2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2012, The GROMACS development team,
6 * check out http://www.gromacs.org for more information.
7 * Copyright (c) 2012, by the GROMACS development team, led by
8 * David van der Spoel, Berk Hess, Erik Lindahl, and including many
9 * others, as listed in the AUTHORS file in the top-level source
10 * directory and at http://www.gromacs.org.
12 * GROMACS is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU Lesser General Public License
14 * as published by the Free Software Foundation; either version 2.1
15 * of the License, or (at your option) any later version.
17 * GROMACS is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 * Lesser General Public License for more details.
22 * You should have received a copy of the GNU Lesser General Public
23 * License along with GROMACS; if not, see
24 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
25 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
27 * If you want to redistribute modifications to GROMACS, please
28 * consider that scientific software is very special. Version
29 * control is crucial - bugs must be traceable. We will be happy to
30 * consider code for inclusion in the official distribution, but
31 * derived work must not be called official GROMACS. Details are found
32 * in the README & COPYING files - if they are missing, get the
33 * official version at http://www.gromacs.org.
35 * To help us fund GROMACS development, we humbly ask that you cite
36 * the research papers on the package. Check out http://www.gromacs.org.
39 /* GMX_MM128_HERE or GMX_MM256_HERE should be set before including this file.
40 * gmx_sse_or_avh.h should be included before including this file.
43 /* Copies PBC shifted i-cell packed atom coordinates to working array */
45 static void icell_set_x_x86_simd128
48 static void icell_set_x_x86_simd256
50 "error: GMX_MM128_HERE or GMX_MM256_HERE not defined"
54 real shx,real shy,real shz,
56 int stride,const real *x,
57 nbnxn_list_work_t *work)
61 nbnxn_x_ci_x86_simd128_t *x_ci;
63 x_ci = work->x_ci_x86_simd128;
65 ia = X_IND_CI_S128(ci);
67 nbnxn_x_ci_x86_simd256_t *x_ci;
69 x_ci = work->x_ci_x86_simd256;
71 ia = X_IND_CI_S256(ci);
74 x_ci->ix_SSE0 = gmx_set1_pr(x[ia + 0*STRIDE_S ] + shx);
75 x_ci->iy_SSE0 = gmx_set1_pr(x[ia + 1*STRIDE_S ] + shy);
76 x_ci->iz_SSE0 = gmx_set1_pr(x[ia + 2*STRIDE_S ] + shz);
77 x_ci->ix_SSE1 = gmx_set1_pr(x[ia + 0*STRIDE_S + 1] + shx);
78 x_ci->iy_SSE1 = gmx_set1_pr(x[ia + 1*STRIDE_S + 1] + shy);
79 x_ci->iz_SSE1 = gmx_set1_pr(x[ia + 2*STRIDE_S + 1] + shz);
80 x_ci->ix_SSE2 = gmx_set1_pr(x[ia + 0*STRIDE_S + 2] + shx);
81 x_ci->iy_SSE2 = gmx_set1_pr(x[ia + 1*STRIDE_S + 2] + shy);
82 x_ci->iz_SSE2 = gmx_set1_pr(x[ia + 2*STRIDE_S + 2] + shz);
83 x_ci->ix_SSE3 = gmx_set1_pr(x[ia + 0*STRIDE_S + 3] + shx);
84 x_ci->iy_SSE3 = gmx_set1_pr(x[ia + 1*STRIDE_S + 3] + shy);
85 x_ci->iz_SSE3 = gmx_set1_pr(x[ia + 2*STRIDE_S + 3] + shz);
88 /* SSE or AVX code for making a pair list of cell ci vs cell cjf-cjl
89 * for coordinates in packed format.
90 * Checks bouding box distances and possibly atom pair distances.
91 * This is an accelerated version of make_cluster_list_simple.
94 static void make_cluster_list_x86_simd128
97 static void make_cluster_list_x86_simd256
99 "error: GMX_MM128_HERE or GMX_MM256_HERE not defined"
102 (const nbnxn_grid_t *gridj,
103 nbnxn_pairlist_t *nbl,
104 int ci,int cjf,int cjl,
105 gmx_bool remove_sub_diag,
110 #ifdef GMX_MM128_HERE
111 const nbnxn_x_ci_x86_simd128_t *work;
113 const nbnxn_x_ci_x86_simd256_t *work;
118 gmx_mm_pr jx_SSE,jy_SSE,jz_SSE;
120 gmx_mm_pr dx_SSE0,dy_SSE0,dz_SSE0;
121 gmx_mm_pr dx_SSE1,dy_SSE1,dz_SSE1;
122 gmx_mm_pr dx_SSE2,dy_SSE2,dz_SSE2;
123 gmx_mm_pr dx_SSE3,dy_SSE3,dz_SSE3;
134 gmx_mm_pr wco_any_SSE01,wco_any_SSE23,wco_any_SSE;
140 int xind_f,xind_l,cj;
142 #ifdef GMX_MM128_HERE
143 cjf = CI_TO_CJ_S128(cjf);
144 cjl = CI_TO_CJ_S128(cjl+1) - 1;
146 work = nbl->work->x_ci_x86_simd128;
148 cjf = CI_TO_CJ_S256(cjf);
149 cjl = CI_TO_CJ_S256(cjl+1) - 1;
151 work = nbl->work->x_ci_x86_simd256;
154 bb_ci = nbl->work->bb_ci;
156 rc2_SSE = gmx_set1_pr(rl2);
159 while (!InRange && cjf <= cjl)
161 d2 = subc_bb_dist2_sse(4,0,bb_ci,cjf,gridj->bbj);
164 /* Check if the distance is within the distance where
165 * we use only the bounding box distance rbb,
166 * or within the cut-off and there is at least one atom pair
167 * within the cut-off.
175 #ifdef GMX_MM128_HERE
176 xind_f = X_IND_CJ_S128(CI_TO_CJ_S128(gridj->cell0) + cjf);
178 xind_f = X_IND_CJ_S256(CI_TO_CJ_S256(gridj->cell0) + cjf);
180 jx_SSE = gmx_load_pr(x_j+xind_f+0*STRIDE_S);
181 jy_SSE = gmx_load_pr(x_j+xind_f+1*STRIDE_S);
182 jz_SSE = gmx_load_pr(x_j+xind_f+2*STRIDE_S);
185 /* Calculate distance */
186 dx_SSE0 = gmx_sub_pr(work->ix_SSE0,jx_SSE);
187 dy_SSE0 = gmx_sub_pr(work->iy_SSE0,jy_SSE);
188 dz_SSE0 = gmx_sub_pr(work->iz_SSE0,jz_SSE);
189 dx_SSE1 = gmx_sub_pr(work->ix_SSE1,jx_SSE);
190 dy_SSE1 = gmx_sub_pr(work->iy_SSE1,jy_SSE);
191 dz_SSE1 = gmx_sub_pr(work->iz_SSE1,jz_SSE);
192 dx_SSE2 = gmx_sub_pr(work->ix_SSE2,jx_SSE);
193 dy_SSE2 = gmx_sub_pr(work->iy_SSE2,jy_SSE);
194 dz_SSE2 = gmx_sub_pr(work->iz_SSE2,jz_SSE);
195 dx_SSE3 = gmx_sub_pr(work->ix_SSE3,jx_SSE);
196 dy_SSE3 = gmx_sub_pr(work->iy_SSE3,jy_SSE);
197 dz_SSE3 = gmx_sub_pr(work->iz_SSE3,jz_SSE);
199 /* rsq = dx*dx+dy*dy+dz*dz */
200 rsq_SSE0 = gmx_calc_rsq_pr(dx_SSE0,dy_SSE0,dz_SSE0);
201 rsq_SSE1 = gmx_calc_rsq_pr(dx_SSE1,dy_SSE1,dz_SSE1);
202 rsq_SSE2 = gmx_calc_rsq_pr(dx_SSE2,dy_SSE2,dz_SSE2);
203 rsq_SSE3 = gmx_calc_rsq_pr(dx_SSE3,dy_SSE3,dz_SSE3);
205 wco_SSE0 = gmx_cmplt_pr(rsq_SSE0,rc2_SSE);
206 wco_SSE1 = gmx_cmplt_pr(rsq_SSE1,rc2_SSE);
207 wco_SSE2 = gmx_cmplt_pr(rsq_SSE2,rc2_SSE);
208 wco_SSE3 = gmx_cmplt_pr(rsq_SSE3,rc2_SSE);
210 wco_any_SSE01 = gmx_or_pr(wco_SSE0,wco_SSE1);
211 wco_any_SSE23 = gmx_or_pr(wco_SSE2,wco_SSE3);
212 wco_any_SSE = gmx_or_pr(wco_any_SSE01,wco_any_SSE23);
214 InRange = gmx_movemask_pr(wco_any_SSE);
216 *ndistc += 4*GMX_X86_SIMD_WIDTH_HERE;
229 while (!InRange && cjl > cjf)
231 d2 = subc_bb_dist2_sse(4,0,bb_ci,cjl,gridj->bbj);
234 /* Check if the distance is within the distance where
235 * we use only the bounding box distance rbb,
236 * or within the cut-off and there is at least one atom pair
237 * within the cut-off.
245 #ifdef GMX_MM128_HERE
246 xind_l = X_IND_CJ_S128(CI_TO_CJ_S128(gridj->cell0) + cjl);
248 xind_l = X_IND_CJ_S256(CI_TO_CJ_S256(gridj->cell0) + cjl);
250 jx_SSE = gmx_load_pr(x_j+xind_l+0*STRIDE_S);
251 jy_SSE = gmx_load_pr(x_j+xind_l+1*STRIDE_S);
252 jz_SSE = gmx_load_pr(x_j+xind_l+2*STRIDE_S);
254 /* Calculate distance */
255 dx_SSE0 = gmx_sub_pr(work->ix_SSE0,jx_SSE);
256 dy_SSE0 = gmx_sub_pr(work->iy_SSE0,jy_SSE);
257 dz_SSE0 = gmx_sub_pr(work->iz_SSE0,jz_SSE);
258 dx_SSE1 = gmx_sub_pr(work->ix_SSE1,jx_SSE);
259 dy_SSE1 = gmx_sub_pr(work->iy_SSE1,jy_SSE);
260 dz_SSE1 = gmx_sub_pr(work->iz_SSE1,jz_SSE);
261 dx_SSE2 = gmx_sub_pr(work->ix_SSE2,jx_SSE);
262 dy_SSE2 = gmx_sub_pr(work->iy_SSE2,jy_SSE);
263 dz_SSE2 = gmx_sub_pr(work->iz_SSE2,jz_SSE);
264 dx_SSE3 = gmx_sub_pr(work->ix_SSE3,jx_SSE);
265 dy_SSE3 = gmx_sub_pr(work->iy_SSE3,jy_SSE);
266 dz_SSE3 = gmx_sub_pr(work->iz_SSE3,jz_SSE);
268 /* rsq = dx*dx+dy*dy+dz*dz */
269 rsq_SSE0 = gmx_calc_rsq_pr(dx_SSE0,dy_SSE0,dz_SSE0);
270 rsq_SSE1 = gmx_calc_rsq_pr(dx_SSE1,dy_SSE1,dz_SSE1);
271 rsq_SSE2 = gmx_calc_rsq_pr(dx_SSE2,dy_SSE2,dz_SSE2);
272 rsq_SSE3 = gmx_calc_rsq_pr(dx_SSE3,dy_SSE3,dz_SSE3);
274 wco_SSE0 = gmx_cmplt_pr(rsq_SSE0,rc2_SSE);
275 wco_SSE1 = gmx_cmplt_pr(rsq_SSE1,rc2_SSE);
276 wco_SSE2 = gmx_cmplt_pr(rsq_SSE2,rc2_SSE);
277 wco_SSE3 = gmx_cmplt_pr(rsq_SSE3,rc2_SSE);
279 wco_any_SSE01 = gmx_or_pr(wco_SSE0,wco_SSE1);
280 wco_any_SSE23 = gmx_or_pr(wco_SSE2,wco_SSE3);
281 wco_any_SSE = gmx_or_pr(wco_any_SSE01,wco_any_SSE23);
283 InRange = gmx_movemask_pr(wco_any_SSE);
285 *ndistc += 4*GMX_X86_SIMD_WIDTH_HERE;
295 for(cj=cjf; cj<=cjl; cj++)
297 /* Store cj and the interaction mask */
298 #ifdef GMX_MM128_HERE
299 nbl->cj[nbl->ncj].cj = CI_TO_CJ_S128(gridj->cell0) + cj;
300 nbl->cj[nbl->ncj].excl = get_imask_x86_simd128(remove_sub_diag,ci,cj);
302 nbl->cj[nbl->ncj].cj = CI_TO_CJ_S256(gridj->cell0) + cj;
303 nbl->cj[nbl->ncj].excl = get_imask_x86_simd256(remove_sub_diag,ci,cj);
307 /* Increase the closing index in i super-cell list */
308 nbl->ci[nbl->nci].cj_ind_end = nbl->ncj;