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
|
/***************************************************************************
zbl.h
-------------------
Trung Dac Nguyen (ORNL)
Class for acceleration of the zbl pair style.
__________________________________________________________________________
This file is part of the LAMMPS Accelerator Library (LAMMPS_AL)
__________________________________________________________________________
begin :
email : ndactrung@gmail.com
***************************************************************************/
#ifndef LAL_ZBL_H
#define LAL_ZBL_H
#include "lal_base_atomic.h"
namespace LAMMPS_AL {
template <class numtyp, class acctyp>
class ZBL : public BaseAtomic<numtyp, acctyp> {
public:
ZBL();
~ZBL();
/// Clear any previous data and set up for a new LAMMPS run
/** \param max_nbors initial number of rows in the neighbor matrix
* \param cell_size cutoff + skin
* \param gpu_split fraction of particles handled by device
*
* Returns:
* - 0 if successful
* - -1 if fix gpu not found
* - -3 if there is an out of memory error
* - -4 if the GPU library was not compiled for GPU
* - -5 Double precision is not supported on card **/
int init(const int ntypes, double **host_cutsq, double **host_sw1,
double **host_sw2, double **host_sw3, double **host_sw4, double **host_sw5,
double **host_d1a, double **host_d2a, double **host_d3a, double **host_d4a,
double **host_zze, double cut_globalsq, double cut_innersq, double cut_inner,
const int nlocal, const int nall, const int max_nbors,
const int maxspecial, const double cell_size,
const double gpu_split, FILE *screen);
/// Clear all host and device data
/** \note This is called at the beginning of the init() routine **/
void clear();
/// Returns memory usage on device per atom
int bytes_per_atom(const int max_nbors) const;
/// Total host memory used by library for pair style
double host_memory_usage() const;
// --------------------------- TYPE DATA --------------------------
/// coeff1.x = sw1, coeff1.y = sw2, coeff1.z = zze, coeff1.w = cutsq
UCL_D_Vec<numtyp4> coeff1;
/// coeff2.x = d1a, coeff2.y = d2a, coeff2.z = d3a, coeff2.w = d4a
UCL_D_Vec<numtyp4> coeff2;
/// coeff3.x = sw3, coeff3.y = sw4, coeff3.z = sw5;
UCL_D_Vec<numtyp4> coeff3;
/// If atom type constants fit in shared memory, use fast kernels
bool shared_types;
numtyp _cut_globalsq;
numtyp _cut_innersq;
numtyp _cut_inner;
/// Number of atom types
int _lj_types;
private:
bool _allocated;
int loop(const int eflag, const int vflag);
};
}
#endif
|
