File: | gromacs/gmxlib/nonbonded/nb_kernel_sse4_1_single/nb_kernel_ElecCSTab_VdwNone_GeomW4W4_sse4_1_single.c |
Location: | line 1308, column 5 |
Description: | Value stored to 'gid' is never read |
1 | /* |
2 | * This file is part of the GROMACS molecular simulation package. |
3 | * |
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. |
8 | * |
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. |
13 | * |
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. |
18 | * |
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. |
23 | * |
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. |
31 | * |
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. |
34 | */ |
35 | /* |
36 | * Note: this file was generated by the GROMACS sse4_1_single kernel generator. |
37 | */ |
38 | #ifdef HAVE_CONFIG_H1 |
39 | #include <config.h> |
40 | #endif |
41 | |
42 | #include <math.h> |
43 | |
44 | #include "../nb_kernel.h" |
45 | #include "types/simple.h" |
46 | #include "gromacs/math/vec.h" |
47 | #include "nrnb.h" |
48 | |
49 | #include "gromacs/simd/math_x86_sse4_1_single.h" |
50 | #include "kernelutil_x86_sse4_1_single.h" |
51 | |
52 | /* |
53 | * Gromacs nonbonded kernel: nb_kernel_ElecCSTab_VdwNone_GeomW4W4_VF_sse4_1_single |
54 | * Electrostatics interaction: CubicSplineTable |
55 | * VdW interaction: None |
56 | * Geometry: Water4-Water4 |
57 | * Calculate force/pot: PotentialAndForce |
58 | */ |
59 | void |
60 | nb_kernel_ElecCSTab_VdwNone_GeomW4W4_VF_sse4_1_single |
61 | (t_nblist * gmx_restrict nlist, |
62 | rvec * gmx_restrict xx, |
63 | rvec * gmx_restrict ff, |
64 | t_forcerec * gmx_restrict fr, |
65 | t_mdatoms * gmx_restrict mdatoms, |
66 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict kernel_data, |
67 | t_nrnb * gmx_restrict nrnb) |
68 | { |
69 | /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or |
70 | * just 0 for non-waters. |
71 | * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different |
72 | * jnr indices corresponding to data put in the four positions in the SIMD register. |
73 | */ |
74 | int i_shift_offset,i_coord_offset,outeriter,inneriter; |
75 | int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx; |
76 | int jnrA,jnrB,jnrC,jnrD; |
77 | int jnrlistA,jnrlistB,jnrlistC,jnrlistD; |
78 | int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD; |
79 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
80 | real rcutoff_scalar; |
81 | real *shiftvec,*fshift,*x,*f; |
82 | real *fjptrA,*fjptrB,*fjptrC,*fjptrD; |
83 | real scratch[4*DIM3]; |
84 | __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall; |
85 | int vdwioffset1; |
86 | __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
87 | int vdwioffset2; |
88 | __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
89 | int vdwioffset3; |
90 | __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3; |
91 | int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D; |
92 | __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
93 | int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D; |
94 | __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
95 | int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D; |
96 | __m128 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3; |
97 | __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11; |
98 | __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12; |
99 | __m128 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13; |
100 | __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21; |
101 | __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22; |
102 | __m128 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23; |
103 | __m128 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31; |
104 | __m128 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32; |
105 | __m128 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33; |
106 | __m128 velec,felec,velecsum,facel,crf,krf,krf2; |
107 | real *charge; |
108 | __m128i vfitab; |
109 | __m128i ifour = _mm_set1_epi32(4); |
110 | __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF; |
111 | real *vftab; |
112 | __m128 dummy_mask,cutoff_mask; |
113 | __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) ); |
114 | __m128 one = _mm_set1_ps(1.0); |
115 | __m128 two = _mm_set1_ps(2.0); |
116 | x = xx[0]; |
117 | f = ff[0]; |
118 | |
119 | nri = nlist->nri; |
120 | iinr = nlist->iinr; |
121 | jindex = nlist->jindex; |
122 | jjnr = nlist->jjnr; |
123 | shiftidx = nlist->shift; |
124 | gid = nlist->gid; |
125 | shiftvec = fr->shift_vec[0]; |
126 | fshift = fr->fshift[0]; |
127 | facel = _mm_set1_ps(fr->epsfac); |
128 | charge = mdatoms->chargeA; |
129 | |
130 | vftab = kernel_data->table_elec->data; |
131 | vftabscale = _mm_set1_ps(kernel_data->table_elec->scale); |
132 | |
133 | /* Setup water-specific parameters */ |
134 | inr = nlist->iinr[0]; |
135 | iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1])); |
136 | iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2])); |
137 | iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3])); |
138 | |
139 | jq1 = _mm_set1_ps(charge[inr+1]); |
140 | jq2 = _mm_set1_ps(charge[inr+2]); |
141 | jq3 = _mm_set1_ps(charge[inr+3]); |
142 | qq11 = _mm_mul_ps(iq1,jq1); |
143 | qq12 = _mm_mul_ps(iq1,jq2); |
144 | qq13 = _mm_mul_ps(iq1,jq3); |
145 | qq21 = _mm_mul_ps(iq2,jq1); |
146 | qq22 = _mm_mul_ps(iq2,jq2); |
147 | qq23 = _mm_mul_ps(iq2,jq3); |
148 | qq31 = _mm_mul_ps(iq3,jq1); |
149 | qq32 = _mm_mul_ps(iq3,jq2); |
150 | qq33 = _mm_mul_ps(iq3,jq3); |
151 | |
152 | /* Avoid stupid compiler warnings */ |
153 | jnrA = jnrB = jnrC = jnrD = 0; |
154 | j_coord_offsetA = 0; |
155 | j_coord_offsetB = 0; |
156 | j_coord_offsetC = 0; |
157 | j_coord_offsetD = 0; |
158 | |
159 | outeriter = 0; |
160 | inneriter = 0; |
161 | |
162 | for(iidx=0;iidx<4*DIM3;iidx++) |
163 | { |
164 | scratch[iidx] = 0.0; |
165 | } |
166 | |
167 | /* Start outer loop over neighborlists */ |
168 | for(iidx=0; iidx<nri; iidx++) |
169 | { |
170 | /* Load shift vector for this list */ |
171 | i_shift_offset = DIM3*shiftidx[iidx]; |
172 | |
173 | /* Load limits for loop over neighbors */ |
174 | j_index_start = jindex[iidx]; |
175 | j_index_end = jindex[iidx+1]; |
176 | |
177 | /* Get outer coordinate index */ |
178 | inr = iinr[iidx]; |
179 | i_coord_offset = DIM3*inr; |
180 | |
181 | /* Load i particle coords and add shift vector */ |
182 | gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM3, |
183 | &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3); |
184 | |
185 | fix1 = _mm_setzero_ps(); |
186 | fiy1 = _mm_setzero_ps(); |
187 | fiz1 = _mm_setzero_ps(); |
188 | fix2 = _mm_setzero_ps(); |
189 | fiy2 = _mm_setzero_ps(); |
190 | fiz2 = _mm_setzero_ps(); |
191 | fix3 = _mm_setzero_ps(); |
192 | fiy3 = _mm_setzero_ps(); |
193 | fiz3 = _mm_setzero_ps(); |
194 | |
195 | /* Reset potential sums */ |
196 | velecsum = _mm_setzero_ps(); |
197 | |
198 | /* Start inner kernel loop */ |
199 | for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4) |
200 | { |
201 | |
202 | /* Get j neighbor index, and coordinate index */ |
203 | jnrA = jjnr[jidx]; |
204 | jnrB = jjnr[jidx+1]; |
205 | jnrC = jjnr[jidx+2]; |
206 | jnrD = jjnr[jidx+3]; |
207 | j_coord_offsetA = DIM3*jnrA; |
208 | j_coord_offsetB = DIM3*jnrB; |
209 | j_coord_offsetC = DIM3*jnrC; |
210 | j_coord_offsetD = DIM3*jnrD; |
211 | |
212 | /* load j atom coordinates */ |
213 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM3,x+j_coord_offsetB+DIM3, |
214 | x+j_coord_offsetC+DIM3,x+j_coord_offsetD+DIM3, |
215 | &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3); |
216 | |
217 | /* Calculate displacement vector */ |
218 | dx11 = _mm_sub_ps(ix1,jx1); |
219 | dy11 = _mm_sub_ps(iy1,jy1); |
220 | dz11 = _mm_sub_ps(iz1,jz1); |
221 | dx12 = _mm_sub_ps(ix1,jx2); |
222 | dy12 = _mm_sub_ps(iy1,jy2); |
223 | dz12 = _mm_sub_ps(iz1,jz2); |
224 | dx13 = _mm_sub_ps(ix1,jx3); |
225 | dy13 = _mm_sub_ps(iy1,jy3); |
226 | dz13 = _mm_sub_ps(iz1,jz3); |
227 | dx21 = _mm_sub_ps(ix2,jx1); |
228 | dy21 = _mm_sub_ps(iy2,jy1); |
229 | dz21 = _mm_sub_ps(iz2,jz1); |
230 | dx22 = _mm_sub_ps(ix2,jx2); |
231 | dy22 = _mm_sub_ps(iy2,jy2); |
232 | dz22 = _mm_sub_ps(iz2,jz2); |
233 | dx23 = _mm_sub_ps(ix2,jx3); |
234 | dy23 = _mm_sub_ps(iy2,jy3); |
235 | dz23 = _mm_sub_ps(iz2,jz3); |
236 | dx31 = _mm_sub_ps(ix3,jx1); |
237 | dy31 = _mm_sub_ps(iy3,jy1); |
238 | dz31 = _mm_sub_ps(iz3,jz1); |
239 | dx32 = _mm_sub_ps(ix3,jx2); |
240 | dy32 = _mm_sub_ps(iy3,jy2); |
241 | dz32 = _mm_sub_ps(iz3,jz2); |
242 | dx33 = _mm_sub_ps(ix3,jx3); |
243 | dy33 = _mm_sub_ps(iy3,jy3); |
244 | dz33 = _mm_sub_ps(iz3,jz3); |
245 | |
246 | /* Calculate squared distance and things based on it */ |
247 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
248 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
249 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
250 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
251 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
252 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
253 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
254 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
255 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
256 | |
257 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
258 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
259 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
260 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
261 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
262 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
263 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
264 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
265 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
266 | |
267 | fjx1 = _mm_setzero_ps(); |
268 | fjy1 = _mm_setzero_ps(); |
269 | fjz1 = _mm_setzero_ps(); |
270 | fjx2 = _mm_setzero_ps(); |
271 | fjy2 = _mm_setzero_ps(); |
272 | fjz2 = _mm_setzero_ps(); |
273 | fjx3 = _mm_setzero_ps(); |
274 | fjy3 = _mm_setzero_ps(); |
275 | fjz3 = _mm_setzero_ps(); |
276 | |
277 | /************************** |
278 | * CALCULATE INTERACTIONS * |
279 | **************************/ |
280 | |
281 | r11 = _mm_mul_ps(rsq11,rinv11); |
282 | |
283 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
284 | rt = _mm_mul_ps(r11,vftabscale); |
285 | vfitab = _mm_cvttps_epi32(rt); |
286 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
287 | vfitab = _mm_slli_epi32(vfitab,2); |
288 | |
289 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
290 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
291 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
292 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
293 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
294 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
295 | Heps = _mm_mul_ps(vfeps,H); |
296 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
297 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
298 | velec = _mm_mul_ps(qq11,VV); |
299 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
300 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
301 | |
302 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
303 | velecsum = _mm_add_ps(velecsum,velec); |
304 | |
305 | fscal = felec; |
306 | |
307 | /* Calculate temporary vectorial force */ |
308 | tx = _mm_mul_ps(fscal,dx11); |
309 | ty = _mm_mul_ps(fscal,dy11); |
310 | tz = _mm_mul_ps(fscal,dz11); |
311 | |
312 | /* Update vectorial force */ |
313 | fix1 = _mm_add_ps(fix1,tx); |
314 | fiy1 = _mm_add_ps(fiy1,ty); |
315 | fiz1 = _mm_add_ps(fiz1,tz); |
316 | |
317 | fjx1 = _mm_add_ps(fjx1,tx); |
318 | fjy1 = _mm_add_ps(fjy1,ty); |
319 | fjz1 = _mm_add_ps(fjz1,tz); |
320 | |
321 | /************************** |
322 | * CALCULATE INTERACTIONS * |
323 | **************************/ |
324 | |
325 | r12 = _mm_mul_ps(rsq12,rinv12); |
326 | |
327 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
328 | rt = _mm_mul_ps(r12,vftabscale); |
329 | vfitab = _mm_cvttps_epi32(rt); |
330 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
331 | vfitab = _mm_slli_epi32(vfitab,2); |
332 | |
333 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
334 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
335 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
336 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
337 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
338 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
339 | Heps = _mm_mul_ps(vfeps,H); |
340 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
341 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
342 | velec = _mm_mul_ps(qq12,VV); |
343 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
344 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
345 | |
346 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
347 | velecsum = _mm_add_ps(velecsum,velec); |
348 | |
349 | fscal = felec; |
350 | |
351 | /* Calculate temporary vectorial force */ |
352 | tx = _mm_mul_ps(fscal,dx12); |
353 | ty = _mm_mul_ps(fscal,dy12); |
354 | tz = _mm_mul_ps(fscal,dz12); |
355 | |
356 | /* Update vectorial force */ |
357 | fix1 = _mm_add_ps(fix1,tx); |
358 | fiy1 = _mm_add_ps(fiy1,ty); |
359 | fiz1 = _mm_add_ps(fiz1,tz); |
360 | |
361 | fjx2 = _mm_add_ps(fjx2,tx); |
362 | fjy2 = _mm_add_ps(fjy2,ty); |
363 | fjz2 = _mm_add_ps(fjz2,tz); |
364 | |
365 | /************************** |
366 | * CALCULATE INTERACTIONS * |
367 | **************************/ |
368 | |
369 | r13 = _mm_mul_ps(rsq13,rinv13); |
370 | |
371 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
372 | rt = _mm_mul_ps(r13,vftabscale); |
373 | vfitab = _mm_cvttps_epi32(rt); |
374 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
375 | vfitab = _mm_slli_epi32(vfitab,2); |
376 | |
377 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
378 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
379 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
380 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
381 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
382 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
383 | Heps = _mm_mul_ps(vfeps,H); |
384 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
385 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
386 | velec = _mm_mul_ps(qq13,VV); |
387 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
388 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
389 | |
390 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
391 | velecsum = _mm_add_ps(velecsum,velec); |
392 | |
393 | fscal = felec; |
394 | |
395 | /* Calculate temporary vectorial force */ |
396 | tx = _mm_mul_ps(fscal,dx13); |
397 | ty = _mm_mul_ps(fscal,dy13); |
398 | tz = _mm_mul_ps(fscal,dz13); |
399 | |
400 | /* Update vectorial force */ |
401 | fix1 = _mm_add_ps(fix1,tx); |
402 | fiy1 = _mm_add_ps(fiy1,ty); |
403 | fiz1 = _mm_add_ps(fiz1,tz); |
404 | |
405 | fjx3 = _mm_add_ps(fjx3,tx); |
406 | fjy3 = _mm_add_ps(fjy3,ty); |
407 | fjz3 = _mm_add_ps(fjz3,tz); |
408 | |
409 | /************************** |
410 | * CALCULATE INTERACTIONS * |
411 | **************************/ |
412 | |
413 | r21 = _mm_mul_ps(rsq21,rinv21); |
414 | |
415 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
416 | rt = _mm_mul_ps(r21,vftabscale); |
417 | vfitab = _mm_cvttps_epi32(rt); |
418 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
419 | vfitab = _mm_slli_epi32(vfitab,2); |
420 | |
421 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
422 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
423 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
424 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
425 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
426 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
427 | Heps = _mm_mul_ps(vfeps,H); |
428 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
429 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
430 | velec = _mm_mul_ps(qq21,VV); |
431 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
432 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
433 | |
434 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
435 | velecsum = _mm_add_ps(velecsum,velec); |
436 | |
437 | fscal = felec; |
438 | |
439 | /* Calculate temporary vectorial force */ |
440 | tx = _mm_mul_ps(fscal,dx21); |
441 | ty = _mm_mul_ps(fscal,dy21); |
442 | tz = _mm_mul_ps(fscal,dz21); |
443 | |
444 | /* Update vectorial force */ |
445 | fix2 = _mm_add_ps(fix2,tx); |
446 | fiy2 = _mm_add_ps(fiy2,ty); |
447 | fiz2 = _mm_add_ps(fiz2,tz); |
448 | |
449 | fjx1 = _mm_add_ps(fjx1,tx); |
450 | fjy1 = _mm_add_ps(fjy1,ty); |
451 | fjz1 = _mm_add_ps(fjz1,tz); |
452 | |
453 | /************************** |
454 | * CALCULATE INTERACTIONS * |
455 | **************************/ |
456 | |
457 | r22 = _mm_mul_ps(rsq22,rinv22); |
458 | |
459 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
460 | rt = _mm_mul_ps(r22,vftabscale); |
461 | vfitab = _mm_cvttps_epi32(rt); |
462 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
463 | vfitab = _mm_slli_epi32(vfitab,2); |
464 | |
465 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
466 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
467 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
468 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
469 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
470 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
471 | Heps = _mm_mul_ps(vfeps,H); |
472 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
473 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
474 | velec = _mm_mul_ps(qq22,VV); |
475 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
476 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
477 | |
478 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
479 | velecsum = _mm_add_ps(velecsum,velec); |
480 | |
481 | fscal = felec; |
482 | |
483 | /* Calculate temporary vectorial force */ |
484 | tx = _mm_mul_ps(fscal,dx22); |
485 | ty = _mm_mul_ps(fscal,dy22); |
486 | tz = _mm_mul_ps(fscal,dz22); |
487 | |
488 | /* Update vectorial force */ |
489 | fix2 = _mm_add_ps(fix2,tx); |
490 | fiy2 = _mm_add_ps(fiy2,ty); |
491 | fiz2 = _mm_add_ps(fiz2,tz); |
492 | |
493 | fjx2 = _mm_add_ps(fjx2,tx); |
494 | fjy2 = _mm_add_ps(fjy2,ty); |
495 | fjz2 = _mm_add_ps(fjz2,tz); |
496 | |
497 | /************************** |
498 | * CALCULATE INTERACTIONS * |
499 | **************************/ |
500 | |
501 | r23 = _mm_mul_ps(rsq23,rinv23); |
502 | |
503 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
504 | rt = _mm_mul_ps(r23,vftabscale); |
505 | vfitab = _mm_cvttps_epi32(rt); |
506 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
507 | vfitab = _mm_slli_epi32(vfitab,2); |
508 | |
509 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
510 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
511 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
512 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
513 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
514 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
515 | Heps = _mm_mul_ps(vfeps,H); |
516 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
517 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
518 | velec = _mm_mul_ps(qq23,VV); |
519 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
520 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
521 | |
522 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
523 | velecsum = _mm_add_ps(velecsum,velec); |
524 | |
525 | fscal = felec; |
526 | |
527 | /* Calculate temporary vectorial force */ |
528 | tx = _mm_mul_ps(fscal,dx23); |
529 | ty = _mm_mul_ps(fscal,dy23); |
530 | tz = _mm_mul_ps(fscal,dz23); |
531 | |
532 | /* Update vectorial force */ |
533 | fix2 = _mm_add_ps(fix2,tx); |
534 | fiy2 = _mm_add_ps(fiy2,ty); |
535 | fiz2 = _mm_add_ps(fiz2,tz); |
536 | |
537 | fjx3 = _mm_add_ps(fjx3,tx); |
538 | fjy3 = _mm_add_ps(fjy3,ty); |
539 | fjz3 = _mm_add_ps(fjz3,tz); |
540 | |
541 | /************************** |
542 | * CALCULATE INTERACTIONS * |
543 | **************************/ |
544 | |
545 | r31 = _mm_mul_ps(rsq31,rinv31); |
546 | |
547 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
548 | rt = _mm_mul_ps(r31,vftabscale); |
549 | vfitab = _mm_cvttps_epi32(rt); |
550 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
551 | vfitab = _mm_slli_epi32(vfitab,2); |
552 | |
553 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
554 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
555 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
556 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
557 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
558 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
559 | Heps = _mm_mul_ps(vfeps,H); |
560 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
561 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
562 | velec = _mm_mul_ps(qq31,VV); |
563 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
564 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
565 | |
566 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
567 | velecsum = _mm_add_ps(velecsum,velec); |
568 | |
569 | fscal = felec; |
570 | |
571 | /* Calculate temporary vectorial force */ |
572 | tx = _mm_mul_ps(fscal,dx31); |
573 | ty = _mm_mul_ps(fscal,dy31); |
574 | tz = _mm_mul_ps(fscal,dz31); |
575 | |
576 | /* Update vectorial force */ |
577 | fix3 = _mm_add_ps(fix3,tx); |
578 | fiy3 = _mm_add_ps(fiy3,ty); |
579 | fiz3 = _mm_add_ps(fiz3,tz); |
580 | |
581 | fjx1 = _mm_add_ps(fjx1,tx); |
582 | fjy1 = _mm_add_ps(fjy1,ty); |
583 | fjz1 = _mm_add_ps(fjz1,tz); |
584 | |
585 | /************************** |
586 | * CALCULATE INTERACTIONS * |
587 | **************************/ |
588 | |
589 | r32 = _mm_mul_ps(rsq32,rinv32); |
590 | |
591 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
592 | rt = _mm_mul_ps(r32,vftabscale); |
593 | vfitab = _mm_cvttps_epi32(rt); |
594 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
595 | vfitab = _mm_slli_epi32(vfitab,2); |
596 | |
597 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
598 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
599 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
600 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
601 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
602 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
603 | Heps = _mm_mul_ps(vfeps,H); |
604 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
605 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
606 | velec = _mm_mul_ps(qq32,VV); |
607 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
608 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
609 | |
610 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
611 | velecsum = _mm_add_ps(velecsum,velec); |
612 | |
613 | fscal = felec; |
614 | |
615 | /* Calculate temporary vectorial force */ |
616 | tx = _mm_mul_ps(fscal,dx32); |
617 | ty = _mm_mul_ps(fscal,dy32); |
618 | tz = _mm_mul_ps(fscal,dz32); |
619 | |
620 | /* Update vectorial force */ |
621 | fix3 = _mm_add_ps(fix3,tx); |
622 | fiy3 = _mm_add_ps(fiy3,ty); |
623 | fiz3 = _mm_add_ps(fiz3,tz); |
624 | |
625 | fjx2 = _mm_add_ps(fjx2,tx); |
626 | fjy2 = _mm_add_ps(fjy2,ty); |
627 | fjz2 = _mm_add_ps(fjz2,tz); |
628 | |
629 | /************************** |
630 | * CALCULATE INTERACTIONS * |
631 | **************************/ |
632 | |
633 | r33 = _mm_mul_ps(rsq33,rinv33); |
634 | |
635 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
636 | rt = _mm_mul_ps(r33,vftabscale); |
637 | vfitab = _mm_cvttps_epi32(rt); |
638 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
639 | vfitab = _mm_slli_epi32(vfitab,2); |
640 | |
641 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
642 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
643 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
644 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
645 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
646 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
647 | Heps = _mm_mul_ps(vfeps,H); |
648 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
649 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
650 | velec = _mm_mul_ps(qq33,VV); |
651 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
652 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
653 | |
654 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
655 | velecsum = _mm_add_ps(velecsum,velec); |
656 | |
657 | fscal = felec; |
658 | |
659 | /* Calculate temporary vectorial force */ |
660 | tx = _mm_mul_ps(fscal,dx33); |
661 | ty = _mm_mul_ps(fscal,dy33); |
662 | tz = _mm_mul_ps(fscal,dz33); |
663 | |
664 | /* Update vectorial force */ |
665 | fix3 = _mm_add_ps(fix3,tx); |
666 | fiy3 = _mm_add_ps(fiy3,ty); |
667 | fiz3 = _mm_add_ps(fiz3,tz); |
668 | |
669 | fjx3 = _mm_add_ps(fjx3,tx); |
670 | fjy3 = _mm_add_ps(fjy3,ty); |
671 | fjz3 = _mm_add_ps(fjz3,tz); |
672 | |
673 | fjptrA = f+j_coord_offsetA; |
674 | fjptrB = f+j_coord_offsetB; |
675 | fjptrC = f+j_coord_offsetC; |
676 | fjptrD = f+j_coord_offsetD; |
677 | |
678 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA+DIM3,fjptrB+DIM3,fjptrC+DIM3,fjptrD+DIM3, |
679 | fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
680 | |
681 | /* Inner loop uses 387 flops */ |
682 | } |
683 | |
684 | if(jidx<j_index_end) |
685 | { |
686 | |
687 | /* Get j neighbor index, and coordinate index */ |
688 | jnrlistA = jjnr[jidx]; |
689 | jnrlistB = jjnr[jidx+1]; |
690 | jnrlistC = jjnr[jidx+2]; |
691 | jnrlistD = jjnr[jidx+3]; |
692 | /* Sign of each element will be negative for non-real atoms. |
693 | * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones, |
694 | * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries. |
695 | */ |
696 | dummy_mask = gmx_mm_castsi128_ps_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())); |
697 | jnrA = (jnrlistA>=0) ? jnrlistA : 0; |
698 | jnrB = (jnrlistB>=0) ? jnrlistB : 0; |
699 | jnrC = (jnrlistC>=0) ? jnrlistC : 0; |
700 | jnrD = (jnrlistD>=0) ? jnrlistD : 0; |
701 | j_coord_offsetA = DIM3*jnrA; |
702 | j_coord_offsetB = DIM3*jnrB; |
703 | j_coord_offsetC = DIM3*jnrC; |
704 | j_coord_offsetD = DIM3*jnrD; |
705 | |
706 | /* load j atom coordinates */ |
707 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM3,x+j_coord_offsetB+DIM3, |
708 | x+j_coord_offsetC+DIM3,x+j_coord_offsetD+DIM3, |
709 | &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3); |
710 | |
711 | /* Calculate displacement vector */ |
712 | dx11 = _mm_sub_ps(ix1,jx1); |
713 | dy11 = _mm_sub_ps(iy1,jy1); |
714 | dz11 = _mm_sub_ps(iz1,jz1); |
715 | dx12 = _mm_sub_ps(ix1,jx2); |
716 | dy12 = _mm_sub_ps(iy1,jy2); |
717 | dz12 = _mm_sub_ps(iz1,jz2); |
718 | dx13 = _mm_sub_ps(ix1,jx3); |
719 | dy13 = _mm_sub_ps(iy1,jy3); |
720 | dz13 = _mm_sub_ps(iz1,jz3); |
721 | dx21 = _mm_sub_ps(ix2,jx1); |
722 | dy21 = _mm_sub_ps(iy2,jy1); |
723 | dz21 = _mm_sub_ps(iz2,jz1); |
724 | dx22 = _mm_sub_ps(ix2,jx2); |
725 | dy22 = _mm_sub_ps(iy2,jy2); |
726 | dz22 = _mm_sub_ps(iz2,jz2); |
727 | dx23 = _mm_sub_ps(ix2,jx3); |
728 | dy23 = _mm_sub_ps(iy2,jy3); |
729 | dz23 = _mm_sub_ps(iz2,jz3); |
730 | dx31 = _mm_sub_ps(ix3,jx1); |
731 | dy31 = _mm_sub_ps(iy3,jy1); |
732 | dz31 = _mm_sub_ps(iz3,jz1); |
733 | dx32 = _mm_sub_ps(ix3,jx2); |
734 | dy32 = _mm_sub_ps(iy3,jy2); |
735 | dz32 = _mm_sub_ps(iz3,jz2); |
736 | dx33 = _mm_sub_ps(ix3,jx3); |
737 | dy33 = _mm_sub_ps(iy3,jy3); |
738 | dz33 = _mm_sub_ps(iz3,jz3); |
739 | |
740 | /* Calculate squared distance and things based on it */ |
741 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
742 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
743 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
744 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
745 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
746 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
747 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
748 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
749 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
750 | |
751 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
752 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
753 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
754 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
755 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
756 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
757 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
758 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
759 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
760 | |
761 | fjx1 = _mm_setzero_ps(); |
762 | fjy1 = _mm_setzero_ps(); |
763 | fjz1 = _mm_setzero_ps(); |
764 | fjx2 = _mm_setzero_ps(); |
765 | fjy2 = _mm_setzero_ps(); |
766 | fjz2 = _mm_setzero_ps(); |
767 | fjx3 = _mm_setzero_ps(); |
768 | fjy3 = _mm_setzero_ps(); |
769 | fjz3 = _mm_setzero_ps(); |
770 | |
771 | /************************** |
772 | * CALCULATE INTERACTIONS * |
773 | **************************/ |
774 | |
775 | r11 = _mm_mul_ps(rsq11,rinv11); |
776 | r11 = _mm_andnot_ps(dummy_mask,r11); |
777 | |
778 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
779 | rt = _mm_mul_ps(r11,vftabscale); |
780 | vfitab = _mm_cvttps_epi32(rt); |
781 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
782 | vfitab = _mm_slli_epi32(vfitab,2); |
783 | |
784 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
785 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
786 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
787 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
788 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
789 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
790 | Heps = _mm_mul_ps(vfeps,H); |
791 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
792 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
793 | velec = _mm_mul_ps(qq11,VV); |
794 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
795 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
796 | |
797 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
798 | velec = _mm_andnot_ps(dummy_mask,velec); |
799 | velecsum = _mm_add_ps(velecsum,velec); |
800 | |
801 | fscal = felec; |
802 | |
803 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
804 | |
805 | /* Calculate temporary vectorial force */ |
806 | tx = _mm_mul_ps(fscal,dx11); |
807 | ty = _mm_mul_ps(fscal,dy11); |
808 | tz = _mm_mul_ps(fscal,dz11); |
809 | |
810 | /* Update vectorial force */ |
811 | fix1 = _mm_add_ps(fix1,tx); |
812 | fiy1 = _mm_add_ps(fiy1,ty); |
813 | fiz1 = _mm_add_ps(fiz1,tz); |
814 | |
815 | fjx1 = _mm_add_ps(fjx1,tx); |
816 | fjy1 = _mm_add_ps(fjy1,ty); |
817 | fjz1 = _mm_add_ps(fjz1,tz); |
818 | |
819 | /************************** |
820 | * CALCULATE INTERACTIONS * |
821 | **************************/ |
822 | |
823 | r12 = _mm_mul_ps(rsq12,rinv12); |
824 | r12 = _mm_andnot_ps(dummy_mask,r12); |
825 | |
826 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
827 | rt = _mm_mul_ps(r12,vftabscale); |
828 | vfitab = _mm_cvttps_epi32(rt); |
829 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
830 | vfitab = _mm_slli_epi32(vfitab,2); |
831 | |
832 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
833 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
834 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
835 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
836 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
837 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
838 | Heps = _mm_mul_ps(vfeps,H); |
839 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
840 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
841 | velec = _mm_mul_ps(qq12,VV); |
842 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
843 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
844 | |
845 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
846 | velec = _mm_andnot_ps(dummy_mask,velec); |
847 | velecsum = _mm_add_ps(velecsum,velec); |
848 | |
849 | fscal = felec; |
850 | |
851 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
852 | |
853 | /* Calculate temporary vectorial force */ |
854 | tx = _mm_mul_ps(fscal,dx12); |
855 | ty = _mm_mul_ps(fscal,dy12); |
856 | tz = _mm_mul_ps(fscal,dz12); |
857 | |
858 | /* Update vectorial force */ |
859 | fix1 = _mm_add_ps(fix1,tx); |
860 | fiy1 = _mm_add_ps(fiy1,ty); |
861 | fiz1 = _mm_add_ps(fiz1,tz); |
862 | |
863 | fjx2 = _mm_add_ps(fjx2,tx); |
864 | fjy2 = _mm_add_ps(fjy2,ty); |
865 | fjz2 = _mm_add_ps(fjz2,tz); |
866 | |
867 | /************************** |
868 | * CALCULATE INTERACTIONS * |
869 | **************************/ |
870 | |
871 | r13 = _mm_mul_ps(rsq13,rinv13); |
872 | r13 = _mm_andnot_ps(dummy_mask,r13); |
873 | |
874 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
875 | rt = _mm_mul_ps(r13,vftabscale); |
876 | vfitab = _mm_cvttps_epi32(rt); |
877 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
878 | vfitab = _mm_slli_epi32(vfitab,2); |
879 | |
880 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
881 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
882 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
883 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
884 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
885 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
886 | Heps = _mm_mul_ps(vfeps,H); |
887 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
888 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
889 | velec = _mm_mul_ps(qq13,VV); |
890 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
891 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
892 | |
893 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
894 | velec = _mm_andnot_ps(dummy_mask,velec); |
895 | velecsum = _mm_add_ps(velecsum,velec); |
896 | |
897 | fscal = felec; |
898 | |
899 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
900 | |
901 | /* Calculate temporary vectorial force */ |
902 | tx = _mm_mul_ps(fscal,dx13); |
903 | ty = _mm_mul_ps(fscal,dy13); |
904 | tz = _mm_mul_ps(fscal,dz13); |
905 | |
906 | /* Update vectorial force */ |
907 | fix1 = _mm_add_ps(fix1,tx); |
908 | fiy1 = _mm_add_ps(fiy1,ty); |
909 | fiz1 = _mm_add_ps(fiz1,tz); |
910 | |
911 | fjx3 = _mm_add_ps(fjx3,tx); |
912 | fjy3 = _mm_add_ps(fjy3,ty); |
913 | fjz3 = _mm_add_ps(fjz3,tz); |
914 | |
915 | /************************** |
916 | * CALCULATE INTERACTIONS * |
917 | **************************/ |
918 | |
919 | r21 = _mm_mul_ps(rsq21,rinv21); |
920 | r21 = _mm_andnot_ps(dummy_mask,r21); |
921 | |
922 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
923 | rt = _mm_mul_ps(r21,vftabscale); |
924 | vfitab = _mm_cvttps_epi32(rt); |
925 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
926 | vfitab = _mm_slli_epi32(vfitab,2); |
927 | |
928 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
929 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
930 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
931 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
932 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
933 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
934 | Heps = _mm_mul_ps(vfeps,H); |
935 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
936 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
937 | velec = _mm_mul_ps(qq21,VV); |
938 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
939 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
940 | |
941 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
942 | velec = _mm_andnot_ps(dummy_mask,velec); |
943 | velecsum = _mm_add_ps(velecsum,velec); |
944 | |
945 | fscal = felec; |
946 | |
947 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
948 | |
949 | /* Calculate temporary vectorial force */ |
950 | tx = _mm_mul_ps(fscal,dx21); |
951 | ty = _mm_mul_ps(fscal,dy21); |
952 | tz = _mm_mul_ps(fscal,dz21); |
953 | |
954 | /* Update vectorial force */ |
955 | fix2 = _mm_add_ps(fix2,tx); |
956 | fiy2 = _mm_add_ps(fiy2,ty); |
957 | fiz2 = _mm_add_ps(fiz2,tz); |
958 | |
959 | fjx1 = _mm_add_ps(fjx1,tx); |
960 | fjy1 = _mm_add_ps(fjy1,ty); |
961 | fjz1 = _mm_add_ps(fjz1,tz); |
962 | |
963 | /************************** |
964 | * CALCULATE INTERACTIONS * |
965 | **************************/ |
966 | |
967 | r22 = _mm_mul_ps(rsq22,rinv22); |
968 | r22 = _mm_andnot_ps(dummy_mask,r22); |
969 | |
970 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
971 | rt = _mm_mul_ps(r22,vftabscale); |
972 | vfitab = _mm_cvttps_epi32(rt); |
973 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
974 | vfitab = _mm_slli_epi32(vfitab,2); |
975 | |
976 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
977 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
978 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
979 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
980 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
981 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
982 | Heps = _mm_mul_ps(vfeps,H); |
983 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
984 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
985 | velec = _mm_mul_ps(qq22,VV); |
986 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
987 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
988 | |
989 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
990 | velec = _mm_andnot_ps(dummy_mask,velec); |
991 | velecsum = _mm_add_ps(velecsum,velec); |
992 | |
993 | fscal = felec; |
994 | |
995 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
996 | |
997 | /* Calculate temporary vectorial force */ |
998 | tx = _mm_mul_ps(fscal,dx22); |
999 | ty = _mm_mul_ps(fscal,dy22); |
1000 | tz = _mm_mul_ps(fscal,dz22); |
1001 | |
1002 | /* Update vectorial force */ |
1003 | fix2 = _mm_add_ps(fix2,tx); |
1004 | fiy2 = _mm_add_ps(fiy2,ty); |
1005 | fiz2 = _mm_add_ps(fiz2,tz); |
1006 | |
1007 | fjx2 = _mm_add_ps(fjx2,tx); |
1008 | fjy2 = _mm_add_ps(fjy2,ty); |
1009 | fjz2 = _mm_add_ps(fjz2,tz); |
1010 | |
1011 | /************************** |
1012 | * CALCULATE INTERACTIONS * |
1013 | **************************/ |
1014 | |
1015 | r23 = _mm_mul_ps(rsq23,rinv23); |
1016 | r23 = _mm_andnot_ps(dummy_mask,r23); |
1017 | |
1018 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1019 | rt = _mm_mul_ps(r23,vftabscale); |
1020 | vfitab = _mm_cvttps_epi32(rt); |
1021 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1022 | vfitab = _mm_slli_epi32(vfitab,2); |
1023 | |
1024 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1025 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1026 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1027 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1028 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1029 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1030 | Heps = _mm_mul_ps(vfeps,H); |
1031 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1032 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
1033 | velec = _mm_mul_ps(qq23,VV); |
1034 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1035 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
1036 | |
1037 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1038 | velec = _mm_andnot_ps(dummy_mask,velec); |
1039 | velecsum = _mm_add_ps(velecsum,velec); |
1040 | |
1041 | fscal = felec; |
1042 | |
1043 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1044 | |
1045 | /* Calculate temporary vectorial force */ |
1046 | tx = _mm_mul_ps(fscal,dx23); |
1047 | ty = _mm_mul_ps(fscal,dy23); |
1048 | tz = _mm_mul_ps(fscal,dz23); |
1049 | |
1050 | /* Update vectorial force */ |
1051 | fix2 = _mm_add_ps(fix2,tx); |
1052 | fiy2 = _mm_add_ps(fiy2,ty); |
1053 | fiz2 = _mm_add_ps(fiz2,tz); |
1054 | |
1055 | fjx3 = _mm_add_ps(fjx3,tx); |
1056 | fjy3 = _mm_add_ps(fjy3,ty); |
1057 | fjz3 = _mm_add_ps(fjz3,tz); |
1058 | |
1059 | /************************** |
1060 | * CALCULATE INTERACTIONS * |
1061 | **************************/ |
1062 | |
1063 | r31 = _mm_mul_ps(rsq31,rinv31); |
1064 | r31 = _mm_andnot_ps(dummy_mask,r31); |
1065 | |
1066 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1067 | rt = _mm_mul_ps(r31,vftabscale); |
1068 | vfitab = _mm_cvttps_epi32(rt); |
1069 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1070 | vfitab = _mm_slli_epi32(vfitab,2); |
1071 | |
1072 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1073 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1074 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1075 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1076 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1077 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1078 | Heps = _mm_mul_ps(vfeps,H); |
1079 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1080 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
1081 | velec = _mm_mul_ps(qq31,VV); |
1082 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1083 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
1084 | |
1085 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1086 | velec = _mm_andnot_ps(dummy_mask,velec); |
1087 | velecsum = _mm_add_ps(velecsum,velec); |
1088 | |
1089 | fscal = felec; |
1090 | |
1091 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1092 | |
1093 | /* Calculate temporary vectorial force */ |
1094 | tx = _mm_mul_ps(fscal,dx31); |
1095 | ty = _mm_mul_ps(fscal,dy31); |
1096 | tz = _mm_mul_ps(fscal,dz31); |
1097 | |
1098 | /* Update vectorial force */ |
1099 | fix3 = _mm_add_ps(fix3,tx); |
1100 | fiy3 = _mm_add_ps(fiy3,ty); |
1101 | fiz3 = _mm_add_ps(fiz3,tz); |
1102 | |
1103 | fjx1 = _mm_add_ps(fjx1,tx); |
1104 | fjy1 = _mm_add_ps(fjy1,ty); |
1105 | fjz1 = _mm_add_ps(fjz1,tz); |
1106 | |
1107 | /************************** |
1108 | * CALCULATE INTERACTIONS * |
1109 | **************************/ |
1110 | |
1111 | r32 = _mm_mul_ps(rsq32,rinv32); |
1112 | r32 = _mm_andnot_ps(dummy_mask,r32); |
1113 | |
1114 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1115 | rt = _mm_mul_ps(r32,vftabscale); |
1116 | vfitab = _mm_cvttps_epi32(rt); |
1117 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1118 | vfitab = _mm_slli_epi32(vfitab,2); |
1119 | |
1120 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1121 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1122 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1123 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1124 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1125 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1126 | Heps = _mm_mul_ps(vfeps,H); |
1127 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1128 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
1129 | velec = _mm_mul_ps(qq32,VV); |
1130 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1131 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
1132 | |
1133 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1134 | velec = _mm_andnot_ps(dummy_mask,velec); |
1135 | velecsum = _mm_add_ps(velecsum,velec); |
1136 | |
1137 | fscal = felec; |
1138 | |
1139 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1140 | |
1141 | /* Calculate temporary vectorial force */ |
1142 | tx = _mm_mul_ps(fscal,dx32); |
1143 | ty = _mm_mul_ps(fscal,dy32); |
1144 | tz = _mm_mul_ps(fscal,dz32); |
1145 | |
1146 | /* Update vectorial force */ |
1147 | fix3 = _mm_add_ps(fix3,tx); |
1148 | fiy3 = _mm_add_ps(fiy3,ty); |
1149 | fiz3 = _mm_add_ps(fiz3,tz); |
1150 | |
1151 | fjx2 = _mm_add_ps(fjx2,tx); |
1152 | fjy2 = _mm_add_ps(fjy2,ty); |
1153 | fjz2 = _mm_add_ps(fjz2,tz); |
1154 | |
1155 | /************************** |
1156 | * CALCULATE INTERACTIONS * |
1157 | **************************/ |
1158 | |
1159 | r33 = _mm_mul_ps(rsq33,rinv33); |
1160 | r33 = _mm_andnot_ps(dummy_mask,r33); |
1161 | |
1162 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1163 | rt = _mm_mul_ps(r33,vftabscale); |
1164 | vfitab = _mm_cvttps_epi32(rt); |
1165 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1166 | vfitab = _mm_slli_epi32(vfitab,2); |
1167 | |
1168 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1169 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1170 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1171 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1172 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1173 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1174 | Heps = _mm_mul_ps(vfeps,H); |
1175 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1176 | VV = _mm_add_ps(Y,_mm_mul_ps(vfeps,Fp)); |
1177 | velec = _mm_mul_ps(qq33,VV); |
1178 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1179 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
1180 | |
1181 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1182 | velec = _mm_andnot_ps(dummy_mask,velec); |
1183 | velecsum = _mm_add_ps(velecsum,velec); |
1184 | |
1185 | fscal = felec; |
1186 | |
1187 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1188 | |
1189 | /* Calculate temporary vectorial force */ |
1190 | tx = _mm_mul_ps(fscal,dx33); |
1191 | ty = _mm_mul_ps(fscal,dy33); |
1192 | tz = _mm_mul_ps(fscal,dz33); |
1193 | |
1194 | /* Update vectorial force */ |
1195 | fix3 = _mm_add_ps(fix3,tx); |
1196 | fiy3 = _mm_add_ps(fiy3,ty); |
1197 | fiz3 = _mm_add_ps(fiz3,tz); |
1198 | |
1199 | fjx3 = _mm_add_ps(fjx3,tx); |
1200 | fjy3 = _mm_add_ps(fjy3,ty); |
1201 | fjz3 = _mm_add_ps(fjz3,tz); |
1202 | |
1203 | fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch; |
1204 | fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch; |
1205 | fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch; |
1206 | fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch; |
1207 | |
1208 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA+DIM3,fjptrB+DIM3,fjptrC+DIM3,fjptrD+DIM3, |
1209 | fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
1210 | |
1211 | /* Inner loop uses 396 flops */ |
1212 | } |
1213 | |
1214 | /* End of innermost loop */ |
1215 | |
1216 | gmx_mm_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3, |
1217 | f+i_coord_offset+DIM3,fshift+i_shift_offset); |
1218 | |
1219 | ggid = gid[iidx]; |
1220 | /* Update potential energies */ |
1221 | gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid); |
1222 | |
1223 | /* Increment number of inner iterations */ |
1224 | inneriter += j_index_end - j_index_start; |
1225 | |
1226 | /* Outer loop uses 19 flops */ |
1227 | } |
1228 | |
1229 | /* Increment number of outer iterations */ |
1230 | outeriter += nri; |
1231 | |
1232 | /* Update outer/inner flops */ |
1233 | |
1234 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_VF,outeriter*19 + inneriter*396)(nrnb)->n[eNR_NBKERNEL_ELEC_W4W4_VF] += outeriter*19 + inneriter *396; |
1235 | } |
1236 | /* |
1237 | * Gromacs nonbonded kernel: nb_kernel_ElecCSTab_VdwNone_GeomW4W4_F_sse4_1_single |
1238 | * Electrostatics interaction: CubicSplineTable |
1239 | * VdW interaction: None |
1240 | * Geometry: Water4-Water4 |
1241 | * Calculate force/pot: Force |
1242 | */ |
1243 | void |
1244 | nb_kernel_ElecCSTab_VdwNone_GeomW4W4_F_sse4_1_single |
1245 | (t_nblist * gmx_restrict nlist, |
1246 | rvec * gmx_restrict xx, |
1247 | rvec * gmx_restrict ff, |
1248 | t_forcerec * gmx_restrict fr, |
1249 | t_mdatoms * gmx_restrict mdatoms, |
1250 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict kernel_data, |
1251 | t_nrnb * gmx_restrict nrnb) |
1252 | { |
1253 | /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or |
1254 | * just 0 for non-waters. |
1255 | * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different |
1256 | * jnr indices corresponding to data put in the four positions in the SIMD register. |
1257 | */ |
1258 | int i_shift_offset,i_coord_offset,outeriter,inneriter; |
1259 | int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx; |
1260 | int jnrA,jnrB,jnrC,jnrD; |
1261 | int jnrlistA,jnrlistB,jnrlistC,jnrlistD; |
1262 | int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD; |
1263 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
1264 | real rcutoff_scalar; |
1265 | real *shiftvec,*fshift,*x,*f; |
1266 | real *fjptrA,*fjptrB,*fjptrC,*fjptrD; |
1267 | real scratch[4*DIM3]; |
1268 | __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall; |
1269 | int vdwioffset1; |
1270 | __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
1271 | int vdwioffset2; |
1272 | __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
1273 | int vdwioffset3; |
1274 | __m128 ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3; |
1275 | int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D; |
1276 | __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
1277 | int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D; |
1278 | __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
1279 | int vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D; |
1280 | __m128 jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3; |
1281 | __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11; |
1282 | __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12; |
1283 | __m128 dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13; |
1284 | __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21; |
1285 | __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22; |
1286 | __m128 dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23; |
1287 | __m128 dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31; |
1288 | __m128 dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32; |
1289 | __m128 dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33; |
1290 | __m128 velec,felec,velecsum,facel,crf,krf,krf2; |
1291 | real *charge; |
1292 | __m128i vfitab; |
1293 | __m128i ifour = _mm_set1_epi32(4); |
1294 | __m128 rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF; |
1295 | real *vftab; |
1296 | __m128 dummy_mask,cutoff_mask; |
1297 | __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) ); |
1298 | __m128 one = _mm_set1_ps(1.0); |
1299 | __m128 two = _mm_set1_ps(2.0); |
1300 | x = xx[0]; |
1301 | f = ff[0]; |
1302 | |
1303 | nri = nlist->nri; |
1304 | iinr = nlist->iinr; |
1305 | jindex = nlist->jindex; |
1306 | jjnr = nlist->jjnr; |
1307 | shiftidx = nlist->shift; |
1308 | gid = nlist->gid; |
Value stored to 'gid' is never read | |
1309 | shiftvec = fr->shift_vec[0]; |
1310 | fshift = fr->fshift[0]; |
1311 | facel = _mm_set1_ps(fr->epsfac); |
1312 | charge = mdatoms->chargeA; |
1313 | |
1314 | vftab = kernel_data->table_elec->data; |
1315 | vftabscale = _mm_set1_ps(kernel_data->table_elec->scale); |
1316 | |
1317 | /* Setup water-specific parameters */ |
1318 | inr = nlist->iinr[0]; |
1319 | iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1])); |
1320 | iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2])); |
1321 | iq3 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3])); |
1322 | |
1323 | jq1 = _mm_set1_ps(charge[inr+1]); |
1324 | jq2 = _mm_set1_ps(charge[inr+2]); |
1325 | jq3 = _mm_set1_ps(charge[inr+3]); |
1326 | qq11 = _mm_mul_ps(iq1,jq1); |
1327 | qq12 = _mm_mul_ps(iq1,jq2); |
1328 | qq13 = _mm_mul_ps(iq1,jq3); |
1329 | qq21 = _mm_mul_ps(iq2,jq1); |
1330 | qq22 = _mm_mul_ps(iq2,jq2); |
1331 | qq23 = _mm_mul_ps(iq2,jq3); |
1332 | qq31 = _mm_mul_ps(iq3,jq1); |
1333 | qq32 = _mm_mul_ps(iq3,jq2); |
1334 | qq33 = _mm_mul_ps(iq3,jq3); |
1335 | |
1336 | /* Avoid stupid compiler warnings */ |
1337 | jnrA = jnrB = jnrC = jnrD = 0; |
1338 | j_coord_offsetA = 0; |
1339 | j_coord_offsetB = 0; |
1340 | j_coord_offsetC = 0; |
1341 | j_coord_offsetD = 0; |
1342 | |
1343 | outeriter = 0; |
1344 | inneriter = 0; |
1345 | |
1346 | for(iidx=0;iidx<4*DIM3;iidx++) |
1347 | { |
1348 | scratch[iidx] = 0.0; |
1349 | } |
1350 | |
1351 | /* Start outer loop over neighborlists */ |
1352 | for(iidx=0; iidx<nri; iidx++) |
1353 | { |
1354 | /* Load shift vector for this list */ |
1355 | i_shift_offset = DIM3*shiftidx[iidx]; |
1356 | |
1357 | /* Load limits for loop over neighbors */ |
1358 | j_index_start = jindex[iidx]; |
1359 | j_index_end = jindex[iidx+1]; |
1360 | |
1361 | /* Get outer coordinate index */ |
1362 | inr = iinr[iidx]; |
1363 | i_coord_offset = DIM3*inr; |
1364 | |
1365 | /* Load i particle coords and add shift vector */ |
1366 | gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM3, |
1367 | &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3); |
1368 | |
1369 | fix1 = _mm_setzero_ps(); |
1370 | fiy1 = _mm_setzero_ps(); |
1371 | fiz1 = _mm_setzero_ps(); |
1372 | fix2 = _mm_setzero_ps(); |
1373 | fiy2 = _mm_setzero_ps(); |
1374 | fiz2 = _mm_setzero_ps(); |
1375 | fix3 = _mm_setzero_ps(); |
1376 | fiy3 = _mm_setzero_ps(); |
1377 | fiz3 = _mm_setzero_ps(); |
1378 | |
1379 | /* Start inner kernel loop */ |
1380 | for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4) |
1381 | { |
1382 | |
1383 | /* Get j neighbor index, and coordinate index */ |
1384 | jnrA = jjnr[jidx]; |
1385 | jnrB = jjnr[jidx+1]; |
1386 | jnrC = jjnr[jidx+2]; |
1387 | jnrD = jjnr[jidx+3]; |
1388 | j_coord_offsetA = DIM3*jnrA; |
1389 | j_coord_offsetB = DIM3*jnrB; |
1390 | j_coord_offsetC = DIM3*jnrC; |
1391 | j_coord_offsetD = DIM3*jnrD; |
1392 | |
1393 | /* load j atom coordinates */ |
1394 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM3,x+j_coord_offsetB+DIM3, |
1395 | x+j_coord_offsetC+DIM3,x+j_coord_offsetD+DIM3, |
1396 | &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3); |
1397 | |
1398 | /* Calculate displacement vector */ |
1399 | dx11 = _mm_sub_ps(ix1,jx1); |
1400 | dy11 = _mm_sub_ps(iy1,jy1); |
1401 | dz11 = _mm_sub_ps(iz1,jz1); |
1402 | dx12 = _mm_sub_ps(ix1,jx2); |
1403 | dy12 = _mm_sub_ps(iy1,jy2); |
1404 | dz12 = _mm_sub_ps(iz1,jz2); |
1405 | dx13 = _mm_sub_ps(ix1,jx3); |
1406 | dy13 = _mm_sub_ps(iy1,jy3); |
1407 | dz13 = _mm_sub_ps(iz1,jz3); |
1408 | dx21 = _mm_sub_ps(ix2,jx1); |
1409 | dy21 = _mm_sub_ps(iy2,jy1); |
1410 | dz21 = _mm_sub_ps(iz2,jz1); |
1411 | dx22 = _mm_sub_ps(ix2,jx2); |
1412 | dy22 = _mm_sub_ps(iy2,jy2); |
1413 | dz22 = _mm_sub_ps(iz2,jz2); |
1414 | dx23 = _mm_sub_ps(ix2,jx3); |
1415 | dy23 = _mm_sub_ps(iy2,jy3); |
1416 | dz23 = _mm_sub_ps(iz2,jz3); |
1417 | dx31 = _mm_sub_ps(ix3,jx1); |
1418 | dy31 = _mm_sub_ps(iy3,jy1); |
1419 | dz31 = _mm_sub_ps(iz3,jz1); |
1420 | dx32 = _mm_sub_ps(ix3,jx2); |
1421 | dy32 = _mm_sub_ps(iy3,jy2); |
1422 | dz32 = _mm_sub_ps(iz3,jz2); |
1423 | dx33 = _mm_sub_ps(ix3,jx3); |
1424 | dy33 = _mm_sub_ps(iy3,jy3); |
1425 | dz33 = _mm_sub_ps(iz3,jz3); |
1426 | |
1427 | /* Calculate squared distance and things based on it */ |
1428 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
1429 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
1430 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
1431 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
1432 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
1433 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
1434 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
1435 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
1436 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
1437 | |
1438 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
1439 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
1440 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
1441 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
1442 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
1443 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
1444 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
1445 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
1446 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
1447 | |
1448 | fjx1 = _mm_setzero_ps(); |
1449 | fjy1 = _mm_setzero_ps(); |
1450 | fjz1 = _mm_setzero_ps(); |
1451 | fjx2 = _mm_setzero_ps(); |
1452 | fjy2 = _mm_setzero_ps(); |
1453 | fjz2 = _mm_setzero_ps(); |
1454 | fjx3 = _mm_setzero_ps(); |
1455 | fjy3 = _mm_setzero_ps(); |
1456 | fjz3 = _mm_setzero_ps(); |
1457 | |
1458 | /************************** |
1459 | * CALCULATE INTERACTIONS * |
1460 | **************************/ |
1461 | |
1462 | r11 = _mm_mul_ps(rsq11,rinv11); |
1463 | |
1464 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1465 | rt = _mm_mul_ps(r11,vftabscale); |
1466 | vfitab = _mm_cvttps_epi32(rt); |
1467 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1468 | vfitab = _mm_slli_epi32(vfitab,2); |
1469 | |
1470 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1471 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1472 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1473 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1474 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1475 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1476 | Heps = _mm_mul_ps(vfeps,H); |
1477 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1478 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1479 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
1480 | |
1481 | fscal = felec; |
1482 | |
1483 | /* Calculate temporary vectorial force */ |
1484 | tx = _mm_mul_ps(fscal,dx11); |
1485 | ty = _mm_mul_ps(fscal,dy11); |
1486 | tz = _mm_mul_ps(fscal,dz11); |
1487 | |
1488 | /* Update vectorial force */ |
1489 | fix1 = _mm_add_ps(fix1,tx); |
1490 | fiy1 = _mm_add_ps(fiy1,ty); |
1491 | fiz1 = _mm_add_ps(fiz1,tz); |
1492 | |
1493 | fjx1 = _mm_add_ps(fjx1,tx); |
1494 | fjy1 = _mm_add_ps(fjy1,ty); |
1495 | fjz1 = _mm_add_ps(fjz1,tz); |
1496 | |
1497 | /************************** |
1498 | * CALCULATE INTERACTIONS * |
1499 | **************************/ |
1500 | |
1501 | r12 = _mm_mul_ps(rsq12,rinv12); |
1502 | |
1503 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1504 | rt = _mm_mul_ps(r12,vftabscale); |
1505 | vfitab = _mm_cvttps_epi32(rt); |
1506 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1507 | vfitab = _mm_slli_epi32(vfitab,2); |
1508 | |
1509 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1510 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1511 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1512 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1513 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1514 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1515 | Heps = _mm_mul_ps(vfeps,H); |
1516 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1517 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1518 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
1519 | |
1520 | fscal = felec; |
1521 | |
1522 | /* Calculate temporary vectorial force */ |
1523 | tx = _mm_mul_ps(fscal,dx12); |
1524 | ty = _mm_mul_ps(fscal,dy12); |
1525 | tz = _mm_mul_ps(fscal,dz12); |
1526 | |
1527 | /* Update vectorial force */ |
1528 | fix1 = _mm_add_ps(fix1,tx); |
1529 | fiy1 = _mm_add_ps(fiy1,ty); |
1530 | fiz1 = _mm_add_ps(fiz1,tz); |
1531 | |
1532 | fjx2 = _mm_add_ps(fjx2,tx); |
1533 | fjy2 = _mm_add_ps(fjy2,ty); |
1534 | fjz2 = _mm_add_ps(fjz2,tz); |
1535 | |
1536 | /************************** |
1537 | * CALCULATE INTERACTIONS * |
1538 | **************************/ |
1539 | |
1540 | r13 = _mm_mul_ps(rsq13,rinv13); |
1541 | |
1542 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1543 | rt = _mm_mul_ps(r13,vftabscale); |
1544 | vfitab = _mm_cvttps_epi32(rt); |
1545 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1546 | vfitab = _mm_slli_epi32(vfitab,2); |
1547 | |
1548 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1549 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1550 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1551 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1552 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1553 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1554 | Heps = _mm_mul_ps(vfeps,H); |
1555 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1556 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1557 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
1558 | |
1559 | fscal = felec; |
1560 | |
1561 | /* Calculate temporary vectorial force */ |
1562 | tx = _mm_mul_ps(fscal,dx13); |
1563 | ty = _mm_mul_ps(fscal,dy13); |
1564 | tz = _mm_mul_ps(fscal,dz13); |
1565 | |
1566 | /* Update vectorial force */ |
1567 | fix1 = _mm_add_ps(fix1,tx); |
1568 | fiy1 = _mm_add_ps(fiy1,ty); |
1569 | fiz1 = _mm_add_ps(fiz1,tz); |
1570 | |
1571 | fjx3 = _mm_add_ps(fjx3,tx); |
1572 | fjy3 = _mm_add_ps(fjy3,ty); |
1573 | fjz3 = _mm_add_ps(fjz3,tz); |
1574 | |
1575 | /************************** |
1576 | * CALCULATE INTERACTIONS * |
1577 | **************************/ |
1578 | |
1579 | r21 = _mm_mul_ps(rsq21,rinv21); |
1580 | |
1581 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1582 | rt = _mm_mul_ps(r21,vftabscale); |
1583 | vfitab = _mm_cvttps_epi32(rt); |
1584 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1585 | vfitab = _mm_slli_epi32(vfitab,2); |
1586 | |
1587 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1588 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1589 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1590 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1591 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1592 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1593 | Heps = _mm_mul_ps(vfeps,H); |
1594 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1595 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1596 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
1597 | |
1598 | fscal = felec; |
1599 | |
1600 | /* Calculate temporary vectorial force */ |
1601 | tx = _mm_mul_ps(fscal,dx21); |
1602 | ty = _mm_mul_ps(fscal,dy21); |
1603 | tz = _mm_mul_ps(fscal,dz21); |
1604 | |
1605 | /* Update vectorial force */ |
1606 | fix2 = _mm_add_ps(fix2,tx); |
1607 | fiy2 = _mm_add_ps(fiy2,ty); |
1608 | fiz2 = _mm_add_ps(fiz2,tz); |
1609 | |
1610 | fjx1 = _mm_add_ps(fjx1,tx); |
1611 | fjy1 = _mm_add_ps(fjy1,ty); |
1612 | fjz1 = _mm_add_ps(fjz1,tz); |
1613 | |
1614 | /************************** |
1615 | * CALCULATE INTERACTIONS * |
1616 | **************************/ |
1617 | |
1618 | r22 = _mm_mul_ps(rsq22,rinv22); |
1619 | |
1620 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1621 | rt = _mm_mul_ps(r22,vftabscale); |
1622 | vfitab = _mm_cvttps_epi32(rt); |
1623 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1624 | vfitab = _mm_slli_epi32(vfitab,2); |
1625 | |
1626 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1627 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1628 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1629 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1630 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1631 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1632 | Heps = _mm_mul_ps(vfeps,H); |
1633 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1634 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1635 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
1636 | |
1637 | fscal = felec; |
1638 | |
1639 | /* Calculate temporary vectorial force */ |
1640 | tx = _mm_mul_ps(fscal,dx22); |
1641 | ty = _mm_mul_ps(fscal,dy22); |
1642 | tz = _mm_mul_ps(fscal,dz22); |
1643 | |
1644 | /* Update vectorial force */ |
1645 | fix2 = _mm_add_ps(fix2,tx); |
1646 | fiy2 = _mm_add_ps(fiy2,ty); |
1647 | fiz2 = _mm_add_ps(fiz2,tz); |
1648 | |
1649 | fjx2 = _mm_add_ps(fjx2,tx); |
1650 | fjy2 = _mm_add_ps(fjy2,ty); |
1651 | fjz2 = _mm_add_ps(fjz2,tz); |
1652 | |
1653 | /************************** |
1654 | * CALCULATE INTERACTIONS * |
1655 | **************************/ |
1656 | |
1657 | r23 = _mm_mul_ps(rsq23,rinv23); |
1658 | |
1659 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1660 | rt = _mm_mul_ps(r23,vftabscale); |
1661 | vfitab = _mm_cvttps_epi32(rt); |
1662 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1663 | vfitab = _mm_slli_epi32(vfitab,2); |
1664 | |
1665 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1666 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1667 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1668 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1669 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1670 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1671 | Heps = _mm_mul_ps(vfeps,H); |
1672 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1673 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1674 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
1675 | |
1676 | fscal = felec; |
1677 | |
1678 | /* Calculate temporary vectorial force */ |
1679 | tx = _mm_mul_ps(fscal,dx23); |
1680 | ty = _mm_mul_ps(fscal,dy23); |
1681 | tz = _mm_mul_ps(fscal,dz23); |
1682 | |
1683 | /* Update vectorial force */ |
1684 | fix2 = _mm_add_ps(fix2,tx); |
1685 | fiy2 = _mm_add_ps(fiy2,ty); |
1686 | fiz2 = _mm_add_ps(fiz2,tz); |
1687 | |
1688 | fjx3 = _mm_add_ps(fjx3,tx); |
1689 | fjy3 = _mm_add_ps(fjy3,ty); |
1690 | fjz3 = _mm_add_ps(fjz3,tz); |
1691 | |
1692 | /************************** |
1693 | * CALCULATE INTERACTIONS * |
1694 | **************************/ |
1695 | |
1696 | r31 = _mm_mul_ps(rsq31,rinv31); |
1697 | |
1698 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1699 | rt = _mm_mul_ps(r31,vftabscale); |
1700 | vfitab = _mm_cvttps_epi32(rt); |
1701 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1702 | vfitab = _mm_slli_epi32(vfitab,2); |
1703 | |
1704 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1705 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1706 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1707 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1708 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1709 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1710 | Heps = _mm_mul_ps(vfeps,H); |
1711 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1712 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1713 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
1714 | |
1715 | fscal = felec; |
1716 | |
1717 | /* Calculate temporary vectorial force */ |
1718 | tx = _mm_mul_ps(fscal,dx31); |
1719 | ty = _mm_mul_ps(fscal,dy31); |
1720 | tz = _mm_mul_ps(fscal,dz31); |
1721 | |
1722 | /* Update vectorial force */ |
1723 | fix3 = _mm_add_ps(fix3,tx); |
1724 | fiy3 = _mm_add_ps(fiy3,ty); |
1725 | fiz3 = _mm_add_ps(fiz3,tz); |
1726 | |
1727 | fjx1 = _mm_add_ps(fjx1,tx); |
1728 | fjy1 = _mm_add_ps(fjy1,ty); |
1729 | fjz1 = _mm_add_ps(fjz1,tz); |
1730 | |
1731 | /************************** |
1732 | * CALCULATE INTERACTIONS * |
1733 | **************************/ |
1734 | |
1735 | r32 = _mm_mul_ps(rsq32,rinv32); |
1736 | |
1737 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1738 | rt = _mm_mul_ps(r32,vftabscale); |
1739 | vfitab = _mm_cvttps_epi32(rt); |
1740 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1741 | vfitab = _mm_slli_epi32(vfitab,2); |
1742 | |
1743 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1744 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1745 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1746 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1747 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1748 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1749 | Heps = _mm_mul_ps(vfeps,H); |
1750 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1751 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1752 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
1753 | |
1754 | fscal = felec; |
1755 | |
1756 | /* Calculate temporary vectorial force */ |
1757 | tx = _mm_mul_ps(fscal,dx32); |
1758 | ty = _mm_mul_ps(fscal,dy32); |
1759 | tz = _mm_mul_ps(fscal,dz32); |
1760 | |
1761 | /* Update vectorial force */ |
1762 | fix3 = _mm_add_ps(fix3,tx); |
1763 | fiy3 = _mm_add_ps(fiy3,ty); |
1764 | fiz3 = _mm_add_ps(fiz3,tz); |
1765 | |
1766 | fjx2 = _mm_add_ps(fjx2,tx); |
1767 | fjy2 = _mm_add_ps(fjy2,ty); |
1768 | fjz2 = _mm_add_ps(fjz2,tz); |
1769 | |
1770 | /************************** |
1771 | * CALCULATE INTERACTIONS * |
1772 | **************************/ |
1773 | |
1774 | r33 = _mm_mul_ps(rsq33,rinv33); |
1775 | |
1776 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1777 | rt = _mm_mul_ps(r33,vftabscale); |
1778 | vfitab = _mm_cvttps_epi32(rt); |
1779 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1780 | vfitab = _mm_slli_epi32(vfitab,2); |
1781 | |
1782 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1783 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1784 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1785 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1786 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1787 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1788 | Heps = _mm_mul_ps(vfeps,H); |
1789 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1790 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1791 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
1792 | |
1793 | fscal = felec; |
1794 | |
1795 | /* Calculate temporary vectorial force */ |
1796 | tx = _mm_mul_ps(fscal,dx33); |
1797 | ty = _mm_mul_ps(fscal,dy33); |
1798 | tz = _mm_mul_ps(fscal,dz33); |
1799 | |
1800 | /* Update vectorial force */ |
1801 | fix3 = _mm_add_ps(fix3,tx); |
1802 | fiy3 = _mm_add_ps(fiy3,ty); |
1803 | fiz3 = _mm_add_ps(fiz3,tz); |
1804 | |
1805 | fjx3 = _mm_add_ps(fjx3,tx); |
1806 | fjy3 = _mm_add_ps(fjy3,ty); |
1807 | fjz3 = _mm_add_ps(fjz3,tz); |
1808 | |
1809 | fjptrA = f+j_coord_offsetA; |
1810 | fjptrB = f+j_coord_offsetB; |
1811 | fjptrC = f+j_coord_offsetC; |
1812 | fjptrD = f+j_coord_offsetD; |
1813 | |
1814 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA+DIM3,fjptrB+DIM3,fjptrC+DIM3,fjptrD+DIM3, |
1815 | fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
1816 | |
1817 | /* Inner loop uses 351 flops */ |
1818 | } |
1819 | |
1820 | if(jidx<j_index_end) |
1821 | { |
1822 | |
1823 | /* Get j neighbor index, and coordinate index */ |
1824 | jnrlistA = jjnr[jidx]; |
1825 | jnrlistB = jjnr[jidx+1]; |
1826 | jnrlistC = jjnr[jidx+2]; |
1827 | jnrlistD = jjnr[jidx+3]; |
1828 | /* Sign of each element will be negative for non-real atoms. |
1829 | * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones, |
1830 | * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries. |
1831 | */ |
1832 | dummy_mask = gmx_mm_castsi128_ps_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())); |
1833 | jnrA = (jnrlistA>=0) ? jnrlistA : 0; |
1834 | jnrB = (jnrlistB>=0) ? jnrlistB : 0; |
1835 | jnrC = (jnrlistC>=0) ? jnrlistC : 0; |
1836 | jnrD = (jnrlistD>=0) ? jnrlistD : 0; |
1837 | j_coord_offsetA = DIM3*jnrA; |
1838 | j_coord_offsetB = DIM3*jnrB; |
1839 | j_coord_offsetC = DIM3*jnrC; |
1840 | j_coord_offsetD = DIM3*jnrD; |
1841 | |
1842 | /* load j atom coordinates */ |
1843 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA+DIM3,x+j_coord_offsetB+DIM3, |
1844 | x+j_coord_offsetC+DIM3,x+j_coord_offsetD+DIM3, |
1845 | &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3); |
1846 | |
1847 | /* Calculate displacement vector */ |
1848 | dx11 = _mm_sub_ps(ix1,jx1); |
1849 | dy11 = _mm_sub_ps(iy1,jy1); |
1850 | dz11 = _mm_sub_ps(iz1,jz1); |
1851 | dx12 = _mm_sub_ps(ix1,jx2); |
1852 | dy12 = _mm_sub_ps(iy1,jy2); |
1853 | dz12 = _mm_sub_ps(iz1,jz2); |
1854 | dx13 = _mm_sub_ps(ix1,jx3); |
1855 | dy13 = _mm_sub_ps(iy1,jy3); |
1856 | dz13 = _mm_sub_ps(iz1,jz3); |
1857 | dx21 = _mm_sub_ps(ix2,jx1); |
1858 | dy21 = _mm_sub_ps(iy2,jy1); |
1859 | dz21 = _mm_sub_ps(iz2,jz1); |
1860 | dx22 = _mm_sub_ps(ix2,jx2); |
1861 | dy22 = _mm_sub_ps(iy2,jy2); |
1862 | dz22 = _mm_sub_ps(iz2,jz2); |
1863 | dx23 = _mm_sub_ps(ix2,jx3); |
1864 | dy23 = _mm_sub_ps(iy2,jy3); |
1865 | dz23 = _mm_sub_ps(iz2,jz3); |
1866 | dx31 = _mm_sub_ps(ix3,jx1); |
1867 | dy31 = _mm_sub_ps(iy3,jy1); |
1868 | dz31 = _mm_sub_ps(iz3,jz1); |
1869 | dx32 = _mm_sub_ps(ix3,jx2); |
1870 | dy32 = _mm_sub_ps(iy3,jy2); |
1871 | dz32 = _mm_sub_ps(iz3,jz2); |
1872 | dx33 = _mm_sub_ps(ix3,jx3); |
1873 | dy33 = _mm_sub_ps(iy3,jy3); |
1874 | dz33 = _mm_sub_ps(iz3,jz3); |
1875 | |
1876 | /* Calculate squared distance and things based on it */ |
1877 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
1878 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
1879 | rsq13 = gmx_mm_calc_rsq_ps(dx13,dy13,dz13); |
1880 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
1881 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
1882 | rsq23 = gmx_mm_calc_rsq_ps(dx23,dy23,dz23); |
1883 | rsq31 = gmx_mm_calc_rsq_ps(dx31,dy31,dz31); |
1884 | rsq32 = gmx_mm_calc_rsq_ps(dx32,dy32,dz32); |
1885 | rsq33 = gmx_mm_calc_rsq_ps(dx33,dy33,dz33); |
1886 | |
1887 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
1888 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
1889 | rinv13 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq13); |
1890 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
1891 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
1892 | rinv23 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq23); |
1893 | rinv31 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq31); |
1894 | rinv32 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq32); |
1895 | rinv33 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq33); |
1896 | |
1897 | fjx1 = _mm_setzero_ps(); |
1898 | fjy1 = _mm_setzero_ps(); |
1899 | fjz1 = _mm_setzero_ps(); |
1900 | fjx2 = _mm_setzero_ps(); |
1901 | fjy2 = _mm_setzero_ps(); |
1902 | fjz2 = _mm_setzero_ps(); |
1903 | fjx3 = _mm_setzero_ps(); |
1904 | fjy3 = _mm_setzero_ps(); |
1905 | fjz3 = _mm_setzero_ps(); |
1906 | |
1907 | /************************** |
1908 | * CALCULATE INTERACTIONS * |
1909 | **************************/ |
1910 | |
1911 | r11 = _mm_mul_ps(rsq11,rinv11); |
1912 | r11 = _mm_andnot_ps(dummy_mask,r11); |
1913 | |
1914 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1915 | rt = _mm_mul_ps(r11,vftabscale); |
1916 | vfitab = _mm_cvttps_epi32(rt); |
1917 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1918 | vfitab = _mm_slli_epi32(vfitab,2); |
1919 | |
1920 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1921 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1922 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1923 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1924 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1925 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1926 | Heps = _mm_mul_ps(vfeps,H); |
1927 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1928 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1929 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq11,FF),_mm_mul_ps(vftabscale,rinv11))); |
1930 | |
1931 | fscal = felec; |
1932 | |
1933 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1934 | |
1935 | /* Calculate temporary vectorial force */ |
1936 | tx = _mm_mul_ps(fscal,dx11); |
1937 | ty = _mm_mul_ps(fscal,dy11); |
1938 | tz = _mm_mul_ps(fscal,dz11); |
1939 | |
1940 | /* Update vectorial force */ |
1941 | fix1 = _mm_add_ps(fix1,tx); |
1942 | fiy1 = _mm_add_ps(fiy1,ty); |
1943 | fiz1 = _mm_add_ps(fiz1,tz); |
1944 | |
1945 | fjx1 = _mm_add_ps(fjx1,tx); |
1946 | fjy1 = _mm_add_ps(fjy1,ty); |
1947 | fjz1 = _mm_add_ps(fjz1,tz); |
1948 | |
1949 | /************************** |
1950 | * CALCULATE INTERACTIONS * |
1951 | **************************/ |
1952 | |
1953 | r12 = _mm_mul_ps(rsq12,rinv12); |
1954 | r12 = _mm_andnot_ps(dummy_mask,r12); |
1955 | |
1956 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1957 | rt = _mm_mul_ps(r12,vftabscale); |
1958 | vfitab = _mm_cvttps_epi32(rt); |
1959 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1960 | vfitab = _mm_slli_epi32(vfitab,2); |
1961 | |
1962 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
1963 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
1964 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
1965 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
1966 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
1967 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
1968 | Heps = _mm_mul_ps(vfeps,H); |
1969 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
1970 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
1971 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq12,FF),_mm_mul_ps(vftabscale,rinv12))); |
1972 | |
1973 | fscal = felec; |
1974 | |
1975 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1976 | |
1977 | /* Calculate temporary vectorial force */ |
1978 | tx = _mm_mul_ps(fscal,dx12); |
1979 | ty = _mm_mul_ps(fscal,dy12); |
1980 | tz = _mm_mul_ps(fscal,dz12); |
1981 | |
1982 | /* Update vectorial force */ |
1983 | fix1 = _mm_add_ps(fix1,tx); |
1984 | fiy1 = _mm_add_ps(fiy1,ty); |
1985 | fiz1 = _mm_add_ps(fiz1,tz); |
1986 | |
1987 | fjx2 = _mm_add_ps(fjx2,tx); |
1988 | fjy2 = _mm_add_ps(fjy2,ty); |
1989 | fjz2 = _mm_add_ps(fjz2,tz); |
1990 | |
1991 | /************************** |
1992 | * CALCULATE INTERACTIONS * |
1993 | **************************/ |
1994 | |
1995 | r13 = _mm_mul_ps(rsq13,rinv13); |
1996 | r13 = _mm_andnot_ps(dummy_mask,r13); |
1997 | |
1998 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
1999 | rt = _mm_mul_ps(r13,vftabscale); |
2000 | vfitab = _mm_cvttps_epi32(rt); |
2001 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2002 | vfitab = _mm_slli_epi32(vfitab,2); |
2003 | |
2004 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
2005 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
2006 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
2007 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
2008 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
2009 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
2010 | Heps = _mm_mul_ps(vfeps,H); |
2011 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
2012 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
2013 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq13,FF),_mm_mul_ps(vftabscale,rinv13))); |
2014 | |
2015 | fscal = felec; |
2016 | |
2017 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2018 | |
2019 | /* Calculate temporary vectorial force */ |
2020 | tx = _mm_mul_ps(fscal,dx13); |
2021 | ty = _mm_mul_ps(fscal,dy13); |
2022 | tz = _mm_mul_ps(fscal,dz13); |
2023 | |
2024 | /* Update vectorial force */ |
2025 | fix1 = _mm_add_ps(fix1,tx); |
2026 | fiy1 = _mm_add_ps(fiy1,ty); |
2027 | fiz1 = _mm_add_ps(fiz1,tz); |
2028 | |
2029 | fjx3 = _mm_add_ps(fjx3,tx); |
2030 | fjy3 = _mm_add_ps(fjy3,ty); |
2031 | fjz3 = _mm_add_ps(fjz3,tz); |
2032 | |
2033 | /************************** |
2034 | * CALCULATE INTERACTIONS * |
2035 | **************************/ |
2036 | |
2037 | r21 = _mm_mul_ps(rsq21,rinv21); |
2038 | r21 = _mm_andnot_ps(dummy_mask,r21); |
2039 | |
2040 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
2041 | rt = _mm_mul_ps(r21,vftabscale); |
2042 | vfitab = _mm_cvttps_epi32(rt); |
2043 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2044 | vfitab = _mm_slli_epi32(vfitab,2); |
2045 | |
2046 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
2047 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
2048 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
2049 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
2050 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
2051 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
2052 | Heps = _mm_mul_ps(vfeps,H); |
2053 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
2054 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
2055 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq21,FF),_mm_mul_ps(vftabscale,rinv21))); |
2056 | |
2057 | fscal = felec; |
2058 | |
2059 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2060 | |
2061 | /* Calculate temporary vectorial force */ |
2062 | tx = _mm_mul_ps(fscal,dx21); |
2063 | ty = _mm_mul_ps(fscal,dy21); |
2064 | tz = _mm_mul_ps(fscal,dz21); |
2065 | |
2066 | /* Update vectorial force */ |
2067 | fix2 = _mm_add_ps(fix2,tx); |
2068 | fiy2 = _mm_add_ps(fiy2,ty); |
2069 | fiz2 = _mm_add_ps(fiz2,tz); |
2070 | |
2071 | fjx1 = _mm_add_ps(fjx1,tx); |
2072 | fjy1 = _mm_add_ps(fjy1,ty); |
2073 | fjz1 = _mm_add_ps(fjz1,tz); |
2074 | |
2075 | /************************** |
2076 | * CALCULATE INTERACTIONS * |
2077 | **************************/ |
2078 | |
2079 | r22 = _mm_mul_ps(rsq22,rinv22); |
2080 | r22 = _mm_andnot_ps(dummy_mask,r22); |
2081 | |
2082 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
2083 | rt = _mm_mul_ps(r22,vftabscale); |
2084 | vfitab = _mm_cvttps_epi32(rt); |
2085 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2086 | vfitab = _mm_slli_epi32(vfitab,2); |
2087 | |
2088 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
2089 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
2090 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
2091 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
2092 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
2093 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
2094 | Heps = _mm_mul_ps(vfeps,H); |
2095 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
2096 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
2097 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq22,FF),_mm_mul_ps(vftabscale,rinv22))); |
2098 | |
2099 | fscal = felec; |
2100 | |
2101 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2102 | |
2103 | /* Calculate temporary vectorial force */ |
2104 | tx = _mm_mul_ps(fscal,dx22); |
2105 | ty = _mm_mul_ps(fscal,dy22); |
2106 | tz = _mm_mul_ps(fscal,dz22); |
2107 | |
2108 | /* Update vectorial force */ |
2109 | fix2 = _mm_add_ps(fix2,tx); |
2110 | fiy2 = _mm_add_ps(fiy2,ty); |
2111 | fiz2 = _mm_add_ps(fiz2,tz); |
2112 | |
2113 | fjx2 = _mm_add_ps(fjx2,tx); |
2114 | fjy2 = _mm_add_ps(fjy2,ty); |
2115 | fjz2 = _mm_add_ps(fjz2,tz); |
2116 | |
2117 | /************************** |
2118 | * CALCULATE INTERACTIONS * |
2119 | **************************/ |
2120 | |
2121 | r23 = _mm_mul_ps(rsq23,rinv23); |
2122 | r23 = _mm_andnot_ps(dummy_mask,r23); |
2123 | |
2124 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
2125 | rt = _mm_mul_ps(r23,vftabscale); |
2126 | vfitab = _mm_cvttps_epi32(rt); |
2127 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2128 | vfitab = _mm_slli_epi32(vfitab,2); |
2129 | |
2130 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
2131 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
2132 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
2133 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
2134 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
2135 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
2136 | Heps = _mm_mul_ps(vfeps,H); |
2137 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
2138 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
2139 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq23,FF),_mm_mul_ps(vftabscale,rinv23))); |
2140 | |
2141 | fscal = felec; |
2142 | |
2143 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2144 | |
2145 | /* Calculate temporary vectorial force */ |
2146 | tx = _mm_mul_ps(fscal,dx23); |
2147 | ty = _mm_mul_ps(fscal,dy23); |
2148 | tz = _mm_mul_ps(fscal,dz23); |
2149 | |
2150 | /* Update vectorial force */ |
2151 | fix2 = _mm_add_ps(fix2,tx); |
2152 | fiy2 = _mm_add_ps(fiy2,ty); |
2153 | fiz2 = _mm_add_ps(fiz2,tz); |
2154 | |
2155 | fjx3 = _mm_add_ps(fjx3,tx); |
2156 | fjy3 = _mm_add_ps(fjy3,ty); |
2157 | fjz3 = _mm_add_ps(fjz3,tz); |
2158 | |
2159 | /************************** |
2160 | * CALCULATE INTERACTIONS * |
2161 | **************************/ |
2162 | |
2163 | r31 = _mm_mul_ps(rsq31,rinv31); |
2164 | r31 = _mm_andnot_ps(dummy_mask,r31); |
2165 | |
2166 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
2167 | rt = _mm_mul_ps(r31,vftabscale); |
2168 | vfitab = _mm_cvttps_epi32(rt); |
2169 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2170 | vfitab = _mm_slli_epi32(vfitab,2); |
2171 | |
2172 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
2173 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
2174 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
2175 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
2176 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
2177 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
2178 | Heps = _mm_mul_ps(vfeps,H); |
2179 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
2180 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
2181 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq31,FF),_mm_mul_ps(vftabscale,rinv31))); |
2182 | |
2183 | fscal = felec; |
2184 | |
2185 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2186 | |
2187 | /* Calculate temporary vectorial force */ |
2188 | tx = _mm_mul_ps(fscal,dx31); |
2189 | ty = _mm_mul_ps(fscal,dy31); |
2190 | tz = _mm_mul_ps(fscal,dz31); |
2191 | |
2192 | /* Update vectorial force */ |
2193 | fix3 = _mm_add_ps(fix3,tx); |
2194 | fiy3 = _mm_add_ps(fiy3,ty); |
2195 | fiz3 = _mm_add_ps(fiz3,tz); |
2196 | |
2197 | fjx1 = _mm_add_ps(fjx1,tx); |
2198 | fjy1 = _mm_add_ps(fjy1,ty); |
2199 | fjz1 = _mm_add_ps(fjz1,tz); |
2200 | |
2201 | /************************** |
2202 | * CALCULATE INTERACTIONS * |
2203 | **************************/ |
2204 | |
2205 | r32 = _mm_mul_ps(rsq32,rinv32); |
2206 | r32 = _mm_andnot_ps(dummy_mask,r32); |
2207 | |
2208 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
2209 | rt = _mm_mul_ps(r32,vftabscale); |
2210 | vfitab = _mm_cvttps_epi32(rt); |
2211 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2212 | vfitab = _mm_slli_epi32(vfitab,2); |
2213 | |
2214 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
2215 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
2216 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
2217 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
2218 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
2219 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
2220 | Heps = _mm_mul_ps(vfeps,H); |
2221 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
2222 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
2223 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq32,FF),_mm_mul_ps(vftabscale,rinv32))); |
2224 | |
2225 | fscal = felec; |
2226 | |
2227 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2228 | |
2229 | /* Calculate temporary vectorial force */ |
2230 | tx = _mm_mul_ps(fscal,dx32); |
2231 | ty = _mm_mul_ps(fscal,dy32); |
2232 | tz = _mm_mul_ps(fscal,dz32); |
2233 | |
2234 | /* Update vectorial force */ |
2235 | fix3 = _mm_add_ps(fix3,tx); |
2236 | fiy3 = _mm_add_ps(fiy3,ty); |
2237 | fiz3 = _mm_add_ps(fiz3,tz); |
2238 | |
2239 | fjx2 = _mm_add_ps(fjx2,tx); |
2240 | fjy2 = _mm_add_ps(fjy2,ty); |
2241 | fjz2 = _mm_add_ps(fjz2,tz); |
2242 | |
2243 | /************************** |
2244 | * CALCULATE INTERACTIONS * |
2245 | **************************/ |
2246 | |
2247 | r33 = _mm_mul_ps(rsq33,rinv33); |
2248 | r33 = _mm_andnot_ps(dummy_mask,r33); |
2249 | |
2250 | /* Calculate table index by multiplying r with table scale and truncate to integer */ |
2251 | rt = _mm_mul_ps(r33,vftabscale); |
2252 | vfitab = _mm_cvttps_epi32(rt); |
2253 | vfeps = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (rt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2254 | vfitab = _mm_slli_epi32(vfitab,2); |
2255 | |
2256 | /* CUBIC SPLINE TABLE ELECTROSTATICS */ |
2257 | Y = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,0)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(0) & 3];})) ); |
2258 | F = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,1)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(1) & 3];})) ); |
2259 | G = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,2)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(2) & 3];})) ); |
2260 | H = _mm_load_ps( vftab + gmx_mm_extract_epi32(vfitab,3)(__extension__ ({ __v4si __a = (__v4si)(vfitab); __a[(3) & 3];})) ); |
2261 | _MM_TRANSPOSE4_PS(Y,F,G,H)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((Y ), (F)); tmp2 = _mm_unpacklo_ps((G), (H)); tmp1 = _mm_unpackhi_ps ((Y), (F)); tmp3 = _mm_unpackhi_ps((G), (H)); (Y) = _mm_movelh_ps (tmp0, tmp2); (F) = _mm_movehl_ps(tmp2, tmp0); (G) = _mm_movelh_ps (tmp1, tmp3); (H) = _mm_movehl_ps(tmp3, tmp1); } while (0); |
2262 | Heps = _mm_mul_ps(vfeps,H); |
2263 | Fp = _mm_add_ps(F,_mm_mul_ps(vfeps,_mm_add_ps(G,Heps))); |
2264 | FF = _mm_add_ps(Fp,_mm_mul_ps(vfeps,_mm_add_ps(G,_mm_add_ps(Heps,Heps)))); |
2265 | felec = _mm_xor_ps(signbit,_mm_mul_ps(_mm_mul_ps(qq33,FF),_mm_mul_ps(vftabscale,rinv33))); |
2266 | |
2267 | fscal = felec; |
2268 | |
2269 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2270 | |
2271 | /* Calculate temporary vectorial force */ |
2272 | tx = _mm_mul_ps(fscal,dx33); |
2273 | ty = _mm_mul_ps(fscal,dy33); |
2274 | tz = _mm_mul_ps(fscal,dz33); |
2275 | |
2276 | /* Update vectorial force */ |
2277 | fix3 = _mm_add_ps(fix3,tx); |
2278 | fiy3 = _mm_add_ps(fiy3,ty); |
2279 | fiz3 = _mm_add_ps(fiz3,tz); |
2280 | |
2281 | fjx3 = _mm_add_ps(fjx3,tx); |
2282 | fjy3 = _mm_add_ps(fjy3,ty); |
2283 | fjz3 = _mm_add_ps(fjz3,tz); |
2284 | |
2285 | fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch; |
2286 | fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch; |
2287 | fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch; |
2288 | fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch; |
2289 | |
2290 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA+DIM3,fjptrB+DIM3,fjptrC+DIM3,fjptrD+DIM3, |
2291 | fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3); |
2292 | |
2293 | /* Inner loop uses 360 flops */ |
2294 | } |
2295 | |
2296 | /* End of innermost loop */ |
2297 | |
2298 | gmx_mm_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3, |
2299 | f+i_coord_offset+DIM3,fshift+i_shift_offset); |
2300 | |
2301 | /* Increment number of inner iterations */ |
2302 | inneriter += j_index_end - j_index_start; |
2303 | |
2304 | /* Outer loop uses 18 flops */ |
2305 | } |
2306 | |
2307 | /* Increment number of outer iterations */ |
2308 | outeriter += nri; |
2309 | |
2310 | /* Update outer/inner flops */ |
2311 | |
2312 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_F,outeriter*18 + inneriter*360)(nrnb)->n[eNR_NBKERNEL_ELEC_W4W4_F] += outeriter*18 + inneriter *360; |
2313 | } |