-
Notifications
You must be signed in to change notification settings - Fork 70
/
BP_GPU.h.bak
88 lines (79 loc) · 4.16 KB
/
BP_GPU.h.bak
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
//#include "/usr/local/cuda/include/cuda_runtime.h"
//#include <cuda_runtime.h>
//#include "/usr/local/cuda/include/cublas_v2.h"
//#include <cublas_v2.h>
#include "/usr/local/cuda-5.0/include/cuda_runtime.h"
#include "/usr/local/cuda-5.0/include/cublas_v2.h"
#include "/usr/local/cuda-5.0/include/curand.h"
//#include <curand.h>
//#include "/usr/local/cuda/include/cuda_runtime.h"
//#include "/usr/local/cuda/include/cublas_v2.h"
//#include "/usr/local/cuda/include/curand.h"
#define MAXLAYER 10
#define MAXCACHEFRAME 200000
struct BP_WorkSpace
{
float *in; /// input data
float *out; /// Output data，输出是float型，它通过softmax变成0,1
//int *targ; /// target data，我要改#############################，target是int，得改成float
float *targ;////////////////////////////////////by yongxu
float *weights[MAXLAYER]; /// weights for layers，指针数组，每个数组都是指针，本质是数组，权重是二维的；；int (*p)[10]; p即为指向数组元素地址的指针，本质是指针
float *bias[MAXLAYER]; /// biases for layers；rbm的输入，输出各有一个bias
float *layer_x[MAXLAYER]; /// Input to layer
float *layer_y[MAXLAYER]; /// Output from layer
float *layer_dedy[MAXLAYER]; /// de/dy
float *layer_dydx[MAXLAYER]; /// dy/dx
float *layer_dedx[MAXLAYER]; /// de/dx
float *layer_ydedx[MAXLAYER];
float *layer_sumdedx[MAXLAYER];
float *delta_bias[MAXLAYER]; // Output bias update
float *delta_weights[MAXLAYER]; // Output bias update
float *DevRandVector; //Dropout随机数存储矩阵
int *DevSeed;//Dropout随机数种子
};
class BP_GPU
{
public:
BP_GPU(int a_GPU_selected, int a_numlayers, int *a_layersizes, int a_bunchsize, float a_lrate, float a_momentum, float a_weightcost,
float **weights, float **bias,int dropoutflag, float visible_omit,float hid_omit);
~BP_GPU();
public:
//void train(int n_frames, const float* in, const int *targ);
//void train(int n_frames, const float* in, const float *targ);////////////////////////////////////////by yongxu
void train(int n_frames, float* in, const float *targ);
//void train_bunch_multi(int n_frames, float** in, int **targ);
//void train_bunch_multi(int n_frames, float** in, float **targ);//////////////////////////////////////by yongxu
void train_bunch_multi(int n_frames, float** in, float **targ);
//void train_bunch_single(int n_frames, const float* in, const int *targ);
//void train_bunch_single(int n_frames, const float* in, const float *targ);//////////////////////////////by yongxu
void train_bunch_single(int n_frames, float* in, const float *targ);
//int CrossValid(int n_frames, const float* in, const int *targ);
float CrossValid(int n_frames, const float* in, const float *targ);///////////////////////////////////////by yongxu
//void cv_bunch_single(int n_frames, const float* in, int *out);
void cv_bunch_single(int n_frames, const float* in, float *out);///////////////////////////////////////////by yongxu
void returnWeights(float **weights, float **bias); /// copy weights and biases from gpu memory to cpu memory
int numlayers;
int layersizes[MAXLAYER];
int bunchsize;
float lrate;
float momentum;
float weightcost;
int dropoutflag;
float visible_omit;
float hid_omit;
private:
void devnew_vf(const char* varname, int n, float **devptr);
void devnew_vi(const char* varname, int n, int **devptr);/////////////////////////////by yongxu
void devfree_vf(const char* varname, float* devptr);
void devfree_vi(const char* varname, int* devptr);
void todev_vf_vf(const char* varname, int n, const float* from, float* devto, cudaStream_t stream);
void fromdev_vf_vf(const char* varname, int n, const float* devfrom, float* to, cudaStream_t stream);
//void todev_vi_vi(const char* varname, int n, const int* from, int* devto, cudaStream_t stream);
//void fromdev_vi_vi(const char* varname, int n, const int* devfrom, int* to, cudaStream_t stream);
BP_WorkSpace *dev; //viaribles for devices
int GPU_total; //devices used num, 表示所有GPU数目
int GPU_selected; //devices selected, 表示采用的GPU数目
cublasHandle_t *handles;
cudaStream_t *streams;
curandGenerator_t *gen;
};