mies-gpu/sparse.cpp at master · evariste-dlr/mies-gpu · GitHub

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
#include "sparse.h"

#include <iostream>
#include <list>

#include <cstring>

sparse_mat * to_sparse_bd( const std::list<float*> & values,
                           const std::list<size_t> & rows)
{
  if (values.size() != rows.size()){
    std::cerr << "[E] Different number of graphs" << std::endl;
    return NULL;
  }

  // Get total nnz values
  uint nnz = 0;
  std::list<float*>::const_iterator it_v = values.begin();
  for (std::list<size_t>::const_iterator it_n=rows.begin(); it_n!=rows.end(); it_n++, it_v++)
    for (uint i=0; i<(*it_n); i++)
      for (uint j=0; j<(*it_n); j++)
        if ((*it_v)[IND(i,j,(*it_n))] != 0.0)
          nnz++;

  // Get total number of nodes
  uint n_nodes = 0;
  for (std::list<size_t>::const_iterator it_n=rows.begin(); it_n!=rows.end(); it_n++)
    n_nodes += (*it_n);

  sparse_mat* mat = (sparse_mat*) malloc(sizeof(sparse_mat));
  mat->val = (float*) malloc(nnz * sizeof(float));
  mat->coo_row_ind = (uint*) malloc(nnz * sizeof(uint));
  mat->coo_col_ind = (uint*) malloc(nnz * sizeof(uint));
  mat->csr_row_ptr = (uint*) malloc((n_nodes+1) * sizeof(uint));
  mat->nnz = nnz;
  mat->rows = n_nodes;
  mat->cols = n_nodes;

  // Fill in the matrix
  uint first_node = 0;
  uint idx = 0;
  bool newline = false;

  it_v = values.begin();
  for (std::list<size_t>::const_iterator it_n=rows.begin(); it_n!=rows.end(); it_n++, it_v++){

    for (uint i=0; i<(*it_n); i++){
      // if the previous line is empty (no newline=false bellow)
      if (newline)
        mat->csr_row_ptr[first_node+i-1] = idx;

      newline = true;
      for (uint j=0; j<(*it_n); j++){

        if ((*it_v)[IND(i,j,(*it_n))] != 0.0){
          mat->val[idx] = (*it_v)[IND(i,j,(*it_n))];
          mat->coo_row_ind[idx] = first_node + i;
          mat->coo_col_ind[idx] = first_node + j;

          if (newline){
            mat->csr_row_ptr[first_node+i] = idx;
            newline = false;
          }

          idx++;
        }

      } // for j
    } // for i

    first_node += (*it_n);
  }// for n

  // If the last line is empty
  if (newline)
    mat->csr_row_ptr[n_nodes-1] = nnz-1;

  // Last row ptr is the last data index
  mat->csr_row_ptr[n_nodes] = nnz-1;

  return mat;
}


void print_sparse( const sparse_mat & mat )
{
  for (uint i=0; i<mat.nnz; i++){
    if (mat.val[i] != 0.0){
      std::cout << "(" << mat.coo_row_ind[i] << "; "
                << mat.coo_col_ind[i] << ") \t -> "
                << mat.val[i] << std::endl;
    }
  }
  std::cout << std::endl;
}


void print_sparse_py( const sparse_mat & mat )
{
  std::cout << "data = [";
  for (int e=0; e<mat.nnz; e++)
    if (mat.val[e] != 0.0)
      std::cout << mat.val[e] << ", ";
  std::cout << "]" << std::endl;

  std::cout << "rows = [";
  for (int e=0; e<mat.nnz; e++)
    if (mat.val[e] != 0.0)
      std::cout << mat.coo_row_ind[e] << ", ";
  std::cout << "]" << std::endl;

  std::cout << "cols = [";
  for (int e=0; e<mat.nnz; e++)
    if (mat.val[e] != 0.0)
      std::cout << mat.coo_col_ind[e] << ", ";
  std::cout << "]" << std::endl;
}


void copy_sparse( sparse_mat * dst, const sparse_mat * src )
{
  dst->val = (float*) malloc(src->nnz * sizeof(float));
  dst->coo_row_ind = (uint*) malloc(src->nnz * sizeof(uint));
  dst->coo_col_ind = (uint*) malloc(src->nnz * sizeof(uint));
  dst->csr_row_ptr = (uint*) malloc((src->rows+1) * sizeof(uint));
  dst->nnz = src->nnz;
  dst->rows = src->rows;
  dst->cols = src->cols;

  memcpy( dst->val, src->val, dst->nnz * sizeof(float) );
  memcpy( dst->coo_row_ind, src->coo_row_ind, dst->nnz * sizeof(uint) );
  memcpy( dst->coo_col_ind, src->coo_col_ind, dst->nnz * sizeof(uint) );
  memcpy( dst->csr_row_ptr, src->csr_row_ptr, (dst->rows+1) * sizeof(uint) );
}


void delete_sparse( sparse_mat * mat )
{
  free(mat->val);
  free(mat->coo_row_ind);
  free(mat->coo_col_ind);
  free(mat->csr_row_ptr);
}