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
|
/*
* Copyright 2008-2009 NVIDIA Corporation
*
* 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.
*/
#ifdef __CUSP_USE_B40C__
#undef B40C_LOG_MEM_BANKS
#undef B40C_LOG_WARP_THREADS
#undef B40C_WARP_THREADS
#undef TallyWarpVote
#undef WarpVoteAll
#undef FastMul
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#ifdef _WIN32
#include <Windows.h>
#include <io.h>
HANDLE hStdOut;
int STDOUT_COPY;
int STDOUT_FILENO;
#define CUSP_CLOSE_STDOUT do{\
hStdOut = GetStdHandle(STD_OUTPUT_HANDLE); \
STDOUT_FILENO = _open_osfhandle((HFILE)hStdOut, O_WRONLY|O_TEXT); \
STDOUT_COPY = _dup(STDOUT_FILENO); \
_close(STDOUT_FILENO); \
}while(0);
#define CUSP_REOPEN_STDOUT do{\
_dup2(STDOUT_COPY, STDOUT_FILENO); \
_close(STDOUT_COPY); \
SetStdHandle(STD_OUTPUT_HANDLE, hStdOut); \
}while(0);
#else
int STDOUT_COPY;
#define CUSP_CLOSE_STDOUT do{\
fflush(stdout); \
STDOUT_COPY = dup(STDOUT_FILENO); \
close(STDOUT_FILENO); \
}while(0);
#define CUSP_REOPEN_STDOUT do{\
dup2(STDOUT_COPY, STDOUT_FILENO); \
close(STDOUT_COPY); \
}while(0);
#endif // end ifdef _WIN32
#include <b40c/graph/bfs/csr_problem.cuh>
#include <b40c/graph/bfs/enactor_hybrid.cuh>
#undef B40C_LOG_MEM_BANKS
#undef B40C_LOG_WARP_THREADS
#undef B40C_WARP_THREADS
#undef TallyWarpVote
#undef WarpVoteAll
#undef FastMul
#endif // end ifdef __CUSP_USE_B40C__
#include <cusp/exception.h>
namespace cusp
{
namespace graph
{
namespace detail
{
namespace device
{
template<bool MARK_PREDECESSORS, typename MatrixType, typename ArrayType>
void breadth_first_search(const MatrixType& G, const typename MatrixType::index_type src,
ArrayType& labels)
{
typedef typename MatrixType::index_type VertexId;
typedef typename MatrixType::index_type SizeT;
#ifdef __CUSP_USE_B40C__
CUSP_CLOSE_STDOUT;
typedef b40c::graph::bfs::CsrProblem<VertexId, SizeT, MARK_PREDECESSORS> CsrProblem;
typedef typename CsrProblem::GraphSlice GraphSlice;
int max_grid_size = 0;
double max_queue_sizing = 1.15;
int nodes = G.num_rows;
int edges = G.num_entries;
CsrProblem csr_problem;
csr_problem.nodes = nodes;
csr_problem.edges = edges;
csr_problem.num_gpus = 1;
if( labels.size() != G.num_rows )
{
throw cusp::runtime_exception("BFS traversal labels is not large enough for result.");
}
// Create a single GPU slice for the currently-set gpu
int gpu;
if (b40c::util::B40CPerror(cudaGetDevice(&gpu), "CsrProblem cudaGetDevice failed", __FILE__, __LINE__))
{
throw cusp::runtime_exception("B40C cudaGetDevice failed.");
}
csr_problem.graph_slices.push_back(new GraphSlice(gpu, 0));
csr_problem.graph_slices[0]->nodes = nodes;
csr_problem.graph_slices[0]->edges = edges;
csr_problem.graph_slices[0]->d_row_offsets = (VertexId *) thrust::raw_pointer_cast(&G.row_offsets[0]);
csr_problem.graph_slices[0]->d_column_indices = (VertexId *) thrust::raw_pointer_cast(&G.column_indices[0]);
csr_problem.graph_slices[0]->d_labels = (VertexId *) thrust::raw_pointer_cast(&labels[0]);
b40c::graph::bfs::EnactorHybrid<false/*INSTRUMENT*/> hybrid(false);
csr_problem.Reset(hybrid.GetFrontierType(), max_queue_sizing);
hybrid.EnactSearch(csr_problem, src, max_grid_size);
// Unset pointers to prevent csr_problem deallocations
csr_problem.graph_slices[0]->d_row_offsets = (VertexId *) NULL;
csr_problem.graph_slices[0]->d_column_indices = (VertexId *) NULL;
csr_problem.graph_slices[0]->d_labels = (VertexId *) NULL;
CUSP_REOPEN_STDOUT;
#else
throw cusp::not_implemented_exception("Device BFS implementation depends on B40C support.");
#endif
}
} // end namespace device
} // end namespace detail
} // end namespace graph
} // end namespace cusp
|
