2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013, 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 sparc64_hpc_ace_double kernel generator.
44 #include "../nb_kernel.h"
45 #include "types/simple.h"
49 #include "kernelutil_sparc64_hpc_ace_double.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecRFCut_VdwNone_GeomP1P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: ReactionField
54 * VdW interaction: None
55 * Geometry: Particle-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecRFCut_VdwNone_GeomP1P1_VF_sparc64_hpc_ace_double
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 refer to j loop unrolling done with double precision SIMD, e.g. for the two 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;
76 int j_coord_offsetA,j_coord_offsetB;
77 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
79 real *shiftvec,*fshift,*x,*f;
80 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
82 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
83 int vdwjidx0A,vdwjidx0B;
84 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
85 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
86 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
89 _fjsp_v2r8 dummy_mask,cutoff_mask;
90 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
91 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
92 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
99 jindex = nlist->jindex;
101 shiftidx = nlist->shift;
103 shiftvec = fr->shift_vec[0];
104 fshift = fr->fshift[0];
105 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
106 charge = mdatoms->chargeA;
107 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
108 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
109 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
111 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
112 rcutoff_scalar = fr->rcoulomb;
113 rcutoff = gmx_fjsp_set1_v2r8(rcutoff_scalar);
114 rcutoff2 = _fjsp_mul_v2r8(rcutoff,rcutoff);
116 /* Avoid stupid compiler warnings */
124 /* Start outer loop over neighborlists */
125 for(iidx=0; iidx<nri; iidx++)
127 /* Load shift vector for this list */
128 i_shift_offset = DIM*shiftidx[iidx];
130 /* Load limits for loop over neighbors */
131 j_index_start = jindex[iidx];
132 j_index_end = jindex[iidx+1];
134 /* Get outer coordinate index */
136 i_coord_offset = DIM*inr;
138 /* Load i particle coords and add shift vector */
139 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
141 fix0 = _fjsp_setzero_v2r8();
142 fiy0 = _fjsp_setzero_v2r8();
143 fiz0 = _fjsp_setzero_v2r8();
145 /* Load parameters for i particles */
146 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
148 /* Reset potential sums */
149 velecsum = _fjsp_setzero_v2r8();
151 /* Start inner kernel loop */
152 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
155 /* Get j neighbor index, and coordinate index */
158 j_coord_offsetA = DIM*jnrA;
159 j_coord_offsetB = DIM*jnrB;
161 /* load j atom coordinates */
162 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
165 /* Calculate displacement vector */
166 dx00 = _fjsp_sub_v2r8(ix0,jx0);
167 dy00 = _fjsp_sub_v2r8(iy0,jy0);
168 dz00 = _fjsp_sub_v2r8(iz0,jz0);
170 /* Calculate squared distance and things based on it */
171 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
173 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
175 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
177 /* Load parameters for j particles */
178 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
180 /**************************
181 * CALCULATE INTERACTIONS *
182 **************************/
184 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
187 /* Compute parameters for interactions between i and j atoms */
188 qq00 = _fjsp_mul_v2r8(iq0,jq0);
190 /* REACTION-FIELD ELECTROSTATICS */
191 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
192 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
194 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
196 /* Update potential sum for this i atom from the interaction with this j atom. */
197 velec = _fjsp_and_v2r8(velec,cutoff_mask);
198 velecsum = _fjsp_add_v2r8(velecsum,velec);
202 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
204 /* Update vectorial force */
205 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
206 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
207 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
209 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
213 /* Inner loop uses 39 flops */
220 j_coord_offsetA = DIM*jnrA;
222 /* load j atom coordinates */
223 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
226 /* Calculate displacement vector */
227 dx00 = _fjsp_sub_v2r8(ix0,jx0);
228 dy00 = _fjsp_sub_v2r8(iy0,jy0);
229 dz00 = _fjsp_sub_v2r8(iz0,jz0);
231 /* Calculate squared distance and things based on it */
232 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
234 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
236 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
238 /* Load parameters for j particles */
239 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
241 /**************************
242 * CALCULATE INTERACTIONS *
243 **************************/
245 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
248 /* Compute parameters for interactions between i and j atoms */
249 qq00 = _fjsp_mul_v2r8(iq0,jq0);
251 /* REACTION-FIELD ELECTROSTATICS */
252 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
253 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
255 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
257 /* Update potential sum for this i atom from the interaction with this j atom. */
258 velec = _fjsp_and_v2r8(velec,cutoff_mask);
259 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
260 velecsum = _fjsp_add_v2r8(velecsum,velec);
264 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
266 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
268 /* Update vectorial force */
269 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
270 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
271 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
273 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
277 /* Inner loop uses 39 flops */
280 /* End of innermost loop */
282 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
283 f+i_coord_offset,fshift+i_shift_offset);
286 /* Update potential energies */
287 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
289 /* Increment number of inner iterations */
290 inneriter += j_index_end - j_index_start;
292 /* Outer loop uses 8 flops */
295 /* Increment number of outer iterations */
298 /* Update outer/inner flops */
300 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VF,outeriter*8 + inneriter*39);
303 * Gromacs nonbonded kernel: nb_kernel_ElecRFCut_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
304 * Electrostatics interaction: ReactionField
305 * VdW interaction: None
306 * Geometry: Particle-Particle
307 * Calculate force/pot: Force
310 nb_kernel_ElecRFCut_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
311 (t_nblist * gmx_restrict nlist,
312 rvec * gmx_restrict xx,
313 rvec * gmx_restrict ff,
314 t_forcerec * gmx_restrict fr,
315 t_mdatoms * gmx_restrict mdatoms,
316 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
317 t_nrnb * gmx_restrict nrnb)
319 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
320 * just 0 for non-waters.
321 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
322 * jnr indices corresponding to data put in the four positions in the SIMD register.
324 int i_shift_offset,i_coord_offset,outeriter,inneriter;
325 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
327 int j_coord_offsetA,j_coord_offsetB;
328 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
330 real *shiftvec,*fshift,*x,*f;
331 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
333 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
334 int vdwjidx0A,vdwjidx0B;
335 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
336 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
337 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
340 _fjsp_v2r8 dummy_mask,cutoff_mask;
341 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
342 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
343 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
350 jindex = nlist->jindex;
352 shiftidx = nlist->shift;
354 shiftvec = fr->shift_vec[0];
355 fshift = fr->fshift[0];
356 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
357 charge = mdatoms->chargeA;
358 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
359 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
360 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
362 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
363 rcutoff_scalar = fr->rcoulomb;
364 rcutoff = gmx_fjsp_set1_v2r8(rcutoff_scalar);
365 rcutoff2 = _fjsp_mul_v2r8(rcutoff,rcutoff);
367 /* Avoid stupid compiler warnings */
375 /* Start outer loop over neighborlists */
376 for(iidx=0; iidx<nri; iidx++)
378 /* Load shift vector for this list */
379 i_shift_offset = DIM*shiftidx[iidx];
381 /* Load limits for loop over neighbors */
382 j_index_start = jindex[iidx];
383 j_index_end = jindex[iidx+1];
385 /* Get outer coordinate index */
387 i_coord_offset = DIM*inr;
389 /* Load i particle coords and add shift vector */
390 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
392 fix0 = _fjsp_setzero_v2r8();
393 fiy0 = _fjsp_setzero_v2r8();
394 fiz0 = _fjsp_setzero_v2r8();
396 /* Load parameters for i particles */
397 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
399 /* Start inner kernel loop */
400 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
403 /* Get j neighbor index, and coordinate index */
406 j_coord_offsetA = DIM*jnrA;
407 j_coord_offsetB = DIM*jnrB;
409 /* load j atom coordinates */
410 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
413 /* Calculate displacement vector */
414 dx00 = _fjsp_sub_v2r8(ix0,jx0);
415 dy00 = _fjsp_sub_v2r8(iy0,jy0);
416 dz00 = _fjsp_sub_v2r8(iz0,jz0);
418 /* Calculate squared distance and things based on it */
419 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
421 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
423 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
425 /* Load parameters for j particles */
426 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
428 /**************************
429 * CALCULATE INTERACTIONS *
430 **************************/
432 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
435 /* Compute parameters for interactions between i and j atoms */
436 qq00 = _fjsp_mul_v2r8(iq0,jq0);
438 /* REACTION-FIELD ELECTROSTATICS */
439 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
441 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
445 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
447 /* Update vectorial force */
448 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
449 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
450 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
452 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
456 /* Inner loop uses 33 flops */
463 j_coord_offsetA = DIM*jnrA;
465 /* load j atom coordinates */
466 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
469 /* Calculate displacement vector */
470 dx00 = _fjsp_sub_v2r8(ix0,jx0);
471 dy00 = _fjsp_sub_v2r8(iy0,jy0);
472 dz00 = _fjsp_sub_v2r8(iz0,jz0);
474 /* Calculate squared distance and things based on it */
475 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
477 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
479 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
481 /* Load parameters for j particles */
482 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
484 /**************************
485 * CALCULATE INTERACTIONS *
486 **************************/
488 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
491 /* Compute parameters for interactions between i and j atoms */
492 qq00 = _fjsp_mul_v2r8(iq0,jq0);
494 /* REACTION-FIELD ELECTROSTATICS */
495 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
497 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
501 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
503 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
505 /* Update vectorial force */
506 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
507 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
508 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
510 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
514 /* Inner loop uses 33 flops */
517 /* End of innermost loop */
519 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
520 f+i_coord_offset,fshift+i_shift_offset);
522 /* Increment number of inner iterations */
523 inneriter += j_index_end - j_index_start;
525 /* Outer loop uses 7 flops */
528 /* Increment number of outer iterations */
531 /* Update outer/inner flops */
533 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_F,outeriter*7 + inneriter*33);