File: | gromacs/gmxlib/nonbonded/nb_kernel_sse4_1_single/nb_kernel_ElecEwSh_VdwNone_GeomW3W3_sse4_1_single.c |
Location: | line 1466, column 22 |
Description: | Value stored to 'signbit' during its initialization 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_ElecEwSh_VdwNone_GeomW3W3_VF_sse4_1_single |
54 | * Electrostatics interaction: Ewald |
55 | * VdW interaction: None |
56 | * Geometry: Water3-Water3 |
57 | * Calculate force/pot: PotentialAndForce |
58 | */ |
59 | void |
60 | nb_kernel_ElecEwSh_VdwNone_GeomW3W3_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 vdwioffset0; |
86 | __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0; |
87 | int vdwioffset1; |
88 | __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
89 | int vdwioffset2; |
90 | __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
91 | int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D; |
92 | __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
93 | int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D; |
94 | __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
95 | int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D; |
96 | __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
97 | __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00; |
98 | __m128 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01; |
99 | __m128 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02; |
100 | __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10; |
101 | __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11; |
102 | __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12; |
103 | __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20; |
104 | __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21; |
105 | __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22; |
106 | __m128 velec,felec,velecsum,facel,crf,krf,krf2; |
107 | real *charge; |
108 | __m128i ewitab; |
109 | __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV; |
110 | real *ewtab; |
111 | __m128 dummy_mask,cutoff_mask; |
112 | __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) ); |
113 | __m128 one = _mm_set1_ps(1.0); |
114 | __m128 two = _mm_set1_ps(2.0); |
115 | x = xx[0]; |
116 | f = ff[0]; |
117 | |
118 | nri = nlist->nri; |
119 | iinr = nlist->iinr; |
120 | jindex = nlist->jindex; |
121 | jjnr = nlist->jjnr; |
122 | shiftidx = nlist->shift; |
123 | gid = nlist->gid; |
124 | shiftvec = fr->shift_vec[0]; |
125 | fshift = fr->fshift[0]; |
126 | facel = _mm_set1_ps(fr->epsfac); |
127 | charge = mdatoms->chargeA; |
128 | |
129 | sh_ewald = _mm_set1_ps(fr->ic->sh_ewald); |
130 | ewtab = fr->ic->tabq_coul_FDV0; |
131 | ewtabscale = _mm_set1_ps(fr->ic->tabq_scale); |
132 | ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale); |
133 | |
134 | /* Setup water-specific parameters */ |
135 | inr = nlist->iinr[0]; |
136 | iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0])); |
137 | iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1])); |
138 | iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2])); |
139 | |
140 | jq0 = _mm_set1_ps(charge[inr+0]); |
141 | jq1 = _mm_set1_ps(charge[inr+1]); |
142 | jq2 = _mm_set1_ps(charge[inr+2]); |
143 | qq00 = _mm_mul_ps(iq0,jq0); |
144 | qq01 = _mm_mul_ps(iq0,jq1); |
145 | qq02 = _mm_mul_ps(iq0,jq2); |
146 | qq10 = _mm_mul_ps(iq1,jq0); |
147 | qq11 = _mm_mul_ps(iq1,jq1); |
148 | qq12 = _mm_mul_ps(iq1,jq2); |
149 | qq20 = _mm_mul_ps(iq2,jq0); |
150 | qq21 = _mm_mul_ps(iq2,jq1); |
151 | qq22 = _mm_mul_ps(iq2,jq2); |
152 | |
153 | /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */ |
154 | rcutoff_scalar = fr->rcoulomb; |
155 | rcutoff = _mm_set1_ps(rcutoff_scalar); |
156 | rcutoff2 = _mm_mul_ps(rcutoff,rcutoff); |
157 | |
158 | /* Avoid stupid compiler warnings */ |
159 | jnrA = jnrB = jnrC = jnrD = 0; |
160 | j_coord_offsetA = 0; |
161 | j_coord_offsetB = 0; |
162 | j_coord_offsetC = 0; |
163 | j_coord_offsetD = 0; |
164 | |
165 | outeriter = 0; |
166 | inneriter = 0; |
167 | |
168 | for(iidx=0;iidx<4*DIM3;iidx++) |
169 | { |
170 | scratch[iidx] = 0.0; |
171 | } |
172 | |
173 | /* Start outer loop over neighborlists */ |
174 | for(iidx=0; iidx<nri; iidx++) |
175 | { |
176 | /* Load shift vector for this list */ |
177 | i_shift_offset = DIM3*shiftidx[iidx]; |
178 | |
179 | /* Load limits for loop over neighbors */ |
180 | j_index_start = jindex[iidx]; |
181 | j_index_end = jindex[iidx+1]; |
182 | |
183 | /* Get outer coordinate index */ |
184 | inr = iinr[iidx]; |
185 | i_coord_offset = DIM3*inr; |
186 | |
187 | /* Load i particle coords and add shift vector */ |
188 | gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset, |
189 | &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2); |
190 | |
191 | fix0 = _mm_setzero_ps(); |
192 | fiy0 = _mm_setzero_ps(); |
193 | fiz0 = _mm_setzero_ps(); |
194 | fix1 = _mm_setzero_ps(); |
195 | fiy1 = _mm_setzero_ps(); |
196 | fiz1 = _mm_setzero_ps(); |
197 | fix2 = _mm_setzero_ps(); |
198 | fiy2 = _mm_setzero_ps(); |
199 | fiz2 = _mm_setzero_ps(); |
200 | |
201 | /* Reset potential sums */ |
202 | velecsum = _mm_setzero_ps(); |
203 | |
204 | /* Start inner kernel loop */ |
205 | for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4) |
206 | { |
207 | |
208 | /* Get j neighbor index, and coordinate index */ |
209 | jnrA = jjnr[jidx]; |
210 | jnrB = jjnr[jidx+1]; |
211 | jnrC = jjnr[jidx+2]; |
212 | jnrD = jjnr[jidx+3]; |
213 | j_coord_offsetA = DIM3*jnrA; |
214 | j_coord_offsetB = DIM3*jnrB; |
215 | j_coord_offsetC = DIM3*jnrC; |
216 | j_coord_offsetD = DIM3*jnrD; |
217 | |
218 | /* load j atom coordinates */ |
219 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
220 | x+j_coord_offsetC,x+j_coord_offsetD, |
221 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2); |
222 | |
223 | /* Calculate displacement vector */ |
224 | dx00 = _mm_sub_ps(ix0,jx0); |
225 | dy00 = _mm_sub_ps(iy0,jy0); |
226 | dz00 = _mm_sub_ps(iz0,jz0); |
227 | dx01 = _mm_sub_ps(ix0,jx1); |
228 | dy01 = _mm_sub_ps(iy0,jy1); |
229 | dz01 = _mm_sub_ps(iz0,jz1); |
230 | dx02 = _mm_sub_ps(ix0,jx2); |
231 | dy02 = _mm_sub_ps(iy0,jy2); |
232 | dz02 = _mm_sub_ps(iz0,jz2); |
233 | dx10 = _mm_sub_ps(ix1,jx0); |
234 | dy10 = _mm_sub_ps(iy1,jy0); |
235 | dz10 = _mm_sub_ps(iz1,jz0); |
236 | dx11 = _mm_sub_ps(ix1,jx1); |
237 | dy11 = _mm_sub_ps(iy1,jy1); |
238 | dz11 = _mm_sub_ps(iz1,jz1); |
239 | dx12 = _mm_sub_ps(ix1,jx2); |
240 | dy12 = _mm_sub_ps(iy1,jy2); |
241 | dz12 = _mm_sub_ps(iz1,jz2); |
242 | dx20 = _mm_sub_ps(ix2,jx0); |
243 | dy20 = _mm_sub_ps(iy2,jy0); |
244 | dz20 = _mm_sub_ps(iz2,jz0); |
245 | dx21 = _mm_sub_ps(ix2,jx1); |
246 | dy21 = _mm_sub_ps(iy2,jy1); |
247 | dz21 = _mm_sub_ps(iz2,jz1); |
248 | dx22 = _mm_sub_ps(ix2,jx2); |
249 | dy22 = _mm_sub_ps(iy2,jy2); |
250 | dz22 = _mm_sub_ps(iz2,jz2); |
251 | |
252 | /* Calculate squared distance and things based on it */ |
253 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
254 | rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01); |
255 | rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02); |
256 | rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10); |
257 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
258 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
259 | rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20); |
260 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
261 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
262 | |
263 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
264 | rinv01 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq01); |
265 | rinv02 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq02); |
266 | rinv10 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq10); |
267 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
268 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
269 | rinv20 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq20); |
270 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
271 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
272 | |
273 | rinvsq00 = _mm_mul_ps(rinv00,rinv00); |
274 | rinvsq01 = _mm_mul_ps(rinv01,rinv01); |
275 | rinvsq02 = _mm_mul_ps(rinv02,rinv02); |
276 | rinvsq10 = _mm_mul_ps(rinv10,rinv10); |
277 | rinvsq11 = _mm_mul_ps(rinv11,rinv11); |
278 | rinvsq12 = _mm_mul_ps(rinv12,rinv12); |
279 | rinvsq20 = _mm_mul_ps(rinv20,rinv20); |
280 | rinvsq21 = _mm_mul_ps(rinv21,rinv21); |
281 | rinvsq22 = _mm_mul_ps(rinv22,rinv22); |
282 | |
283 | fjx0 = _mm_setzero_ps(); |
284 | fjy0 = _mm_setzero_ps(); |
285 | fjz0 = _mm_setzero_ps(); |
286 | fjx1 = _mm_setzero_ps(); |
287 | fjy1 = _mm_setzero_ps(); |
288 | fjz1 = _mm_setzero_ps(); |
289 | fjx2 = _mm_setzero_ps(); |
290 | fjy2 = _mm_setzero_ps(); |
291 | fjz2 = _mm_setzero_ps(); |
292 | |
293 | /************************** |
294 | * CALCULATE INTERACTIONS * |
295 | **************************/ |
296 | |
297 | if (gmx_mm_any_lt(rsq00,rcutoff2)) |
298 | { |
299 | |
300 | r00 = _mm_mul_ps(rsq00,rinv00); |
301 | |
302 | /* EWALD ELECTROSTATICS */ |
303 | |
304 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
305 | ewrt = _mm_mul_ps(r00,ewtabscale); |
306 | ewitab = _mm_cvttps_epi32(ewrt); |
307 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
308 | ewitab = _mm_slli_epi32(ewitab,2); |
309 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
310 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
311 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
312 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
313 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
314 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
315 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
316 | velec = _mm_mul_ps(qq00,_mm_sub_ps(_mm_sub_ps(rinv00,sh_ewald),velec)); |
317 | felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec)); |
318 | |
319 | cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2); |
320 | |
321 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
322 | velec = _mm_and_ps(velec,cutoff_mask); |
323 | velecsum = _mm_add_ps(velecsum,velec); |
324 | |
325 | fscal = felec; |
326 | |
327 | fscal = _mm_and_ps(fscal,cutoff_mask); |
328 | |
329 | /* Calculate temporary vectorial force */ |
330 | tx = _mm_mul_ps(fscal,dx00); |
331 | ty = _mm_mul_ps(fscal,dy00); |
332 | tz = _mm_mul_ps(fscal,dz00); |
333 | |
334 | /* Update vectorial force */ |
335 | fix0 = _mm_add_ps(fix0,tx); |
336 | fiy0 = _mm_add_ps(fiy0,ty); |
337 | fiz0 = _mm_add_ps(fiz0,tz); |
338 | |
339 | fjx0 = _mm_add_ps(fjx0,tx); |
340 | fjy0 = _mm_add_ps(fjy0,ty); |
341 | fjz0 = _mm_add_ps(fjz0,tz); |
342 | |
343 | } |
344 | |
345 | /************************** |
346 | * CALCULATE INTERACTIONS * |
347 | **************************/ |
348 | |
349 | if (gmx_mm_any_lt(rsq01,rcutoff2)) |
350 | { |
351 | |
352 | r01 = _mm_mul_ps(rsq01,rinv01); |
353 | |
354 | /* EWALD ELECTROSTATICS */ |
355 | |
356 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
357 | ewrt = _mm_mul_ps(r01,ewtabscale); |
358 | ewitab = _mm_cvttps_epi32(ewrt); |
359 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
360 | ewitab = _mm_slli_epi32(ewitab,2); |
361 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
362 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
363 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
364 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
365 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
366 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
367 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
368 | velec = _mm_mul_ps(qq01,_mm_sub_ps(_mm_sub_ps(rinv01,sh_ewald),velec)); |
369 | felec = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec)); |
370 | |
371 | cutoff_mask = _mm_cmplt_ps(rsq01,rcutoff2); |
372 | |
373 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
374 | velec = _mm_and_ps(velec,cutoff_mask); |
375 | velecsum = _mm_add_ps(velecsum,velec); |
376 | |
377 | fscal = felec; |
378 | |
379 | fscal = _mm_and_ps(fscal,cutoff_mask); |
380 | |
381 | /* Calculate temporary vectorial force */ |
382 | tx = _mm_mul_ps(fscal,dx01); |
383 | ty = _mm_mul_ps(fscal,dy01); |
384 | tz = _mm_mul_ps(fscal,dz01); |
385 | |
386 | /* Update vectorial force */ |
387 | fix0 = _mm_add_ps(fix0,tx); |
388 | fiy0 = _mm_add_ps(fiy0,ty); |
389 | fiz0 = _mm_add_ps(fiz0,tz); |
390 | |
391 | fjx1 = _mm_add_ps(fjx1,tx); |
392 | fjy1 = _mm_add_ps(fjy1,ty); |
393 | fjz1 = _mm_add_ps(fjz1,tz); |
394 | |
395 | } |
396 | |
397 | /************************** |
398 | * CALCULATE INTERACTIONS * |
399 | **************************/ |
400 | |
401 | if (gmx_mm_any_lt(rsq02,rcutoff2)) |
402 | { |
403 | |
404 | r02 = _mm_mul_ps(rsq02,rinv02); |
405 | |
406 | /* EWALD ELECTROSTATICS */ |
407 | |
408 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
409 | ewrt = _mm_mul_ps(r02,ewtabscale); |
410 | ewitab = _mm_cvttps_epi32(ewrt); |
411 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
412 | ewitab = _mm_slli_epi32(ewitab,2); |
413 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
414 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
415 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
416 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
417 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
418 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
419 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
420 | velec = _mm_mul_ps(qq02,_mm_sub_ps(_mm_sub_ps(rinv02,sh_ewald),velec)); |
421 | felec = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec)); |
422 | |
423 | cutoff_mask = _mm_cmplt_ps(rsq02,rcutoff2); |
424 | |
425 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
426 | velec = _mm_and_ps(velec,cutoff_mask); |
427 | velecsum = _mm_add_ps(velecsum,velec); |
428 | |
429 | fscal = felec; |
430 | |
431 | fscal = _mm_and_ps(fscal,cutoff_mask); |
432 | |
433 | /* Calculate temporary vectorial force */ |
434 | tx = _mm_mul_ps(fscal,dx02); |
435 | ty = _mm_mul_ps(fscal,dy02); |
436 | tz = _mm_mul_ps(fscal,dz02); |
437 | |
438 | /* Update vectorial force */ |
439 | fix0 = _mm_add_ps(fix0,tx); |
440 | fiy0 = _mm_add_ps(fiy0,ty); |
441 | fiz0 = _mm_add_ps(fiz0,tz); |
442 | |
443 | fjx2 = _mm_add_ps(fjx2,tx); |
444 | fjy2 = _mm_add_ps(fjy2,ty); |
445 | fjz2 = _mm_add_ps(fjz2,tz); |
446 | |
447 | } |
448 | |
449 | /************************** |
450 | * CALCULATE INTERACTIONS * |
451 | **************************/ |
452 | |
453 | if (gmx_mm_any_lt(rsq10,rcutoff2)) |
454 | { |
455 | |
456 | r10 = _mm_mul_ps(rsq10,rinv10); |
457 | |
458 | /* EWALD ELECTROSTATICS */ |
459 | |
460 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
461 | ewrt = _mm_mul_ps(r10,ewtabscale); |
462 | ewitab = _mm_cvttps_epi32(ewrt); |
463 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
464 | ewitab = _mm_slli_epi32(ewitab,2); |
465 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
466 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
467 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
468 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
469 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
470 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
471 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
472 | velec = _mm_mul_ps(qq10,_mm_sub_ps(_mm_sub_ps(rinv10,sh_ewald),velec)); |
473 | felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec)); |
474 | |
475 | cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2); |
476 | |
477 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
478 | velec = _mm_and_ps(velec,cutoff_mask); |
479 | velecsum = _mm_add_ps(velecsum,velec); |
480 | |
481 | fscal = felec; |
482 | |
483 | fscal = _mm_and_ps(fscal,cutoff_mask); |
484 | |
485 | /* Calculate temporary vectorial force */ |
486 | tx = _mm_mul_ps(fscal,dx10); |
487 | ty = _mm_mul_ps(fscal,dy10); |
488 | tz = _mm_mul_ps(fscal,dz10); |
489 | |
490 | /* Update vectorial force */ |
491 | fix1 = _mm_add_ps(fix1,tx); |
492 | fiy1 = _mm_add_ps(fiy1,ty); |
493 | fiz1 = _mm_add_ps(fiz1,tz); |
494 | |
495 | fjx0 = _mm_add_ps(fjx0,tx); |
496 | fjy0 = _mm_add_ps(fjy0,ty); |
497 | fjz0 = _mm_add_ps(fjz0,tz); |
498 | |
499 | } |
500 | |
501 | /************************** |
502 | * CALCULATE INTERACTIONS * |
503 | **************************/ |
504 | |
505 | if (gmx_mm_any_lt(rsq11,rcutoff2)) |
506 | { |
507 | |
508 | r11 = _mm_mul_ps(rsq11,rinv11); |
509 | |
510 | /* EWALD ELECTROSTATICS */ |
511 | |
512 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
513 | ewrt = _mm_mul_ps(r11,ewtabscale); |
514 | ewitab = _mm_cvttps_epi32(ewrt); |
515 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
516 | ewitab = _mm_slli_epi32(ewitab,2); |
517 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
518 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
519 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
520 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
521 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
522 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
523 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
524 | velec = _mm_mul_ps(qq11,_mm_sub_ps(_mm_sub_ps(rinv11,sh_ewald),velec)); |
525 | felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec)); |
526 | |
527 | cutoff_mask = _mm_cmplt_ps(rsq11,rcutoff2); |
528 | |
529 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
530 | velec = _mm_and_ps(velec,cutoff_mask); |
531 | velecsum = _mm_add_ps(velecsum,velec); |
532 | |
533 | fscal = felec; |
534 | |
535 | fscal = _mm_and_ps(fscal,cutoff_mask); |
536 | |
537 | /* Calculate temporary vectorial force */ |
538 | tx = _mm_mul_ps(fscal,dx11); |
539 | ty = _mm_mul_ps(fscal,dy11); |
540 | tz = _mm_mul_ps(fscal,dz11); |
541 | |
542 | /* Update vectorial force */ |
543 | fix1 = _mm_add_ps(fix1,tx); |
544 | fiy1 = _mm_add_ps(fiy1,ty); |
545 | fiz1 = _mm_add_ps(fiz1,tz); |
546 | |
547 | fjx1 = _mm_add_ps(fjx1,tx); |
548 | fjy1 = _mm_add_ps(fjy1,ty); |
549 | fjz1 = _mm_add_ps(fjz1,tz); |
550 | |
551 | } |
552 | |
553 | /************************** |
554 | * CALCULATE INTERACTIONS * |
555 | **************************/ |
556 | |
557 | if (gmx_mm_any_lt(rsq12,rcutoff2)) |
558 | { |
559 | |
560 | r12 = _mm_mul_ps(rsq12,rinv12); |
561 | |
562 | /* EWALD ELECTROSTATICS */ |
563 | |
564 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
565 | ewrt = _mm_mul_ps(r12,ewtabscale); |
566 | ewitab = _mm_cvttps_epi32(ewrt); |
567 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
568 | ewitab = _mm_slli_epi32(ewitab,2); |
569 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
570 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
571 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
572 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
573 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
574 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
575 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
576 | velec = _mm_mul_ps(qq12,_mm_sub_ps(_mm_sub_ps(rinv12,sh_ewald),velec)); |
577 | felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec)); |
578 | |
579 | cutoff_mask = _mm_cmplt_ps(rsq12,rcutoff2); |
580 | |
581 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
582 | velec = _mm_and_ps(velec,cutoff_mask); |
583 | velecsum = _mm_add_ps(velecsum,velec); |
584 | |
585 | fscal = felec; |
586 | |
587 | fscal = _mm_and_ps(fscal,cutoff_mask); |
588 | |
589 | /* Calculate temporary vectorial force */ |
590 | tx = _mm_mul_ps(fscal,dx12); |
591 | ty = _mm_mul_ps(fscal,dy12); |
592 | tz = _mm_mul_ps(fscal,dz12); |
593 | |
594 | /* Update vectorial force */ |
595 | fix1 = _mm_add_ps(fix1,tx); |
596 | fiy1 = _mm_add_ps(fiy1,ty); |
597 | fiz1 = _mm_add_ps(fiz1,tz); |
598 | |
599 | fjx2 = _mm_add_ps(fjx2,tx); |
600 | fjy2 = _mm_add_ps(fjy2,ty); |
601 | fjz2 = _mm_add_ps(fjz2,tz); |
602 | |
603 | } |
604 | |
605 | /************************** |
606 | * CALCULATE INTERACTIONS * |
607 | **************************/ |
608 | |
609 | if (gmx_mm_any_lt(rsq20,rcutoff2)) |
610 | { |
611 | |
612 | r20 = _mm_mul_ps(rsq20,rinv20); |
613 | |
614 | /* EWALD ELECTROSTATICS */ |
615 | |
616 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
617 | ewrt = _mm_mul_ps(r20,ewtabscale); |
618 | ewitab = _mm_cvttps_epi32(ewrt); |
619 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
620 | ewitab = _mm_slli_epi32(ewitab,2); |
621 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
622 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
623 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
624 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
625 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
626 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
627 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
628 | velec = _mm_mul_ps(qq20,_mm_sub_ps(_mm_sub_ps(rinv20,sh_ewald),velec)); |
629 | felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec)); |
630 | |
631 | cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2); |
632 | |
633 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
634 | velec = _mm_and_ps(velec,cutoff_mask); |
635 | velecsum = _mm_add_ps(velecsum,velec); |
636 | |
637 | fscal = felec; |
638 | |
639 | fscal = _mm_and_ps(fscal,cutoff_mask); |
640 | |
641 | /* Calculate temporary vectorial force */ |
642 | tx = _mm_mul_ps(fscal,dx20); |
643 | ty = _mm_mul_ps(fscal,dy20); |
644 | tz = _mm_mul_ps(fscal,dz20); |
645 | |
646 | /* Update vectorial force */ |
647 | fix2 = _mm_add_ps(fix2,tx); |
648 | fiy2 = _mm_add_ps(fiy2,ty); |
649 | fiz2 = _mm_add_ps(fiz2,tz); |
650 | |
651 | fjx0 = _mm_add_ps(fjx0,tx); |
652 | fjy0 = _mm_add_ps(fjy0,ty); |
653 | fjz0 = _mm_add_ps(fjz0,tz); |
654 | |
655 | } |
656 | |
657 | /************************** |
658 | * CALCULATE INTERACTIONS * |
659 | **************************/ |
660 | |
661 | if (gmx_mm_any_lt(rsq21,rcutoff2)) |
662 | { |
663 | |
664 | r21 = _mm_mul_ps(rsq21,rinv21); |
665 | |
666 | /* EWALD ELECTROSTATICS */ |
667 | |
668 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
669 | ewrt = _mm_mul_ps(r21,ewtabscale); |
670 | ewitab = _mm_cvttps_epi32(ewrt); |
671 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
672 | ewitab = _mm_slli_epi32(ewitab,2); |
673 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
674 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
675 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
676 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
677 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
678 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
679 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
680 | velec = _mm_mul_ps(qq21,_mm_sub_ps(_mm_sub_ps(rinv21,sh_ewald),velec)); |
681 | felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec)); |
682 | |
683 | cutoff_mask = _mm_cmplt_ps(rsq21,rcutoff2); |
684 | |
685 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
686 | velec = _mm_and_ps(velec,cutoff_mask); |
687 | velecsum = _mm_add_ps(velecsum,velec); |
688 | |
689 | fscal = felec; |
690 | |
691 | fscal = _mm_and_ps(fscal,cutoff_mask); |
692 | |
693 | /* Calculate temporary vectorial force */ |
694 | tx = _mm_mul_ps(fscal,dx21); |
695 | ty = _mm_mul_ps(fscal,dy21); |
696 | tz = _mm_mul_ps(fscal,dz21); |
697 | |
698 | /* Update vectorial force */ |
699 | fix2 = _mm_add_ps(fix2,tx); |
700 | fiy2 = _mm_add_ps(fiy2,ty); |
701 | fiz2 = _mm_add_ps(fiz2,tz); |
702 | |
703 | fjx1 = _mm_add_ps(fjx1,tx); |
704 | fjy1 = _mm_add_ps(fjy1,ty); |
705 | fjz1 = _mm_add_ps(fjz1,tz); |
706 | |
707 | } |
708 | |
709 | /************************** |
710 | * CALCULATE INTERACTIONS * |
711 | **************************/ |
712 | |
713 | if (gmx_mm_any_lt(rsq22,rcutoff2)) |
714 | { |
715 | |
716 | r22 = _mm_mul_ps(rsq22,rinv22); |
717 | |
718 | /* EWALD ELECTROSTATICS */ |
719 | |
720 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
721 | ewrt = _mm_mul_ps(r22,ewtabscale); |
722 | ewitab = _mm_cvttps_epi32(ewrt); |
723 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
724 | ewitab = _mm_slli_epi32(ewitab,2); |
725 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
726 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
727 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
728 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
729 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
730 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
731 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
732 | velec = _mm_mul_ps(qq22,_mm_sub_ps(_mm_sub_ps(rinv22,sh_ewald),velec)); |
733 | felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec)); |
734 | |
735 | cutoff_mask = _mm_cmplt_ps(rsq22,rcutoff2); |
736 | |
737 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
738 | velec = _mm_and_ps(velec,cutoff_mask); |
739 | velecsum = _mm_add_ps(velecsum,velec); |
740 | |
741 | fscal = felec; |
742 | |
743 | fscal = _mm_and_ps(fscal,cutoff_mask); |
744 | |
745 | /* Calculate temporary vectorial force */ |
746 | tx = _mm_mul_ps(fscal,dx22); |
747 | ty = _mm_mul_ps(fscal,dy22); |
748 | tz = _mm_mul_ps(fscal,dz22); |
749 | |
750 | /* Update vectorial force */ |
751 | fix2 = _mm_add_ps(fix2,tx); |
752 | fiy2 = _mm_add_ps(fiy2,ty); |
753 | fiz2 = _mm_add_ps(fiz2,tz); |
754 | |
755 | fjx2 = _mm_add_ps(fjx2,tx); |
756 | fjy2 = _mm_add_ps(fjy2,ty); |
757 | fjz2 = _mm_add_ps(fjz2,tz); |
758 | |
759 | } |
760 | |
761 | fjptrA = f+j_coord_offsetA; |
762 | fjptrB = f+j_coord_offsetB; |
763 | fjptrC = f+j_coord_offsetC; |
764 | fjptrD = f+j_coord_offsetD; |
765 | |
766 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
767 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2); |
768 | |
769 | /* Inner loop uses 414 flops */ |
770 | } |
771 | |
772 | if(jidx<j_index_end) |
773 | { |
774 | |
775 | /* Get j neighbor index, and coordinate index */ |
776 | jnrlistA = jjnr[jidx]; |
777 | jnrlistB = jjnr[jidx+1]; |
778 | jnrlistC = jjnr[jidx+2]; |
779 | jnrlistD = jjnr[jidx+3]; |
780 | /* Sign of each element will be negative for non-real atoms. |
781 | * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones, |
782 | * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries. |
783 | */ |
784 | dummy_mask = gmx_mm_castsi128_ps_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())); |
785 | jnrA = (jnrlistA>=0) ? jnrlistA : 0; |
786 | jnrB = (jnrlistB>=0) ? jnrlistB : 0; |
787 | jnrC = (jnrlistC>=0) ? jnrlistC : 0; |
788 | jnrD = (jnrlistD>=0) ? jnrlistD : 0; |
789 | j_coord_offsetA = DIM3*jnrA; |
790 | j_coord_offsetB = DIM3*jnrB; |
791 | j_coord_offsetC = DIM3*jnrC; |
792 | j_coord_offsetD = DIM3*jnrD; |
793 | |
794 | /* load j atom coordinates */ |
795 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
796 | x+j_coord_offsetC,x+j_coord_offsetD, |
797 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2); |
798 | |
799 | /* Calculate displacement vector */ |
800 | dx00 = _mm_sub_ps(ix0,jx0); |
801 | dy00 = _mm_sub_ps(iy0,jy0); |
802 | dz00 = _mm_sub_ps(iz0,jz0); |
803 | dx01 = _mm_sub_ps(ix0,jx1); |
804 | dy01 = _mm_sub_ps(iy0,jy1); |
805 | dz01 = _mm_sub_ps(iz0,jz1); |
806 | dx02 = _mm_sub_ps(ix0,jx2); |
807 | dy02 = _mm_sub_ps(iy0,jy2); |
808 | dz02 = _mm_sub_ps(iz0,jz2); |
809 | dx10 = _mm_sub_ps(ix1,jx0); |
810 | dy10 = _mm_sub_ps(iy1,jy0); |
811 | dz10 = _mm_sub_ps(iz1,jz0); |
812 | dx11 = _mm_sub_ps(ix1,jx1); |
813 | dy11 = _mm_sub_ps(iy1,jy1); |
814 | dz11 = _mm_sub_ps(iz1,jz1); |
815 | dx12 = _mm_sub_ps(ix1,jx2); |
816 | dy12 = _mm_sub_ps(iy1,jy2); |
817 | dz12 = _mm_sub_ps(iz1,jz2); |
818 | dx20 = _mm_sub_ps(ix2,jx0); |
819 | dy20 = _mm_sub_ps(iy2,jy0); |
820 | dz20 = _mm_sub_ps(iz2,jz0); |
821 | dx21 = _mm_sub_ps(ix2,jx1); |
822 | dy21 = _mm_sub_ps(iy2,jy1); |
823 | dz21 = _mm_sub_ps(iz2,jz1); |
824 | dx22 = _mm_sub_ps(ix2,jx2); |
825 | dy22 = _mm_sub_ps(iy2,jy2); |
826 | dz22 = _mm_sub_ps(iz2,jz2); |
827 | |
828 | /* Calculate squared distance and things based on it */ |
829 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
830 | rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01); |
831 | rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02); |
832 | rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10); |
833 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
834 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
835 | rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20); |
836 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
837 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
838 | |
839 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
840 | rinv01 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq01); |
841 | rinv02 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq02); |
842 | rinv10 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq10); |
843 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
844 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
845 | rinv20 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq20); |
846 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
847 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
848 | |
849 | rinvsq00 = _mm_mul_ps(rinv00,rinv00); |
850 | rinvsq01 = _mm_mul_ps(rinv01,rinv01); |
851 | rinvsq02 = _mm_mul_ps(rinv02,rinv02); |
852 | rinvsq10 = _mm_mul_ps(rinv10,rinv10); |
853 | rinvsq11 = _mm_mul_ps(rinv11,rinv11); |
854 | rinvsq12 = _mm_mul_ps(rinv12,rinv12); |
855 | rinvsq20 = _mm_mul_ps(rinv20,rinv20); |
856 | rinvsq21 = _mm_mul_ps(rinv21,rinv21); |
857 | rinvsq22 = _mm_mul_ps(rinv22,rinv22); |
858 | |
859 | fjx0 = _mm_setzero_ps(); |
860 | fjy0 = _mm_setzero_ps(); |
861 | fjz0 = _mm_setzero_ps(); |
862 | fjx1 = _mm_setzero_ps(); |
863 | fjy1 = _mm_setzero_ps(); |
864 | fjz1 = _mm_setzero_ps(); |
865 | fjx2 = _mm_setzero_ps(); |
866 | fjy2 = _mm_setzero_ps(); |
867 | fjz2 = _mm_setzero_ps(); |
868 | |
869 | /************************** |
870 | * CALCULATE INTERACTIONS * |
871 | **************************/ |
872 | |
873 | if (gmx_mm_any_lt(rsq00,rcutoff2)) |
874 | { |
875 | |
876 | r00 = _mm_mul_ps(rsq00,rinv00); |
877 | r00 = _mm_andnot_ps(dummy_mask,r00); |
878 | |
879 | /* EWALD ELECTROSTATICS */ |
880 | |
881 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
882 | ewrt = _mm_mul_ps(r00,ewtabscale); |
883 | ewitab = _mm_cvttps_epi32(ewrt); |
884 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
885 | ewitab = _mm_slli_epi32(ewitab,2); |
886 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
887 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
888 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
889 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
890 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
891 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
892 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
893 | velec = _mm_mul_ps(qq00,_mm_sub_ps(_mm_sub_ps(rinv00,sh_ewald),velec)); |
894 | felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec)); |
895 | |
896 | cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2); |
897 | |
898 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
899 | velec = _mm_and_ps(velec,cutoff_mask); |
900 | velec = _mm_andnot_ps(dummy_mask,velec); |
901 | velecsum = _mm_add_ps(velecsum,velec); |
902 | |
903 | fscal = felec; |
904 | |
905 | fscal = _mm_and_ps(fscal,cutoff_mask); |
906 | |
907 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
908 | |
909 | /* Calculate temporary vectorial force */ |
910 | tx = _mm_mul_ps(fscal,dx00); |
911 | ty = _mm_mul_ps(fscal,dy00); |
912 | tz = _mm_mul_ps(fscal,dz00); |
913 | |
914 | /* Update vectorial force */ |
915 | fix0 = _mm_add_ps(fix0,tx); |
916 | fiy0 = _mm_add_ps(fiy0,ty); |
917 | fiz0 = _mm_add_ps(fiz0,tz); |
918 | |
919 | fjx0 = _mm_add_ps(fjx0,tx); |
920 | fjy0 = _mm_add_ps(fjy0,ty); |
921 | fjz0 = _mm_add_ps(fjz0,tz); |
922 | |
923 | } |
924 | |
925 | /************************** |
926 | * CALCULATE INTERACTIONS * |
927 | **************************/ |
928 | |
929 | if (gmx_mm_any_lt(rsq01,rcutoff2)) |
930 | { |
931 | |
932 | r01 = _mm_mul_ps(rsq01,rinv01); |
933 | r01 = _mm_andnot_ps(dummy_mask,r01); |
934 | |
935 | /* EWALD ELECTROSTATICS */ |
936 | |
937 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
938 | ewrt = _mm_mul_ps(r01,ewtabscale); |
939 | ewitab = _mm_cvttps_epi32(ewrt); |
940 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
941 | ewitab = _mm_slli_epi32(ewitab,2); |
942 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
943 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
944 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
945 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
946 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
947 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
948 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
949 | velec = _mm_mul_ps(qq01,_mm_sub_ps(_mm_sub_ps(rinv01,sh_ewald),velec)); |
950 | felec = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec)); |
951 | |
952 | cutoff_mask = _mm_cmplt_ps(rsq01,rcutoff2); |
953 | |
954 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
955 | velec = _mm_and_ps(velec,cutoff_mask); |
956 | velec = _mm_andnot_ps(dummy_mask,velec); |
957 | velecsum = _mm_add_ps(velecsum,velec); |
958 | |
959 | fscal = felec; |
960 | |
961 | fscal = _mm_and_ps(fscal,cutoff_mask); |
962 | |
963 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
964 | |
965 | /* Calculate temporary vectorial force */ |
966 | tx = _mm_mul_ps(fscal,dx01); |
967 | ty = _mm_mul_ps(fscal,dy01); |
968 | tz = _mm_mul_ps(fscal,dz01); |
969 | |
970 | /* Update vectorial force */ |
971 | fix0 = _mm_add_ps(fix0,tx); |
972 | fiy0 = _mm_add_ps(fiy0,ty); |
973 | fiz0 = _mm_add_ps(fiz0,tz); |
974 | |
975 | fjx1 = _mm_add_ps(fjx1,tx); |
976 | fjy1 = _mm_add_ps(fjy1,ty); |
977 | fjz1 = _mm_add_ps(fjz1,tz); |
978 | |
979 | } |
980 | |
981 | /************************** |
982 | * CALCULATE INTERACTIONS * |
983 | **************************/ |
984 | |
985 | if (gmx_mm_any_lt(rsq02,rcutoff2)) |
986 | { |
987 | |
988 | r02 = _mm_mul_ps(rsq02,rinv02); |
989 | r02 = _mm_andnot_ps(dummy_mask,r02); |
990 | |
991 | /* EWALD ELECTROSTATICS */ |
992 | |
993 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
994 | ewrt = _mm_mul_ps(r02,ewtabscale); |
995 | ewitab = _mm_cvttps_epi32(ewrt); |
996 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
997 | ewitab = _mm_slli_epi32(ewitab,2); |
998 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
999 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
1000 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
1001 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
1002 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
1003 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
1004 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
1005 | velec = _mm_mul_ps(qq02,_mm_sub_ps(_mm_sub_ps(rinv02,sh_ewald),velec)); |
1006 | felec = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec)); |
1007 | |
1008 | cutoff_mask = _mm_cmplt_ps(rsq02,rcutoff2); |
1009 | |
1010 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1011 | velec = _mm_and_ps(velec,cutoff_mask); |
1012 | velec = _mm_andnot_ps(dummy_mask,velec); |
1013 | velecsum = _mm_add_ps(velecsum,velec); |
1014 | |
1015 | fscal = felec; |
1016 | |
1017 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1018 | |
1019 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1020 | |
1021 | /* Calculate temporary vectorial force */ |
1022 | tx = _mm_mul_ps(fscal,dx02); |
1023 | ty = _mm_mul_ps(fscal,dy02); |
1024 | tz = _mm_mul_ps(fscal,dz02); |
1025 | |
1026 | /* Update vectorial force */ |
1027 | fix0 = _mm_add_ps(fix0,tx); |
1028 | fiy0 = _mm_add_ps(fiy0,ty); |
1029 | fiz0 = _mm_add_ps(fiz0,tz); |
1030 | |
1031 | fjx2 = _mm_add_ps(fjx2,tx); |
1032 | fjy2 = _mm_add_ps(fjy2,ty); |
1033 | fjz2 = _mm_add_ps(fjz2,tz); |
1034 | |
1035 | } |
1036 | |
1037 | /************************** |
1038 | * CALCULATE INTERACTIONS * |
1039 | **************************/ |
1040 | |
1041 | if (gmx_mm_any_lt(rsq10,rcutoff2)) |
1042 | { |
1043 | |
1044 | r10 = _mm_mul_ps(rsq10,rinv10); |
1045 | r10 = _mm_andnot_ps(dummy_mask,r10); |
1046 | |
1047 | /* EWALD ELECTROSTATICS */ |
1048 | |
1049 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1050 | ewrt = _mm_mul_ps(r10,ewtabscale); |
1051 | ewitab = _mm_cvttps_epi32(ewrt); |
1052 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1053 | ewitab = _mm_slli_epi32(ewitab,2); |
1054 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
1055 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
1056 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
1057 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
1058 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
1059 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
1060 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
1061 | velec = _mm_mul_ps(qq10,_mm_sub_ps(_mm_sub_ps(rinv10,sh_ewald),velec)); |
1062 | felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec)); |
1063 | |
1064 | cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2); |
1065 | |
1066 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1067 | velec = _mm_and_ps(velec,cutoff_mask); |
1068 | velec = _mm_andnot_ps(dummy_mask,velec); |
1069 | velecsum = _mm_add_ps(velecsum,velec); |
1070 | |
1071 | fscal = felec; |
1072 | |
1073 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1074 | |
1075 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1076 | |
1077 | /* Calculate temporary vectorial force */ |
1078 | tx = _mm_mul_ps(fscal,dx10); |
1079 | ty = _mm_mul_ps(fscal,dy10); |
1080 | tz = _mm_mul_ps(fscal,dz10); |
1081 | |
1082 | /* Update vectorial force */ |
1083 | fix1 = _mm_add_ps(fix1,tx); |
1084 | fiy1 = _mm_add_ps(fiy1,ty); |
1085 | fiz1 = _mm_add_ps(fiz1,tz); |
1086 | |
1087 | fjx0 = _mm_add_ps(fjx0,tx); |
1088 | fjy0 = _mm_add_ps(fjy0,ty); |
1089 | fjz0 = _mm_add_ps(fjz0,tz); |
1090 | |
1091 | } |
1092 | |
1093 | /************************** |
1094 | * CALCULATE INTERACTIONS * |
1095 | **************************/ |
1096 | |
1097 | if (gmx_mm_any_lt(rsq11,rcutoff2)) |
1098 | { |
1099 | |
1100 | r11 = _mm_mul_ps(rsq11,rinv11); |
1101 | r11 = _mm_andnot_ps(dummy_mask,r11); |
1102 | |
1103 | /* EWALD ELECTROSTATICS */ |
1104 | |
1105 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1106 | ewrt = _mm_mul_ps(r11,ewtabscale); |
1107 | ewitab = _mm_cvttps_epi32(ewrt); |
1108 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1109 | ewitab = _mm_slli_epi32(ewitab,2); |
1110 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
1111 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
1112 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
1113 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
1114 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
1115 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
1116 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
1117 | velec = _mm_mul_ps(qq11,_mm_sub_ps(_mm_sub_ps(rinv11,sh_ewald),velec)); |
1118 | felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec)); |
1119 | |
1120 | cutoff_mask = _mm_cmplt_ps(rsq11,rcutoff2); |
1121 | |
1122 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1123 | velec = _mm_and_ps(velec,cutoff_mask); |
1124 | velec = _mm_andnot_ps(dummy_mask,velec); |
1125 | velecsum = _mm_add_ps(velecsum,velec); |
1126 | |
1127 | fscal = felec; |
1128 | |
1129 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1130 | |
1131 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1132 | |
1133 | /* Calculate temporary vectorial force */ |
1134 | tx = _mm_mul_ps(fscal,dx11); |
1135 | ty = _mm_mul_ps(fscal,dy11); |
1136 | tz = _mm_mul_ps(fscal,dz11); |
1137 | |
1138 | /* Update vectorial force */ |
1139 | fix1 = _mm_add_ps(fix1,tx); |
1140 | fiy1 = _mm_add_ps(fiy1,ty); |
1141 | fiz1 = _mm_add_ps(fiz1,tz); |
1142 | |
1143 | fjx1 = _mm_add_ps(fjx1,tx); |
1144 | fjy1 = _mm_add_ps(fjy1,ty); |
1145 | fjz1 = _mm_add_ps(fjz1,tz); |
1146 | |
1147 | } |
1148 | |
1149 | /************************** |
1150 | * CALCULATE INTERACTIONS * |
1151 | **************************/ |
1152 | |
1153 | if (gmx_mm_any_lt(rsq12,rcutoff2)) |
1154 | { |
1155 | |
1156 | r12 = _mm_mul_ps(rsq12,rinv12); |
1157 | r12 = _mm_andnot_ps(dummy_mask,r12); |
1158 | |
1159 | /* EWALD ELECTROSTATICS */ |
1160 | |
1161 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1162 | ewrt = _mm_mul_ps(r12,ewtabscale); |
1163 | ewitab = _mm_cvttps_epi32(ewrt); |
1164 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1165 | ewitab = _mm_slli_epi32(ewitab,2); |
1166 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
1167 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
1168 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
1169 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
1170 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
1171 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
1172 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
1173 | velec = _mm_mul_ps(qq12,_mm_sub_ps(_mm_sub_ps(rinv12,sh_ewald),velec)); |
1174 | felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec)); |
1175 | |
1176 | cutoff_mask = _mm_cmplt_ps(rsq12,rcutoff2); |
1177 | |
1178 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1179 | velec = _mm_and_ps(velec,cutoff_mask); |
1180 | velec = _mm_andnot_ps(dummy_mask,velec); |
1181 | velecsum = _mm_add_ps(velecsum,velec); |
1182 | |
1183 | fscal = felec; |
1184 | |
1185 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1186 | |
1187 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1188 | |
1189 | /* Calculate temporary vectorial force */ |
1190 | tx = _mm_mul_ps(fscal,dx12); |
1191 | ty = _mm_mul_ps(fscal,dy12); |
1192 | tz = _mm_mul_ps(fscal,dz12); |
1193 | |
1194 | /* Update vectorial force */ |
1195 | fix1 = _mm_add_ps(fix1,tx); |
1196 | fiy1 = _mm_add_ps(fiy1,ty); |
1197 | fiz1 = _mm_add_ps(fiz1,tz); |
1198 | |
1199 | fjx2 = _mm_add_ps(fjx2,tx); |
1200 | fjy2 = _mm_add_ps(fjy2,ty); |
1201 | fjz2 = _mm_add_ps(fjz2,tz); |
1202 | |
1203 | } |
1204 | |
1205 | /************************** |
1206 | * CALCULATE INTERACTIONS * |
1207 | **************************/ |
1208 | |
1209 | if (gmx_mm_any_lt(rsq20,rcutoff2)) |
1210 | { |
1211 | |
1212 | r20 = _mm_mul_ps(rsq20,rinv20); |
1213 | r20 = _mm_andnot_ps(dummy_mask,r20); |
1214 | |
1215 | /* EWALD ELECTROSTATICS */ |
1216 | |
1217 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1218 | ewrt = _mm_mul_ps(r20,ewtabscale); |
1219 | ewitab = _mm_cvttps_epi32(ewrt); |
1220 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1221 | ewitab = _mm_slli_epi32(ewitab,2); |
1222 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
1223 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
1224 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
1225 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
1226 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
1227 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
1228 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
1229 | velec = _mm_mul_ps(qq20,_mm_sub_ps(_mm_sub_ps(rinv20,sh_ewald),velec)); |
1230 | felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec)); |
1231 | |
1232 | cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2); |
1233 | |
1234 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1235 | velec = _mm_and_ps(velec,cutoff_mask); |
1236 | velec = _mm_andnot_ps(dummy_mask,velec); |
1237 | velecsum = _mm_add_ps(velecsum,velec); |
1238 | |
1239 | fscal = felec; |
1240 | |
1241 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1242 | |
1243 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1244 | |
1245 | /* Calculate temporary vectorial force */ |
1246 | tx = _mm_mul_ps(fscal,dx20); |
1247 | ty = _mm_mul_ps(fscal,dy20); |
1248 | tz = _mm_mul_ps(fscal,dz20); |
1249 | |
1250 | /* Update vectorial force */ |
1251 | fix2 = _mm_add_ps(fix2,tx); |
1252 | fiy2 = _mm_add_ps(fiy2,ty); |
1253 | fiz2 = _mm_add_ps(fiz2,tz); |
1254 | |
1255 | fjx0 = _mm_add_ps(fjx0,tx); |
1256 | fjy0 = _mm_add_ps(fjy0,ty); |
1257 | fjz0 = _mm_add_ps(fjz0,tz); |
1258 | |
1259 | } |
1260 | |
1261 | /************************** |
1262 | * CALCULATE INTERACTIONS * |
1263 | **************************/ |
1264 | |
1265 | if (gmx_mm_any_lt(rsq21,rcutoff2)) |
1266 | { |
1267 | |
1268 | r21 = _mm_mul_ps(rsq21,rinv21); |
1269 | r21 = _mm_andnot_ps(dummy_mask,r21); |
1270 | |
1271 | /* EWALD ELECTROSTATICS */ |
1272 | |
1273 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1274 | ewrt = _mm_mul_ps(r21,ewtabscale); |
1275 | ewitab = _mm_cvttps_epi32(ewrt); |
1276 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1277 | ewitab = _mm_slli_epi32(ewitab,2); |
1278 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
1279 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
1280 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
1281 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
1282 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
1283 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
1284 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
1285 | velec = _mm_mul_ps(qq21,_mm_sub_ps(_mm_sub_ps(rinv21,sh_ewald),velec)); |
1286 | felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec)); |
1287 | |
1288 | cutoff_mask = _mm_cmplt_ps(rsq21,rcutoff2); |
1289 | |
1290 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1291 | velec = _mm_and_ps(velec,cutoff_mask); |
1292 | velec = _mm_andnot_ps(dummy_mask,velec); |
1293 | velecsum = _mm_add_ps(velecsum,velec); |
1294 | |
1295 | fscal = felec; |
1296 | |
1297 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1298 | |
1299 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1300 | |
1301 | /* Calculate temporary vectorial force */ |
1302 | tx = _mm_mul_ps(fscal,dx21); |
1303 | ty = _mm_mul_ps(fscal,dy21); |
1304 | tz = _mm_mul_ps(fscal,dz21); |
1305 | |
1306 | /* Update vectorial force */ |
1307 | fix2 = _mm_add_ps(fix2,tx); |
1308 | fiy2 = _mm_add_ps(fiy2,ty); |
1309 | fiz2 = _mm_add_ps(fiz2,tz); |
1310 | |
1311 | fjx1 = _mm_add_ps(fjx1,tx); |
1312 | fjy1 = _mm_add_ps(fjy1,ty); |
1313 | fjz1 = _mm_add_ps(fjz1,tz); |
1314 | |
1315 | } |
1316 | |
1317 | /************************** |
1318 | * CALCULATE INTERACTIONS * |
1319 | **************************/ |
1320 | |
1321 | if (gmx_mm_any_lt(rsq22,rcutoff2)) |
1322 | { |
1323 | |
1324 | r22 = _mm_mul_ps(rsq22,rinv22); |
1325 | r22 = _mm_andnot_ps(dummy_mask,r22); |
1326 | |
1327 | /* EWALD ELECTROSTATICS */ |
1328 | |
1329 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1330 | ewrt = _mm_mul_ps(r22,ewtabscale); |
1331 | ewitab = _mm_cvttps_epi32(ewrt); |
1332 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1333 | ewitab = _mm_slli_epi32(ewitab,2); |
1334 | ewtabF = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})) ); |
1335 | ewtabD = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})) ); |
1336 | ewtabV = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})) ); |
1337 | ewtabFn = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})) ); |
1338 | _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn)do { __m128 tmp3, tmp2, tmp1, tmp0; tmp0 = _mm_unpacklo_ps((ewtabF ), (ewtabD)); tmp2 = _mm_unpacklo_ps((ewtabV), (ewtabFn)); tmp1 = _mm_unpackhi_ps((ewtabF), (ewtabD)); tmp3 = _mm_unpackhi_ps ((ewtabV), (ewtabFn)); (ewtabF) = _mm_movelh_ps(tmp0, tmp2); ( ewtabD) = _mm_movehl_ps(tmp2, tmp0); (ewtabV) = _mm_movelh_ps (tmp1, tmp3); (ewtabFn) = _mm_movehl_ps(tmp3, tmp1); } while ( 0); |
1339 | felec = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD)); |
1340 | velec = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec))); |
1341 | velec = _mm_mul_ps(qq22,_mm_sub_ps(_mm_sub_ps(rinv22,sh_ewald),velec)); |
1342 | felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec)); |
1343 | |
1344 | cutoff_mask = _mm_cmplt_ps(rsq22,rcutoff2); |
1345 | |
1346 | /* Update potential sum for this i atom from the interaction with this j atom. */ |
1347 | velec = _mm_and_ps(velec,cutoff_mask); |
1348 | velec = _mm_andnot_ps(dummy_mask,velec); |
1349 | velecsum = _mm_add_ps(velecsum,velec); |
1350 | |
1351 | fscal = felec; |
1352 | |
1353 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1354 | |
1355 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
1356 | |
1357 | /* Calculate temporary vectorial force */ |
1358 | tx = _mm_mul_ps(fscal,dx22); |
1359 | ty = _mm_mul_ps(fscal,dy22); |
1360 | tz = _mm_mul_ps(fscal,dz22); |
1361 | |
1362 | /* Update vectorial force */ |
1363 | fix2 = _mm_add_ps(fix2,tx); |
1364 | fiy2 = _mm_add_ps(fiy2,ty); |
1365 | fiz2 = _mm_add_ps(fiz2,tz); |
1366 | |
1367 | fjx2 = _mm_add_ps(fjx2,tx); |
1368 | fjy2 = _mm_add_ps(fjy2,ty); |
1369 | fjz2 = _mm_add_ps(fjz2,tz); |
1370 | |
1371 | } |
1372 | |
1373 | fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch; |
1374 | fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch; |
1375 | fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch; |
1376 | fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch; |
1377 | |
1378 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
1379 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2); |
1380 | |
1381 | /* Inner loop uses 423 flops */ |
1382 | } |
1383 | |
1384 | /* End of innermost loop */ |
1385 | |
1386 | gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2, |
1387 | f+i_coord_offset,fshift+i_shift_offset); |
1388 | |
1389 | ggid = gid[iidx]; |
1390 | /* Update potential energies */ |
1391 | gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid); |
1392 | |
1393 | /* Increment number of inner iterations */ |
1394 | inneriter += j_index_end - j_index_start; |
1395 | |
1396 | /* Outer loop uses 19 flops */ |
1397 | } |
1398 | |
1399 | /* Increment number of outer iterations */ |
1400 | outeriter += nri; |
1401 | |
1402 | /* Update outer/inner flops */ |
1403 | |
1404 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3W3_VF,outeriter*19 + inneriter*423)(nrnb)->n[eNR_NBKERNEL_ELEC_W3W3_VF] += outeriter*19 + inneriter *423; |
1405 | } |
1406 | /* |
1407 | * Gromacs nonbonded kernel: nb_kernel_ElecEwSh_VdwNone_GeomW3W3_F_sse4_1_single |
1408 | * Electrostatics interaction: Ewald |
1409 | * VdW interaction: None |
1410 | * Geometry: Water3-Water3 |
1411 | * Calculate force/pot: Force |
1412 | */ |
1413 | void |
1414 | nb_kernel_ElecEwSh_VdwNone_GeomW3W3_F_sse4_1_single |
1415 | (t_nblist * gmx_restrict nlist, |
1416 | rvec * gmx_restrict xx, |
1417 | rvec * gmx_restrict ff, |
1418 | t_forcerec * gmx_restrict fr, |
1419 | t_mdatoms * gmx_restrict mdatoms, |
1420 | nb_kernel_data_t gmx_unused__attribute__ ((unused)) * gmx_restrict kernel_data, |
1421 | t_nrnb * gmx_restrict nrnb) |
1422 | { |
1423 | /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or |
1424 | * just 0 for non-waters. |
1425 | * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different |
1426 | * jnr indices corresponding to data put in the four positions in the SIMD register. |
1427 | */ |
1428 | int i_shift_offset,i_coord_offset,outeriter,inneriter; |
1429 | int j_index_start,j_index_end,jidx,nri,inr,ggid,iidx; |
1430 | int jnrA,jnrB,jnrC,jnrD; |
1431 | int jnrlistA,jnrlistB,jnrlistC,jnrlistD; |
1432 | int j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD; |
1433 | int *iinr,*jindex,*jjnr,*shiftidx,*gid; |
1434 | real rcutoff_scalar; |
1435 | real *shiftvec,*fshift,*x,*f; |
1436 | real *fjptrA,*fjptrB,*fjptrC,*fjptrD; |
1437 | real scratch[4*DIM3]; |
1438 | __m128 tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall; |
1439 | int vdwioffset0; |
1440 | __m128 ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0; |
1441 | int vdwioffset1; |
1442 | __m128 ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1; |
1443 | int vdwioffset2; |
1444 | __m128 ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2; |
1445 | int vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D; |
1446 | __m128 jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0; |
1447 | int vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D; |
1448 | __m128 jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1; |
1449 | int vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D; |
1450 | __m128 jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2; |
1451 | __m128 dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00; |
1452 | __m128 dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01; |
1453 | __m128 dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02; |
1454 | __m128 dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10; |
1455 | __m128 dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11; |
1456 | __m128 dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12; |
1457 | __m128 dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20; |
1458 | __m128 dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21; |
1459 | __m128 dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22; |
1460 | __m128 velec,felec,velecsum,facel,crf,krf,krf2; |
1461 | real *charge; |
1462 | __m128i ewitab; |
1463 | __m128 ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV; |
1464 | real *ewtab; |
1465 | __m128 dummy_mask,cutoff_mask; |
1466 | __m128 signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) ); |
Value stored to 'signbit' during its initialization is never read | |
1467 | __m128 one = _mm_set1_ps(1.0); |
1468 | __m128 two = _mm_set1_ps(2.0); |
1469 | x = xx[0]; |
1470 | f = ff[0]; |
1471 | |
1472 | nri = nlist->nri; |
1473 | iinr = nlist->iinr; |
1474 | jindex = nlist->jindex; |
1475 | jjnr = nlist->jjnr; |
1476 | shiftidx = nlist->shift; |
1477 | gid = nlist->gid; |
1478 | shiftvec = fr->shift_vec[0]; |
1479 | fshift = fr->fshift[0]; |
1480 | facel = _mm_set1_ps(fr->epsfac); |
1481 | charge = mdatoms->chargeA; |
1482 | |
1483 | sh_ewald = _mm_set1_ps(fr->ic->sh_ewald); |
1484 | ewtab = fr->ic->tabq_coul_F; |
1485 | ewtabscale = _mm_set1_ps(fr->ic->tabq_scale); |
1486 | ewtabhalfspace = _mm_set1_ps(0.5/fr->ic->tabq_scale); |
1487 | |
1488 | /* Setup water-specific parameters */ |
1489 | inr = nlist->iinr[0]; |
1490 | iq0 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0])); |
1491 | iq1 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1])); |
1492 | iq2 = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2])); |
1493 | |
1494 | jq0 = _mm_set1_ps(charge[inr+0]); |
1495 | jq1 = _mm_set1_ps(charge[inr+1]); |
1496 | jq2 = _mm_set1_ps(charge[inr+2]); |
1497 | qq00 = _mm_mul_ps(iq0,jq0); |
1498 | qq01 = _mm_mul_ps(iq0,jq1); |
1499 | qq02 = _mm_mul_ps(iq0,jq2); |
1500 | qq10 = _mm_mul_ps(iq1,jq0); |
1501 | qq11 = _mm_mul_ps(iq1,jq1); |
1502 | qq12 = _mm_mul_ps(iq1,jq2); |
1503 | qq20 = _mm_mul_ps(iq2,jq0); |
1504 | qq21 = _mm_mul_ps(iq2,jq1); |
1505 | qq22 = _mm_mul_ps(iq2,jq2); |
1506 | |
1507 | /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */ |
1508 | rcutoff_scalar = fr->rcoulomb; |
1509 | rcutoff = _mm_set1_ps(rcutoff_scalar); |
1510 | rcutoff2 = _mm_mul_ps(rcutoff,rcutoff); |
1511 | |
1512 | /* Avoid stupid compiler warnings */ |
1513 | jnrA = jnrB = jnrC = jnrD = 0; |
1514 | j_coord_offsetA = 0; |
1515 | j_coord_offsetB = 0; |
1516 | j_coord_offsetC = 0; |
1517 | j_coord_offsetD = 0; |
1518 | |
1519 | outeriter = 0; |
1520 | inneriter = 0; |
1521 | |
1522 | for(iidx=0;iidx<4*DIM3;iidx++) |
1523 | { |
1524 | scratch[iidx] = 0.0; |
1525 | } |
1526 | |
1527 | /* Start outer loop over neighborlists */ |
1528 | for(iidx=0; iidx<nri; iidx++) |
1529 | { |
1530 | /* Load shift vector for this list */ |
1531 | i_shift_offset = DIM3*shiftidx[iidx]; |
1532 | |
1533 | /* Load limits for loop over neighbors */ |
1534 | j_index_start = jindex[iidx]; |
1535 | j_index_end = jindex[iidx+1]; |
1536 | |
1537 | /* Get outer coordinate index */ |
1538 | inr = iinr[iidx]; |
1539 | i_coord_offset = DIM3*inr; |
1540 | |
1541 | /* Load i particle coords and add shift vector */ |
1542 | gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset, |
1543 | &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2); |
1544 | |
1545 | fix0 = _mm_setzero_ps(); |
1546 | fiy0 = _mm_setzero_ps(); |
1547 | fiz0 = _mm_setzero_ps(); |
1548 | fix1 = _mm_setzero_ps(); |
1549 | fiy1 = _mm_setzero_ps(); |
1550 | fiz1 = _mm_setzero_ps(); |
1551 | fix2 = _mm_setzero_ps(); |
1552 | fiy2 = _mm_setzero_ps(); |
1553 | fiz2 = _mm_setzero_ps(); |
1554 | |
1555 | /* Start inner kernel loop */ |
1556 | for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4) |
1557 | { |
1558 | |
1559 | /* Get j neighbor index, and coordinate index */ |
1560 | jnrA = jjnr[jidx]; |
1561 | jnrB = jjnr[jidx+1]; |
1562 | jnrC = jjnr[jidx+2]; |
1563 | jnrD = jjnr[jidx+3]; |
1564 | j_coord_offsetA = DIM3*jnrA; |
1565 | j_coord_offsetB = DIM3*jnrB; |
1566 | j_coord_offsetC = DIM3*jnrC; |
1567 | j_coord_offsetD = DIM3*jnrD; |
1568 | |
1569 | /* load j atom coordinates */ |
1570 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
1571 | x+j_coord_offsetC,x+j_coord_offsetD, |
1572 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2); |
1573 | |
1574 | /* Calculate displacement vector */ |
1575 | dx00 = _mm_sub_ps(ix0,jx0); |
1576 | dy00 = _mm_sub_ps(iy0,jy0); |
1577 | dz00 = _mm_sub_ps(iz0,jz0); |
1578 | dx01 = _mm_sub_ps(ix0,jx1); |
1579 | dy01 = _mm_sub_ps(iy0,jy1); |
1580 | dz01 = _mm_sub_ps(iz0,jz1); |
1581 | dx02 = _mm_sub_ps(ix0,jx2); |
1582 | dy02 = _mm_sub_ps(iy0,jy2); |
1583 | dz02 = _mm_sub_ps(iz0,jz2); |
1584 | dx10 = _mm_sub_ps(ix1,jx0); |
1585 | dy10 = _mm_sub_ps(iy1,jy0); |
1586 | dz10 = _mm_sub_ps(iz1,jz0); |
1587 | dx11 = _mm_sub_ps(ix1,jx1); |
1588 | dy11 = _mm_sub_ps(iy1,jy1); |
1589 | dz11 = _mm_sub_ps(iz1,jz1); |
1590 | dx12 = _mm_sub_ps(ix1,jx2); |
1591 | dy12 = _mm_sub_ps(iy1,jy2); |
1592 | dz12 = _mm_sub_ps(iz1,jz2); |
1593 | dx20 = _mm_sub_ps(ix2,jx0); |
1594 | dy20 = _mm_sub_ps(iy2,jy0); |
1595 | dz20 = _mm_sub_ps(iz2,jz0); |
1596 | dx21 = _mm_sub_ps(ix2,jx1); |
1597 | dy21 = _mm_sub_ps(iy2,jy1); |
1598 | dz21 = _mm_sub_ps(iz2,jz1); |
1599 | dx22 = _mm_sub_ps(ix2,jx2); |
1600 | dy22 = _mm_sub_ps(iy2,jy2); |
1601 | dz22 = _mm_sub_ps(iz2,jz2); |
1602 | |
1603 | /* Calculate squared distance and things based on it */ |
1604 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
1605 | rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01); |
1606 | rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02); |
1607 | rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10); |
1608 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
1609 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
1610 | rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20); |
1611 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
1612 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
1613 | |
1614 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
1615 | rinv01 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq01); |
1616 | rinv02 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq02); |
1617 | rinv10 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq10); |
1618 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
1619 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
1620 | rinv20 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq20); |
1621 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
1622 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
1623 | |
1624 | rinvsq00 = _mm_mul_ps(rinv00,rinv00); |
1625 | rinvsq01 = _mm_mul_ps(rinv01,rinv01); |
1626 | rinvsq02 = _mm_mul_ps(rinv02,rinv02); |
1627 | rinvsq10 = _mm_mul_ps(rinv10,rinv10); |
1628 | rinvsq11 = _mm_mul_ps(rinv11,rinv11); |
1629 | rinvsq12 = _mm_mul_ps(rinv12,rinv12); |
1630 | rinvsq20 = _mm_mul_ps(rinv20,rinv20); |
1631 | rinvsq21 = _mm_mul_ps(rinv21,rinv21); |
1632 | rinvsq22 = _mm_mul_ps(rinv22,rinv22); |
1633 | |
1634 | fjx0 = _mm_setzero_ps(); |
1635 | fjy0 = _mm_setzero_ps(); |
1636 | fjz0 = _mm_setzero_ps(); |
1637 | fjx1 = _mm_setzero_ps(); |
1638 | fjy1 = _mm_setzero_ps(); |
1639 | fjz1 = _mm_setzero_ps(); |
1640 | fjx2 = _mm_setzero_ps(); |
1641 | fjy2 = _mm_setzero_ps(); |
1642 | fjz2 = _mm_setzero_ps(); |
1643 | |
1644 | /************************** |
1645 | * CALCULATE INTERACTIONS * |
1646 | **************************/ |
1647 | |
1648 | if (gmx_mm_any_lt(rsq00,rcutoff2)) |
1649 | { |
1650 | |
1651 | r00 = _mm_mul_ps(rsq00,rinv00); |
1652 | |
1653 | /* EWALD ELECTROSTATICS */ |
1654 | |
1655 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1656 | ewrt = _mm_mul_ps(r00,ewtabscale); |
1657 | ewitab = _mm_cvttps_epi32(ewrt); |
1658 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1659 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1660 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1661 | &ewtabF,&ewtabFn); |
1662 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1663 | felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec)); |
1664 | |
1665 | cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2); |
1666 | |
1667 | fscal = felec; |
1668 | |
1669 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1670 | |
1671 | /* Calculate temporary vectorial force */ |
1672 | tx = _mm_mul_ps(fscal,dx00); |
1673 | ty = _mm_mul_ps(fscal,dy00); |
1674 | tz = _mm_mul_ps(fscal,dz00); |
1675 | |
1676 | /* Update vectorial force */ |
1677 | fix0 = _mm_add_ps(fix0,tx); |
1678 | fiy0 = _mm_add_ps(fiy0,ty); |
1679 | fiz0 = _mm_add_ps(fiz0,tz); |
1680 | |
1681 | fjx0 = _mm_add_ps(fjx0,tx); |
1682 | fjy0 = _mm_add_ps(fjy0,ty); |
1683 | fjz0 = _mm_add_ps(fjz0,tz); |
1684 | |
1685 | } |
1686 | |
1687 | /************************** |
1688 | * CALCULATE INTERACTIONS * |
1689 | **************************/ |
1690 | |
1691 | if (gmx_mm_any_lt(rsq01,rcutoff2)) |
1692 | { |
1693 | |
1694 | r01 = _mm_mul_ps(rsq01,rinv01); |
1695 | |
1696 | /* EWALD ELECTROSTATICS */ |
1697 | |
1698 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1699 | ewrt = _mm_mul_ps(r01,ewtabscale); |
1700 | ewitab = _mm_cvttps_epi32(ewrt); |
1701 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1702 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1703 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1704 | &ewtabF,&ewtabFn); |
1705 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1706 | felec = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec)); |
1707 | |
1708 | cutoff_mask = _mm_cmplt_ps(rsq01,rcutoff2); |
1709 | |
1710 | fscal = felec; |
1711 | |
1712 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1713 | |
1714 | /* Calculate temporary vectorial force */ |
1715 | tx = _mm_mul_ps(fscal,dx01); |
1716 | ty = _mm_mul_ps(fscal,dy01); |
1717 | tz = _mm_mul_ps(fscal,dz01); |
1718 | |
1719 | /* Update vectorial force */ |
1720 | fix0 = _mm_add_ps(fix0,tx); |
1721 | fiy0 = _mm_add_ps(fiy0,ty); |
1722 | fiz0 = _mm_add_ps(fiz0,tz); |
1723 | |
1724 | fjx1 = _mm_add_ps(fjx1,tx); |
1725 | fjy1 = _mm_add_ps(fjy1,ty); |
1726 | fjz1 = _mm_add_ps(fjz1,tz); |
1727 | |
1728 | } |
1729 | |
1730 | /************************** |
1731 | * CALCULATE INTERACTIONS * |
1732 | **************************/ |
1733 | |
1734 | if (gmx_mm_any_lt(rsq02,rcutoff2)) |
1735 | { |
1736 | |
1737 | r02 = _mm_mul_ps(rsq02,rinv02); |
1738 | |
1739 | /* EWALD ELECTROSTATICS */ |
1740 | |
1741 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1742 | ewrt = _mm_mul_ps(r02,ewtabscale); |
1743 | ewitab = _mm_cvttps_epi32(ewrt); |
1744 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1745 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1746 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1747 | &ewtabF,&ewtabFn); |
1748 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1749 | felec = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec)); |
1750 | |
1751 | cutoff_mask = _mm_cmplt_ps(rsq02,rcutoff2); |
1752 | |
1753 | fscal = felec; |
1754 | |
1755 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1756 | |
1757 | /* Calculate temporary vectorial force */ |
1758 | tx = _mm_mul_ps(fscal,dx02); |
1759 | ty = _mm_mul_ps(fscal,dy02); |
1760 | tz = _mm_mul_ps(fscal,dz02); |
1761 | |
1762 | /* Update vectorial force */ |
1763 | fix0 = _mm_add_ps(fix0,tx); |
1764 | fiy0 = _mm_add_ps(fiy0,ty); |
1765 | fiz0 = _mm_add_ps(fiz0,tz); |
1766 | |
1767 | fjx2 = _mm_add_ps(fjx2,tx); |
1768 | fjy2 = _mm_add_ps(fjy2,ty); |
1769 | fjz2 = _mm_add_ps(fjz2,tz); |
1770 | |
1771 | } |
1772 | |
1773 | /************************** |
1774 | * CALCULATE INTERACTIONS * |
1775 | **************************/ |
1776 | |
1777 | if (gmx_mm_any_lt(rsq10,rcutoff2)) |
1778 | { |
1779 | |
1780 | r10 = _mm_mul_ps(rsq10,rinv10); |
1781 | |
1782 | /* EWALD ELECTROSTATICS */ |
1783 | |
1784 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1785 | ewrt = _mm_mul_ps(r10,ewtabscale); |
1786 | ewitab = _mm_cvttps_epi32(ewrt); |
1787 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1788 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1789 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1790 | &ewtabF,&ewtabFn); |
1791 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1792 | felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec)); |
1793 | |
1794 | cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2); |
1795 | |
1796 | fscal = felec; |
1797 | |
1798 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1799 | |
1800 | /* Calculate temporary vectorial force */ |
1801 | tx = _mm_mul_ps(fscal,dx10); |
1802 | ty = _mm_mul_ps(fscal,dy10); |
1803 | tz = _mm_mul_ps(fscal,dz10); |
1804 | |
1805 | /* Update vectorial force */ |
1806 | fix1 = _mm_add_ps(fix1,tx); |
1807 | fiy1 = _mm_add_ps(fiy1,ty); |
1808 | fiz1 = _mm_add_ps(fiz1,tz); |
1809 | |
1810 | fjx0 = _mm_add_ps(fjx0,tx); |
1811 | fjy0 = _mm_add_ps(fjy0,ty); |
1812 | fjz0 = _mm_add_ps(fjz0,tz); |
1813 | |
1814 | } |
1815 | |
1816 | /************************** |
1817 | * CALCULATE INTERACTIONS * |
1818 | **************************/ |
1819 | |
1820 | if (gmx_mm_any_lt(rsq11,rcutoff2)) |
1821 | { |
1822 | |
1823 | r11 = _mm_mul_ps(rsq11,rinv11); |
1824 | |
1825 | /* EWALD ELECTROSTATICS */ |
1826 | |
1827 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1828 | ewrt = _mm_mul_ps(r11,ewtabscale); |
1829 | ewitab = _mm_cvttps_epi32(ewrt); |
1830 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1831 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1832 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1833 | &ewtabF,&ewtabFn); |
1834 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1835 | felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec)); |
1836 | |
1837 | cutoff_mask = _mm_cmplt_ps(rsq11,rcutoff2); |
1838 | |
1839 | fscal = felec; |
1840 | |
1841 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1842 | |
1843 | /* Calculate temporary vectorial force */ |
1844 | tx = _mm_mul_ps(fscal,dx11); |
1845 | ty = _mm_mul_ps(fscal,dy11); |
1846 | tz = _mm_mul_ps(fscal,dz11); |
1847 | |
1848 | /* Update vectorial force */ |
1849 | fix1 = _mm_add_ps(fix1,tx); |
1850 | fiy1 = _mm_add_ps(fiy1,ty); |
1851 | fiz1 = _mm_add_ps(fiz1,tz); |
1852 | |
1853 | fjx1 = _mm_add_ps(fjx1,tx); |
1854 | fjy1 = _mm_add_ps(fjy1,ty); |
1855 | fjz1 = _mm_add_ps(fjz1,tz); |
1856 | |
1857 | } |
1858 | |
1859 | /************************** |
1860 | * CALCULATE INTERACTIONS * |
1861 | **************************/ |
1862 | |
1863 | if (gmx_mm_any_lt(rsq12,rcutoff2)) |
1864 | { |
1865 | |
1866 | r12 = _mm_mul_ps(rsq12,rinv12); |
1867 | |
1868 | /* EWALD ELECTROSTATICS */ |
1869 | |
1870 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1871 | ewrt = _mm_mul_ps(r12,ewtabscale); |
1872 | ewitab = _mm_cvttps_epi32(ewrt); |
1873 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1874 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1875 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1876 | &ewtabF,&ewtabFn); |
1877 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1878 | felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec)); |
1879 | |
1880 | cutoff_mask = _mm_cmplt_ps(rsq12,rcutoff2); |
1881 | |
1882 | fscal = felec; |
1883 | |
1884 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1885 | |
1886 | /* Calculate temporary vectorial force */ |
1887 | tx = _mm_mul_ps(fscal,dx12); |
1888 | ty = _mm_mul_ps(fscal,dy12); |
1889 | tz = _mm_mul_ps(fscal,dz12); |
1890 | |
1891 | /* Update vectorial force */ |
1892 | fix1 = _mm_add_ps(fix1,tx); |
1893 | fiy1 = _mm_add_ps(fiy1,ty); |
1894 | fiz1 = _mm_add_ps(fiz1,tz); |
1895 | |
1896 | fjx2 = _mm_add_ps(fjx2,tx); |
1897 | fjy2 = _mm_add_ps(fjy2,ty); |
1898 | fjz2 = _mm_add_ps(fjz2,tz); |
1899 | |
1900 | } |
1901 | |
1902 | /************************** |
1903 | * CALCULATE INTERACTIONS * |
1904 | **************************/ |
1905 | |
1906 | if (gmx_mm_any_lt(rsq20,rcutoff2)) |
1907 | { |
1908 | |
1909 | r20 = _mm_mul_ps(rsq20,rinv20); |
1910 | |
1911 | /* EWALD ELECTROSTATICS */ |
1912 | |
1913 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1914 | ewrt = _mm_mul_ps(r20,ewtabscale); |
1915 | ewitab = _mm_cvttps_epi32(ewrt); |
1916 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1917 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1918 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1919 | &ewtabF,&ewtabFn); |
1920 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1921 | felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec)); |
1922 | |
1923 | cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2); |
1924 | |
1925 | fscal = felec; |
1926 | |
1927 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1928 | |
1929 | /* Calculate temporary vectorial force */ |
1930 | tx = _mm_mul_ps(fscal,dx20); |
1931 | ty = _mm_mul_ps(fscal,dy20); |
1932 | tz = _mm_mul_ps(fscal,dz20); |
1933 | |
1934 | /* Update vectorial force */ |
1935 | fix2 = _mm_add_ps(fix2,tx); |
1936 | fiy2 = _mm_add_ps(fiy2,ty); |
1937 | fiz2 = _mm_add_ps(fiz2,tz); |
1938 | |
1939 | fjx0 = _mm_add_ps(fjx0,tx); |
1940 | fjy0 = _mm_add_ps(fjy0,ty); |
1941 | fjz0 = _mm_add_ps(fjz0,tz); |
1942 | |
1943 | } |
1944 | |
1945 | /************************** |
1946 | * CALCULATE INTERACTIONS * |
1947 | **************************/ |
1948 | |
1949 | if (gmx_mm_any_lt(rsq21,rcutoff2)) |
1950 | { |
1951 | |
1952 | r21 = _mm_mul_ps(rsq21,rinv21); |
1953 | |
1954 | /* EWALD ELECTROSTATICS */ |
1955 | |
1956 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
1957 | ewrt = _mm_mul_ps(r21,ewtabscale); |
1958 | ewitab = _mm_cvttps_epi32(ewrt); |
1959 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
1960 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
1961 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
1962 | &ewtabF,&ewtabFn); |
1963 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
1964 | felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec)); |
1965 | |
1966 | cutoff_mask = _mm_cmplt_ps(rsq21,rcutoff2); |
1967 | |
1968 | fscal = felec; |
1969 | |
1970 | fscal = _mm_and_ps(fscal,cutoff_mask); |
1971 | |
1972 | /* Calculate temporary vectorial force */ |
1973 | tx = _mm_mul_ps(fscal,dx21); |
1974 | ty = _mm_mul_ps(fscal,dy21); |
1975 | tz = _mm_mul_ps(fscal,dz21); |
1976 | |
1977 | /* Update vectorial force */ |
1978 | fix2 = _mm_add_ps(fix2,tx); |
1979 | fiy2 = _mm_add_ps(fiy2,ty); |
1980 | fiz2 = _mm_add_ps(fiz2,tz); |
1981 | |
1982 | fjx1 = _mm_add_ps(fjx1,tx); |
1983 | fjy1 = _mm_add_ps(fjy1,ty); |
1984 | fjz1 = _mm_add_ps(fjz1,tz); |
1985 | |
1986 | } |
1987 | |
1988 | /************************** |
1989 | * CALCULATE INTERACTIONS * |
1990 | **************************/ |
1991 | |
1992 | if (gmx_mm_any_lt(rsq22,rcutoff2)) |
1993 | { |
1994 | |
1995 | r22 = _mm_mul_ps(rsq22,rinv22); |
1996 | |
1997 | /* EWALD ELECTROSTATICS */ |
1998 | |
1999 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2000 | ewrt = _mm_mul_ps(r22,ewtabscale); |
2001 | ewitab = _mm_cvttps_epi32(ewrt); |
2002 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2003 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2004 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2005 | &ewtabF,&ewtabFn); |
2006 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2007 | felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec)); |
2008 | |
2009 | cutoff_mask = _mm_cmplt_ps(rsq22,rcutoff2); |
2010 | |
2011 | fscal = felec; |
2012 | |
2013 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2014 | |
2015 | /* Calculate temporary vectorial force */ |
2016 | tx = _mm_mul_ps(fscal,dx22); |
2017 | ty = _mm_mul_ps(fscal,dy22); |
2018 | tz = _mm_mul_ps(fscal,dz22); |
2019 | |
2020 | /* Update vectorial force */ |
2021 | fix2 = _mm_add_ps(fix2,tx); |
2022 | fiy2 = _mm_add_ps(fiy2,ty); |
2023 | fiz2 = _mm_add_ps(fiz2,tz); |
2024 | |
2025 | fjx2 = _mm_add_ps(fjx2,tx); |
2026 | fjy2 = _mm_add_ps(fjy2,ty); |
2027 | fjz2 = _mm_add_ps(fjz2,tz); |
2028 | |
2029 | } |
2030 | |
2031 | fjptrA = f+j_coord_offsetA; |
2032 | fjptrB = f+j_coord_offsetB; |
2033 | fjptrC = f+j_coord_offsetC; |
2034 | fjptrD = f+j_coord_offsetD; |
2035 | |
2036 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
2037 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2); |
2038 | |
2039 | /* Inner loop uses 351 flops */ |
2040 | } |
2041 | |
2042 | if(jidx<j_index_end) |
2043 | { |
2044 | |
2045 | /* Get j neighbor index, and coordinate index */ |
2046 | jnrlistA = jjnr[jidx]; |
2047 | jnrlistB = jjnr[jidx+1]; |
2048 | jnrlistC = jjnr[jidx+2]; |
2049 | jnrlistD = jjnr[jidx+3]; |
2050 | /* Sign of each element will be negative for non-real atoms. |
2051 | * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones, |
2052 | * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries. |
2053 | */ |
2054 | dummy_mask = gmx_mm_castsi128_ps_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())); |
2055 | jnrA = (jnrlistA>=0) ? jnrlistA : 0; |
2056 | jnrB = (jnrlistB>=0) ? jnrlistB : 0; |
2057 | jnrC = (jnrlistC>=0) ? jnrlistC : 0; |
2058 | jnrD = (jnrlistD>=0) ? jnrlistD : 0; |
2059 | j_coord_offsetA = DIM3*jnrA; |
2060 | j_coord_offsetB = DIM3*jnrB; |
2061 | j_coord_offsetC = DIM3*jnrC; |
2062 | j_coord_offsetD = DIM3*jnrD; |
2063 | |
2064 | /* load j atom coordinates */ |
2065 | gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB, |
2066 | x+j_coord_offsetC,x+j_coord_offsetD, |
2067 | &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2); |
2068 | |
2069 | /* Calculate displacement vector */ |
2070 | dx00 = _mm_sub_ps(ix0,jx0); |
2071 | dy00 = _mm_sub_ps(iy0,jy0); |
2072 | dz00 = _mm_sub_ps(iz0,jz0); |
2073 | dx01 = _mm_sub_ps(ix0,jx1); |
2074 | dy01 = _mm_sub_ps(iy0,jy1); |
2075 | dz01 = _mm_sub_ps(iz0,jz1); |
2076 | dx02 = _mm_sub_ps(ix0,jx2); |
2077 | dy02 = _mm_sub_ps(iy0,jy2); |
2078 | dz02 = _mm_sub_ps(iz0,jz2); |
2079 | dx10 = _mm_sub_ps(ix1,jx0); |
2080 | dy10 = _mm_sub_ps(iy1,jy0); |
2081 | dz10 = _mm_sub_ps(iz1,jz0); |
2082 | dx11 = _mm_sub_ps(ix1,jx1); |
2083 | dy11 = _mm_sub_ps(iy1,jy1); |
2084 | dz11 = _mm_sub_ps(iz1,jz1); |
2085 | dx12 = _mm_sub_ps(ix1,jx2); |
2086 | dy12 = _mm_sub_ps(iy1,jy2); |
2087 | dz12 = _mm_sub_ps(iz1,jz2); |
2088 | dx20 = _mm_sub_ps(ix2,jx0); |
2089 | dy20 = _mm_sub_ps(iy2,jy0); |
2090 | dz20 = _mm_sub_ps(iz2,jz0); |
2091 | dx21 = _mm_sub_ps(ix2,jx1); |
2092 | dy21 = _mm_sub_ps(iy2,jy1); |
2093 | dz21 = _mm_sub_ps(iz2,jz1); |
2094 | dx22 = _mm_sub_ps(ix2,jx2); |
2095 | dy22 = _mm_sub_ps(iy2,jy2); |
2096 | dz22 = _mm_sub_ps(iz2,jz2); |
2097 | |
2098 | /* Calculate squared distance and things based on it */ |
2099 | rsq00 = gmx_mm_calc_rsq_ps(dx00,dy00,dz00); |
2100 | rsq01 = gmx_mm_calc_rsq_ps(dx01,dy01,dz01); |
2101 | rsq02 = gmx_mm_calc_rsq_ps(dx02,dy02,dz02); |
2102 | rsq10 = gmx_mm_calc_rsq_ps(dx10,dy10,dz10); |
2103 | rsq11 = gmx_mm_calc_rsq_ps(dx11,dy11,dz11); |
2104 | rsq12 = gmx_mm_calc_rsq_ps(dx12,dy12,dz12); |
2105 | rsq20 = gmx_mm_calc_rsq_ps(dx20,dy20,dz20); |
2106 | rsq21 = gmx_mm_calc_rsq_ps(dx21,dy21,dz21); |
2107 | rsq22 = gmx_mm_calc_rsq_ps(dx22,dy22,dz22); |
2108 | |
2109 | rinv00 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq00); |
2110 | rinv01 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq01); |
2111 | rinv02 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq02); |
2112 | rinv10 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq10); |
2113 | rinv11 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq11); |
2114 | rinv12 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq12); |
2115 | rinv20 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq20); |
2116 | rinv21 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq21); |
2117 | rinv22 = gmx_mm_invsqrt_psgmx_simd_invsqrt_f(rsq22); |
2118 | |
2119 | rinvsq00 = _mm_mul_ps(rinv00,rinv00); |
2120 | rinvsq01 = _mm_mul_ps(rinv01,rinv01); |
2121 | rinvsq02 = _mm_mul_ps(rinv02,rinv02); |
2122 | rinvsq10 = _mm_mul_ps(rinv10,rinv10); |
2123 | rinvsq11 = _mm_mul_ps(rinv11,rinv11); |
2124 | rinvsq12 = _mm_mul_ps(rinv12,rinv12); |
2125 | rinvsq20 = _mm_mul_ps(rinv20,rinv20); |
2126 | rinvsq21 = _mm_mul_ps(rinv21,rinv21); |
2127 | rinvsq22 = _mm_mul_ps(rinv22,rinv22); |
2128 | |
2129 | fjx0 = _mm_setzero_ps(); |
2130 | fjy0 = _mm_setzero_ps(); |
2131 | fjz0 = _mm_setzero_ps(); |
2132 | fjx1 = _mm_setzero_ps(); |
2133 | fjy1 = _mm_setzero_ps(); |
2134 | fjz1 = _mm_setzero_ps(); |
2135 | fjx2 = _mm_setzero_ps(); |
2136 | fjy2 = _mm_setzero_ps(); |
2137 | fjz2 = _mm_setzero_ps(); |
2138 | |
2139 | /************************** |
2140 | * CALCULATE INTERACTIONS * |
2141 | **************************/ |
2142 | |
2143 | if (gmx_mm_any_lt(rsq00,rcutoff2)) |
2144 | { |
2145 | |
2146 | r00 = _mm_mul_ps(rsq00,rinv00); |
2147 | r00 = _mm_andnot_ps(dummy_mask,r00); |
2148 | |
2149 | /* EWALD ELECTROSTATICS */ |
2150 | |
2151 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2152 | ewrt = _mm_mul_ps(r00,ewtabscale); |
2153 | ewitab = _mm_cvttps_epi32(ewrt); |
2154 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2155 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2156 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2157 | &ewtabF,&ewtabFn); |
2158 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2159 | felec = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec)); |
2160 | |
2161 | cutoff_mask = _mm_cmplt_ps(rsq00,rcutoff2); |
2162 | |
2163 | fscal = felec; |
2164 | |
2165 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2166 | |
2167 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2168 | |
2169 | /* Calculate temporary vectorial force */ |
2170 | tx = _mm_mul_ps(fscal,dx00); |
2171 | ty = _mm_mul_ps(fscal,dy00); |
2172 | tz = _mm_mul_ps(fscal,dz00); |
2173 | |
2174 | /* Update vectorial force */ |
2175 | fix0 = _mm_add_ps(fix0,tx); |
2176 | fiy0 = _mm_add_ps(fiy0,ty); |
2177 | fiz0 = _mm_add_ps(fiz0,tz); |
2178 | |
2179 | fjx0 = _mm_add_ps(fjx0,tx); |
2180 | fjy0 = _mm_add_ps(fjy0,ty); |
2181 | fjz0 = _mm_add_ps(fjz0,tz); |
2182 | |
2183 | } |
2184 | |
2185 | /************************** |
2186 | * CALCULATE INTERACTIONS * |
2187 | **************************/ |
2188 | |
2189 | if (gmx_mm_any_lt(rsq01,rcutoff2)) |
2190 | { |
2191 | |
2192 | r01 = _mm_mul_ps(rsq01,rinv01); |
2193 | r01 = _mm_andnot_ps(dummy_mask,r01); |
2194 | |
2195 | /* EWALD ELECTROSTATICS */ |
2196 | |
2197 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2198 | ewrt = _mm_mul_ps(r01,ewtabscale); |
2199 | ewitab = _mm_cvttps_epi32(ewrt); |
2200 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2201 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2202 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2203 | &ewtabF,&ewtabFn); |
2204 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2205 | felec = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec)); |
2206 | |
2207 | cutoff_mask = _mm_cmplt_ps(rsq01,rcutoff2); |
2208 | |
2209 | fscal = felec; |
2210 | |
2211 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2212 | |
2213 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2214 | |
2215 | /* Calculate temporary vectorial force */ |
2216 | tx = _mm_mul_ps(fscal,dx01); |
2217 | ty = _mm_mul_ps(fscal,dy01); |
2218 | tz = _mm_mul_ps(fscal,dz01); |
2219 | |
2220 | /* Update vectorial force */ |
2221 | fix0 = _mm_add_ps(fix0,tx); |
2222 | fiy0 = _mm_add_ps(fiy0,ty); |
2223 | fiz0 = _mm_add_ps(fiz0,tz); |
2224 | |
2225 | fjx1 = _mm_add_ps(fjx1,tx); |
2226 | fjy1 = _mm_add_ps(fjy1,ty); |
2227 | fjz1 = _mm_add_ps(fjz1,tz); |
2228 | |
2229 | } |
2230 | |
2231 | /************************** |
2232 | * CALCULATE INTERACTIONS * |
2233 | **************************/ |
2234 | |
2235 | if (gmx_mm_any_lt(rsq02,rcutoff2)) |
2236 | { |
2237 | |
2238 | r02 = _mm_mul_ps(rsq02,rinv02); |
2239 | r02 = _mm_andnot_ps(dummy_mask,r02); |
2240 | |
2241 | /* EWALD ELECTROSTATICS */ |
2242 | |
2243 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2244 | ewrt = _mm_mul_ps(r02,ewtabscale); |
2245 | ewitab = _mm_cvttps_epi32(ewrt); |
2246 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2247 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2248 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2249 | &ewtabF,&ewtabFn); |
2250 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2251 | felec = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec)); |
2252 | |
2253 | cutoff_mask = _mm_cmplt_ps(rsq02,rcutoff2); |
2254 | |
2255 | fscal = felec; |
2256 | |
2257 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2258 | |
2259 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2260 | |
2261 | /* Calculate temporary vectorial force */ |
2262 | tx = _mm_mul_ps(fscal,dx02); |
2263 | ty = _mm_mul_ps(fscal,dy02); |
2264 | tz = _mm_mul_ps(fscal,dz02); |
2265 | |
2266 | /* Update vectorial force */ |
2267 | fix0 = _mm_add_ps(fix0,tx); |
2268 | fiy0 = _mm_add_ps(fiy0,ty); |
2269 | fiz0 = _mm_add_ps(fiz0,tz); |
2270 | |
2271 | fjx2 = _mm_add_ps(fjx2,tx); |
2272 | fjy2 = _mm_add_ps(fjy2,ty); |
2273 | fjz2 = _mm_add_ps(fjz2,tz); |
2274 | |
2275 | } |
2276 | |
2277 | /************************** |
2278 | * CALCULATE INTERACTIONS * |
2279 | **************************/ |
2280 | |
2281 | if (gmx_mm_any_lt(rsq10,rcutoff2)) |
2282 | { |
2283 | |
2284 | r10 = _mm_mul_ps(rsq10,rinv10); |
2285 | r10 = _mm_andnot_ps(dummy_mask,r10); |
2286 | |
2287 | /* EWALD ELECTROSTATICS */ |
2288 | |
2289 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2290 | ewrt = _mm_mul_ps(r10,ewtabscale); |
2291 | ewitab = _mm_cvttps_epi32(ewrt); |
2292 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2293 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2294 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2295 | &ewtabF,&ewtabFn); |
2296 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2297 | felec = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec)); |
2298 | |
2299 | cutoff_mask = _mm_cmplt_ps(rsq10,rcutoff2); |
2300 | |
2301 | fscal = felec; |
2302 | |
2303 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2304 | |
2305 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2306 | |
2307 | /* Calculate temporary vectorial force */ |
2308 | tx = _mm_mul_ps(fscal,dx10); |
2309 | ty = _mm_mul_ps(fscal,dy10); |
2310 | tz = _mm_mul_ps(fscal,dz10); |
2311 | |
2312 | /* Update vectorial force */ |
2313 | fix1 = _mm_add_ps(fix1,tx); |
2314 | fiy1 = _mm_add_ps(fiy1,ty); |
2315 | fiz1 = _mm_add_ps(fiz1,tz); |
2316 | |
2317 | fjx0 = _mm_add_ps(fjx0,tx); |
2318 | fjy0 = _mm_add_ps(fjy0,ty); |
2319 | fjz0 = _mm_add_ps(fjz0,tz); |
2320 | |
2321 | } |
2322 | |
2323 | /************************** |
2324 | * CALCULATE INTERACTIONS * |
2325 | **************************/ |
2326 | |
2327 | if (gmx_mm_any_lt(rsq11,rcutoff2)) |
2328 | { |
2329 | |
2330 | r11 = _mm_mul_ps(rsq11,rinv11); |
2331 | r11 = _mm_andnot_ps(dummy_mask,r11); |
2332 | |
2333 | /* EWALD ELECTROSTATICS */ |
2334 | |
2335 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2336 | ewrt = _mm_mul_ps(r11,ewtabscale); |
2337 | ewitab = _mm_cvttps_epi32(ewrt); |
2338 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2339 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2340 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2341 | &ewtabF,&ewtabFn); |
2342 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2343 | felec = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec)); |
2344 | |
2345 | cutoff_mask = _mm_cmplt_ps(rsq11,rcutoff2); |
2346 | |
2347 | fscal = felec; |
2348 | |
2349 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2350 | |
2351 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2352 | |
2353 | /* Calculate temporary vectorial force */ |
2354 | tx = _mm_mul_ps(fscal,dx11); |
2355 | ty = _mm_mul_ps(fscal,dy11); |
2356 | tz = _mm_mul_ps(fscal,dz11); |
2357 | |
2358 | /* Update vectorial force */ |
2359 | fix1 = _mm_add_ps(fix1,tx); |
2360 | fiy1 = _mm_add_ps(fiy1,ty); |
2361 | fiz1 = _mm_add_ps(fiz1,tz); |
2362 | |
2363 | fjx1 = _mm_add_ps(fjx1,tx); |
2364 | fjy1 = _mm_add_ps(fjy1,ty); |
2365 | fjz1 = _mm_add_ps(fjz1,tz); |
2366 | |
2367 | } |
2368 | |
2369 | /************************** |
2370 | * CALCULATE INTERACTIONS * |
2371 | **************************/ |
2372 | |
2373 | if (gmx_mm_any_lt(rsq12,rcutoff2)) |
2374 | { |
2375 | |
2376 | r12 = _mm_mul_ps(rsq12,rinv12); |
2377 | r12 = _mm_andnot_ps(dummy_mask,r12); |
2378 | |
2379 | /* EWALD ELECTROSTATICS */ |
2380 | |
2381 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2382 | ewrt = _mm_mul_ps(r12,ewtabscale); |
2383 | ewitab = _mm_cvttps_epi32(ewrt); |
2384 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2385 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2386 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2387 | &ewtabF,&ewtabFn); |
2388 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2389 | felec = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec)); |
2390 | |
2391 | cutoff_mask = _mm_cmplt_ps(rsq12,rcutoff2); |
2392 | |
2393 | fscal = felec; |
2394 | |
2395 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2396 | |
2397 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2398 | |
2399 | /* Calculate temporary vectorial force */ |
2400 | tx = _mm_mul_ps(fscal,dx12); |
2401 | ty = _mm_mul_ps(fscal,dy12); |
2402 | tz = _mm_mul_ps(fscal,dz12); |
2403 | |
2404 | /* Update vectorial force */ |
2405 | fix1 = _mm_add_ps(fix1,tx); |
2406 | fiy1 = _mm_add_ps(fiy1,ty); |
2407 | fiz1 = _mm_add_ps(fiz1,tz); |
2408 | |
2409 | fjx2 = _mm_add_ps(fjx2,tx); |
2410 | fjy2 = _mm_add_ps(fjy2,ty); |
2411 | fjz2 = _mm_add_ps(fjz2,tz); |
2412 | |
2413 | } |
2414 | |
2415 | /************************** |
2416 | * CALCULATE INTERACTIONS * |
2417 | **************************/ |
2418 | |
2419 | if (gmx_mm_any_lt(rsq20,rcutoff2)) |
2420 | { |
2421 | |
2422 | r20 = _mm_mul_ps(rsq20,rinv20); |
2423 | r20 = _mm_andnot_ps(dummy_mask,r20); |
2424 | |
2425 | /* EWALD ELECTROSTATICS */ |
2426 | |
2427 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2428 | ewrt = _mm_mul_ps(r20,ewtabscale); |
2429 | ewitab = _mm_cvttps_epi32(ewrt); |
2430 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2431 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2432 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2433 | &ewtabF,&ewtabFn); |
2434 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2435 | felec = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec)); |
2436 | |
2437 | cutoff_mask = _mm_cmplt_ps(rsq20,rcutoff2); |
2438 | |
2439 | fscal = felec; |
2440 | |
2441 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2442 | |
2443 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2444 | |
2445 | /* Calculate temporary vectorial force */ |
2446 | tx = _mm_mul_ps(fscal,dx20); |
2447 | ty = _mm_mul_ps(fscal,dy20); |
2448 | tz = _mm_mul_ps(fscal,dz20); |
2449 | |
2450 | /* Update vectorial force */ |
2451 | fix2 = _mm_add_ps(fix2,tx); |
2452 | fiy2 = _mm_add_ps(fiy2,ty); |
2453 | fiz2 = _mm_add_ps(fiz2,tz); |
2454 | |
2455 | fjx0 = _mm_add_ps(fjx0,tx); |
2456 | fjy0 = _mm_add_ps(fjy0,ty); |
2457 | fjz0 = _mm_add_ps(fjz0,tz); |
2458 | |
2459 | } |
2460 | |
2461 | /************************** |
2462 | * CALCULATE INTERACTIONS * |
2463 | **************************/ |
2464 | |
2465 | if (gmx_mm_any_lt(rsq21,rcutoff2)) |
2466 | { |
2467 | |
2468 | r21 = _mm_mul_ps(rsq21,rinv21); |
2469 | r21 = _mm_andnot_ps(dummy_mask,r21); |
2470 | |
2471 | /* EWALD ELECTROSTATICS */ |
2472 | |
2473 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2474 | ewrt = _mm_mul_ps(r21,ewtabscale); |
2475 | ewitab = _mm_cvttps_epi32(ewrt); |
2476 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2477 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2478 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2479 | &ewtabF,&ewtabFn); |
2480 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2481 | felec = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec)); |
2482 | |
2483 | cutoff_mask = _mm_cmplt_ps(rsq21,rcutoff2); |
2484 | |
2485 | fscal = felec; |
2486 | |
2487 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2488 | |
2489 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2490 | |
2491 | /* Calculate temporary vectorial force */ |
2492 | tx = _mm_mul_ps(fscal,dx21); |
2493 | ty = _mm_mul_ps(fscal,dy21); |
2494 | tz = _mm_mul_ps(fscal,dz21); |
2495 | |
2496 | /* Update vectorial force */ |
2497 | fix2 = _mm_add_ps(fix2,tx); |
2498 | fiy2 = _mm_add_ps(fiy2,ty); |
2499 | fiz2 = _mm_add_ps(fiz2,tz); |
2500 | |
2501 | fjx1 = _mm_add_ps(fjx1,tx); |
2502 | fjy1 = _mm_add_ps(fjy1,ty); |
2503 | fjz1 = _mm_add_ps(fjz1,tz); |
2504 | |
2505 | } |
2506 | |
2507 | /************************** |
2508 | * CALCULATE INTERACTIONS * |
2509 | **************************/ |
2510 | |
2511 | if (gmx_mm_any_lt(rsq22,rcutoff2)) |
2512 | { |
2513 | |
2514 | r22 = _mm_mul_ps(rsq22,rinv22); |
2515 | r22 = _mm_andnot_ps(dummy_mask,r22); |
2516 | |
2517 | /* EWALD ELECTROSTATICS */ |
2518 | |
2519 | /* Calculate Ewald table index by multiplying r with scale and truncate to integer */ |
2520 | ewrt = _mm_mul_ps(r22,ewtabscale); |
2521 | ewitab = _mm_cvttps_epi32(ewrt); |
2522 | eweps = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR)__extension__ ({ __m128 __X = (ewrt); (__m128) __builtin_ia32_roundps ((__v4sf)__X, ((0x00 | 0x01))); })); |
2523 | gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(0) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,1)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(1) & 3];})), |
2524 | ewtab + gmx_mm_extract_epi32(ewitab,2)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(2) & 3];})),ewtab + gmx_mm_extract_epi32(ewitab,3)(__extension__ ({ __v4si __a = (__v4si)(ewitab); __a[(3) & 3];})), |
2525 | &ewtabF,&ewtabFn); |
2526 | felec = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn)); |
2527 | felec = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec)); |
2528 | |
2529 | cutoff_mask = _mm_cmplt_ps(rsq22,rcutoff2); |
2530 | |
2531 | fscal = felec; |
2532 | |
2533 | fscal = _mm_and_ps(fscal,cutoff_mask); |
2534 | |
2535 | fscal = _mm_andnot_ps(dummy_mask,fscal); |
2536 | |
2537 | /* Calculate temporary vectorial force */ |
2538 | tx = _mm_mul_ps(fscal,dx22); |
2539 | ty = _mm_mul_ps(fscal,dy22); |
2540 | tz = _mm_mul_ps(fscal,dz22); |
2541 | |
2542 | /* Update vectorial force */ |
2543 | fix2 = _mm_add_ps(fix2,tx); |
2544 | fiy2 = _mm_add_ps(fiy2,ty); |
2545 | fiz2 = _mm_add_ps(fiz2,tz); |
2546 | |
2547 | fjx2 = _mm_add_ps(fjx2,tx); |
2548 | fjy2 = _mm_add_ps(fjy2,ty); |
2549 | fjz2 = _mm_add_ps(fjz2,tz); |
2550 | |
2551 | } |
2552 | |
2553 | fjptrA = (jnrlistA>=0) ? f+j_coord_offsetA : scratch; |
2554 | fjptrB = (jnrlistB>=0) ? f+j_coord_offsetB : scratch; |
2555 | fjptrC = (jnrlistC>=0) ? f+j_coord_offsetC : scratch; |
2556 | fjptrD = (jnrlistD>=0) ? f+j_coord_offsetD : scratch; |
2557 | |
2558 | gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD, |
2559 | fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2); |
2560 | |
2561 | /* Inner loop uses 360 flops */ |
2562 | } |
2563 | |
2564 | /* End of innermost loop */ |
2565 | |
2566 | gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2, |
2567 | f+i_coord_offset,fshift+i_shift_offset); |
2568 | |
2569 | /* Increment number of inner iterations */ |
2570 | inneriter += j_index_end - j_index_start; |
2571 | |
2572 | /* Outer loop uses 18 flops */ |
2573 | } |
2574 | |
2575 | /* Increment number of outer iterations */ |
2576 | outeriter += nri; |
2577 | |
2578 | /* Update outer/inner flops */ |
2579 | |
2580 | inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3W3_F,outeriter*18 + inneriter*360)(nrnb)->n[eNR_NBKERNEL_ELEC_W3W3_F] += outeriter*18 + inneriter *360; |
2581 | } |