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/legacyheaders/vec.h"
49 #include "kernelutil_sparc64_hpc_ace_double.h"
52 * Gromacs nonbonded kernel: nb_kernel_ElecNone_VdwLJ_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_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;
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 /* Avoid stupid compiler warnings */
121 /* Start outer loop over neighborlists */
122 for(iidx=0; iidx<nri; iidx++)
124 /* Load shift vector for this list */
125 i_shift_offset = DIM*shiftidx[iidx];
127 /* Load limits for loop over neighbors */
128 j_index_start = jindex[iidx];
129 j_index_end = jindex[iidx+1];
131 /* Get outer coordinate index */
133 i_coord_offset = DIM*inr;
135 /* Load i particle coords and add shift vector */
136 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
138 fix0 = _fjsp_setzero_v2r8();
139 fiy0 = _fjsp_setzero_v2r8();
140 fiz0 = _fjsp_setzero_v2r8();
142 /* Load parameters for i particles */
143 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
145 /* Reset potential sums */
146 vvdwsum = _fjsp_setzero_v2r8();
148 /* Start inner kernel loop */
149 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
152 /* Get j neighbor index, and coordinate index */
155 j_coord_offsetA = DIM*jnrA;
156 j_coord_offsetB = DIM*jnrB;
158 /* load j atom coordinates */
159 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
162 /* Calculate displacement vector */
163 dx00 = _fjsp_sub_v2r8(ix0,jx0);
164 dy00 = _fjsp_sub_v2r8(iy0,jy0);
165 dz00 = _fjsp_sub_v2r8(iz0,jz0);
167 /* Calculate squared distance and things based on it */
168 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
170 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
172 /* Load parameters for j particles */
173 vdwjidx0A = 2*vdwtype[jnrA+0];
174 vdwjidx0B = 2*vdwtype[jnrB+0];
176 /**************************
177 * CALCULATE INTERACTIONS *
178 **************************/
180 /* Compute parameters for interactions between i and j atoms */
181 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
182 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
184 /* LENNARD-JONES DISPERSION/REPULSION */
186 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
187 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
188 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
189 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
190 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
192 /* Update potential sum for this i atom from the interaction with this j atom. */
193 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
197 /* Update vectorial force */
198 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
199 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
200 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
202 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
204 /* Inner loop uses 35 flops */
211 j_coord_offsetA = DIM*jnrA;
213 /* load j atom coordinates */
214 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
217 /* Calculate displacement vector */
218 dx00 = _fjsp_sub_v2r8(ix0,jx0);
219 dy00 = _fjsp_sub_v2r8(iy0,jy0);
220 dz00 = _fjsp_sub_v2r8(iz0,jz0);
222 /* Calculate squared distance and things based on it */
223 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
225 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
227 /* Load parameters for j particles */
228 vdwjidx0A = 2*vdwtype[jnrA+0];
230 /**************************
231 * CALCULATE INTERACTIONS *
232 **************************/
234 /* Compute parameters for interactions between i and j atoms */
235 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
236 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
238 /* LENNARD-JONES DISPERSION/REPULSION */
240 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
241 vvdw6 = _fjsp_mul_v2r8(c6_00,rinvsix);
242 vvdw12 = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
243 vvdw = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
244 fvdw = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
246 /* Update potential sum for this i atom from the interaction with this j atom. */
247 vvdw = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
248 vvdwsum = _fjsp_add_v2r8(vvdwsum,vvdw);
252 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
254 /* Update vectorial force */
255 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
256 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
257 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
259 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
261 /* Inner loop uses 35 flops */
264 /* End of innermost loop */
266 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
267 f+i_coord_offset,fshift+i_shift_offset);
270 /* Update potential energies */
271 gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
273 /* Increment number of inner iterations */
274 inneriter += j_index_end - j_index_start;
276 /* Outer loop uses 7 flops */
279 /* Increment number of outer iterations */
282 /* Update outer/inner flops */
284 inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*35);
287 * Gromacs nonbonded kernel: nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
288 * Electrostatics interaction: None
289 * VdW interaction: LennardJones
290 * Geometry: Particle-Particle
291 * Calculate force/pot: Force
294 nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
295 (t_nblist * gmx_restrict nlist,
296 rvec * gmx_restrict xx,
297 rvec * gmx_restrict ff,
298 t_forcerec * gmx_restrict fr,
299 t_mdatoms * gmx_restrict mdatoms,
300 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
301 t_nrnb * gmx_restrict nrnb)
303 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
304 * just 0 for non-waters.
305 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
306 * jnr indices corresponding to data put in the four positions in the SIMD register.
308 int i_shift_offset,i_coord_offset,outeriter,inneriter;
309 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
311 int j_coord_offsetA,j_coord_offsetB;
312 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
314 real *shiftvec,*fshift,*x,*f;
315 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
317 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
318 int vdwjidx0A,vdwjidx0B;
319 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
320 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
322 _fjsp_v2r8 rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
325 _fjsp_v2r8 one_sixth = gmx_fjsp_set1_v2r8(1.0/6.0);
326 _fjsp_v2r8 one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
328 _fjsp_v2r8 dummy_mask,cutoff_mask;
329 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
330 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
331 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
338 jindex = nlist->jindex;
340 shiftidx = nlist->shift;
342 shiftvec = fr->shift_vec[0];
343 fshift = fr->fshift[0];
344 nvdwtype = fr->ntype;
346 vdwtype = mdatoms->typeA;
348 /* Avoid stupid compiler warnings */
356 /* Start outer loop over neighborlists */
357 for(iidx=0; iidx<nri; iidx++)
359 /* Load shift vector for this list */
360 i_shift_offset = DIM*shiftidx[iidx];
362 /* Load limits for loop over neighbors */
363 j_index_start = jindex[iidx];
364 j_index_end = jindex[iidx+1];
366 /* Get outer coordinate index */
368 i_coord_offset = DIM*inr;
370 /* Load i particle coords and add shift vector */
371 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
373 fix0 = _fjsp_setzero_v2r8();
374 fiy0 = _fjsp_setzero_v2r8();
375 fiz0 = _fjsp_setzero_v2r8();
377 /* Load parameters for i particles */
378 vdwioffset0 = 2*nvdwtype*vdwtype[inr+0];
380 /* Start inner kernel loop */
381 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
384 /* Get j neighbor index, and coordinate index */
387 j_coord_offsetA = DIM*jnrA;
388 j_coord_offsetB = DIM*jnrB;
390 /* load j atom coordinates */
391 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
394 /* Calculate displacement vector */
395 dx00 = _fjsp_sub_v2r8(ix0,jx0);
396 dy00 = _fjsp_sub_v2r8(iy0,jy0);
397 dz00 = _fjsp_sub_v2r8(iz0,jz0);
399 /* Calculate squared distance and things based on it */
400 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
402 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
404 /* Load parameters for j particles */
405 vdwjidx0A = 2*vdwtype[jnrA+0];
406 vdwjidx0B = 2*vdwtype[jnrB+0];
408 /**************************
409 * CALCULATE INTERACTIONS *
410 **************************/
412 /* Compute parameters for interactions between i and j atoms */
413 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
414 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
416 /* LENNARD-JONES DISPERSION/REPULSION */
418 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
419 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
423 /* Update vectorial force */
424 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
425 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
426 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
428 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
430 /* Inner loop uses 30 flops */
437 j_coord_offsetA = DIM*jnrA;
439 /* load j atom coordinates */
440 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
443 /* Calculate displacement vector */
444 dx00 = _fjsp_sub_v2r8(ix0,jx0);
445 dy00 = _fjsp_sub_v2r8(iy0,jy0);
446 dz00 = _fjsp_sub_v2r8(iz0,jz0);
448 /* Calculate squared distance and things based on it */
449 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
451 rinvsq00 = gmx_fjsp_inv_v2r8(rsq00);
453 /* Load parameters for j particles */
454 vdwjidx0A = 2*vdwtype[jnrA+0];
456 /**************************
457 * CALCULATE INTERACTIONS *
458 **************************/
460 /* Compute parameters for interactions between i and j atoms */
461 gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
462 vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
464 /* LENNARD-JONES DISPERSION/REPULSION */
466 rinvsix = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
467 fvdw = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
471 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
473 /* Update vectorial force */
474 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
475 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
476 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
478 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
480 /* Inner loop uses 30 flops */
483 /* End of innermost loop */
485 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
486 f+i_coord_offset,fshift+i_shift_offset);
488 /* Increment number of inner iterations */
489 inneriter += j_index_end - j_index_start;
491 /* Outer loop uses 6 flops */
494 /* Increment number of outer iterations */
497 /* Update outer/inner flops */
499 inc_nrnb(nrnb,eNR_NBKERNEL_VDW_F,outeriter*6 + inneriter*30);