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
|
// **************************************************************************
// mie.cu
// -------------------
// Trung Dac Nguyen (ORNL)
//
// Device code for acceleration of the mie pair style
//
// __________________________________________________________________________
// This file is part of the LAMMPS Accelerator Library (LAMMPS_AL)
// __________________________________________________________________________
//
// begin :
// email : nguyentd@ornl.gov
// ***************************************************************************
#if defined(NV_KERNEL) || defined(USE_HIP)
#include "lal_aux_fun1.h"
#ifndef _DOUBLE_DOUBLE
_texture( pos_tex,float4);
#else
_texture_2d( pos_tex,int4);
#endif
#else
#define pos_tex x_
#endif
__kernel void k_mie(const __global numtyp4 *restrict x_,
const __global numtyp4 *restrict mie1,
const __global numtyp4 *restrict mie3,
const int lj_types,
const __global numtyp *restrict sp_lj_in,
const __global int *dev_nbor,
const __global int *dev_packed,
__global acctyp4 *restrict ans,
__global acctyp *restrict engv,
const int eflag, const int vflag, const int inum,
const int nbor_pitch, const int t_per_atom) {
int tid, ii, offset;
atom_info(t_per_atom,ii,tid,offset);
__local numtyp sp_lj[4];
int n_stride;
local_allocate_store_pair();
sp_lj[0]=sp_lj_in[0];
sp_lj[1]=sp_lj_in[1];
sp_lj[2]=sp_lj_in[2];
sp_lj[3]=sp_lj_in[3];
acctyp4 f;
f.x=(acctyp)0; f.y=(acctyp)0; f.z=(acctyp)0;
acctyp energy, virial[6];
if (EVFLAG) {
energy=(acctyp)0;
for (int i=0; i<6; i++) virial[i]=(acctyp)0;
}
if (ii<inum) {
int nbor, nbor_end;
int i, numj;
nbor_info(dev_nbor,dev_packed,nbor_pitch,t_per_atom,ii,offset,i,numj,
n_stride,nbor_end,nbor);
numtyp4 ix; fetch4(ix,i,pos_tex); //x_[i];
int itype=ix.w;
numtyp factor_lj;
for ( ; nbor<nbor_end; nbor+=n_stride) {
int j=dev_packed[nbor];
factor_lj = sp_lj[sbmask(j)];
j &= NEIGHMASK;
numtyp4 jx; fetch4(jx,j,pos_tex); //x_[j];
int jtype=jx.w;
// Compute r12
numtyp delx = ix.x-jx.x;
numtyp dely = ix.y-jx.y;
numtyp delz = ix.z-jx.z;
numtyp rsq = delx*delx+dely*dely+delz*delz;
int mtype=itype*lj_types+jtype;
if (rsq<mie3[mtype].w) {
numtyp r2inv = ucl_recip(rsq);
numtyp rgamA = pow(r2inv,(mie1[mtype].z/(numtyp)2.0));
numtyp rgamR = pow(r2inv,(mie1[mtype].w/(numtyp)2.0));
numtyp forcemie = (mie1[mtype].x*rgamR - mie1[mtype].y*rgamA);
numtyp force = factor_lj*forcemie*r2inv;
f.x+=delx*force;
f.y+=dely*force;
f.z+=delz*force;
if (EVFLAG && eflag) {
numtyp e=(mie3[mtype].x*rgamR - mie3[mtype].y*rgamA) -
mie3[mtype].z;
energy+=factor_lj*e;
}
if (EVFLAG && vflag) {
virial[0] += delx*delx*force;
virial[1] += dely*dely*force;
virial[2] += delz*delz*force;
virial[3] += delx*dely*force;
virial[4] += delx*delz*force;
virial[5] += dely*delz*force;
}
}
} // for nbor
} // if ii
store_answers(f,energy,virial,ii,inum,tid,t_per_atom,offset,eflag,vflag,
ans,engv);
}
__kernel void k_mie_fast(const __global numtyp4 *restrict x_,
const __global numtyp4 *restrict mie1_in,
const __global numtyp4 *restrict mie3_in,
const __global numtyp *restrict sp_lj_in,
const __global int *dev_nbor,
const __global int *dev_packed,
__global acctyp4 *restrict ans,
__global acctyp *restrict engv,
const int eflag, const int vflag, const int inum,
const int nbor_pitch, const int t_per_atom) {
int tid, ii, offset;
atom_info(t_per_atom,ii,tid,offset);
__local numtyp4 mie1[MAX_SHARED_TYPES*MAX_SHARED_TYPES];
__local numtyp4 mie3[MAX_SHARED_TYPES*MAX_SHARED_TYPES];
__local numtyp sp_lj[4];
int n_stride;
local_allocate_store_pair();
if (tid<4)
sp_lj[tid]=sp_lj_in[tid];
if (tid<MAX_SHARED_TYPES*MAX_SHARED_TYPES) {
mie1[tid]=mie1_in[tid];
mie3[tid]=mie3_in[tid];
}
acctyp4 f;
f.x=(acctyp)0; f.y=(acctyp)0; f.z=(acctyp)0;
acctyp energy, virial[6];
if (EVFLAG) {
energy=(acctyp)0;
for (int i=0; i<6; i++) virial[i]=(acctyp)0;
}
__syncthreads();
if (ii<inum) {
int nbor, nbor_end;
int i, numj;
nbor_info(dev_nbor,dev_packed,nbor_pitch,t_per_atom,ii,offset,i,numj,
n_stride,nbor_end,nbor);
numtyp4 ix; fetch4(ix,i,pos_tex); //x_[i];
int iw=ix.w;
int itype=fast_mul((int)MAX_SHARED_TYPES,iw);
numtyp factor_lj;
for ( ; nbor<nbor_end; nbor+=n_stride) {
int j=dev_packed[nbor];
factor_lj = sp_lj[sbmask(j)];
j &= NEIGHMASK;
numtyp4 jx; fetch4(jx,j,pos_tex); //x_[j];
int mtype=itype+jx.w;
// Compute r12
numtyp delx = ix.x-jx.x;
numtyp dely = ix.y-jx.y;
numtyp delz = ix.z-jx.z;
numtyp rsq = delx*delx+dely*dely+delz*delz;
if (rsq<mie3[mtype].w) {
numtyp r2inv = ucl_recip(rsq);
numtyp rgamA = pow(r2inv,(mie1[mtype].z/(numtyp)2.0));
numtyp rgamR = pow(r2inv,(mie1[mtype].w/(numtyp)2.0));
numtyp forcemie = (mie1[mtype].x*rgamR - mie1[mtype].y*rgamA);
numtyp force = factor_lj*forcemie*r2inv;
f.x+=delx*force;
f.y+=dely*force;
f.z+=delz*force;
if (EVFLAG && eflag) {
numtyp e=(mie3[mtype].x*rgamR - mie3[mtype].y*rgamA) -
mie3[mtype].z;
energy+=factor_lj*e;
}
if (EVFLAG && vflag) {
virial[0] += delx*delx*force;
virial[1] += dely*dely*force;
virial[2] += delz*delz*force;
virial[3] += delx*dely*force;
virial[4] += delx*delz*force;
virial[5] += dely*delz*force;
}
}
} // for nbor
} // if ii
store_answers(f,energy,virial,ii,inum,tid,t_per_atom,offset,eflag,vflag,
ans,engv);
}
|
