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_ElecEw_VdwNone_GeomP1P1_VF_sparc64_hpc_ace_double
53 * Electrostatics interaction: Ewald
54 * VdW interaction: None
55 * Geometry: Particle-Particle
56 * Calculate force/pot: PotentialAndForce
59 nb_kernel_ElecEw_VdwNone_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;
88 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
91 _fjsp_v2r8 dummy_mask,cutoff_mask;
92 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
93 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
94 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
101 jindex = nlist->jindex;
103 shiftidx = nlist->shift;
105 shiftvec = fr->shift_vec[0];
106 fshift = fr->fshift[0];
107 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
108 charge = mdatoms->chargeA;
110 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
111 ewtab = fr->ic->tabq_coul_FDV0;
112 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
113 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
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));
147 /* Reset potential sums */
148 velecsum = _fjsp_setzero_v2r8();
150 /* Start inner kernel loop */
151 for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
154 /* Get j neighbor index, and coordinate index */
157 j_coord_offsetA = DIM*jnrA;
158 j_coord_offsetB = DIM*jnrB;
160 /* load j atom coordinates */
161 gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
164 /* Calculate displacement vector */
165 dx00 = _fjsp_sub_v2r8(ix0,jx0);
166 dy00 = _fjsp_sub_v2r8(iy0,jy0);
167 dz00 = _fjsp_sub_v2r8(iz0,jz0);
169 /* Calculate squared distance and things based on it */
170 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
172 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
174 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
176 /* Load parameters for j particles */
177 jq0 = gmx_fjsp_load_2real_swizzle_v2r8(charge+jnrA+0,charge+jnrB+0);
179 /**************************
180 * CALCULATE INTERACTIONS *
181 **************************/
183 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
185 /* Compute parameters for interactions between i and j atoms */
186 qq00 = _fjsp_mul_v2r8(iq0,jq0);
188 /* EWALD ELECTROSTATICS */
190 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
191 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
192 itab_tmp = _fjsp_dtox_v2r8(ewrt);
193 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
194 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
196 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
197 ewtabD = _fjsp_load_v2r8( ewtab + 4*ewconv.i[1] );
198 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
199 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
200 ewtabFn = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[1] +2);
201 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
202 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
203 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
204 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(rinv00,velec));
205 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
207 /* Update potential sum for this i atom from the interaction with this j atom. */
208 velecsum = _fjsp_add_v2r8(velecsum,velec);
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 44 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);
247 /**************************
248 * CALCULATE INTERACTIONS *
249 **************************/
251 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
253 /* Compute parameters for interactions between i and j atoms */
254 qq00 = _fjsp_mul_v2r8(iq0,jq0);
256 /* EWALD ELECTROSTATICS */
258 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
259 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
260 itab_tmp = _fjsp_dtox_v2r8(ewrt);
261 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
262 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
264 ewtabF = _fjsp_load_v2r8( ewtab + 4*ewconv.i[0] );
265 ewtabD = _fjsp_setzero_v2r8();
266 GMX_FJSP_TRANSPOSE2_V2R8(ewtabF,ewtabD);
267 ewtabV = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(), ewtab + 4*ewconv.i[0] +2);
268 ewtabFn = _fjsp_setzero_v2r8();
269 GMX_FJSP_TRANSPOSE2_V2R8(ewtabV,ewtabFn);
270 felec = _fjsp_madd_v2r8(eweps,ewtabD,ewtabF);
271 velec = _fjsp_nmsub_v2r8(_fjsp_mul_v2r8(ewtabhalfspace,eweps) ,_fjsp_add_v2r8(ewtabF,felec), ewtabV);
272 velec = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(rinv00,velec));
273 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
275 /* Update potential sum for this i atom from the interaction with this j atom. */
276 velec = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
277 velecsum = _fjsp_add_v2r8(velecsum,velec);
281 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
283 /* Update vectorial force */
284 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
285 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
286 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
288 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
290 /* Inner loop uses 44 flops */
293 /* End of innermost loop */
295 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
296 f+i_coord_offset,fshift+i_shift_offset);
299 /* Update potential energies */
300 gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
302 /* Increment number of inner iterations */
303 inneriter += j_index_end - j_index_start;
305 /* Outer loop uses 8 flops */
308 /* Increment number of outer iterations */
311 /* Update outer/inner flops */
313 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VF,outeriter*8 + inneriter*44);
316 * Gromacs nonbonded kernel: nb_kernel_ElecEw_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
317 * Electrostatics interaction: Ewald
318 * VdW interaction: None
319 * Geometry: Particle-Particle
320 * Calculate force/pot: Force
323 nb_kernel_ElecEw_VdwNone_GeomP1P1_F_sparc64_hpc_ace_double
324 (t_nblist * gmx_restrict nlist,
325 rvec * gmx_restrict xx,
326 rvec * gmx_restrict ff,
327 t_forcerec * gmx_restrict fr,
328 t_mdatoms * gmx_restrict mdatoms,
329 nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
330 t_nrnb * gmx_restrict nrnb)
332 /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
333 * just 0 for non-waters.
334 * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
335 * jnr indices corresponding to data put in the four positions in the SIMD register.
337 int i_shift_offset,i_coord_offset,outeriter,inneriter;
338 int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
340 int j_coord_offsetA,j_coord_offsetB;
341 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
343 real *shiftvec,*fshift,*x,*f;
344 _fjsp_v2r8 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
346 _fjsp_v2r8 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
347 int vdwjidx0A,vdwjidx0B;
348 _fjsp_v2r8 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
349 _fjsp_v2r8 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
350 _fjsp_v2r8 velec,felec,velecsum,facel,crf,krf,krf2;
352 _fjsp_v2r8 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
355 _fjsp_v2r8 dummy_mask,cutoff_mask;
356 _fjsp_v2r8 one = gmx_fjsp_set1_v2r8(1.0);
357 _fjsp_v2r8 two = gmx_fjsp_set1_v2r8(2.0);
358 union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
365 jindex = nlist->jindex;
367 shiftidx = nlist->shift;
369 shiftvec = fr->shift_vec[0];
370 fshift = fr->fshift[0];
371 facel = gmx_fjsp_set1_v2r8(fr->epsfac);
372 charge = mdatoms->chargeA;
374 sh_ewald = gmx_fjsp_set1_v2r8(fr->ic->sh_ewald);
375 ewtab = fr->ic->tabq_coul_F;
376 ewtabscale = gmx_fjsp_set1_v2r8(fr->ic->tabq_scale);
377 ewtabhalfspace = gmx_fjsp_set1_v2r8(0.5/fr->ic->tabq_scale);
379 /* Avoid stupid compiler warnings */
387 /* Start outer loop over neighborlists */
388 for(iidx=0; iidx<nri; iidx++)
390 /* Load shift vector for this list */
391 i_shift_offset = DIM*shiftidx[iidx];
393 /* Load limits for loop over neighbors */
394 j_index_start = jindex[iidx];
395 j_index_end = jindex[iidx+1];
397 /* Get outer coordinate index */
399 i_coord_offset = DIM*inr;
401 /* Load i particle coords and add shift vector */
402 gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
404 fix0 = _fjsp_setzero_v2r8();
405 fiy0 = _fjsp_setzero_v2r8();
406 fiz0 = _fjsp_setzero_v2r8();
408 /* Load parameters for i particles */
409 iq0 = _fjsp_mul_v2r8(facel,gmx_fjsp_load1_v2r8(charge+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);
440 /**************************
441 * CALCULATE INTERACTIONS *
442 **************************/
444 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
446 /* Compute parameters for interactions between i and j atoms */
447 qq00 = _fjsp_mul_v2r8(iq0,jq0);
449 /* EWALD ELECTROSTATICS */
451 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
452 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
453 itab_tmp = _fjsp_dtox_v2r8(ewrt);
454 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
455 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
457 gmx_fjsp_load_2pair_swizzle_v2r8(ewtab+ewconv.i[0],ewtab+ewconv.i[1],
459 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
460 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
464 /* Update vectorial force */
465 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
466 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
467 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
469 gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
471 /* Inner loop uses 39 flops */
478 j_coord_offsetA = DIM*jnrA;
480 /* load j atom coordinates */
481 gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
484 /* Calculate displacement vector */
485 dx00 = _fjsp_sub_v2r8(ix0,jx0);
486 dy00 = _fjsp_sub_v2r8(iy0,jy0);
487 dz00 = _fjsp_sub_v2r8(iz0,jz0);
489 /* Calculate squared distance and things based on it */
490 rsq00 = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
492 rinv00 = gmx_fjsp_invsqrt_v2r8(rsq00);
494 rinvsq00 = _fjsp_mul_v2r8(rinv00,rinv00);
496 /* Load parameters for j particles */
497 jq0 = _fjsp_loadl_v2r8(_fjsp_setzero_v2r8(),charge+jnrA+0);
499 /**************************
500 * CALCULATE INTERACTIONS *
501 **************************/
503 r00 = _fjsp_mul_v2r8(rsq00,rinv00);
505 /* Compute parameters for interactions between i and j atoms */
506 qq00 = _fjsp_mul_v2r8(iq0,jq0);
508 /* EWALD ELECTROSTATICS */
510 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
511 ewrt = _fjsp_mul_v2r8(r00,ewtabscale);
512 itab_tmp = _fjsp_dtox_v2r8(ewrt);
513 eweps = _fjsp_sub_v2r8(ewrt,_fjsp_xtod_v2r8(itab_tmp));
514 _fjsp_store_v2r8(&ewconv.simd,itab_tmp);
516 gmx_fjsp_load_1pair_swizzle_v2r8(ewtab+ewconv.i[0],&ewtabF,&ewtabFn);
517 felec = _fjsp_madd_v2r8(eweps,ewtabFn,_fjsp_nmsub_v2r8(eweps,ewtabF,ewtabF));
518 felec = _fjsp_mul_v2r8(_fjsp_mul_v2r8(qq00,rinv00),_fjsp_sub_v2r8(rinvsq00,felec));
522 fscal = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
524 /* Update vectorial force */
525 fix0 = _fjsp_madd_v2r8(dx00,fscal,fix0);
526 fiy0 = _fjsp_madd_v2r8(dy00,fscal,fiy0);
527 fiz0 = _fjsp_madd_v2r8(dz00,fscal,fiz0);
529 gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
531 /* Inner loop uses 39 flops */
534 /* End of innermost loop */
536 gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
537 f+i_coord_offset,fshift+i_shift_offset);
539 /* Increment number of inner iterations */
540 inneriter += j_index_end - j_index_start;
542 /* Outer loop uses 7 flops */
545 /* Increment number of outer iterations */
548 /* Update outer/inner flops */
550 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_F,outeriter*7 + inneriter*39);