
503 rinvsq_S3 = gmx_mul_pr(rinv_S3, rinv_S3);
506 /* Note that here we calculate force*r, not the usual force/r.
507 * This allows avoiding masking the reaction-field contribution,
508 * as frcoul is later multiplied by rinvsq which has been
509 * masked with the cut-off check.
513 /* Only add 1/r for non-excluded atom pairs */
514 rinv_ex_S0 = gmx_blendzero_pr(rinv_S0, int_S0);
515 rinv_ex_S1 = gmx_blendzero_pr(rinv_S1, int_S1);
516 rinv_ex_S2 = gmx_blendzero_pr(rinv_S2, int_S2);
517 rinv_ex_S3 = gmx_blendzero_pr(rinv_S3, int_S3);
519 /* No exclusion forces, we always need 1/r */
520 #define rinv_ex_S0 rinv_S0
521 #define rinv_ex_S1 rinv_S1
522 #define rinv_ex_S2 rinv_S2
523 #define rinv_ex_S3 rinv_S3
527 /* Electrostatic interactions */
528 frcoul_S0 = gmx_mul_pr(qq_S0, gmx_add_pr(rinv_ex_S0, gmx_mul_pr(rsq_S0, mrc_3_S)));
529 frcoul_S1 = gmx_mul_pr(qq_S1, gmx_add_pr(rinv_ex_S1, gmx_mul_pr(rsq_S1, mrc_3_S)));
530 frcoul_S2 = gmx_mul_pr(qq_S2, gmx_add_pr(rinv_ex_S2, gmx_mul_pr(rsq_S2, mrc_3_S)));
531 frcoul_S3 = gmx_mul_pr(qq_S3, gmx_add_pr(rinv_ex_S3, gmx_mul_pr(rsq_S3, mrc_3_S)));
534 vcoul_S0 = gmx_mul_pr(qq_S0, gmx_add_pr(rinv_ex_S0, gmx_add_pr(gmx_mul_pr(rsq_S0, hrc_3_S), moh_rc_S)));
535 vcoul_S1 = gmx_mul_pr(qq_S1, gmx_add_pr(rinv_ex_S1, gmx_add_pr(gmx_mul_pr(rsq_S1, hrc_3_S), moh_rc_S)));
536 vcoul_S2 = gmx_mul_pr(qq_S2, gmx_add_pr(rinv_ex_S2, gmx_add_pr(gmx_mul_pr(rsq_S2, hrc_3_S), moh_rc_S)));
537 vcoul_S3 = gmx_mul_pr(qq_S3, gmx_add_pr(rinv_ex_S3, gmx_add_pr(gmx_mul_pr(rsq_S3, hrc_3_S), moh_rc_S)));
541 #ifdef CALC_COUL_EWALD
542 /* We need to mask (or limit) rsq for the cut-off,
543 * as large distances can cause an overflow in gmx_pmecorrF/V.
545 #ifndef NBNXN_CUTOFF_USE_BLENDV
546 brsq_S0 = gmx_mul_pr(beta2_S, gmx_blendzero_pr(rsq_S0, wco_S0));
547 brsq_S1 = gmx_mul_pr(beta2_S, gmx_blendzero_pr(rsq_S1, wco_S1));
548 brsq_S2 = gmx_mul_pr(beta2_S, gmx_blendzero_pr(rsq_S2, wco_S2));
549 brsq_S3 = gmx_mul_pr(beta2_S, gmx_blendzero_pr(rsq_S3, wco_S3));
551 /* Strangely, putting mul on a separate line is slower (icc 13) */
552 brsq_S0 = gmx_mul_pr(beta2_S, gmx_blendv_pr(rsq_S0, zero_S, gmx_sub_pr(rc2_S, rsq_S0)));
553 brsq_S1 = gmx_mul_pr(beta2_S, gmx_blendv_pr(rsq_S1, zero_S, gmx_sub_pr(rc2_S, rsq_S1)));
554 brsq_S2 = gmx_mul_pr(beta2_S, gmx_blendv_pr(rsq_S2, zero_S, gmx_sub_pr(rc2_S, rsq_S2)));
555 brsq_S3 = gmx_mul_pr(beta2_S, gmx_blendv_pr(rsq_S3, zero_S, gmx_sub_pr(rc2_S, rsq_S3)));
557 ewcorr_S0 = gmx_mul_pr(gmx_pmecorrF_pr(brsq_S0), beta_S);
558 ewcorr_S1 = gmx_mul_pr(gmx_pmecorrF_pr(brsq_S1), beta_S);
559 ewcorr_S2 = gmx_mul_pr(gmx_pmecorrF_pr(brsq_S2), beta_S);
560 ewcorr_S3 = gmx_mul_pr(gmx_pmecorrF_pr(brsq_S3), beta_S);
561 frcoul_S0 = gmx_mul_pr(qq_S0, gmx_add_pr(rinv_ex_S0, gmx_mul_pr(ewcorr_S0, brsq_S0)));
562 frcoul_S1 = gmx_mul_pr(qq_S1, gmx_add_pr(rinv_ex_S1, gmx_mul_pr(ewcorr_S1, brsq_S1)));
563 frcoul_S2 = gmx_mul_pr(qq_S2, gmx_add_pr(rinv_ex_S2, gmx_mul_pr(ewcorr_S2, brsq_S2)));
564 frcoul_S3 = gmx_mul_pr(qq_S3, gmx_add_pr(rinv_ex_S3, gmx_mul_pr(ewcorr_S3, brsq_S3)));
567 vc_sub_S0 = gmx_mul_pr(gmx_pmecorrV_pr(brsq_S0), beta_S);
568 vc_sub_S1 = gmx_mul_pr(gmx_pmecorrV_pr(brsq_S1), beta_S);
569 vc_sub_S2 = gmx_mul_pr(gmx_pmecorrV_pr(brsq_S2), beta_S);
570 vc_sub_S3 = gmx_mul_pr(gmx_pmecorrV_pr(brsq_S3), beta_S);
573 #endif /* CALC_COUL_EWALD */
576 /* Electrostatic interactions */
577 r_S0 = gmx_mul_pr(rsq_S0, rinv_S0);
578 r_S1 = gmx_mul_pr(rsq_S1, rinv_S1);
579 r_S2 = gmx_mul_pr(rsq_S2, rinv_S2);
580 r_S3 = gmx_mul_pr(rsq_S3, rinv_S3);
581 /* Convert r to scaled table units */
582 rs_S0 = gmx_mul_pr(r_S0, invtsp_S);
583 rs_S1 = gmx_mul_pr(r_S1, invtsp_S);
584 rs_S2 = gmx_mul_pr(r_S2, invtsp_S);
585 rs_S3 = gmx_mul_pr(r_S3, invtsp_S);
586 /* Truncate scaled r to an int */
587 ti_S0 = gmx_cvttpr_epi32(rs_S0);
588 ti_S1 = gmx_cvttpr_epi32(rs_S1);
589 ti_S2 = gmx_cvttpr_epi32(rs_S2);
590 ti_S3 = gmx_cvttpr_epi32(rs_S3);
591 #ifdef GMX_HAVE_SIMD_FLOOR
592 /* SSE4.1 floor is faster than gmx_cvtepi32_ps int->float cast */
593 rf_S0 = gmx_floor_pr(rs_S0);
594 rf_S1 = gmx_floor_pr(rs_S1);
595 rf_S2 = gmx_floor_pr(rs_S2);
596 rf_S3 = gmx_floor_pr(rs_S3);
598 rf_S0 = gmx_cvtepi32_pr(ti_S0);
599 rf_S1 = gmx_cvtepi32_pr(ti_S1);
600 rf_S2 = gmx_cvtepi32_pr(ti_S2);
601 rf_S3 = gmx_cvtepi32_pr(ti_S3);
603 frac_S0 = gmx_sub_pr(rs_S0, rf_S0);
604 frac_S1 = gmx_sub_pr(rs_S1, rf_S1);
605 frac_S2 = gmx_sub_pr(rs_S2, rf_S2);
606 frac_S3 = gmx_sub_pr(rs_S3, rf_S3);
608 /* Load and interpolate table forces and possibly energies.
609 * Force and energy can be combined in one table, stride 4: FDV0
610 * or in two separate tables with stride 1: F and V
611 * Currently single precision uses FDV0, double F and V.
613 #ifndef CALC_ENERGIES
614 load_table_f(tab_coul_F, ti_S0, ti0, ctab0_S0, ctab1_S0);
615 load_table_f(tab_coul_F, ti_S1, ti1, ctab0_S1, ctab1_S1);
616 load_table_f(tab_coul_F, ti_S2, ti2, ctab0_S2, ctab1_S2);
617 load_table_f(tab_coul_F, ti_S3, ti3, ctab0_S3, ctab1_S3);
620 load_table_f_v(tab_coul_F, ti_S0, ti0, ctab0_S0, ctab1_S0, ctabv_S0);
621 load_table_f_v(tab_coul_F, ti_S1, ti1, ctab0_S1, ctab1_S1, ctabv_S1);
622 load_table_f_v(tab_coul_F, ti_S2, ti2, ctab0_S2, ctab1_S2, ctabv_S2);
623 load_table_f_v(tab_coul_F, ti_S3, ti3, ctab0_S3, ctab1_S3, ctabv_S3);
625 load_table_f_v(tab_coul_F, tab_coul_V, ti_S0, ti0, ctab0_S0, ctab1_S0, ctabv_S0);
626 load_table_f_v(tab_coul_F, tab_coul_V, ti_S1, ti1, ctab0_S1, ctab1_S1, ctabv_S1);
627 load_table_f_v(tab_coul_F, tab_coul_V, ti_S2, ti2, ctab0_S2, ctab1_S2, ctabv_S2);
628 load_table_f_v(tab_coul_F, tab_coul_V, ti_S3, ti3, ctab0_S3, ctab1_S3, ctabv_S3);
631 fsub_S0 = gmx_add_pr(ctab0_S0, gmx_mul_pr(frac_S0, ctab1_S0));
632 fsub_S1 = gmx_add_pr(ctab0_S1, gmx_mul_pr(frac_S1, ctab1_S1));
633 fsub_S2 = gmx_add_pr(ctab0_S2, gmx_mul_pr(frac_S2, ctab1_S2));
634 fsub_S3 = gmx_add_pr(ctab0_S3, gmx_mul_pr(frac_S3, ctab1_S3));
635 frcoul_S0 = gmx_mul_pr(qq_S0, gmx_sub_pr(rinv_ex_S0, gmx_mul_pr(fsub_S0, r_S0)));
636 frcoul_S1 = gmx_mul_pr(qq_S1, gmx_sub_pr(rinv_ex_S1, gmx_mul_pr(fsub_S1, r_S1)));
637 frcoul_S2 = gmx_mul_pr(qq_S2, gmx_sub_pr(rinv_ex_S2, gmx_mul_pr(fsub_S2, r_S2)));
638 frcoul_S3 = gmx_mul_pr(qq_S3, gmx_sub_pr(rinv_ex_S3, gmx_mul_pr(fsub_S3, r_S3)));
641 vc_sub_S0 = gmx_add_pr(ctabv_S0, gmx_mul_pr(gmx_mul_pr(mhalfsp_S, frac_S0), gmx_add_pr(ctab0_S0, fsub_S0)));
642 vc_sub_S1 = gmx_add_pr(ctabv_S1, gmx_mul_pr(gmx_mul_pr(mhalfsp_S, frac_S1), gmx_add_pr(ctab0_S1, fsub_S1)));
643 vc_sub_S2 = gmx_add_pr(ctabv_S2, gmx_mul_pr(gmx_mul_pr(mhalfsp_S, frac_S2), gmx_add_pr(ctab0_S2, fsub_S2)));
644 vc_sub_S3 = gmx_add_pr(ctabv_S3, gmx_mul_pr(gmx_mul_pr(mhalfsp_S, frac_S3), gmx_add_pr(ctab0_S3, fsub_S3)));
646 #endif /* CALC_COUL_TAB */
648 #if defined CALC_ENERGIES && (defined CALC_COUL_EWALD || defined CALC_COUL_TAB)
649 #ifndef NO_SHIFT_EWALD
650 /* Add Ewald potential shift to vc_sub for convenience */
652 vc_sub_S0 = gmx_add_pr(vc_sub_S0, gmx_blendzero_pr(sh_ewald_S, int_S0));
653 vc_sub_S1 = gmx_add_pr(vc_sub_S1, gmx_blendzero_pr(sh_ewald_S, int_S1));
654 vc_sub_S2 = gmx_add_pr(vc_sub_S2, gmx_blendzero_pr(sh_ewald_S, int_S2));
655 vc_sub_S3 = gmx_add_pr(vc_sub_S3, gmx_blendzero_pr(sh_ewald_S, int_S3));
657 vc_sub_S0 = gmx_add_pr(vc_sub_S0, sh_ewald_S);
658 vc_sub_S1 = gmx_add_pr(vc_sub_S1, sh_ewald_S);
659 vc_sub_S2 = gmx_add_pr(vc_sub_S2, sh_ewald_S);
660 vc_sub_S3 = gmx_add_pr(vc_sub_S3, sh_ewald_S);
664 vcoul_S0 = gmx_mul_pr(qq_S0, gmx_sub_pr(rinv_ex_S0, vc_sub_S0));
665 vcoul_S1 = gmx_mul_pr(qq_S1, gmx_sub_pr(rinv_ex_S1, vc_sub_S1));
666 vcoul_S2 = gmx_mul_pr(qq_S2, gmx_sub_pr(rinv_ex_S2, vc_sub_S2));
667 vcoul_S3 = gmx_mul_pr(qq_S3, gmx_sub_pr(rinv_ex_S3, vc_sub_S3));
672 /* Mask energy for cut-off and diagonal */
673 vcoul_S0 = gmx_blendzero_pr(vcoul_S0, wco_S0);
674 vcoul_S1 = gmx_blendzero_pr(vcoul_S1, wco_S1);
675 vcoul_S2 = gmx_blendzero_pr(vcoul_S2, wco_S2);
676 vcoul_S3 = gmx_blendzero_pr(vcoul_S3, wco_S3);
679 #endif /* CALC_COULOMB */
682 /* Lennard-Jones interaction */
684 #ifdef VDW_CUTOFF_CHECK
685 wco_vdw_S0 = gmx_cmplt_pr(rsq_S0, rcvdw2_S);
686 wco_vdw_S1 = gmx_cmplt_pr(rsq_S1, rcvdw2_S);
688 wco_vdw_S2 = gmx_cmplt_pr(rsq_S2, rcvdw2_S);
689 wco_vdw_S3 = gmx_cmplt_pr(rsq_S3, rcvdw2_S);
692 /* Same cut-off for Coulomb and VdW, reuse the registers */
693 #define wco_vdw_S0 wco_S0
694 #define wco_vdw_S1 wco_S1
695 #define wco_vdw_S2 wco_S2
696 #define wco_vdw_S3 wco_S3
700 rinvsix_S0 = gmx_mul_pr(rinvsq_S0, gmx_mul_pr(rinvsq_S0, rinvsq_S0));
701 rinvsix_S1 = gmx_mul_pr(rinvsq_S1, gmx_mul_pr(rinvsq_S1, rinvsq_S1));
703 rinvsix_S0 = gmx_blendzero_pr(rinvsix_S0, int_S0);
704 rinvsix_S1 = gmx_blendzero_pr(rinvsix_S1, int_S1);
707 rinvsix_S2 = gmx_mul_pr(rinvsq_S2, gmx_mul_pr(rinvsq_S2, rinvsq_S2));
708 rinvsix_S3 = gmx_mul_pr(rinvsq_S3, gmx_mul_pr(rinvsq_S3, rinvsq_S3));
710 rinvsix_S2 = gmx_blendzero_pr(rinvsix_S2, int_S2);
711 rinvsix_S3 = gmx_blendzero_pr(rinvsix_S3, int_S3);
714 #ifdef VDW_CUTOFF_CHECK
715 rinvsix_S0 = gmx_blendzero_pr(rinvsix_S0, wco_vdw_S0);
716 rinvsix_S1 = gmx_blendzero_pr(rinvsix_S1, wco_vdw_S1);
718 rinvsix_S2 = gmx_blendzero_pr(rinvsix_S2, wco_vdw_S2);
719 rinvsix_S3 = gmx_blendzero_pr(rinvsix_S3, wco_vdw_S3);
722 FrLJ6_S0 = gmx_mul_pr(c6_S0, rinvsix_S0);
723 FrLJ6_S1 = gmx_mul_pr(c6_S1, rinvsix_S1);
725 FrLJ6_S2 = gmx_mul_pr(c6_S2, rinvsix_S2);
726 FrLJ6_S3 = gmx_mul_pr(c6_S3, rinvsix_S3);
728 FrLJ12_S0 = gmx_mul_pr(c12_S0, gmx_mul_pr(rinvsix_S0, rinvsix_S0));
729 FrLJ12_S1 = gmx_mul_pr(c12_S1, gmx_mul_pr(rinvsix_S1, rinvsix_S1));
731 FrLJ12_S2 = gmx_mul_pr(c12_S2, gmx_mul_pr(rinvsix_S2, rinvsix_S2));
732 FrLJ12_S3 = gmx_mul_pr(c12_S3, gmx_mul_pr(rinvsix_S3, rinvsix_S3));
734 #endif /* not LJ_COMB_LB */
737 sir_S0 = gmx_mul_pr(sig_S0, rinv_S0);
738 sir_S1 = gmx_mul_pr(sig_S1, rinv_S1);
740 sir_S2 = gmx_mul_pr(sig_S2, rinv_S2);
741 sir_S3 = gmx_mul_pr(sig_S3, rinv_S3);
743 sir2_S0 = gmx_mul_pr(sir_S0, sir_S0);
744 sir2_S1 = gmx_mul_pr(sir_S1, sir_S1);
746 sir2_S2 = gmx_mul_pr(sir_S2, sir_S2);
747 sir2_S3 = gmx_mul_pr(sir_S3, sir_S3);
749 sir6_S0 = gmx_mul_pr(sir2_S0, gmx_mul_pr(sir2_S0, sir2_S0));
750 sir6_S1 = gmx_mul_pr(sir2_S1, gmx_mul_pr(sir2_S1, sir2_S1));
752 sir6_S0 = gmx_blendzero_pr(sir6_S0, int_S0);
753 sir6_S1 = gmx_blendzero_pr(sir6_S1, int_S1);
756 sir6_S2 = gmx_mul_pr(sir2_S2, gmx_mul_pr(sir2_S2, sir2_S2));
757 sir6_S3 = gmx_mul_pr(sir2_S3, gmx_mul_pr(sir2_S3, sir2_S3));
759 sir6_S2 = gmx_blendzero_pr(sir6_S2, int_S2);
760 sir6_S3 = gmx_blendzero_pr(sir6_S3, int_S3);
763 #ifdef VDW_CUTOFF_CHECK
764 sir6_S0 = gmx_blendzero_pr(sir6_S0, wco_vdw_S0);
765 sir6_S1 = gmx_blendzero_pr(sir6_S1, wco_vdw_S1);
767 sir6_S2 = gmx_blendzero_pr(sir6_S2, wco_vdw_S2);
768 sir6_S3 = gmx_blendzero_pr(sir6_S3, wco_vdw_S3);
771 FrLJ6_S0 = gmx_mul_pr(eps_S0, sir6_S0);
772 FrLJ6_S1 = gmx_mul_pr(eps_S1, sir6_S1);
774 FrLJ6_S2 = gmx_mul_pr(eps_S2, sir6_S2);
775 FrLJ6_S3 = gmx_mul_pr(eps_S3, sir6_S3);
777 FrLJ12_S0 = gmx_mul_pr(FrLJ6_S0, sir6_S0);
778 FrLJ12_S1 = gmx_mul_pr(FrLJ6_S1, sir6_S1);
780 FrLJ12_S2 = gmx_mul_pr(FrLJ6_S2, sir6_S2);
781 FrLJ12_S3 = gmx_mul_pr(FrLJ6_S3, sir6_S3);
783 #if defined CALC_ENERGIES
784 /* We need C6 and C12 to calculate the LJ potential shift */
785 sig2_S0 = gmx_mul_pr(sig_S0, sig_S0);
786 sig2_S1 = gmx_mul_pr(sig_S1, sig_S1);
788 sig2_S2 = gmx_mul_pr(sig_S2, sig_S2);
789 sig2_S3 = gmx_mul_pr(sig_S3, sig_S3);
791 sig6_S0 = gmx_mul_pr(sig2_S0, gmx_mul_pr(sig2_S0, sig2_S0));
792 sig6_S1 = gmx_mul_pr(sig2_S1, gmx_mul_pr(sig2_S1, sig2_S1));
794 sig6_S2 = gmx_mul_pr(sig2_S2, gmx_mul_pr(sig2_S2, sig2_S2));
795 sig6_S3 = gmx_mul_pr(sig2_S3, gmx_mul_pr(sig2_S3, sig2_S3));
797 c6_S0 = gmx_mul_pr(eps_S0, sig6_S0);
798 c6_S1 = gmx_mul_pr(eps_S1, sig6_S1);
800 c6_S2 = gmx_mul_pr(eps_S2, sig6_S2);
801 c6_S3 = gmx_mul_pr(eps_S3, sig6_S3);
803 c12_S0 = gmx_mul_pr(c6_S0, sig6_S0);
804 c12_S1 = gmx_mul_pr(c6_S1, sig6_S1);
806 c12_S2 = gmx_mul_pr(c6_S2, sig6_S2);
807 c12_S3 = gmx_mul_pr(c6_S3, sig6_S3);
810 #endif /* LJ_COMB_LB */
816 /* Extract the group pair index per j pair.
817 * Energy groups are stored per i-cluster, so things get
818 * complicated when the i- and j-cluster size don't match.
823 egps_j = nbat->energrp[cj>>1];
824 egp_jj[0] = ((egps_j >> ((cj & 1)*egps_jshift)) & egps_jmask)*egps_jstride;
826 /* We assume UNROLLI <= UNROLLJ */
828 for (jdi = 0; jdi < UNROLLJ/UNROLLI; jdi++)
831 egps_j = nbat->energrp[cj*(UNROLLJ/UNROLLI)+jdi];
832 for (jj = 0; jj < (UNROLLI/2); jj++)
834 egp_jj[jdi*(UNROLLI/2)+jj] = ((egps_j >> (jj*egps_jshift)) & egps_jmask)*egps_jstride;
842 #ifndef ENERGY_GROUPS
843 vctot_S = gmx_add_pr(vctot_S, gmx_sum4_pr(vcoul_S0, vcoul_S1, vcoul_S2, vcoul_S3));
845 add_ener_grp(vcoul_S0, vctp[0], egp_jj);
846 add_ener_grp(vcoul_S1, vctp[1], egp_jj);
847 add_ener_grp(vcoul_S2, vctp[2], egp_jj);
848 add_ener_grp(vcoul_S3, vctp[3], egp_jj);
853 /* Calculate the LJ energies */
854 VLJ6_S0 = gmx_mul_pr(sixth_S, gmx_sub_pr(FrLJ6_S0, gmx_mul_pr(c6_S0, sh_invrc6_S)));
855 VLJ6_S1 = gmx_mul_pr(sixth_S, gmx_sub_pr(FrLJ6_S1, gmx_mul_pr(c6_S1, sh_invrc6_S)));
857 VLJ6_S2 = gmx_mul_pr(sixth_S, gmx_sub_pr(FrLJ6_S2, gmx_mul_pr(c6_S2, sh_invrc6_S)));
858 VLJ6_S3 = gmx_mul_pr(sixth_S, gmx_sub_pr(FrLJ6_S3, gmx_mul_pr(c6_S3, sh_invrc6_S)));
860 VLJ12_S0 = gmx_mul_pr(twelveth_S, gmx_sub_pr(FrLJ12_S0, gmx_mul_pr(c12_S0, sh_invrc12_S)));
861 VLJ12_S1 = gmx_mul_pr(twelveth_S, gmx_sub_pr(FrLJ12_S1, gmx_mul_pr(c12_S1, sh_invrc12_S)));
863 VLJ12_S2 = gmx_mul_pr(twelveth_S, gmx_sub_pr(FrLJ12_S2, gmx_mul_pr(c12_S2, sh_invrc12_S)));
864 VLJ12_S3 = gmx_mul_pr(twelveth_S, gmx_sub_pr(FrLJ12_S3, gmx_mul_pr(c12_S3, sh_invrc12_S)));
867 VLJ_S0 = gmx_sub_pr(VLJ12_S0, VLJ6_S0);
868 VLJ_S1 = gmx_sub_pr(VLJ12_S1, VLJ6_S1);
870 VLJ_S2 = gmx_sub_pr(VLJ12_S2, VLJ6_S2);
871 VLJ_S3 = gmx_sub_pr(VLJ12_S3, VLJ6_S3);
873 /* The potential shift should be removed for pairs beyond cut-off */
874 VLJ_S0 = gmx_blendzero_pr(VLJ_S0, wco_vdw_S0);
875 VLJ_S1 = gmx_blendzero_pr(VLJ_S1, wco_vdw_S1);
877 VLJ_S2 = gmx_blendzero_pr(VLJ_S2, wco_vdw_S2);
878 VLJ_S3 = gmx_blendzero_pr(VLJ_S3, wco_vdw_S3);
881 /* The potential shift should be removed for excluded pairs */
882 VLJ_S0 = gmx_blendzero_pr(VLJ_S0, int_S0);
883 VLJ_S1 = gmx_blendzero_pr(VLJ_S1, int_S1);
885 VLJ_S2 = gmx_blendzero_pr(VLJ_S2, int_S2);
886 VLJ_S3 = gmx_blendzero_pr(VLJ_S3, int_S3);
889 #ifndef ENERGY_GROUPS
890 Vvdwtot_S = gmx_add_pr(Vvdwtot_S,
892 gmx_sum4_pr(VLJ_S0, VLJ_S1, VLJ_S2, VLJ_S3)
894 gmx_add_pr(VLJ_S0, VLJ_S1)
898 add_ener_grp(VLJ_S0, vvdwtp[0], egp_jj);
899 add_ener_grp(VLJ_S1, vvdwtp[1], egp_jj);
901 add_ener_grp(VLJ_S2, vvdwtp[2], egp_jj);
902 add_ener_grp(VLJ_S3, vvdwtp[3], egp_jj);
906 #endif /* CALC_ENERGIES */
909 fscal_S0 = gmx_mul_pr(rinvsq_S0,
911 gmx_add_pr(frcoul_S0,
915 gmx_sub_pr(FrLJ12_S0, FrLJ6_S0)));
916 fscal_S1 = gmx_mul_pr(rinvsq_S1,
918 gmx_add_pr(frcoul_S1,
922 gmx_sub_pr(FrLJ12_S1, FrLJ6_S1)));
924 fscal_S0 = gmx_mul_pr(rinvsq_S0, frcoul_S0);
925 fscal_S1 = gmx_mul_pr(rinvsq_S1, frcoul_S1);
927 #if defined CALC_LJ && !defined HALF_LJ
928 fscal_S2 = gmx_mul_pr(rinvsq_S2,
930 gmx_add_pr(frcoul_S2,
934 gmx_sub_pr(FrLJ12_S2, FrLJ6_S2)));
935 fscal_S3 = gmx_mul_pr(rinvsq_S3,
937 gmx_add_pr(frcoul_S3,
941 gmx_sub_pr(FrLJ12_S3, FrLJ6_S3)));
943 /* Atom 2 and 3 don't have LJ, so only add Coulomb forces */
944 fscal_S2 = gmx_mul_pr(rinvsq_S2, frcoul_S2);
945 fscal_S3 = gmx_mul_pr(rinvsq_S3, frcoul_S3);
948 /* Calculate temporary vectorial force */
949 tx_S0 = gmx_mul_pr(fscal_S0, dx_S0);
950 tx_S1 = gmx_mul_pr(fscal_S1, dx_S1);
951 tx_S2 = gmx_mul_pr(fscal_S2, dx_S2);
952 tx_S3 = gmx_mul_pr(fscal_S3, dx_S3);
953 ty_S0 = gmx_mul_pr(fscal_S0, dy_S0);
954 ty_S1 = gmx_mul_pr(fscal_S1, dy_S1);
955 ty_S2 = gmx_mul_pr(fscal_S2, dy_S2);
956 ty_S3 = gmx_mul_pr(fscal_S3, dy_S3);
957 tz_S0 = gmx_mul_pr(fscal_S0, dz_S0);
958 tz_S1 = gmx_mul_pr(fscal_S1, dz_S1);
959 tz_S2 = gmx_mul_pr(fscal_S2, dz_S2);
960 tz_S3 = gmx_mul_pr(fscal_S3, dz_S3);
962 /* Increment i atom force */
963 fix_S0 = gmx_add_pr(fix_S0, tx_S0);
964 fix_S1 = gmx_add_pr(fix_S1, tx_S1);
965 fix_S2 = gmx_add_pr(fix_S2, tx_S2);
966 fix_S3 = gmx_add_pr(fix_S3, tx_S3);
967 fiy_S0 = gmx_add_pr(fiy_S0, ty_S0);
968 fiy_S1 = gmx_add_pr(fiy_S1, ty_S1);
969 fiy_S2 = gmx_add_pr(fiy_S2, ty_S2);
970 fiy_S3 = gmx_add_pr(fiy_S3, ty_S3);
971 fiz_S0 = gmx_add_pr(fiz_S0, tz_S0);
972 fiz_S1 = gmx_add_pr(fiz_S1, tz_S1);
973 fiz_S2 = gmx_add_pr(fiz_S2, tz_S2);
974 fiz_S3 = gmx_add_pr(fiz_S3, tz_S3);
976 /* Decrement j atom force */
978 gmx_sub_pr( gmx_load_pr(f+ajx), gmx_sum4_pr(tx_S0, tx_S1, tx_S2, tx_S3) ));
980 gmx_sub_pr( gmx_load_pr(f+ajy), gmx_sum4_pr(ty_S0, ty_S1, ty_S2, ty_S3) ));
982 gmx_sub_pr( gmx_load_pr(f+ajz), gmx_sum4_pr(tz_S0, tz_S1, tz_S2, tz_S3) ));
995 #undef NBNXN_CUTOFF_USE_BLENDV