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.
44 #include "../nb_kernel.h"
45 #include "gromacs/legacyheaders/types/simple.h"
46 #include "gromacs/math/vec.h"
47 #include "gromacs/legacyheaders/nrnb.h"
49 #include "kernelutil_sparc64_hpc_ace_double.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: Coulomb
54 * VdW interaction: LennardJones
55 * Geometry: Particle-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecCoul_VdwLJ_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 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
92 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
93 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
95 _fjsp_v2r8 dummy_mask,cutoff_mask;
96 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
97 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
98 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
105 jindex = nlist->jindex;
107 shiftidx = nlist->shift;
109 shiftvec = fr->shift_vec[0];
110 fshift = fr->fshift[0];
111 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
112 charge = mdatoms->chargeA;
113 nvdwtype = fr->ntype;
115 vdwtype = mdatoms->typeA;
117 /* Avoid stupid compiler warnings */
125 /* Start outer loop over neighborlists */
126 for(iidx=0; iidx<nri; iidx++)
128 /* Load shift vector for this list */
129 i_shift_offset = DIM*shiftidx[iidx];
131 /* Load limits for loop over neighbors */
132 j_index_start = jindex[iidx];
133 j_index_end = jindex[iidx+1];
135 /* Get outer coordinate index */
137 i_coord_offset = DIM*inr;
139 /* Load i particle coords and add shift vector */
140 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
142 fix0 = _fjsp_setzero_v2r8();
143 fiy0 = _fjsp_setzero_v2r8();
144 fiz0 = _fjsp_setzero_v2r8();
146 /* Load parameters for i particles */
147 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
148 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
150 /* Reset potential sums */
151 velecsum = _fjsp_setzero_v2r8();
152 vvdwsum = _fjsp_setzero_v2r8();
154 /* Start inner kernel loop */
155 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
158 /* Get j neighbor index, and coordinate index */
161 j_coord_offsetA = DIM*jnrA;
162 j_coord_offsetB = DIM*jnrB;
164 /* load j atom coordinates */
165 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
168 /* Calculate displacement vector */
169 dx00 = _fjsp_sub_v2r8(ix0,jx0);
170 dy00 = _fjsp_sub_v2r8(iy0,jy0);
171 dz00 = _fjsp_sub_v2r8(iz0,jz0);
173 /* Calculate squared distance and things based on it */
174 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
176 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
178 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
180 /* Load parameters for j particles */
181 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
182 vdwjidx0A = 2*vdwtype[jnrA+0];
183 vdwjidx0B = 2*vdwtype[jnrB+0];
185 /**************************
186 * CALCULATE INTERACTIONS *
187 **************************/
189 /* Compute parameters for interactions between i and j atoms */
190 qq00 = _fjsp_mul_v2r8(iq0,jq0);
191 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
192 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
194 /* COULOMB ELECTROSTATICS */
195 velec = _fjsp_mul_v2r8(qq00,rinv00);
196 felec = _fjsp_mul_v2r8(velec,rinvsq00);
198 /* LENNARD-JONES DISPERSION/REPULSION */
200 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
201 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
202 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
203 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
204 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
206 /* Update potential sum for this i atom from the interaction with this j atom. */
207 velecsum = _fjsp_add_v2r8(velecsum,velec);
208 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
210 fscal = _fjsp_add_v2r8(felec,fvdw);
212 /* Update vectorial force */
213 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
214 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
215 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
217 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
219 /* Inner loop uses 43 flops */
226 j_coord_offsetA = DIM*jnrA;
228 /* load j atom coordinates */
229 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
232 /* Calculate displacement vector */
233 dx00 = _fjsp_sub_v2r8(ix0,jx0);
234 dy00 = _fjsp_sub_v2r8(iy0,jy0);
235 dz00 = _fjsp_sub_v2r8(iz0,jz0);
237 /* Calculate squared distance and things based on it */
238 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
240 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
242 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
244 /* Load parameters for j particles */
245 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
246 vdwjidx0A = 2*vdwtype[jnrA+0];
248 /**************************
249 * CALCULATE INTERACTIONS *
250 **************************/
252 /* Compute parameters for interactions between i and j atoms */
253 qq00 = _fjsp_mul_v2r8(iq0,jq0);
254 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
255 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
257 /* COULOMB ELECTROSTATICS */
258 velec = _fjsp_mul_v2r8(qq00,rinv00);
259 felec = _fjsp_mul_v2r8(velec,rinvsq00);
261 /* LENNARD-JONES DISPERSION/REPULSION */
263 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
264 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
265 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
266 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
267 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
269 /* Update potential sum for this i atom from the interaction with this j atom. */
270 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
271 velecsum = _fjsp_add_v2r8(velecsum,velec);
272 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
273 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
275 fscal = _fjsp_add_v2r8(felec,fvdw);
277 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
279 /* Update vectorial force */
280 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
281 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
282 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
284 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
286 /* Inner loop uses 43 flops */
289 /* End of innermost loop */
291 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
292 f+i_coord_offset,fshift+i_shift_offset);
295 /* Update potential energies */
296 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
297 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
299 /* Increment number of inner iterations */
300 inneriter += j_index_end - j_index_start;
302 /* Outer loop uses 9 flops */
305 /* Increment number of outer iterations */
308 /* Update outer/inner flops */
310 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*9 + inneriter*43);
313 * Gromacs nonbonded kernel: nb_kernel_ElecCoul_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
314 * Electrostatics interaction: Coulomb
315 * VdW interaction: LennardJones
316 * Geometry: Particle-Particle
317 * Calculate force/pot: Force
320 nb_kernel_ElecCoul_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
321 (t_nblist * gmx_restrict nlist,
322 rvec * gmx_restrict xx,
323 rvec * gmx_restrict ff,
324 t_forcerec * gmx_restrict fr,
325 t_mdatoms * gmx_restrict mdatoms,
326 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
327 t_nrnb * gmx_restrict nrnb)
329 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
330 * just 0 for non-waters.
331 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
332 * jnr indices corresponding to data put in the four positions in the SIMD register.
334 int i_shift_offset,i_coord_offset,outeriter,inneriter;
335 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
337 int j_coord_offsetA,j_coord_offsetB;
338 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
340 real *shiftvec,*fshift,*x,*f;
341 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
343 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
344 int vdwjidx0A,vdwjidx0B;
345 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
346 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
347 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
350 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
353 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
354 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
356 _fjsp_v2r8 dummy_mask,cutoff_mask;
357 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
358 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
359 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
366 jindex = nlist->jindex;
368 shiftidx = nlist->shift;
370 shiftvec = fr->shift_vec[0];
371 fshift = fr->fshift[0];
372 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
373 charge = mdatoms->chargeA;
374 nvdwtype = fr->ntype;
376 vdwtype = mdatoms->typeA;
378 /* Avoid stupid compiler warnings */
386 /* Start outer loop over neighborlists */
387 for(iidx=0; iidx<nri; iidx++)
389 /* Load shift vector for this list */
390 i_shift_offset = DIM*shiftidx[iidx];
392 /* Load limits for loop over neighbors */
393 j_index_start = jindex[iidx];
394 j_index_end = jindex[iidx+1];
396 /* Get outer coordinate index */
398 i_coord_offset = DIM*inr;
400 /* Load i particle coords and add shift vector */
401 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
403 fix0 = _fjsp_setzero_v2r8();
404 fiy0 = _fjsp_setzero_v2r8();
405 fiz0 = _fjsp_setzero_v2r8();
407 /* Load parameters for i particles */
408 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+inr+0));
409 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
411 /* Start inner kernel loop */
412 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
415 /* Get j neighbor index, and coordinate index */
418 j_coord_offsetA = DIM*jnrA;
419 j_coord_offsetB = DIM*jnrB;
421 /* load j atom coordinates */
422 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
425 /* Calculate displacement vector */
426 dx00 = _fjsp_sub_v2r8(ix0,jx0);
427 dy00 = _fjsp_sub_v2r8(iy0,jy0);
428 dz00 = _fjsp_sub_v2r8(iz0,jz0);
430 /* Calculate squared distance and things based on it */
431 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
433 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
435 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
437 /* Load parameters for j particles */
438 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
439 vdwjidx0A = 2*vdwtype[jnrA+0];
440 vdwjidx0B = 2*vdwtype[jnrB+0];
442 /**************************
443 * CALCULATE INTERACTIONS *
444 **************************/
446 /* Compute parameters for interactions between i and j atoms */
447 qq00 = _fjsp_mul_v2r8(iq0,jq0);
448 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
449 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
451 /* COULOMB ELECTROSTATICS */
452 velec = _fjsp_mul_v2r8(qq00,rinv00);
453 felec = _fjsp_mul_v2r8(velec,rinvsq00);
455 /* LENNARD-JONES DISPERSION/REPULSION */
457 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
458 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
460 fscal = _fjsp_add_v2r8(felec,fvdw);
462 /* Update vectorial force */
463 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
464 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
465 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
467 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
469 /* Inner loop uses 37 flops */
476 j_coord_offsetA = DIM*jnrA;
478 /* load j atom coordinates */
479 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
482 /* Calculate displacement vector */
483 dx00 = _fjsp_sub_v2r8(ix0,jx0);
484 dy00 = _fjsp_sub_v2r8(iy0,jy0);
485 dz00 = _fjsp_sub_v2r8(iz0,jz0);
487 /* Calculate squared distance and things based on it */
488 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
490 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
492 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
494 /* Load parameters for j particles */
495 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
496 vdwjidx0A = 2*vdwtype[jnrA+0];
498 /**************************
499 * CALCULATE INTERACTIONS *
500 **************************/
502 /* Compute parameters for interactions between i and j atoms */
503 qq00 = _fjsp_mul_v2r8(iq0,jq0);
504 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
505 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
507 /* COULOMB ELECTROSTATICS */
508 velec = _fjsp_mul_v2r8(qq00,rinv00);
509 felec = _fjsp_mul_v2r8(velec,rinvsq00);
511 /* LENNARD-JONES DISPERSION/REPULSION */
513 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
514 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
516 fscal = _fjsp_add_v2r8(felec,fvdw);
518 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
520 /* Update vectorial force */
521 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
522 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
523 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
525 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
527 /* Inner loop uses 37 flops */
530 /* End of innermost loop */
532 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
533 f+i_coord_offset,fshift+i_shift_offset);
535 /* Increment number of inner iterations */
536 inneriter += j_index_end - j_index_start;
538 /* Outer loop uses 7 flops */
541 /* Increment number of outer iterations */
544 /* Update outer/inner flops */
546 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*7 + inneriter*37);