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 "types/simple.h"
46 #include "gromacs/math/vec.h"
49 #include "kernelutil_sparc64_hpc_ace_double.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecNone_VdwLJSh_GeomP1P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: None
54 * VdW interaction: LennardJones
55 * Geometry: Particle-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecNone_VdwLJSh_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;
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 nvdwtype = fr->ntype;
111 vdwtype = mdatoms->typeA;
113 rcutoff_scalar = fr->rvdw;
114 rcutoff = gmx_fjsp_set1_v2r8(rcutoff_scalar);
115 rcutoff2 = _fjsp_mul_v2r8(rcutoff,rcutoff);
117 sh_vdw_invrcut6 = gmx_fjsp_set1_v2r8(fr->ic->sh_invrc6);
118 rvdw = gmx_fjsp_set1_v2r8(fr->rvdw);
120 /* Avoid stupid compiler warnings */
128 /* Start outer loop over neighborlists */
129 for(iidx=0; iidx<nri; iidx++)
131 /* Load shift vector for this list */
132 i_shift_offset = DIM*shiftidx[iidx];
134 /* Load limits for loop over neighbors */
135 j_index_start = jindex[iidx];
136 j_index_end = jindex[iidx+1];
138 /* Get outer coordinate index */
140 i_coord_offset = DIM*inr;
142 /* Load i particle coords and add shift vector */
143 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
145 fix0 = _fjsp_setzero_v2r8();
146 fiy0 = _fjsp_setzero_v2r8();
147 fiz0 = _fjsp_setzero_v2r8();
149 /* Load parameters for i particles */
150 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
152 /* Reset potential sums */
153 vvdwsum = _fjsp_setzero_v2r8();
155 /* Start inner kernel loop */
156 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
159 /* Get j neighbor index, and coordinate index */
162 j_coord_offsetA = DIM*jnrA;
163 j_coord_offsetB = DIM*jnrB;
165 /* load j atom coordinates */
166 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
169 /* Calculate displacement vector */
170 dx00 = _fjsp_sub_v2r8(ix0,jx0);
171 dy00 = _fjsp_sub_v2r8(iy0,jy0);
172 dz00 = _fjsp_sub_v2r8(iz0,jz0);
174 /* Calculate squared distance and things based on it */
175 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
177 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
179 /* Load parameters for j particles */
180 vdwjidx0A = 2*vdwtype[jnrA+0];
181 vdwjidx0B = 2*vdwtype[jnrB+0];
183 /**************************
184 * CALCULATE INTERACTIONS *
185 **************************/
187 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
190 /* Compute parameters for interactions between i and j atoms */
191 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
192 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
194 /* LENNARD-JONES DISPERSION/REPULSION */
196 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
197 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
198 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
199 vvdw = _fjsp_msub_v2r8(_fjsp_nmsub_v2r8(c12_00,_fjsp_mul_v2r8(sh_vdw_invrcut6,sh_vdw_invrcut6),vvdw12),one_twelfth,
200 _fjsp_mul_v2r8(_fjsp_nmsub_v2r8( c6_00,sh_vdw_invrcut6,vvdw6),one_sixth));
201 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
203 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
205 /* Update potential sum for this i atom from the interaction with this j atom. */
206 vvdw = _fjsp_and_v2r8(vvdw,cutoff_mask);
207 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
211 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
213 /* Update vectorial force */
214 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
215 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
216 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
218 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
222 /* Inner loop uses 44 flops */
229 j_coord_offsetA = DIM*jnrA;
231 /* load j atom coordinates */
232 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
235 /* Calculate displacement vector */
236 dx00 = _fjsp_sub_v2r8(ix0,jx0);
237 dy00 = _fjsp_sub_v2r8(iy0,jy0);
238 dz00 = _fjsp_sub_v2r8(iz0,jz0);
240 /* Calculate squared distance and things based on it */
241 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
243 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
245 /* Load parameters for j particles */
246 vdwjidx0A = 2*vdwtype[jnrA+0];
248 /**************************
249 * CALCULATE INTERACTIONS *
250 **************************/
252 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
255 /* Compute parameters for interactions between i and j atoms */
256 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
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(_fjsp_nmsub_v2r8(c12_00,_fjsp_mul_v2r8(sh_vdw_invrcut6,sh_vdw_invrcut6),vvdw12),one_twelfth,
264 _fjsp_mul_v2r8(_fjsp_nmsub_v2r8( c6_00,sh_vdw_invrcut6,vvdw6),one_sixth));
265 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
267 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
269 /* Update potential sum for this i atom from the interaction with this j atom. */
270 vvdw = _fjsp_and_v2r8(vvdw,cutoff_mask);
271 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
272 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
276 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
278 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
280 /* Update vectorial force */
281 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
282 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
283 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
285 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
289 /* Inner loop uses 44 flops */
292 /* End of innermost loop */
294 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
295 f+i_coord_offset,fshift+i_shift_offset);
298 /* Update potential energies */
299 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
301 /* Increment number of inner iterations */
302 inneriter += j_index_end - j_index_start;
304 /* Outer loop uses 7 flops */
307 /* Increment number of outer iterations */
310 /* Update outer/inner flops */
312 inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*44);
315 * Gromacs nonbonded kernel: nb_kernel_ElecNone_VdwLJSh_GeomP1P1_F_sparc64_hpc_ace_double
316 * Electrostatics interaction: None
317 * VdW interaction: LennardJones
318 * Geometry: Particle-Particle
319 * Calculate force/pot: Force
322 nb_kernel_ElecNone_VdwLJSh_GeomP1P1_F_sparc64_hpc_ace_double
323 (t_nblist * gmx_restrict nlist,
324 rvec * gmx_restrict xx,
325 rvec * gmx_restrict ff,
326 t_forcerec * gmx_restrict fr,
327 t_mdatoms * gmx_restrict mdatoms,
328 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
329 t_nrnb * gmx_restrict nrnb)
331 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
332 * just 0 for non-waters.
333 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
334 * jnr indices corresponding to data put in the four positions in the SIMD register.
336 int i_shift_offset,i_coord_offset,outeriter,inneriter;
337 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
339 int j_coord_offsetA,j_coord_offsetB;
340 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
342 real *shiftvec,*fshift,*x,*f;
343 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
345 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
346 int vdwjidx0A,vdwjidx0B;
347 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
348 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
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 nvdwtype = fr->ntype;
374 vdwtype = mdatoms->typeA;
376 rcutoff_scalar = fr->rvdw;
377 rcutoff = gmx_fjsp_set1_v2r8(rcutoff_scalar);
378 rcutoff2 = _fjsp_mul_v2r8(rcutoff,rcutoff);
380 sh_vdw_invrcut6 = gmx_fjsp_set1_v2r8(fr->ic->sh_invrc6);
381 rvdw = gmx_fjsp_set1_v2r8(fr->rvdw);
383 /* Avoid stupid compiler warnings */
391 /* Start outer loop over neighborlists */
392 for(iidx=0; iidx<nri; iidx++)
394 /* Load shift vector for this list */
395 i_shift_offset = DIM*shiftidx[iidx];
397 /* Load limits for loop over neighbors */
398 j_index_start = jindex[iidx];
399 j_index_end = jindex[iidx+1];
401 /* Get outer coordinate index */
403 i_coord_offset = DIM*inr;
405 /* Load i particle coords and add shift vector */
406 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
408 fix0 = _fjsp_setzero_v2r8();
409 fiy0 = _fjsp_setzero_v2r8();
410 fiz0 = _fjsp_setzero_v2r8();
412 /* Load parameters for i particles */
413 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
415 /* Start inner kernel loop */
416 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
419 /* Get j neighbor index, and coordinate index */
422 j_coord_offsetA = DIM*jnrA;
423 j_coord_offsetB = DIM*jnrB;
425 /* load j atom coordinates */
426 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
429 /* Calculate displacement vector */
430 dx00 = _fjsp_sub_v2r8(ix0,jx0);
431 dy00 = _fjsp_sub_v2r8(iy0,jy0);
432 dz00 = _fjsp_sub_v2r8(iz0,jz0);
434 /* Calculate squared distance and things based on it */
435 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
437 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
439 /* Load parameters for j particles */
440 vdwjidx0A = 2*vdwtype[jnrA+0];
441 vdwjidx0B = 2*vdwtype[jnrB+0];
443 /**************************
444 * CALCULATE INTERACTIONS *
445 **************************/
447 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
450 /* Compute parameters for interactions between i and j atoms */
451 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
452 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
454 /* LENNARD-JONES DISPERSION/REPULSION */
456 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
457 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
459 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
463 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
465 /* Update vectorial force */
466 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
467 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
468 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
470 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
474 /* Inner loop uses 33 flops */
481 j_coord_offsetA = DIM*jnrA;
483 /* load j atom coordinates */
484 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
487 /* Calculate displacement vector */
488 dx00 = _fjsp_sub_v2r8(ix0,jx0);
489 dy00 = _fjsp_sub_v2r8(iy0,jy0);
490 dz00 = _fjsp_sub_v2r8(iz0,jz0);
492 /* Calculate squared distance and things based on it */
493 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
495 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
497 /* Load parameters for j particles */
498 vdwjidx0A = 2*vdwtype[jnrA+0];
500 /**************************
501 * CALCULATE INTERACTIONS *
502 **************************/
504 if (gmx_fjsp_any_lt_v2r8(rsq00,rcutoff2))
507 /* Compute parameters for interactions between i and j atoms */
508 gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
510 /* LENNARD-JONES DISPERSION/REPULSION */
512 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
513 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
515 cutoff_mask = _fjsp_cmplt_v2r8(rsq00,rcutoff2);
519 fscal = _fjsp_and_v2r8(fscal,cutoff_mask);
521 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
523 /* Update vectorial force */
524 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
525 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
526 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
528 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
532 /* Inner loop uses 33 flops */
535 /* End of innermost loop */
537 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
538 f+i_coord_offset,fshift+i_shift_offset);
540 /* Increment number of inner iterations */
541 inneriter += j_index_end - j_index_start;
543 /* Outer loop uses 6 flops */
546 /* Increment number of outer iterations */
549 /* Update outer/inner flops */
551 inc_nrnb(nrnb,eNR_NBKERNEL_VDW_F,outeriter*6 + inneriter*33);