-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathHoloConst.cu
1330 lines (1158 loc) · 40.3 KB
/
HoloConst.cu
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#include <iostream>
#include <cuda.h>
#include <math.h>
#include "gpu_utils.h"
#include "cuda_utils.h"
#include "HoloConst.h"
struct HoloConstSettings_t {
// Solvents
double mO_div_mH2O;
double mH_div_mH2O;
double ra, rc, rb, ra_inv, rc2;
int nsolvent;
const solvent_t* __restrict__ solvent_ind;
// Pairs
int npair;
int npair_type;
int2* pair_ind;
bond_t* pair_indtype;
double *pair_constr;
double *pair_mass;
// Trips
int ntrip;
int ntrip_type;
int3* trip_ind;
angle_t* trip_indtype;
double *trip_constr;
double *trip_mass;
// Quads
int nquad;
int nquad_type;
int4* quad_ind;
dihe_t* quad_indtype;
double *quad_constr;
double *quad_mass;
// Buffer for all constraints and masses
int ntot_type;
double *constr_mass;
double shake_tol;
int max_niter;
const double* __restrict__ xyz[2][3];
double* __restrict__ xyz2[3];
};
static HoloConstSettings_t h_setup;
static __constant__ HoloConstSettings_t d_setup;
// x0 = xyz_texref[0][0]
// y0 = xyz_texref[0][1]
// z0 = xyz_texref[0][2]
// x1 = xyz_texref[1][0]
// y1 = xyz_texref[1][1]
// z1 = xyz_texref[1][2]
static texture<int2, 1, cudaReadModeElementType> xyz_texref00;
static texture<int2, 1, cudaReadModeElementType> xyz_texref01;
static texture<int2, 1, cudaReadModeElementType> xyz_texref02;
static texture<int2, 1, cudaReadModeElementType> xyz_texref10;
static texture<int2, 1, cudaReadModeElementType> xyz_texref11;
static texture<int2, 1, cudaReadModeElementType> xyz_texref12;
texture<int2, 1, cudaReadModeElementType>& get_xyz_texref(const int i, const int j) {
switch(i) {
case 0:
switch(j) {
case 0:
return xyz_texref00;
case 1:
return xyz_texref01;
case 2:
return xyz_texref02;
default:
std::cerr << "get_xyz_texref, index out of bounds: i=" << i << " j=" << j << std::endl;
exit(1);
}
case 1:
switch(j) {
case 0:
return xyz_texref10;
case 1:
return xyz_texref11;
case 2:
return xyz_texref12;
default:
std::cerr << "get_xyz_texref, index out of bounds: i=" << i << " j=" << j << std::endl;
exit(1);
}
default:
std::cerr << "get_xyz_texref, index out of bounds: i=" << i << " j=" << j << std::endl;
exit(1);
}
}
static int2* xyz_texref_pointer[2][3];
static int texref_size[2][3];
template <int i, int j>
__forceinline__ __device__ double load_coord(const int ind) {
#if __CUDA_ARCH__ < 350
int2 val;
if (i == 0) {
if (j == 0) {
val = tex1Dfetch(xyz_texref00, ind);
} else if (j == 1) {
val = tex1Dfetch(xyz_texref01, ind);
} else if (j == 2) {
val = tex1Dfetch(xyz_texref02, ind);
}
} else if (i == 1) {
if (j == 0) {
val = tex1Dfetch(xyz_texref10, ind);
} else if (j == 1) {
val = tex1Dfetch(xyz_texref11, ind);
} else if (j == 2) {
val = tex1Dfetch(xyz_texref12, ind);
}
}
return __hiloint2double(val.y, val.x);
#else
return __ldg(&d_setup.xyz[i][j][ind]);
#endif
}
template<bool use_indexed>
__forceinline__ __device__ void pair_calc(const int imol,
const double* __restrict__ sh_constr,
const double* __restrict__ sh_mass) {
int i, j, itype;
if (use_indexed) {
bond_t bond = d_setup.pair_indtype[imol];
i = bond.i;
j = bond.j;
itype = bond.itype;
} else {
int2 ind = d_setup.pair_ind[imol];
i = ind.x;
j = ind.y;
itype = imol;
}
// Load coordinates
double x0i = load_coord<0, 0>(i);
double y0i = load_coord<0, 1>(i);
double z0i = load_coord<0, 2>(i);
double x0j = load_coord<0, 0>(j);
double y0j = load_coord<0, 1>(j);
double z0j = load_coord<0, 2>(j);
double x1i = load_coord<1, 0>(i);
double y1i = load_coord<1, 1>(i);
double z1i = load_coord<1, 2>(i);
double x1j = load_coord<1, 0>(j);
double y1j = load_coord<1, 1>(j);
double z1j = load_coord<1, 2>(j);
double xpij = x1i - x1j;
double ypij = y1i - y1j;
double zpij = z1i - z1j;
double rijsq = xpij*xpij + ypij*ypij + zpij*zpij;
double diff;
if (use_indexed) {
diff = sh_constr[itype] - rijsq;
} else {
diff = d_setup.pair_constr[itype] - rijsq;
}
double xrij = x0i - x0j;
double yrij = y0i - y0j;
double zrij = z0i - z0j;
double rrijsq = xrij*xrij + yrij*yrij + zrij*zrij;
double rijrijp = xrij*xpij + yrij*ypij + zrij*zpij;
double lambda = 2.0*(-rijrijp + sqrt(rijrijp*rijrijp+rrijsq*diff))/(rrijsq);
double hmassi_val, hmassj_val;
if (use_indexed) {
hmassi_val = sh_mass[itype*2];
hmassj_val = sh_mass[itype*2+1];
} else {
hmassi_val = d_setup.pair_mass[itype*2];
hmassj_val = d_setup.pair_mass[itype*2+1];
}
x1i += hmassi_val*lambda*xrij;
y1i += hmassi_val*lambda*yrij;
z1i += hmassi_val*lambda*zrij;
x1j -= hmassj_val*lambda*xrij;
y1j -= hmassj_val*lambda*yrij;
z1j -= hmassj_val*lambda*zrij;
// Store results
d_setup.xyz2[0][i] = x1i;
d_setup.xyz2[1][i] = y1i;
d_setup.xyz2[2][i] = z1i;
d_setup.xyz2[0][j] = x1j;
d_setup.xyz2[1][j] = y1j;
d_setup.xyz2[2][j] = z1j;
}
template<bool use_indexed>
__forceinline__ __device__ void trip_calc(const int imol,
const double* __restrict__ sh_constr,
const double* __restrict__ sh_mass) {
int i, j, k, itype;
if (use_indexed) {
angle_t angle = d_setup.trip_indtype[imol];
i = angle.i;
j = angle.j;
k = angle.k;
itype = angle.itype;
} else {
int3 ind = d_setup.trip_ind[imol];
i = ind.x;
j = ind.y;
k = ind.z;
itype = imol;
}
// Load coordinates
double x0i = load_coord<0, 0>(i);
double y0i = load_coord<0, 1>(i);
double z0i = load_coord<0, 2>(i);
double x0j = load_coord<0, 0>(j);
double y0j = load_coord<0, 1>(j);
double z0j = load_coord<0, 2>(j);
double x0k = load_coord<0, 0>(k);
double y0k = load_coord<0, 1>(k);
double z0k = load_coord<0, 2>(k);
double x1i = load_coord<1, 0>(i);
double y1i = load_coord<1, 1>(i);
double z1i = load_coord<1, 2>(i);
double x1j = load_coord<1, 0>(j);
double y1j = load_coord<1, 1>(j);
double z1j = load_coord<1, 2>(j);
double x1k = load_coord<1, 0>(k);
double y1k = load_coord<1, 1>(k);
double z1k = load_coord<1, 2>(k);
double xrij = x0i - x0j;
double yrij = y0i - y0j;
double zrij = z0i - z0j;
double xrik = x0i - x0k;
double yrik = y0i - y0k;
double zrik = z0i - z0k;
double rrijsq = xrij*xrij + yrij*yrij + zrij*zrij;
double rriksq = xrik*xrik + yrik*yrik + zrik*zrik;
double rijrik = xrij*xrik + yrij*yrik + zrij*zrik;
double mmi, mmj, mmk, mij, mik;
if (use_indexed) {
mmi = sh_mass[itype*5];
mmj = sh_mass[itype*5+1];
mmk = sh_mass[itype*5+2];
mij = sh_mass[itype*5+3];
mik = sh_mass[itype*5+4];
} else {
mmi = d_setup.trip_mass[itype*5];
mmj = d_setup.trip_mass[itype*5+1];
mmk = d_setup.trip_mass[itype*5+2];
mij = d_setup.trip_mass[itype*5+3];
mik = d_setup.trip_mass[itype*5+4];
}
double acorr1 = mij*mij*rrijsq;
double acorr2 = mij*mmi*2.0*rijrik;
double acorr3 = mmi*mmi*rriksq;
double acorr4 = mmi*mmi*rrijsq;
double acorr5 = mik*mmi*2.0*rijrik;
double acorr6 = mik*mik*rriksq;
double xpij = x1i - x1j;
double ypij = y1i - y1j;
double zpij = z1i - z1j;
double xpik = x1i - x1k;
double ypik = y1i - y1k;
double zpik = z1i - z1k;
double rijsq = xpij*xpij + ypij*ypij + zpij*zpij;
double riksq = xpik*xpik + ypik*ypik + zpik*zpik;
double dij, dik;
if (use_indexed) {
dij = sh_constr[itype*2] - rijsq;
dik = sh_constr[itype*2+1] - riksq;
} else {
dij = d_setup.trip_constr[itype*2] - rijsq;
dik = d_setup.trip_constr[itype*2+1] - riksq;
}
double rijrijp = xrij*xpij + yrij*ypij + zrij*zpij;
double rijrikp = xrij*xpik + yrij*ypik + zrij*zpik;
double rikrijp = xpij*xrik + ypij*yrik + zpij*zrik;
double rikrikp = xrik*xpik + yrik*ypik + zrik*zpik;
double dinv=0.5/(rijrijp*rikrikp*mij*mik - rijrikp*rikrijp*mmi*mmi);
double a12 = dinv*( rikrikp*mik*(dij) - rikrijp*mmi*(dik));
double a13 = dinv*(-mmi*rijrikp*(dij) + rijrijp*mij*(dik));
double a120 = 0.0;
double a130 = 0.0;
int niter = 0;
while ((fabs(a120-a12) > d_setup.shake_tol ||
fabs(a130-a13) > d_setup.shake_tol) &&
(niter < d_setup.max_niter)) {
a120 = a12;
a130 = a13;
double a12corr = acorr1*a12*a12 + acorr2*a12*a13 + acorr3*a13*a13;
double a13corr = acorr4*a12*a12 + acorr5*a12*a13 + acorr6*a13*a13;
a12 = dinv*( rikrikp*mik*(dij-a12corr) - rikrijp*mmi*(dik-a13corr));
a13 = dinv*(-mmi*rijrikp*(dij-a12corr) + rijrijp*mij*(dik-a13corr));
niter++;
}
x1i += mmi*(xrij*a12+xrik*a13);
x1j -= mmj*(xrij*a12);
x1k -= mmk*(xrik*a13);
y1i += mmi*(yrij*a12+yrik*a13);
y1j -= mmj*(yrij*a12);
y1k -= mmk*(yrik*a13);
z1i += mmi*(zrij*a12+zrik*a13);
z1j -= mmj*(zrij*a12);
z1k -= mmk*(zrik*a13);
d_setup.xyz2[0][i] = x1i;
d_setup.xyz2[1][i] = y1i;
d_setup.xyz2[2][i] = z1i;
d_setup.xyz2[0][j] = x1j;
d_setup.xyz2[1][j] = y1j;
d_setup.xyz2[2][j] = z1j;
d_setup.xyz2[0][k] = x1k;
d_setup.xyz2[1][k] = y1k;
d_setup.xyz2[2][k] = z1k;
}
template<bool use_indexed>
__forceinline__ __device__ void quad_calc(const int imol,
const double* __restrict__ sh_constr,
const double* __restrict__ sh_mass) {
int i, j, k, l, itype;
if (use_indexed) {
dihe_t dihe = d_setup.quad_indtype[imol];
i = dihe.i;
j = dihe.j;
k = dihe.k;
l = dihe.l;
itype = dihe.itype;
} else {
int4 ind = d_setup.quad_ind[imol];
i = ind.x;
j = ind.y;
k = ind.z;
l = ind.w;
itype = imol;
}
// Load coordinates
double x0i = load_coord<0, 0>(i);
double y0i = load_coord<0, 1>(i);
double z0i = load_coord<0, 2>(i);
double x0j = load_coord<0, 0>(j);
double y0j = load_coord<0, 1>(j);
double z0j = load_coord<0, 2>(j);
double x0k = load_coord<0, 0>(k);
double y0k = load_coord<0, 1>(k);
double z0k = load_coord<0, 2>(k);
double x0l = load_coord<0, 0>(l);
double y0l = load_coord<0, 1>(l);
double z0l = load_coord<0, 2>(l);
double x1i = load_coord<1, 0>(i);
double y1i = load_coord<1, 1>(i);
double z1i = load_coord<1, 2>(i);
double x1j = load_coord<1, 0>(j);
double y1j = load_coord<1, 1>(j);
double z1j = load_coord<1, 2>(j);
double x1k = load_coord<1, 0>(k);
double y1k = load_coord<1, 1>(k);
double z1k = load_coord<1, 2>(k);
double x1l = load_coord<1, 0>(l);
double y1l = load_coord<1, 1>(l);
double z1l = load_coord<1, 2>(l);
double xrij = x0i - x0j;
double yrij = y0i - y0j;
double zrij = z0i - z0j;
double xrik = x0i - x0k;
double yrik = y0i - y0k;
double zrik = z0i - z0k;
double xril = x0i - x0l;
double yril = y0i - y0l;
double zril = z0i - z0l;
//i2 = ammi_ind(iconst)
double rrijsq = xrij*xrij + yrij*yrij + zrij*zrij;
double rriksq = xrik*xrik + yrik*yrik + zrik*zrik;
double rrilsq = xril*xril + yril*yril + zril*zril;
double rijrik = xrij*xrik + yrij*yrik + zrij*zrik;
double rijril = xrij*xril + yrij*yril + zrij*zril;
double rikril = xrik*xril + yrik*yril + zrik*zril;
double mmi, mmj, mmk, mml, mij, mik, mil;
if (use_indexed) {
mmi = sh_mass[itype*7];
mmj = sh_mass[itype*7+1];
mmk = sh_mass[itype*7+2];
mml = sh_mass[itype*7+3];
mij = sh_mass[itype*7+4];
mik = sh_mass[itype*7+5];
mil = sh_mass[itype*7+6];
} else {
mmi = d_setup.quad_mass[itype*7];
mmj = d_setup.quad_mass[itype*7+1];
mmk = d_setup.quad_mass[itype*7+2];
mml = d_setup.quad_mass[itype*7+3];
mij = d_setup.quad_mass[itype*7+4];
mik = d_setup.quad_mass[itype*7+5];
mil = d_setup.quad_mass[itype*7+6];
}
double acorr1 = mij*mij*rrijsq;
double acorr2 = 2.0*mij*mmi*rijrik;
double acorr3 = mmi*mmi*rriksq;
double acorr4 = mmi*mmi*rrijsq;
double acorr5 = 2.0*mik*mmi*rijrik;
double acorr6 = mik*mik*rriksq;
double acorr7 = 2.0*mij*mmi*rijril;
double acorr8 = 2.0*mmi*mmi*rikril;
double acorr9 = mmi*mmi*rrilsq;
double acorr10 = 2.0*mmi*mmi*rijril;
double acorr11 = 2.0*mmi*mik*rikril;
double acorr12 = 2.0*mmi*mmi*rijrik;
double acorr13 = 2.0*mmi*mil*rijril;
double acorr14 = 2.0*mmi*mil*rikril;
double acorr15 = mil*mil*rrilsq;
double xpij = x1i - x1j;
double ypij = y1i - y1j;
double zpij = z1i - z1j;
double xpik = x1i - x1k;
double ypik = y1i - y1k;
double zpik = z1i - z1k;
double xpil = x1i - x1l;
double ypil = y1i - y1l;
double zpil = z1i - z1l;
double rijsq = xpij*xpij + ypij*ypij + zpij*zpij;
double riksq = xpik*xpik + ypik*ypik + zpik*zpik;
double rilsq = xpil*xpil + ypil*ypil + zpil*zpil;
double dij, dik, dil;
if (use_indexed) {
dij = sh_constr[itype*3] - rijsq;
dik = sh_constr[itype*3+1] - riksq;
dil = sh_constr[itype*3+2] - rilsq;
} else {
dij = d_setup.quad_constr[itype*3] - rijsq;
dik = d_setup.quad_constr[itype*3+1] - riksq;
dil = d_setup.quad_constr[itype*3+2] - rilsq;
}
double rijrijp = xrij*xpij + yrij*ypij + zrij*zpij;
double rijrikp = xrij*xpik + yrij*ypik + zrij*zpik;
double rijrilp = xrij*xpil + yrij*ypil + zrij*zpil;
double rikrijp = xrik*xpij + yrik*ypij + zrik*zpij;
double rikrikp = xrik*xpik + yrik*ypik + zrik*zpik;
double rikrilp = xrik*xpil + yrik*ypil + zrik*zpil;
double rilrijp = xril*xpij + yril*ypij + zril*zpij;
double rilrikp = xril*xpik + yril*ypik + zril*zpik;
double rilrilp = xril*xpil + yril*ypil + zril*zpil;
double d1 = mik*mil*rikrikp*rilrilp - mmi*mmi*rikrilp*rilrikp;
double d2 = mmi*mil*rikrijp*rilrilp - mmi*mmi*rikrilp*rilrijp;
double d3 = mmi*mmi*rikrijp*rilrikp - mik*mmi*rikrikp*rilrijp;
double d4 = mmi*mil*rijrikp*rilrilp - mmi*mmi*rijrilp*rilrikp;
double d5 = mij*mil*rijrijp*rilrilp - mmi*mmi*rijrilp*rilrijp;
double d6 = mij*mmi*rijrijp*rilrikp - mmi*mmi*rijrikp*rilrijp;
double d7 = mmi*mmi*rijrikp*rikrilp - mmi*mik*rijrilp*rikrikp;
double d8 = mij*mmi*rijrijp*rikrilp - mmi*mmi*rijrilp*rikrijp;
double d9 = mij*mik*rijrijp*rikrikp - mmi*mmi*rijrikp*rikrijp;
double dinv = 0.5/(rijrijp*mij*d1 - mmi*rijrikp*d2 + mmi*rijrilp*d3);
double a12 = dinv*( d1*dij - d2*dik + d3*dil);
double a13 = dinv*(-d4*dij + d5*dik - d6*dil);
double a14 = dinv*( d7*dij - d8*dik + d9*dil);
double a120 = 0.0;
double a130 = 0.0;
double a140 = 0.0;
int niter = 0;
while ((fabs(a120-a12) > d_setup.shake_tol ||
fabs(a130-a13) > d_setup.shake_tol ||
fabs(a140-a14) > d_setup.shake_tol) &&
(niter < d_setup.max_niter)) {
a120 = a12;
a130 = a13;
a140 = a14;
double a12corr = acorr1*a12*a12 + acorr2*a12*a13 + acorr3*a13*a13
+ acorr7*a12*a14 + acorr8*a13*a14 + acorr9*a14*a14;
double a13corr = acorr4*a12*a12 + acorr5*a12*a13 + acorr6*a13*a13
+ acorr10*a12*a14 + acorr11*a13*a14 + acorr9*a14*a14;
double a14corr = acorr4*a12*a12 + acorr12*a12*a13 + acorr3*a13*a13
+ acorr13*a12*a14 + acorr14*a13*a14 + acorr15*a14*a14;
a12 = dinv*( d1*(dij-a12corr) - d2*(dik-a13corr) + d3*(dil-a14corr));
a13 = dinv*( -d4*(dij-a12corr) + d5*(dik-a13corr) - d6*(dil-a14corr));
a14 = dinv*( d7*(dij-a12corr) - d8*(dik-a13corr) + d9*(dil-a14corr));
niter++;
}
x1i += mmi*(xrij*a12+xrik*a13+xril*a14);
y1i += mmi*(yrij*a12+yrik*a13+yril*a14);
z1i += mmi*(zrij*a12+zrik*a13+zril*a14);
x1j -= mmj*(xrij*a12);
y1j -= mmj*(yrij*a12);
z1j -= mmj*(zrij*a12);
x1k -= mmk*(xrik*a13);
y1k -= mmk*(yrik*a13);
z1k -= mmk*(zrik*a13);
x1l -= mml*(xril*a14);
y1l -= mml*(yril*a14);
z1l -= mml*(zril*a14);
d_setup.xyz2[0][i] = x1i;
d_setup.xyz2[1][i] = y1i;
d_setup.xyz2[2][i] = z1i;
d_setup.xyz2[0][j] = x1j;
d_setup.xyz2[1][j] = y1j;
d_setup.xyz2[2][j] = z1j;
d_setup.xyz2[0][k] = x1k;
d_setup.xyz2[1][k] = y1k;
d_setup.xyz2[2][k] = z1k;
d_setup.xyz2[0][l] = x1l;
d_setup.xyz2[1][l] = y1l;
d_setup.xyz2[2][l] = z1l;
}
__forceinline__ __device__ void solvent_calc(int imol) {
solvent_t ind = d_setup.solvent_ind[imol];
// Load coordinates
double x0i = load_coord<0, 0>(ind.i);
double y0i = load_coord<0, 1>(ind.i);
double z0i = load_coord<0, 2>(ind.i);
double x0j = load_coord<0, 0>(ind.j);
double y0j = load_coord<0, 1>(ind.j);
double z0j = load_coord<0, 2>(ind.j);
double x0k = load_coord<0, 0>(ind.k);
double y0k = load_coord<0, 1>(ind.k);
double z0k = load_coord<0, 2>(ind.k);
double x1i = load_coord<1, 0>(ind.i);
double y1i = load_coord<1, 1>(ind.i);
double z1i = load_coord<1, 2>(ind.i);
double x1j = load_coord<1, 0>(ind.j);
double y1j = load_coord<1, 1>(ind.j);
double z1j = load_coord<1, 2>(ind.j);
double x1k = load_coord<1, 0>(ind.k);
double y1k = load_coord<1, 1>(ind.k);
double z1k = load_coord<1, 2>(ind.k);
//
// Convert to primed coordinates
//
// Calculate center of mass for (x1, y1, z1)
double xcm = x1i*d_setup.mO_div_mH2O + (x1j + x1k)*d_setup.mH_div_mH2O;
double ycm = y1i*d_setup.mO_div_mH2O + (y1j + y1k)*d_setup.mH_div_mH2O;
double zcm = z1i*d_setup.mO_div_mH2O + (z1j + z1k)*d_setup.mH_div_mH2O;
// Calculate (x1, y1, z1) with center of mass at origin
double xa1 = x1i - xcm;
double ya1 = y1i - ycm;
double za1 = z1i - zcm;
double xb1 = x1j - xcm;
double yb1 = y1j - ycm;
double zb1 = z1j - zcm;
double xc1 = x1k - xcm;
double yc1 = y1k - ycm;
double zc1 = z1k - zcm;
double xb0 = x0j - x0i;
double yb0 = y0j - y0i;
double zb0 = z0j - z0i;
double xc0 = x0k - x0i;
double yc0 = y0k - y0i;
double zc0 = z0k - z0i;
// (xb0, yb0, zb0), (xc0, yc0, zc0), and (xa1, ya1, za1) define the primed coordinate set:
// * X'Y' plane is parallel to the plane defined by (xb0, yb0, zb0) and (xc0, yc0, zc0)
// * Y'Z' plane contains (xa1, ya1, za1)
double xakszd = yb0 * zc0 - zb0 * yc0;
double yakszd = zb0 * xc0 - xb0 * zc0;
double zakszd = xb0 * yc0 - yb0 * xc0;
double xaksxd = ya1 * zakszd - za1 * yakszd;
double yaksxd = za1 * xakszd - xa1 * zakszd;
double zaksxd = xa1 * yakszd - ya1 * xakszd;
double xaksyd = yakszd * zaksxd - zakszd * yaksxd;
double yaksyd = zakszd * xaksxd - xakszd * zaksxd;
double zaksyd = xakszd * yaksxd - yakszd * xaksxd;
double axlng_inv = rsqrt(xaksxd * xaksxd + yaksxd * yaksxd + zaksxd * zaksxd);
double aylng_inv = rsqrt(xaksyd * xaksyd + yaksyd * yaksyd + zaksyd * zaksyd);
double azlng_inv = rsqrt(xakszd * xakszd + yakszd * yakszd + zakszd * zakszd);
double trans11 = xaksxd * axlng_inv;
double trans21 = yaksxd * axlng_inv;
double trans31 = zaksxd * axlng_inv;
double trans12 = xaksyd * aylng_inv;
double trans22 = yaksyd * aylng_inv;
double trans32 = zaksyd * aylng_inv;
double trans13 = xakszd * azlng_inv;
double trans23 = yakszd * azlng_inv;
double trans33 = zakszd * azlng_inv;
// Calculate necessary primed coordinates
double xb0p = trans11 * xb0 + trans21 * yb0 + trans31 * zb0;
double yb0p = trans12 * xb0 + trans22 * yb0 + trans32 * zb0;
double xc0p = trans11 * xc0 + trans21 * yc0 + trans31 * zc0;
double yc0p = trans12 * xc0 + trans22 * yc0 + trans32 * zc0;
double za1p = trans13 * xa1 + trans23 * ya1 + trans33 * za1;
double xb1p = trans11 * xb1 + trans21 * yb1 + trans31 * zb1;
double yb1p = trans12 * xb1 + trans22 * yb1 + trans32 * zb1;
double zb1p = trans13 * xb1 + trans23 * yb1 + trans33 * zb1;
double xc1p = trans11 * xc1 + trans21 * yc1 + trans31 * zc1;
double yc1p = trans12 * xc1 + trans22 * yc1 + trans32 * zc1;
double zc1p = trans13 * xc1 + trans23 * yc1 + trans33 * zc1;
//
// Calculate rotation angles (phi, psi, theta)
//
double sinphi = za1p * d_setup.ra_inv;
double cosphi = sqrt(1.0 - sinphi*sinphi);
double sinpsi = (zb1p - zc1p) / (d_setup.rc2 * cosphi);
double cospsi = sqrt(1.0 - sinpsi*sinpsi);
double ya2p = d_setup.ra * cosphi;
double xb2p = - d_setup.rc * cospsi;
double yb2p = - d_setup.rb * cosphi - d_setup.rc *sinpsi * sinphi;
double yc2p = - d_setup.rb * cosphi + d_setup.rc *sinpsi * sinphi;
// xb2p = -half * sqrt(hhhh - (yb2p-yc2p) * (yb2p-yc2p) - (zb1p-zc1p) * (zb1p-zc1p))
double alpha = (xb2p * (xb0p-xc0p) + yb0p * yb2p + yc0p * yc2p);
double beta = (xb2p * (yc0p-yb0p) + xb0p * yb2p + xc0p * yc2p);
double gamma = xb0p * yb1p - xb1p * yb0p + xc0p * yc1p - xc1p * yc0p;
double alpha_beta = alpha * alpha + beta * beta;
double sintheta = (alpha*gamma - beta * sqrt(alpha_beta - gamma*gamma)) / alpha_beta;
double costheta = sqrt(1.0 - sintheta*sintheta);
double xa3p = -ya2p * sintheta;
double ya3p = ya2p * costheta;
double za3p = za1p;
double xb3p = xb2p * costheta - yb2p * sintheta;
double yb3p = xb2p * sintheta + yb2p * costheta;
double zb3p = zb1p;
double xc3p = -xb2p * costheta - yc2p * sintheta;
double yc3p = -xb2p * sintheta + yc2p * costheta;
double zc3p = zc1p;
d_setup.xyz2[0][ind.i] = xcm + trans11 * xa3p + trans12 * ya3p + trans13 * za3p;
d_setup.xyz2[1][ind.i] = ycm + trans21 * xa3p + trans22 * ya3p + trans23 * za3p;
d_setup.xyz2[2][ind.i] = zcm + trans31 * xa3p + trans32 * ya3p + trans33 * za3p;
d_setup.xyz2[0][ind.j] = xcm + trans11 * xb3p + trans12 * yb3p + trans13 * zb3p;
d_setup.xyz2[1][ind.j] = ycm + trans21 * xb3p + trans22 * yb3p + trans23 * zb3p;
d_setup.xyz2[2][ind.j] = zcm + trans31 * xb3p + trans32 * yb3p + trans33 * zb3p;
d_setup.xyz2[0][ind.k] = xcm + trans11 * xc3p + trans12 * yc3p + trans13 * zc3p;
d_setup.xyz2[1][ind.k] = ycm + trans21 * xc3p + trans22 * yc3p + trans23 * zc3p;
d_setup.xyz2[2][ind.k] = zcm + trans31 * xc3p + trans32 * yc3p + trans33 * zc3p;
}
template<bool use_indexed>
__global__ void all_kernels() {
// Shared memory, only used when use_indexed = true
// Requires: npair_type*sizeof(double)*3 + ntrip_type*sizeof(double)*7 + nquad_type*sizeof(double)*10
extern __shared__ double sh_constr_mass[];
double* __restrict__ sh_pair_constr;
double* __restrict__ sh_pair_mass;
double* __restrict__ sh_trip_constr;
double* __restrict__ sh_trip_mass;
double* __restrict__ sh_quad_constr;
double* __restrict__ sh_quad_mass;
if (use_indexed) {
int pos = 0;
sh_pair_constr = &sh_constr_mass[pos];
pos += d_setup.npair_type;
sh_pair_mass = &sh_constr_mass[pos];
pos += d_setup.npair_type*2;
sh_trip_constr = &sh_constr_mass[pos];
pos += d_setup.ntrip_type*2;
sh_trip_mass = &sh_constr_mass[pos];
pos += d_setup.ntrip_type*5;
sh_quad_constr = &sh_constr_mass[pos];
pos += d_setup.nquad_type*3;
sh_quad_mass = &sh_constr_mass[pos];
int t;
t = threadIdx.x;
while (t < d_setup.ntot_type) {
sh_constr_mass[t] = d_setup.constr_mass[t];
t += blockDim.x;
}
__syncthreads();
}
const int imol = threadIdx.x + blockDim.x*blockIdx.x;
if (imol < d_setup.nsolvent) {
solvent_calc(imol);
} else if (imol < d_setup.nsolvent + d_setup.npair) {
pair_calc<use_indexed>(imol - d_setup.nsolvent,
sh_pair_constr, sh_pair_mass);
} else if (imol < d_setup.nsolvent + d_setup.npair + d_setup.ntrip) {
trip_calc<use_indexed>(imol - d_setup.nsolvent - d_setup.npair,
sh_trip_constr, sh_trip_mass);
} else if (imol < d_setup.nsolvent + d_setup.npair + d_setup.ntrip + d_setup.nquad) {
quad_calc<use_indexed>(imol - d_setup.nsolvent - d_setup.npair - d_setup.ntrip,
sh_quad_constr, sh_quad_mass);
}
}
__global__ void setup_list_kernel(const int npair, const int* __restrict__ pair_tbl,
const bond_t* __restrict__ global_pair_indtype,
const int ntrip, const int* __restrict__ trip_tbl,
const angle_t* __restrict__ global_trip_indtype,
const int nquad, const int* __restrict__ quad_tbl,
const dihe_t* __restrict__ global_quad_indtype,
const int nsolvent, const int* __restrict__ solvent_tbl,
const solvent_t* __restrict__ global_solvent_ind,
const int* __restrict__ glo2loc,
bond_t* __restrict__ pair_indtype,
angle_t* __restrict__ trip_indtype,
dihe_t* __restrict__ quad_indtype,
solvent_t* __restrict__ solvent_ind) {
int i = threadIdx.x + blockIdx.x*blockDim.x;
if (i < npair) {
bond_t pair = global_pair_indtype[pair_tbl[i]];
pair.i = glo2loc[pair.i];
pair.j = glo2loc[pair.j];
pair_indtype[i] = pair;
} else if (i < npair + ntrip) {
i -= npair;
angle_t trip = global_trip_indtype[trip_tbl[i]];
trip.i = glo2loc[trip.i];
trip.j = glo2loc[trip.j];
trip.k = glo2loc[trip.k];
trip_indtype[i] = trip;
} else if (i < npair + ntrip + nquad) {
i -= npair + ntrip;
dihe_t quad = global_quad_indtype[quad_tbl[i]];
quad.i = glo2loc[quad.i];
quad.j = glo2loc[quad.j];
quad.k = glo2loc[quad.k];
quad.l = glo2loc[quad.l];
quad_indtype[i] = quad;
} else if (i < npair + ntrip + nquad + nsolvent) {
i -= npair + ntrip + nquad;
solvent_t solvent = global_solvent_ind[solvent_tbl[i]];
solvent.i = glo2loc[solvent.i];
solvent.j = glo2loc[solvent.j];
solvent.k = glo2loc[solvent.k];
solvent_ind[i] = solvent;
}
}
//################################################################################################
//################################################################################################
//################################################################################################
//
// Class creator
//
HoloConst::HoloConst() {
max_niter = 1000;
shake_tol = 1.0e-8;
nsolvent = 0;
solvent_ind_len = 0;
solvent_ind = NULL;
npair = 0;
pair_ind_len = 0;
pair_ind = NULL;
pair_indtype_len = 0;
pair_indtype = NULL;
npair_type = 0;
pair_constr = NULL;
pair_mass = NULL;
ntrip = 0;
trip_ind_len = 0;
trip_ind = NULL;
trip_indtype_len = 0;
trip_indtype = NULL;
ntrip_type = 0;
trip_constr = NULL;
trip_mass = NULL;
nquad = 0;
quad_ind_len = 0;
quad_ind = NULL;
quad_indtype_len = 0;
quad_indtype = NULL;
nquad_type = 0;
quad_constr = NULL;
quad_mass = NULL;
ntot_type = 0;
constr_mass_len = 0;
constr_mass = NULL;
for (int i=0;i < 2;i++) {
for (int j=0;j < 3;j++) {
xyz_texref_pointer[i][j] = NULL;
texref_size[i][j] = 0;
}
}
if (get_cuda_arch() < 350) {
use_textures = true;
} else {
use_textures = false;
}
use_indexed = false;
use_settle = false;
}
//
// Class destructor
//
HoloConst::~HoloConst() {
for (int i=0;i < 2;i++)
for (int j=0;j < 3;j++)
if (xyz_texref_pointer[i][j] != NULL) {
cudaCheck(cudaUnbindTexture(get_xyz_texref(i,j)));
xyz_texref_pointer[i][j] = NULL;
}
if (solvent_ind != NULL) deallocate<solvent_t>(&solvent_ind);
if (pair_ind != NULL) deallocate<int2>(&pair_ind);
if (pair_indtype != NULL) deallocate<bond_t>(&pair_indtype);
if (trip_ind != NULL) deallocate<int3>(&trip_ind);
if (trip_indtype != NULL) deallocate<angle_t>(&trip_indtype);
if (quad_ind != NULL) deallocate<int4>(&quad_ind);
if (quad_indtype != NULL) deallocate<dihe_t>(&quad_indtype);
if (constr_mass != NULL) deallocate<double>(&constr_mass);
}
//
// Setup
//
void HoloConst::setup_solvent_parameters(double mO, double mH, double rOHsq, double rHHsq) {
double mH2O = mO + mH + mH;
mO_div_mH2O = mO/mH2O;
mH_div_mH2O = mH/mH2O;
// Setup ra, rb, rc
ra = mH_div_mH2O*sqrt(4.0*rOHsq - rHHsq);
ra_inv = 1.0/ra;
rb = ra*mO/(2.0*mH);
rc = sqrt(rHHsq)/2.0;
rc2 = 2.0*rc;
}
//
// Setup SETTLE parameters
//
// massP = mass of heavy atoms
// massH = mass of light (hydrogen) atoms
// rPHsq = squared distance between P and H
// rHHsq = squared distance between H and H
//
void HoloConst::setup_settle_parameters(const int nsettle,
const double* h_massP, const double* h_massH,
const double* h_rPHsq, const double* h_rHHsq) {
use_settle = true;
//this->nsettle = nsettle;
}
void HoloConst::realloc_constr_mass(const int npair_type, const int ntrip_type, const int nquad_type) {
ntot_type = npair_type*3 + ntrip_type*7 + nquad_type*10;
reallocate<double>(&constr_mass, &constr_mass_len, ntot_type, 1.0f);
int pos = 0;
pair_constr = &constr_mass[pos];
pos += npair_type;
pair_mass = &constr_mass[pos];
pos += npair_type*2;
trip_constr = &constr_mass[pos];
pos += ntrip_type*2;
trip_mass = &constr_mass[pos];
pos += ntrip_type*5;
quad_constr = &constr_mass[pos];
pos += nquad_type*3;
quad_mass = &constr_mass[pos];
}
//
// Setups all ind, mass, and constr arrays
//
void HoloConst::setup_ind_mass_constr(const int npair, const int2 *h_pair_ind,
const double *h_pair_constr, const double *h_pair_mass,
const int ntrip, const int3 *h_trip_ind,
const double *h_trip_constr, const double *h_trip_mass,
const int nquad, const int4 *h_quad_ind,
const double *h_quad_constr, const double *h_quad_mass,
const int nsolvent, const solvent_t *h_solvent_ind) {
use_indexed = false;
realloc_constr_mass(npair, ntrip, nquad);
// Copy ind, mass, and constr from CPU to GPU
set_pair(npair, h_pair_ind);
set_pair_type(npair, h_pair_constr, h_pair_mass);
set_trip(ntrip, h_trip_ind);
set_trip_type(ntrip, h_trip_constr, h_trip_mass);
set_quad(nquad, h_quad_ind);
set_quad_type(nquad, h_quad_constr, h_quad_mass);
set_solvent(nsolvent, h_solvent_ind);
}
//
// Setup indexed
//
void HoloConst::setup_indexed(const int npair, const bond_t* h_pair_indtype,
const int npair_type, const double* h_pair_constr, const double* h_pair_mass,
const int ntrip, const angle_t* h_trip_indtype,
const int ntrip_type, const double* h_trip_constr, const double* h_trip_mass,
const int nquad, const dihe_t* h_quad_indtype,
const int nquad_type, const double* h_quad_constr, const double* h_quad_mass,
const int nsolvent, const solvent_t* h_solvent_ind) {
use_indexed = true;
realloc_constr_mass(npair_type, ntrip_type, nquad_type);
set_pair(npair, h_pair_indtype);
set_pair_type(npair_type, h_pair_constr, h_pair_mass);
set_trip(ntrip, h_trip_indtype);
set_trip_type(ntrip_type, h_trip_constr, h_trip_mass);
set_quad(nquad, h_quad_indtype);
set_quad_type(nquad_type, h_quad_constr, h_quad_mass);
set_solvent(nsolvent, h_solvent_ind);
}
//
// Setup the list of holonomic constraints
//
// pair_tbl[0..npair-1] contains the indices that are constrained
// global_pair_indtype[] is the global list of indices
//
void HoloConst::setup_list(const int* glo2loc,
const int npair, const int* pair_tbl, const bond_t* global_pair_indtype,
const int ntrip, const int* trip_tbl, const angle_t* global_trip_indtype,
const int nquad, const int* quad_tbl, const dihe_t* global_quad_indtype,
const int nsolvent, const int* solvent_tbl, const solvent_t* global_solvent_ind,
cudaStream_t stream) {
this->npair = npair;
this->ntrip = ntrip;
this->nquad = nquad;
this->nsolvent = nsolvent;
int ntot = nsolvent + npair + ntrip + nquad;
if (ntot == 0) return;
reallocate<bond_t>(&pair_indtype, &pair_indtype_len, npair, 1.2f);
reallocate<angle_t>(&trip_indtype, &trip_indtype_len, ntrip, 1.2f);
reallocate<dihe_t>(&quad_indtype, &quad_indtype_len, nquad, 1.2f);
reallocate<solvent_t>(&solvent_ind, &solvent_ind_len, nsolvent, 1.2f);
int nthread = 512;
int nblock = (ntot - 1)/nthread + 1;
setup_list_kernel<<< nblock, nthread, 0, stream >>>