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
|
/***************************************************************************
amoeba_ext.cpp
-------------------
Trung Dac Nguyen (Northwestern)
Functions for LAMMPS access to amoeba acceleration routines.
__________________________________________________________________________
This file is part of the LAMMPS Accelerator Library (LAMMPS_AL)
__________________________________________________________________________
begin :
email : trung.nguyen@northwestern.edu
***************************************************************************/
#include <iostream>
#include <cassert>
#include <cmath>
#include "lal_amoeba.h"
using namespace std;
using namespace LAMMPS_AL;
static Amoeba<PRECISION,ACC_PRECISION> AMOEBAMF;
// ---------------------------------------------------------------------------
// Allocate memory on host and device and copy constants to device
// ---------------------------------------------------------------------------
int amoeba_gpu_init(const int ntypes, const int max_amtype, const int max_amclass,
const double *host_pdamp, const double *host_thole,
const double *host_dirdamp, const int *host_amtype2class,
const double *host_special_hal,
const double *host_special_repel,
const double *host_special_disp,
const double *host_special_mpole,
const double *host_special_polar_wscale,
const double *host_special_polar_piscale,
const double *host_special_polar_pscale,
const double *host_csix, const double *host_adisp,
const int nlocal, const int nall, const int max_nbors,
const int maxspecial, const int maxspecial15,
const double cell_size, int &gpu_mode, FILE *screen,
const double polar_dscale, const double polar_uscale) {
AMOEBAMF.clear();
gpu_mode=AMOEBAMF.device->gpu_mode();
double gpu_split=AMOEBAMF.device->particle_split();
int first_gpu=AMOEBAMF.device->first_device();
int last_gpu=AMOEBAMF.device->last_device();
int world_me=AMOEBAMF.device->world_me();
int gpu_rank=AMOEBAMF.device->gpu_rank();
int procs_per_gpu=AMOEBAMF.device->procs_per_gpu();
AMOEBAMF.device->init_message(screen,"amoeba",first_gpu,last_gpu);
bool message=false;
if (AMOEBAMF.device->replica_me()==0 && screen)
message=true;
if (message) {
fprintf(screen,"Initializing GPU and compiling on process 0...");
fflush(screen);
}
int init_ok=0;
if (world_me==0)
init_ok=AMOEBAMF.init(ntypes, max_amtype, max_amclass,
host_pdamp, host_thole, host_dirdamp,
host_amtype2class, host_special_hal,
host_special_repel, host_special_disp,
host_special_mpole, host_special_polar_wscale,
host_special_polar_piscale, host_special_polar_pscale,
host_csix, host_adisp, nlocal, nall, max_nbors,
maxspecial, maxspecial15, cell_size, gpu_split,
screen, polar_dscale, polar_uscale);
AMOEBAMF.device->world_barrier();
if (message)
fprintf(screen,"Done.\n");
for (int i=0; i<procs_per_gpu; i++) {
if (message) {
if (last_gpu-first_gpu==0)
fprintf(screen,"Initializing GPU %d on core %d...",first_gpu,i);
else
fprintf(screen,"Initializing GPUs %d-%d on core %d...",first_gpu,
last_gpu,i);
fflush(screen);
}
if (gpu_rank==i && world_me!=0)
init_ok=AMOEBAMF.init(ntypes, max_amtype, max_amclass,
host_pdamp, host_thole, host_dirdamp,
host_amtype2class, host_special_hal,
host_special_repel, host_special_disp,
host_special_mpole, host_special_polar_wscale,
host_special_polar_piscale, host_special_polar_pscale,
host_csix, host_adisp, nlocal, nall, max_nbors,
maxspecial, maxspecial15, cell_size, gpu_split,
screen, polar_dscale, polar_uscale);
AMOEBAMF.device->gpu_barrier();
if (message)
fprintf(screen,"Done.\n");
}
if (message)
fprintf(screen,"\n");
if (init_ok==0)
AMOEBAMF.estimate_gpu_overhead();
return init_ok;
}
void amoeba_gpu_clear() {
AMOEBAMF.clear();
}
int** amoeba_gpu_precompute(const int ago, const int inum_full, const int nall,
double **host_x, int *host_type, int *host_amtype,
int *host_amgroup, double **host_rpole, double ** /*host_uind*/,
double ** /*host_uinp*/, double * /*host_pval*/,
double *sublo, double *subhi, tagint *tag,
int **nspecial, tagint **special,
int *nspecial15, tagint **special15,
const bool eflag_in, const bool vflag_in,
const bool eatom, const bool vatom, int &host_start,
int **ilist, int **jnum, const double cpu_time,
bool &success, double *host_q, double *boxlo, double *prd) {
return AMOEBAMF.precompute(ago, inum_full, nall, host_x, host_type,
host_amtype, host_amgroup, host_rpole,
nullptr, nullptr, nullptr, sublo, subhi, tag,
nspecial, special, nspecial15, special15,
eflag_in, vflag_in, eatom, vatom,
host_start, ilist, jnum, cpu_time,
success, host_q, boxlo, prd);
}
void amoeba_gpu_compute_multipole_real(const int ago, const int inum_full,
const int nall, double **host_x, int *host_type,
int *host_amtype, int *host_amgroup, double **host_rpole,
double *sublo, double *subhi, tagint *tag, int **nspecial,
tagint **special, int *nspecial15, tagint** special15,
const bool eflag, const bool vflag, const bool eatom,
const bool vatom, int &host_start,
int **ilist, int **jnum, const double cpu_time,
bool &success, const double aewald, const double felec, const double off2,
double *host_q, double *boxlo, double *prd, void **tep_ptr) {
AMOEBAMF.compute_multipole_real(ago, inum_full, nall, host_x, host_type,
host_amtype, host_amgroup, host_rpole, nullptr, sublo, subhi,
tag, nspecial, special, nspecial15, special15,
eflag, vflag, eatom, vatom, host_start, ilist, jnum,
cpu_time, success, aewald, felec, off2, host_q, boxlo, prd, tep_ptr);
}
void amoeba_gpu_compute_udirect2b(int *host_amtype, int *host_amgroup, double **host_rpole,
double **host_uind, double **host_uinp,
const double aewald, const double off2, void **fieldp_ptr) {
AMOEBAMF.compute_udirect2b(host_amtype, host_amgroup, host_rpole, host_uind, host_uinp, nullptr,
aewald, off2, fieldp_ptr);
}
void amoeba_gpu_compute_umutual2b(int *host_amtype, int *host_amgroup, double **host_rpole,
double **host_uind, double **host_uinp,
const double aewald, const double off2, void **fieldp_ptr) {
AMOEBAMF.compute_umutual2b(host_amtype, host_amgroup, host_rpole, host_uind, host_uinp, nullptr,
aewald, off2, fieldp_ptr);
}
void amoeba_gpu_update_fieldp(void **fieldp_ptr) {
AMOEBAMF.update_fieldp(fieldp_ptr);
}
void amoeba_gpu_compute_polar_real(int *host_amtype, int *host_amgroup, double **host_rpole,
double **host_uind, double **host_uinp,
const bool eflag_in, const bool vflag_in,
const bool eatom, const bool vatom,
const double aewald, const double felec, const double off2,
void **tep_ptr) {
AMOEBAMF.compute_polar_real(host_amtype, host_amgroup, host_rpole, host_uind, host_uinp, nullptr,
eflag_in, vflag_in, eatom, vatom, aewald, felec, off2, tep_ptr);
}
void amoeba_gpu_precompute_kspace(const int inum_full, const int bsorder,
double ***host_thetai1, double ***host_thetai2,
double ***host_thetai3, int** igrid,
const int nzlo_out, const int nzhi_out,
const int nylo_out, const int nyhi_out,
const int nxlo_out, const int nxhi_out) {
AMOEBAMF.precompute_kspace(inum_full, bsorder, host_thetai1, host_thetai2, host_thetai3, igrid,
nzlo_out, nzhi_out, nylo_out, nyhi_out, nxlo_out, nxhi_out);
}
void amoeba_gpu_fphi_uind(double ****host_grid_brick, void **host_fdip_phi1,
void **host_fdip_phi2, void **host_fdip_sum_phi) {
AMOEBAMF.compute_fphi_uind(host_grid_brick, host_fdip_phi1,
host_fdip_phi2, host_fdip_sum_phi);
}
void amoeba_gpu_fphi_mpole(double ***host_grid_brick, void **host_fphi, const double felec) {
AMOEBAMF.compute_fphi_mpole(host_grid_brick, host_fphi, felec);
}
void amoeba_setup_fft(const int numel, const int element_type) {
AMOEBAMF.setup_fft(numel, element_type);
}
void amoeba_compute_fft1d(void* in, void* out, const int numel, const int mode) {
AMOEBAMF.compute_fft1d(in, out, numel, mode);
}
double amoeba_gpu_bytes() {
return AMOEBAMF.host_memory_usage();
}
|
