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_ElecCoul_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
51 * Electrostatics interaction: Coulomb
52 * VdW interaction: LennardJones
53 * Geometry: Particle-Particle
54 * Calculate force/pot: PotentialAndForce
57 nb_kernel_ElecCoul_VdwLJ_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 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
90 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
91 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
93 _fjsp_v2r8 dummy_mask,cutoff_mask;
94 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
95 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
96 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
103 jindex = nlist->jindex;
105 shiftidx = nlist->shift;
107 shiftvec = fr->shift_vec[0];
108 fshift = fr->fshift[0];
109 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
110 charge = mdatoms->chargeA;
111 nvdwtype = fr->ntype;
113 vdwtype = mdatoms->typeA;
115 /* Avoid stupid compiler warnings */
123 /* Start outer loop over neighborlists */
124 for(iidx=0; iidx<nri; iidx++)
126 /* Load shift vector for this list */
127 i_shift_offset = DIM*shiftidx[iidx];
129 /* Load limits for loop over neighbors */
130 j_index_start = jindex[iidx];
131 j_index_end = jindex[iidx+1];
133 /* Get outer coordinate index */
135 i_coord_offset = DIM*inr;
137 /* Load i particle coords and add shift vector */
138 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
140 fix0 = _fjsp_setzero_v2r8();
141 fiy0 = _fjsp_setzero_v2r8();
142 fiz0 = _fjsp_setzero_v2r8();
144 /* Load parameters for i particles */
145 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
146 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
148 /* Reset potential sums */
149 velecsum = _fjsp_setzero_v2r8();
150 vvdwsum = _fjsp_setzero_v2r8();
152 /* Start inner kernel loop */
153 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
156 /* Get j neighbor index, and coordinate index */
159 j_coord_offsetA = DIM*jnrA;
160 j_coord_offsetB = DIM*jnrB;
162 /* load j atom coordinates */
163 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
166 /* Calculate displacement vector */
167 dx00 = _fjsp_sub_v2r8(ix0,jx0);
168 dy00 = _fjsp_sub_v2r8(iy0,jy0);
169 dz00 = _fjsp_sub_v2r8(iz0,jz0);
171 /* Calculate squared distance and things based on it */
172 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
174 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
176 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
178 /* Load parameters for j particles */
179 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
180 vdwjidx0A = 2*vdwtype[jnrA+0];
181 vdwjidx0B = 2*vdwtype[jnrB+0];
183 /**************************
184 * CALCULATE INTERACTIONS *
185 **************************/
187 /* Compute parameters for interactions between i and j atoms */
188 qq00 = _fjsp_mul_v2r8(iq0,jq0);
189 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
190 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
192 /* COULOMB ELECTROSTATICS */
193 velec = _fjsp_mul_v2r8(qq00,rinv00);
194 felec = _fjsp_mul_v2r8(velec,rinvsq00);
196 /* LENNARD-JONES DISPERSION/REPULSION */
198 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
199 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
200 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
201 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
202 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
204 /* Update potential sum for this i atom from the interaction with this j atom. */
205 velecsum = _fjsp_add_v2r8(velecsum,velec);
206 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
208 fscal = _fjsp_add_v2r8(felec,fvdw);
210 /* Update vectorial force */
211 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
212 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
213 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
215 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
217 /* Inner loop uses 43 flops */
224 j_coord_offsetA = DIM*jnrA;
226 /* load j atom coordinates */
227 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
230 /* Calculate displacement vector */
231 dx00 = _fjsp_sub_v2r8(ix0,jx0);
232 dy00 = _fjsp_sub_v2r8(iy0,jy0);
233 dz00 = _fjsp_sub_v2r8(iz0,jz0);
235 /* Calculate squared distance and things based on it */
236 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
238 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
240 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
242 /* Load parameters for j particles */
243 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
244 vdwjidx0A = 2*vdwtype[jnrA+0];
246 /**************************
247 * CALCULATE INTERACTIONS *
248 **************************/
250 /* Compute parameters for interactions between i and j atoms */
251 qq00 = _fjsp_mul_v2r8(iq0,jq0);
252 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
254 /* COULOMB ELECTROSTATICS */
255 velec = _fjsp_mul_v2r8(qq00,rinv00);
256 felec = _fjsp_mul_v2r8(velec,rinvsq00);
258 /* LENNARD-JONES DISPERSION/REPULSION */
260 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
261 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
262 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
263 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
264 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
266 /* Update potential sum for this i atom from the interaction with this j atom. */
267 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
268 velecsum = _fjsp_add_v2r8(velecsum,velec);
269 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
270 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
272 fscal = _fjsp_add_v2r8(felec,fvdw);
274 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
276 /* Update vectorial force */
277 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
278 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
279 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
281 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
283 /* Inner loop uses 43 flops */
286 /* End of innermost loop */
288 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
289 f+i_coord_offset,fshift+i_shift_offset);
292 /* Update potential energies */
293 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
294 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
296 /* Increment number of inner iterations */
297 inneriter += j_index_end - j_index_start;
299 /* Outer loop uses 9 flops */
302 /* Increment number of outer iterations */
305 /* Update outer/inner flops */
307 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*9 + inneriter*43);
310 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
311 * Electrostatics interaction: Coulomb
312 * VdW interaction: LennardJones
313 * Geometry: Particle-Particle
314 * Calculate force/pot: Force
317 nb_kernel_ElecCoul_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
318 (t_nblist * gmx_restrict nlist,
319 rvec * gmx_restrict xx,
320 rvec * gmx_restrict ff,
321 t_forcerec * gmx_restrict fr,
322 t_mdatoms * gmx_restrict mdatoms,
323 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
324 t_nrnb * gmx_restrict nrnb)
326 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
327 * just 0 for non-waters.
328 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
329 * jnr indices corresponding to data put in the four positions in the SIMD register.
331 int i_shift_offset,i_coord_offset,outeriter,inneriter;
332 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
334 int j_coord_offsetA,j_coord_offsetB;
335 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
337 real *shiftvec,*fshift,*x,*f;
338 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
340 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
341 int vdwjidx0A,vdwjidx0B;
342 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
343 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
344 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
347 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
350 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
351 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
353 _fjsp_v2r8 dummy_mask,cutoff_mask;
354 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
355 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
356 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
363 jindex = nlist->jindex;
365 shiftidx = nlist->shift;
367 shiftvec = fr->shift_vec[0];
368 fshift = fr->fshift[0];
369 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
370 charge = mdatoms->chargeA;
371 nvdwtype = fr->ntype;
373 vdwtype = mdatoms->typeA;
375 /* Avoid stupid compiler warnings */
383 /* Start outer loop over neighborlists */
384 for(iidx=0; iidx<nri; iidx++)
386 /* Load shift vector for this list */
387 i_shift_offset = DIM*shiftidx[iidx];
389 /* Load limits for loop over neighbors */
390 j_index_start = jindex[iidx];
391 j_index_end = jindex[iidx+1];
393 /* Get outer coordinate index */
395 i_coord_offset = DIM*inr;
397 /* Load i particle coords and add shift vector */
398 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
400 fix0 = _fjsp_setzero_v2r8();
401 fiy0 = _fjsp_setzero_v2r8();
402 fiz0 = _fjsp_setzero_v2r8();
404 /* Load parameters for i particles */
405 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
406 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
408 /* Start inner kernel loop */
409 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
412 /* Get j neighbor index, and coordinate index */
415 j_coord_offsetA = DIM*jnrA;
416 j_coord_offsetB = DIM*jnrB;
418 /* load j atom coordinates */
419 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
422 /* Calculate displacement vector */
423 dx00 = _fjsp_sub_v2r8(ix0,jx0);
424 dy00 = _fjsp_sub_v2r8(iy0,jy0);
425 dz00 = _fjsp_sub_v2r8(iz0,jz0);
427 /* Calculate squared distance and things based on it */
428 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
430 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
432 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
434 /* Load parameters for j particles */
435 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
436 vdwjidx0A = 2*vdwtype[jnrA+0];
437 vdwjidx0B = 2*vdwtype[jnrB+0];
439 /**************************
440 * CALCULATE INTERACTIONS *
441 **************************/
443 /* Compute parameters for interactions between i and j atoms */
444 qq00 = _fjsp_mul_v2r8(iq0,jq0);
445 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
446 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
448 /* COULOMB ELECTROSTATICS */
449 velec = _fjsp_mul_v2r8(qq00,rinv00);
450 felec = _fjsp_mul_v2r8(velec,rinvsq00);
452 /* LENNARD-JONES DISPERSION/REPULSION */
454 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
455 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
457 fscal = _fjsp_add_v2r8(felec,fvdw);
459 /* Update vectorial force */
460 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
461 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
462 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
464 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
466 /* Inner loop uses 37 flops */
473 j_coord_offsetA = DIM*jnrA;
475 /* load j atom coordinates */
476 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
479 /* Calculate displacement vector */
480 dx00 = _fjsp_sub_v2r8(ix0,jx0);
481 dy00 = _fjsp_sub_v2r8(iy0,jy0);
482 dz00 = _fjsp_sub_v2r8(iz0,jz0);
484 /* Calculate squared distance and things based on it */
485 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
487 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
489 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
491 /* Load parameters for j particles */
492 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
493 vdwjidx0A = 2*vdwtype[jnrA+0];
495 /**************************
496 * CALCULATE INTERACTIONS *
497 **************************/
499 /* Compute parameters for interactions between i and j atoms */
500 qq00 = _fjsp_mul_v2r8(iq0,jq0);
501 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
503 /* COULOMB ELECTROSTATICS */
504 velec = _fjsp_mul_v2r8(qq00,rinv00);
505 felec = _fjsp_mul_v2r8(velec,rinvsq00);
507 /* LENNARD-JONES DISPERSION/REPULSION */
509 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
510 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
512 fscal = _fjsp_add_v2r8(felec,fvdw);
514 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
516 /* Update vectorial force */
517 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
518 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
519 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
521 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
523 /* Inner loop uses 37 flops */
526 /* End of innermost loop */
528 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
529 f+i_coord_offset,fshift+i_shift_offset);
531 /* Increment number of inner iterations */
532 inneriter += j_index_end - j_index_start;
534 /* Outer loop uses 7 flops */
537 /* Increment number of outer iterations */
540 /* Update outer/inner flops */
542 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*7 + inneriter*37);