2 * This file is part of the GROMACS molecular simulation package.
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.
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.
42 #include "../nb_kernel.h"
43 #include "gromacs/legacyheaders/types/simple.h"
44 #include "gromacs/math/vec.h"
45 #include "gromacs/legacyheaders/nrnb.h"
47 #include "kernelutil_sparc64_hpc_ace_double.h"
50 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomP1P1_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: ReactionField
52 * VdW interaction: None
53 * Geometry: Particle-Particle
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecRF_VdwNone_GeomP1P1_VF_sparc64_hpc_ace_double
58 (t_nblist * gmx_restrict nlist,
59 rvec * gmx_restrict xx,
60 rvec * gmx_restrict ff,
61 t_forcerec * gmx_restrict fr,
62 t_mdatoms * gmx_restrict mdatoms,
63 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
64 t_nrnb * gmx_restrict nrnb)
66 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
67 * just 0 for non-waters.
68 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
69 * jnr indices corresponding to data put in the four positions in the SIMD register.
71 int i_shift_offset,i_coord_offset,outeriter,inneriter;
72 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
74 int j_coord_offsetA,j_coord_offsetB;
75 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
77 real *shiftvec,*fshift,*x,*f;
78 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
80 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
81 int vdwjidx0A,vdwjidx0B;
82 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
83 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
84 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
87 _fjsp_v2r8 dummy_mask,cutoff_mask;
88 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
89 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
90 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
97 jindex = nlist->jindex;
99 shiftidx = nlist->shift;
101 shiftvec = fr->shift_vec[0];
102 fshift = fr->fshift[0];
103 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
104 charge = mdatoms->chargeA;
105 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
106 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
107 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
109 /* Avoid stupid compiler warnings */
117 /* Start outer loop over neighborlists */
118 for(iidx=0; iidx<nri; iidx++)
120 /* Load shift vector for this list */
121 i_shift_offset = DIM*shiftidx[iidx];
123 /* Load limits for loop over neighbors */
124 j_index_start = jindex[iidx];
125 j_index_end = jindex[iidx+1];
127 /* Get outer coordinate index */
129 i_coord_offset = DIM*inr;
131 /* Load i particle coords and add shift vector */
132 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
134 fix0 = _fjsp_setzero_v2r8();
135 fiy0 = _fjsp_setzero_v2r8();
136 fiz0 = _fjsp_setzero_v2r8();
138 /* Load parameters for i particles */
139 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
141 /* Reset potential sums */
142 velecsum = _fjsp_setzero_v2r8();
144 /* Start inner kernel loop */
145 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
148 /* Get j neighbor index, and coordinate index */
151 j_coord_offsetA = DIM*jnrA;
152 j_coord_offsetB = DIM*jnrB;
154 /* load j atom coordinates */
155 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
158 /* Calculate displacement vector */
159 dx00 = _fjsp_sub_v2r8(ix0,jx0);
160 dy00 = _fjsp_sub_v2r8(iy0,jy0);
161 dz00 = _fjsp_sub_v2r8(iz0,jz0);
163 /* Calculate squared distance and things based on it */
164 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
166 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
168 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
170 /* Load parameters for j particles */
171 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
173 /**************************
174 * CALCULATE INTERACTIONS *
175 **************************/
177 /* Compute parameters for interactions between i and j atoms */
178 qq00 = _fjsp_mul_v2r8(iq0,jq0);
180 /* REACTION-FIELD ELECTROSTATICS */
181 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
182 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
184 /* Update potential sum for this i atom from the interaction with this j atom. */
185 velecsum = _fjsp_add_v2r8(velecsum,velec);
189 /* Update vectorial force */
190 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
191 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
192 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
194 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
196 /* Inner loop uses 35 flops */
203 j_coord_offsetA = DIM*jnrA;
205 /* load j atom coordinates */
206 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
209 /* Calculate displacement vector */
210 dx00 = _fjsp_sub_v2r8(ix0,jx0);
211 dy00 = _fjsp_sub_v2r8(iy0,jy0);
212 dz00 = _fjsp_sub_v2r8(iz0,jz0);
214 /* Calculate squared distance and things based on it */
215 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
217 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
219 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
221 /* Load parameters for j particles */
222 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
224 /**************************
225 * CALCULATE INTERACTIONS *
226 **************************/
228 /* Compute parameters for interactions between i and j atoms */
229 qq00 = _fjsp_mul_v2r8(iq0,jq0);
231 /* REACTION-FIELD ELECTROSTATICS */
232 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
233 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
235 /* Update potential sum for this i atom from the interaction with this j atom. */
236 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
237 velecsum = _fjsp_add_v2r8(velecsum,velec);
241 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
243 /* Update vectorial force */
244 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
245 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
246 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
248 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
250 /* Inner loop uses 35 flops */
253 /* End of innermost loop */
255 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
256 f+i_coord_offset,fshift+i_shift_offset);
259 /* Update potential energies */
260 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
262 /* Increment number of inner iterations */
263 inneriter += j_index_end - j_index_start;
265 /* Outer loop uses 8 flops */
268 /* Increment number of outer iterations */
271 /* Update outer/inner flops */
273 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VF,outeriter*8 + inneriter*35);
276 * Gromacs nonbonded kernel: nb_kernel_ElecRF_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
277 * Electrostatics interaction: ReactionField
278 * VdW interaction: None
279 * Geometry: Particle-Particle
280 * Calculate force/pot: Force
283 nb_kernel_ElecRF_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
284 (t_nblist * gmx_restrict nlist,
285 rvec * gmx_restrict xx,
286 rvec * gmx_restrict ff,
287 t_forcerec * gmx_restrict fr,
288 t_mdatoms * gmx_restrict mdatoms,
289 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
290 t_nrnb * gmx_restrict nrnb)
292 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
293 * just 0 for non-waters.
294 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
295 * jnr indices corresponding to data put in the four positions in the SIMD register.
297 int i_shift_offset,i_coord_offset,outeriter,inneriter;
298 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
300 int j_coord_offsetA,j_coord_offsetB;
301 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
303 real *shiftvec,*fshift,*x,*f;
304 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
306 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
307 int vdwjidx0A,vdwjidx0B;
308 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
309 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
310 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
313 _fjsp_v2r8 dummy_mask,cutoff_mask;
314 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
315 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
316 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
323 jindex = nlist->jindex;
325 shiftidx = nlist->shift;
327 shiftvec = fr->shift_vec[0];
328 fshift = fr->fshift[0];
329 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
330 charge = mdatoms->chargeA;
331 krf = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
332 krf2 = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
333 crf = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
335 /* Avoid stupid compiler warnings */
343 /* Start outer loop over neighborlists */
344 for(iidx=0; iidx<nri; iidx++)
346 /* Load shift vector for this list */
347 i_shift_offset = DIM*shiftidx[iidx];
349 /* Load limits for loop over neighbors */
350 j_index_start = jindex[iidx];
351 j_index_end = jindex[iidx+1];
353 /* Get outer coordinate index */
355 i_coord_offset = DIM*inr;
357 /* Load i particle coords and add shift vector */
358 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
360 fix0 = _fjsp_setzero_v2r8();
361 fiy0 = _fjsp_setzero_v2r8();
362 fiz0 = _fjsp_setzero_v2r8();
364 /* Load parameters for i particles */
365 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
367 /* Start inner kernel loop */
368 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
371 /* Get j neighbor index, and coordinate index */
374 j_coord_offsetA = DIM*jnrA;
375 j_coord_offsetB = DIM*jnrB;
377 /* load j atom coordinates */
378 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
381 /* Calculate displacement vector */
382 dx00 = _fjsp_sub_v2r8(ix0,jx0);
383 dy00 = _fjsp_sub_v2r8(iy0,jy0);
384 dz00 = _fjsp_sub_v2r8(iz0,jz0);
386 /* Calculate squared distance and things based on it */
387 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
389 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
391 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
393 /* Load parameters for j particles */
394 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
396 /**************************
397 * CALCULATE INTERACTIONS *
398 **************************/
400 /* Compute parameters for interactions between i and j atoms */
401 qq00 = _fjsp_mul_v2r8(iq0,jq0);
403 /* REACTION-FIELD ELECTROSTATICS */
404 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
408 /* Update vectorial force */
409 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
410 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
411 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
413 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
415 /* Inner loop uses 30 flops */
422 j_coord_offsetA = DIM*jnrA;
424 /* load j atom coordinates */
425 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
428 /* Calculate displacement vector */
429 dx00 = _fjsp_sub_v2r8(ix0,jx0);
430 dy00 = _fjsp_sub_v2r8(iy0,jy0);
431 dz00 = _fjsp_sub_v2r8(iz0,jz0);
433 /* Calculate squared distance and things based on it */
434 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
436 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
438 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
440 /* Load parameters for j particles */
441 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
443 /**************************
444 * CALCULATE INTERACTIONS *
445 **************************/
447 /* Compute parameters for interactions between i and j atoms */
448 qq00 = _fjsp_mul_v2r8(iq0,jq0);
450 /* REACTION-FIELD ELECTROSTATICS */
451 felec = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
455 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
457 /* Update vectorial force */
458 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
459 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
460 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
462 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
464 /* Inner loop uses 30 flops */
467 /* End of innermost loop */
469 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
470 f+i_coord_offset,fshift+i_shift_offset);
472 /* Increment number of inner iterations */
473 inneriter += j_index_end - j_index_start;
475 /* Outer loop uses 7 flops */
478 /* Increment number of outer iterations */
481 /* Update outer/inner flops */
483 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_F,outeriter*7 + inneriter*30);