2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014,2015, 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.
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.
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.
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.
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.
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.
36 * Note: this file was generated by the GROMACS sse2_single kernel generator.
44 #include "../nb_kernel.h"
45 #include "gromacs/math/vec.h"
46 #include "gromacs/legacyheaders/nrnb.h"
48 #include "gromacs/simd/math_x86_sse2_single.h"
49 #include "kernelutil_x86_sse2_single.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomP1P1_VF_sse2_single
53 * Electrostatics interaction: ReactionField
54 * VdW interaction: None
55 * Geometry: Particle-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecRF_VdwNone_GeomP1P1_VF_sse2_single
60 (t_nblist
* gmx_restrict nlist
,
61 rvec
* gmx_restrict xx
,
62 rvec
* gmx_restrict ff
,
63 t_forcerec
* gmx_restrict fr
,
64 t_mdatoms
* gmx_restrict mdatoms
,
65 nb_kernel_data_t gmx_unused
* gmx_restrict kernel_data
,
66 t_nrnb
* gmx_restrict nrnb
)
68 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
69 * just 0 for non-waters.
70 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
71 * jnr indices corresponding to data put in the four positions in the SIMD register.
73 int i_shift_offset
,i_coord_offset
,outeriter
,inneriter
;
74 int j_index_start
,j_index_end
,jidx
,nri
,inr
,ggid
,iidx
;
75 int jnrA
,jnrB
,jnrC
,jnrD
;
76 int jnrlistA
,jnrlistB
,jnrlistC
,jnrlistD
;
77 int j_coord_offsetA
,j_coord_offsetB
,j_coord_offsetC
,j_coord_offsetD
;
78 int *iinr
,*jindex
,*jjnr
,*shiftidx
,*gid
;
80 real
*shiftvec
,*fshift
,*x
,*f
;
81 real
*fjptrA
,*fjptrB
,*fjptrC
,*fjptrD
;
83 __m128 tx
,ty
,tz
,fscal
,rcutoff
,rcutoff2
,jidxall
;
85 __m128 ix0
,iy0
,iz0
,fix0
,fiy0
,fiz0
,iq0
,isai0
;
86 int vdwjidx0A
,vdwjidx0B
,vdwjidx0C
,vdwjidx0D
;
87 __m128 jx0
,jy0
,jz0
,fjx0
,fjy0
,fjz0
,jq0
,isaj0
;
88 __m128 dx00
,dy00
,dz00
,rsq00
,rinv00
,rinvsq00
,r00
,qq00
,c6_00
,c12_00
;
89 __m128 velec
,felec
,velecsum
,facel
,crf
,krf
,krf2
;
91 __m128 dummy_mask
,cutoff_mask
;
92 __m128 signbit
= _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
93 __m128 one
= _mm_set1_ps(1.0);
94 __m128 two
= _mm_set1_ps(2.0);
100 jindex
= nlist
->jindex
;
102 shiftidx
= nlist
->shift
;
104 shiftvec
= fr
->shift_vec
[0];
105 fshift
= fr
->fshift
[0];
106 facel
= _mm_set1_ps(fr
->epsfac
);
107 charge
= mdatoms
->chargeA
;
108 krf
= _mm_set1_ps(fr
->ic
->k_rf
);
109 krf2
= _mm_set1_ps(fr
->ic
->k_rf
*2.0);
110 crf
= _mm_set1_ps(fr
->ic
->c_rf
);
112 /* Avoid stupid compiler warnings */
113 jnrA
= jnrB
= jnrC
= jnrD
= 0;
122 for(iidx
=0;iidx
<4*DIM
;iidx
++)
127 /* Start outer loop over neighborlists */
128 for(iidx
=0; iidx
<nri
; iidx
++)
130 /* Load shift vector for this list */
131 i_shift_offset
= DIM
*shiftidx
[iidx
];
133 /* Load limits for loop over neighbors */
134 j_index_start
= jindex
[iidx
];
135 j_index_end
= jindex
[iidx
+1];
137 /* Get outer coordinate index */
139 i_coord_offset
= DIM
*inr
;
141 /* Load i particle coords and add shift vector */
142 gmx_mm_load_shift_and_1rvec_broadcast_ps(shiftvec
+i_shift_offset
,x
+i_coord_offset
,&ix0
,&iy0
,&iz0
);
144 fix0
= _mm_setzero_ps();
145 fiy0
= _mm_setzero_ps();
146 fiz0
= _mm_setzero_ps();
148 /* Load parameters for i particles */
149 iq0
= _mm_mul_ps(facel
,_mm_load1_ps(charge
+inr
+0));
151 /* Reset potential sums */
152 velecsum
= _mm_setzero_ps();
154 /* Start inner kernel loop */
155 for(jidx
=j_index_start
; jidx
<j_index_end
&& jjnr
[jidx
+3]>=0; jidx
+=4)
158 /* Get j neighbor index, and coordinate index */
163 j_coord_offsetA
= DIM
*jnrA
;
164 j_coord_offsetB
= DIM
*jnrB
;
165 j_coord_offsetC
= DIM
*jnrC
;
166 j_coord_offsetD
= DIM
*jnrD
;
168 /* load j atom coordinates */
169 gmx_mm_load_1rvec_4ptr_swizzle_ps(x
+j_coord_offsetA
,x
+j_coord_offsetB
,
170 x
+j_coord_offsetC
,x
+j_coord_offsetD
,
173 /* Calculate displacement vector */
174 dx00
= _mm_sub_ps(ix0
,jx0
);
175 dy00
= _mm_sub_ps(iy0
,jy0
);
176 dz00
= _mm_sub_ps(iz0
,jz0
);
178 /* Calculate squared distance and things based on it */
179 rsq00
= gmx_mm_calc_rsq_ps(dx00
,dy00
,dz00
);
181 rinv00
= gmx_mm_invsqrt_ps(rsq00
);
183 rinvsq00
= _mm_mul_ps(rinv00
,rinv00
);
185 /* Load parameters for j particles */
186 jq0
= gmx_mm_load_4real_swizzle_ps(charge
+jnrA
+0,charge
+jnrB
+0,
187 charge
+jnrC
+0,charge
+jnrD
+0);
189 /**************************
190 * CALCULATE INTERACTIONS *
191 **************************/
193 /* Compute parameters for interactions between i and j atoms */
194 qq00
= _mm_mul_ps(iq0
,jq0
);
196 /* REACTION-FIELD ELECTROSTATICS */
197 velec
= _mm_mul_ps(qq00
,_mm_sub_ps(_mm_add_ps(rinv00
,_mm_mul_ps(krf
,rsq00
)),crf
));
198 felec
= _mm_mul_ps(qq00
,_mm_sub_ps(_mm_mul_ps(rinv00
,rinvsq00
),krf2
));
200 /* Update potential sum for this i atom from the interaction with this j atom. */
201 velecsum
= _mm_add_ps(velecsum
,velec
);
205 /* Calculate temporary vectorial force */
206 tx
= _mm_mul_ps(fscal
,dx00
);
207 ty
= _mm_mul_ps(fscal
,dy00
);
208 tz
= _mm_mul_ps(fscal
,dz00
);
210 /* Update vectorial force */
211 fix0
= _mm_add_ps(fix0
,tx
);
212 fiy0
= _mm_add_ps(fiy0
,ty
);
213 fiz0
= _mm_add_ps(fiz0
,tz
);
215 fjptrA
= f
+j_coord_offsetA
;
216 fjptrB
= f
+j_coord_offsetB
;
217 fjptrC
= f
+j_coord_offsetC
;
218 fjptrD
= f
+j_coord_offsetD
;
219 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA
,fjptrB
,fjptrC
,fjptrD
,tx
,ty
,tz
);
221 /* Inner loop uses 32 flops */
227 /* Get j neighbor index, and coordinate index */
228 jnrlistA
= jjnr
[jidx
];
229 jnrlistB
= jjnr
[jidx
+1];
230 jnrlistC
= jjnr
[jidx
+2];
231 jnrlistD
= jjnr
[jidx
+3];
232 /* Sign of each element will be negative for non-real atoms.
233 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
234 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
236 dummy_mask
= gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i
*)(jjnr
+jidx
)),_mm_setzero_si128()));
237 jnrA
= (jnrlistA
>=0) ? jnrlistA
: 0;
238 jnrB
= (jnrlistB
>=0) ? jnrlistB
: 0;
239 jnrC
= (jnrlistC
>=0) ? jnrlistC
: 0;
240 jnrD
= (jnrlistD
>=0) ? jnrlistD
: 0;
241 j_coord_offsetA
= DIM
*jnrA
;
242 j_coord_offsetB
= DIM
*jnrB
;
243 j_coord_offsetC
= DIM
*jnrC
;
244 j_coord_offsetD
= DIM
*jnrD
;
246 /* load j atom coordinates */
247 gmx_mm_load_1rvec_4ptr_swizzle_ps(x
+j_coord_offsetA
,x
+j_coord_offsetB
,
248 x
+j_coord_offsetC
,x
+j_coord_offsetD
,
251 /* Calculate displacement vector */
252 dx00
= _mm_sub_ps(ix0
,jx0
);
253 dy00
= _mm_sub_ps(iy0
,jy0
);
254 dz00
= _mm_sub_ps(iz0
,jz0
);
256 /* Calculate squared distance and things based on it */
257 rsq00
= gmx_mm_calc_rsq_ps(dx00
,dy00
,dz00
);
259 rinv00
= gmx_mm_invsqrt_ps(rsq00
);
261 rinvsq00
= _mm_mul_ps(rinv00
,rinv00
);
263 /* Load parameters for j particles */
264 jq0
= gmx_mm_load_4real_swizzle_ps(charge
+jnrA
+0,charge
+jnrB
+0,
265 charge
+jnrC
+0,charge
+jnrD
+0);
267 /**************************
268 * CALCULATE INTERACTIONS *
269 **************************/
271 /* Compute parameters for interactions between i and j atoms */
272 qq00
= _mm_mul_ps(iq0
,jq0
);
274 /* REACTION-FIELD ELECTROSTATICS */
275 velec
= _mm_mul_ps(qq00
,_mm_sub_ps(_mm_add_ps(rinv00
,_mm_mul_ps(krf
,rsq00
)),crf
));
276 felec
= _mm_mul_ps(qq00
,_mm_sub_ps(_mm_mul_ps(rinv00
,rinvsq00
),krf2
));
278 /* Update potential sum for this i atom from the interaction with this j atom. */
279 velec
= _mm_andnot_ps(dummy_mask
,velec
);
280 velecsum
= _mm_add_ps(velecsum
,velec
);
284 fscal
= _mm_andnot_ps(dummy_mask
,fscal
);
286 /* Calculate temporary vectorial force */
287 tx
= _mm_mul_ps(fscal
,dx00
);
288 ty
= _mm_mul_ps(fscal
,dy00
);
289 tz
= _mm_mul_ps(fscal
,dz00
);
291 /* Update vectorial force */
292 fix0
= _mm_add_ps(fix0
,tx
);
293 fiy0
= _mm_add_ps(fiy0
,ty
);
294 fiz0
= _mm_add_ps(fiz0
,tz
);
296 fjptrA
= (jnrlistA
>=0) ? f
+j_coord_offsetA
: scratch
;
297 fjptrB
= (jnrlistB
>=0) ? f
+j_coord_offsetB
: scratch
;
298 fjptrC
= (jnrlistC
>=0) ? f
+j_coord_offsetC
: scratch
;
299 fjptrD
= (jnrlistD
>=0) ? f
+j_coord_offsetD
: scratch
;
300 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA
,fjptrB
,fjptrC
,fjptrD
,tx
,ty
,tz
);
302 /* Inner loop uses 32 flops */
305 /* End of innermost loop */
307 gmx_mm_update_iforce_1atom_swizzle_ps(fix0
,fiy0
,fiz0
,
308 f
+i_coord_offset
,fshift
+i_shift_offset
);
311 /* Update potential energies */
312 gmx_mm_update_1pot_ps(velecsum
,kernel_data
->energygrp_elec
+ggid
);
314 /* Increment number of inner iterations */
315 inneriter
+= j_index_end
- j_index_start
;
317 /* Outer loop uses 8 flops */
320 /* Increment number of outer iterations */
323 /* Update outer/inner flops */
325 inc_nrnb(nrnb
,eNR_NBKERNEL_ELEC_VF
,outeriter
*8 + inneriter
*32);
328 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomP1P1_F_sse2_single
329 * Electrostatics interaction: ReactionField
330 * VdW interaction: None
331 * Geometry: Particle-Particle
332 * Calculate force/pot: Force
335 nb_kernel_ElecRF_VdwNone_GeomP1P1_F_sse2_single
336 (t_nblist
* gmx_restrict nlist
,
337 rvec
* gmx_restrict xx
,
338 rvec
* gmx_restrict ff
,
339 t_forcerec
* gmx_restrict fr
,
340 t_mdatoms
* gmx_restrict mdatoms
,
341 nb_kernel_data_t gmx_unused
* gmx_restrict kernel_data
,
342 t_nrnb
* gmx_restrict nrnb
)
344 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
345 * just 0 for non-waters.
346 * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
347 * jnr indices corresponding to data put in the four positions in the SIMD register.
349 int i_shift_offset
,i_coord_offset
,outeriter
,inneriter
;
350 int j_index_start
,j_index_end
,jidx
,nri
,inr
,ggid
,iidx
;
351 int jnrA
,jnrB
,jnrC
,jnrD
;
352 int jnrlistA
,jnrlistB
,jnrlistC
,jnrlistD
;
353 int j_coord_offsetA
,j_coord_offsetB
,j_coord_offsetC
,j_coord_offsetD
;
354 int *iinr
,*jindex
,*jjnr
,*shiftidx
,*gid
;
356 real
*shiftvec
,*fshift
,*x
,*f
;
357 real
*fjptrA
,*fjptrB
,*fjptrC
,*fjptrD
;
359 __m128 tx
,ty
,tz
,fscal
,rcutoff
,rcutoff2
,jidxall
;
361 __m128 ix0
,iy0
,iz0
,fix0
,fiy0
,fiz0
,iq0
,isai0
;
362 int vdwjidx0A
,vdwjidx0B
,vdwjidx0C
,vdwjidx0D
;
363 __m128 jx0
,jy0
,jz0
,fjx0
,fjy0
,fjz0
,jq0
,isaj0
;
364 __m128 dx00
,dy00
,dz00
,rsq00
,rinv00
,rinvsq00
,r00
,qq00
,c6_00
,c12_00
;
365 __m128 velec
,felec
,velecsum
,facel
,crf
,krf
,krf2
;
367 __m128 dummy_mask
,cutoff_mask
;
368 __m128 signbit
= _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
369 __m128 one
= _mm_set1_ps(1.0);
370 __m128 two
= _mm_set1_ps(2.0);
376 jindex
= nlist
->jindex
;
378 shiftidx
= nlist
->shift
;
380 shiftvec
= fr
->shift_vec
[0];
381 fshift
= fr
->fshift
[0];
382 facel
= _mm_set1_ps(fr
->epsfac
);
383 charge
= mdatoms
->chargeA
;
384 krf
= _mm_set1_ps(fr
->ic
->k_rf
);
385 krf2
= _mm_set1_ps(fr
->ic
->k_rf
*2.0);
386 crf
= _mm_set1_ps(fr
->ic
->c_rf
);
388 /* Avoid stupid compiler warnings */
389 jnrA
= jnrB
= jnrC
= jnrD
= 0;
398 for(iidx
=0;iidx
<4*DIM
;iidx
++)
403 /* Start outer loop over neighborlists */
404 for(iidx
=0; iidx
<nri
; iidx
++)
406 /* Load shift vector for this list */
407 i_shift_offset
= DIM
*shiftidx
[iidx
];
409 /* Load limits for loop over neighbors */
410 j_index_start
= jindex
[iidx
];
411 j_index_end
= jindex
[iidx
+1];
413 /* Get outer coordinate index */
415 i_coord_offset
= DIM
*inr
;
417 /* Load i particle coords and add shift vector */
418 gmx_mm_load_shift_and_1rvec_broadcast_ps(shiftvec
+i_shift_offset
,x
+i_coord_offset
,&ix0
,&iy0
,&iz0
);
420 fix0
= _mm_setzero_ps();
421 fiy0
= _mm_setzero_ps();
422 fiz0
= _mm_setzero_ps();
424 /* Load parameters for i particles */
425 iq0
= _mm_mul_ps(facel
,_mm_load1_ps(charge
+inr
+0));
427 /* Start inner kernel loop */
428 for(jidx
=j_index_start
; jidx
<j_index_end
&& jjnr
[jidx
+3]>=0; jidx
+=4)
431 /* Get j neighbor index, and coordinate index */
436 j_coord_offsetA
= DIM
*jnrA
;
437 j_coord_offsetB
= DIM
*jnrB
;
438 j_coord_offsetC
= DIM
*jnrC
;
439 j_coord_offsetD
= DIM
*jnrD
;
441 /* load j atom coordinates */
442 gmx_mm_load_1rvec_4ptr_swizzle_ps(x
+j_coord_offsetA
,x
+j_coord_offsetB
,
443 x
+j_coord_offsetC
,x
+j_coord_offsetD
,
446 /* Calculate displacement vector */
447 dx00
= _mm_sub_ps(ix0
,jx0
);
448 dy00
= _mm_sub_ps(iy0
,jy0
);
449 dz00
= _mm_sub_ps(iz0
,jz0
);
451 /* Calculate squared distance and things based on it */
452 rsq00
= gmx_mm_calc_rsq_ps(dx00
,dy00
,dz00
);
454 rinv00
= gmx_mm_invsqrt_ps(rsq00
);
456 rinvsq00
= _mm_mul_ps(rinv00
,rinv00
);
458 /* Load parameters for j particles */
459 jq0
= gmx_mm_load_4real_swizzle_ps(charge
+jnrA
+0,charge
+jnrB
+0,
460 charge
+jnrC
+0,charge
+jnrD
+0);
462 /**************************
463 * CALCULATE INTERACTIONS *
464 **************************/
466 /* Compute parameters for interactions between i and j atoms */
467 qq00
= _mm_mul_ps(iq0
,jq0
);
469 /* REACTION-FIELD ELECTROSTATICS */
470 felec
= _mm_mul_ps(qq00
,_mm_sub_ps(_mm_mul_ps(rinv00
,rinvsq00
),krf2
));
474 /* Calculate temporary vectorial force */
475 tx
= _mm_mul_ps(fscal
,dx00
);
476 ty
= _mm_mul_ps(fscal
,dy00
);
477 tz
= _mm_mul_ps(fscal
,dz00
);
479 /* Update vectorial force */
480 fix0
= _mm_add_ps(fix0
,tx
);
481 fiy0
= _mm_add_ps(fiy0
,ty
);
482 fiz0
= _mm_add_ps(fiz0
,tz
);
484 fjptrA
= f
+j_coord_offsetA
;
485 fjptrB
= f
+j_coord_offsetB
;
486 fjptrC
= f
+j_coord_offsetC
;
487 fjptrD
= f
+j_coord_offsetD
;
488 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA
,fjptrB
,fjptrC
,fjptrD
,tx
,ty
,tz
);
490 /* Inner loop uses 27 flops */
496 /* Get j neighbor index, and coordinate index */
497 jnrlistA
= jjnr
[jidx
];
498 jnrlistB
= jjnr
[jidx
+1];
499 jnrlistC
= jjnr
[jidx
+2];
500 jnrlistD
= jjnr
[jidx
+3];
501 /* Sign of each element will be negative for non-real atoms.
502 * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
503 * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
505 dummy_mask
= gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i
*)(jjnr
+jidx
)),_mm_setzero_si128()));
506 jnrA
= (jnrlistA
>=0) ? jnrlistA
: 0;
507 jnrB
= (jnrlistB
>=0) ? jnrlistB
: 0;
508 jnrC
= (jnrlistC
>=0) ? jnrlistC
: 0;
509 jnrD
= (jnrlistD
>=0) ? jnrlistD
: 0;
510 j_coord_offsetA
= DIM
*jnrA
;
511 j_coord_offsetB
= DIM
*jnrB
;
512 j_coord_offsetC
= DIM
*jnrC
;
513 j_coord_offsetD
= DIM
*jnrD
;
515 /* load j atom coordinates */
516 gmx_mm_load_1rvec_4ptr_swizzle_ps(x
+j_coord_offsetA
,x
+j_coord_offsetB
,
517 x
+j_coord_offsetC
,x
+j_coord_offsetD
,
520 /* Calculate displacement vector */
521 dx00
= _mm_sub_ps(ix0
,jx0
);
522 dy00
= _mm_sub_ps(iy0
,jy0
);
523 dz00
= _mm_sub_ps(iz0
,jz0
);
525 /* Calculate squared distance and things based on it */
526 rsq00
= gmx_mm_calc_rsq_ps(dx00
,dy00
,dz00
);
528 rinv00
= gmx_mm_invsqrt_ps(rsq00
);
530 rinvsq00
= _mm_mul_ps(rinv00
,rinv00
);
532 /* Load parameters for j particles */
533 jq0
= gmx_mm_load_4real_swizzle_ps(charge
+jnrA
+0,charge
+jnrB
+0,
534 charge
+jnrC
+0,charge
+jnrD
+0);
536 /**************************
537 * CALCULATE INTERACTIONS *
538 **************************/
540 /* Compute parameters for interactions between i and j atoms */
541 qq00
= _mm_mul_ps(iq0
,jq0
);
543 /* REACTION-FIELD ELECTROSTATICS */
544 felec
= _mm_mul_ps(qq00
,_mm_sub_ps(_mm_mul_ps(rinv00
,rinvsq00
),krf2
));
548 fscal
= _mm_andnot_ps(dummy_mask
,fscal
);
550 /* Calculate temporary vectorial force */
551 tx
= _mm_mul_ps(fscal
,dx00
);
552 ty
= _mm_mul_ps(fscal
,dy00
);
553 tz
= _mm_mul_ps(fscal
,dz00
);
555 /* Update vectorial force */
556 fix0
= _mm_add_ps(fix0
,tx
);
557 fiy0
= _mm_add_ps(fiy0
,ty
);
558 fiz0
= _mm_add_ps(fiz0
,tz
);
560 fjptrA
= (jnrlistA
>=0) ? f
+j_coord_offsetA
: scratch
;
561 fjptrB
= (jnrlistB
>=0) ? f
+j_coord_offsetB
: scratch
;
562 fjptrC
= (jnrlistC
>=0) ? f
+j_coord_offsetC
: scratch
;
563 fjptrD
= (jnrlistD
>=0) ? f
+j_coord_offsetD
: scratch
;
564 gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA
,fjptrB
,fjptrC
,fjptrD
,tx
,ty
,tz
);
566 /* Inner loop uses 27 flops */
569 /* End of innermost loop */
571 gmx_mm_update_iforce_1atom_swizzle_ps(fix0
,fiy0
,fiz0
,
572 f
+i_coord_offset
,fshift
+i_shift_offset
);
574 /* Increment number of inner iterations */
575 inneriter
+= j_index_end
- j_index_start
;
577 /* Outer loop uses 7 flops */
580 /* Increment number of outer iterations */
583 /* Update outer/inner flops */
585 inc_nrnb(nrnb
,eNR_NBKERNEL_ELEC_F
,outeriter
*7 + inneriter
*27);