forked from leopwma/spilady
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcheck_spin_GPU.cu
160 lines (131 loc) · 5.58 KB
/
check_spin_GPU.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
/********************************************************************************
*
* Copyright (C) 2015 Culham Centre for Fusion Energy,
* United Kingdom Atomic Energy Authority, Oxfordshire OX14 3DB, UK
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
********************************************************************************
*
* Program: SPILADY - A Spin-Lattice Dynamics Simulation Program
* Version: 1.0
* Date: Aug 2015
* Author: Pui-Wai (Leo) MA
* Contact: [email protected]
* Address: Culham Centre for Fusion Energy, OX14 3DB, United Kingdom
*
********************************************************************************/
#if (defined SDH || defined SDHL || defined SLDH || defined SLDHL || defined SLDNC) && defined GPU
#include "spilady.h"
#include "prototype_GPU.h"
/****************************************************************************
* GPU prototypes
****************************************************************************/
__global__ void LP1ChSp_part1(struct varGPU *var_ptr_d, struct atom_struct *first_atom_ptr_d,
double *ave_sx_ptr_d, double *ave_sy_ptr_d, double *ave_sz_ptr_d);
__global__ void LP1ChSp_part2(double *ave_sx_ptr_d, double *ave_sy_ptr_d, double *ave_sz_ptr_d);
/****************************************************************************
* CPU codes
****************************************************************************/
void check_spin_GPU(int current_step){
size_t size = no_of_MP*no_of_threads*sizeof(double);
ave_s = vec_zero();
ave_m = vec_zero();
double *ave_sx_ptr_d;
double *ave_sy_ptr_d;
double *ave_sz_ptr_d;
cudaMalloc((void**)&ave_sx_ptr_d, size);
cudaMalloc((void**)&ave_sy_ptr_d, size);
cudaMalloc((void**)&ave_sz_ptr_d, size);
LP1ChSp_part1<<<no_of_MP,no_of_threads>>>(var_ptr_d, first_atom_ptr_d,
ave_sx_ptr_d, ave_sy_ptr_d, ave_sz_ptr_d);
LP1ChSp_part2<<<no_of_MP,no_of_threads>>>(ave_sx_ptr_d, ave_sy_ptr_d, ave_sz_ptr_d);
cudaMemcpy(&ave_s.x, ave_sx_ptr_d, sizeof(double), cudaMemcpyDeviceToHost);
cudaMemcpy(&ave_s.y, ave_sy_ptr_d, sizeof(double), cudaMemcpyDeviceToHost);
cudaMemcpy(&ave_s.z, ave_sz_ptr_d, sizeof(double), cudaMemcpyDeviceToHost);
ave_s = vec_divide(ave_s, natom);
double ave_s0 = vec_length(ave_s);
ave_m = vec_times(-el_g, ave_s);
double ave_m0 = vec_length(ave_m);
char out_spn_front[] = "spn-";
char out_spn[256];
strcpy(out_spn,out_spn_front);
strcat(out_spn,out_body);
strcat(out_spn,".dat");
ofstream out_file(out_spn,ios::app);
out_file << setiosflags(ios::scientific) << setprecision(15);
out_file << current_step
<< " " << total_time
<< " " << ave_s.x
<< " " << ave_s.y
<< " " << ave_s.z
<< " " << ave_s0
<< " " << ave_m.x
<< " " << ave_m.y
<< " " << ave_m.z
<< " " << ave_m0
<< '\n';
out_file.close();
cudaFree(ave_sx_ptr_d);
cudaFree(ave_sy_ptr_d);
cudaFree(ave_sz_ptr_d);
}
void check_spin(int current_step){
check_spin_GPU(current_step);
}
/****************************************************************************
* GPU codes
****************************************************************************/
__global__ void LP1ChSp_part1(struct varGPU *var_ptr_d, struct atom_struct *first_atom_ptr_d,
double *ave_sx_ptr_d, double *ave_sy_ptr_d, double *ave_sz_ptr_d){
int i = blockIdx.x*blockDim.x + threadIdx.x;
*(ave_sx_ptr_d + i) = 0.0;
*(ave_sy_ptr_d + i) = 0.0;
*(ave_sz_ptr_d + i) = 0.0;
int area = blockDim.x*gridDim.x;
int k = (var_ptr_d->natom - 1)/area + 1;
for (int j = 0; j < k; ++j){
int m = i + j*area;
if (m < var_ptr_d->natom) {
struct atom_struct *atom_ptr;
atom_ptr = first_atom_ptr_d + m;
*(ave_sx_ptr_d + i) += atom_ptr->s.x;
*(ave_sy_ptr_d + i) += atom_ptr->s.y;
*(ave_sz_ptr_d + i) += atom_ptr->s.z;
}
}
__syncthreads();
}
__global__ void LP1ChSp_part2(double *ave_sx_ptr_d, double *ave_sy_ptr_d, double *ave_sz_ptr_d){
int depth = blockIdx.x*blockDim.x;
if (threadIdx.x == 0){
for (int j = 1; j < blockDim.x; ++j) *(ave_sx_ptr_d + depth) += *(ave_sx_ptr_d + depth + j);
}
if (threadIdx.x == 1){
for (int j = 1; j < blockDim.x; ++j) *(ave_sy_ptr_d + depth) += *(ave_sy_ptr_d + depth + j);
}
if (threadIdx.x == 2){
for (int j = 1; j < blockDim.x; ++j) *(ave_sz_ptr_d + depth) += *(ave_sz_ptr_d + depth + j);
}
__threadfence();
if (blockIdx.x == 0 && threadIdx.x == 0){
for (int j = 1; j < gridDim.x; ++j) *ave_sx_ptr_d += *(ave_sx_ptr_d + j*blockDim.x);
}
if (blockIdx.x == 0 && threadIdx.x == 1){
for (int j = 1; j < gridDim.x; ++j) *ave_sy_ptr_d += *(ave_sy_ptr_d + j*blockDim.x);
}
if (blockIdx.x == 0 && threadIdx.x == 2){
for (int j = 1; j < gridDim.x; ++j) *ave_sz_ptr_d += *(ave_sz_ptr_d + j*blockDim.x);
}
}
#endif