forked from RoyalCaliber/vertexAPI2
-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathbfs.cu
263 lines (218 loc) · 6.36 KB
/
bfs.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
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
/******************************************************************************
Copyright 2013 Royal Caliber LLC. (http://www.royal-caliber.com)
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.
******************************************************************************/
//BFS using vertexAPI2
#include "util.cuh"
#include "graphio.h"
#include "refgas.h"
#include "gpugas.h"
//nvcc doesn't like the __device__ variable to be a static member inside BFS
//so these are both outside.
int g_iterationCount;
__device__ __constant__ int g_iterationCountGPU;
struct BFS
{
struct VertexData
{
int depth;
};
struct EdgeData {}; //nothing
typedef int GatherResult;
static const int gatherZero = INT_MAX - 1;
__host__ __device__
static int gatherReduce(const int& left, const int& right)
{
return 0; //do nothing
}
__host__ __device__
static int gatherMap(
const VertexData* dst, const VertexData *src, const EdgeData* edge)
{
return 0; //do nothing
}
__host__ __device__
static bool apply(VertexData* vert, int dist)
{
if( vert->depth == -1 )
{
#ifdef __CUDA_ARCH__
vert->depth = g_iterationCountGPU;
#else
vert->depth = g_iterationCount;
#endif
return true;
}
return false;
}
__host__ __device__
static void scatter(
const VertexData* src, const VertexData *dst, EdgeData* edge)
{
//nothing
}
};
template<bool GPU>
void setIterationCount(int v)
{
if( GPU )
cudaMemcpyToSymbol(g_iterationCountGPU, &v, sizeof(v));
else
g_iterationCount = v;
}
template<typename Engine, bool GPU>
float run(int nVertices, BFS::VertexData* vertexData, int nEdges
, const int *srcs, const int *dsts, int sourceVertex)
{
Engine engine;
int iteration;
GpuTimer gpu_timer;
float elapsed = 0.0f;
// average elapsed time of 10 runs
for (int itr = 0; itr < 10; ++itr)
{
// reset the graph
for(int i = 0; i < nVertices; ++i) vertexData[i].depth = -1;
engine.setGraph(nVertices, vertexData, nEdges, 0, &srcs[0], &dsts[0]);
engine.setActive(sourceVertex, sourceVertex+1);
iteration = 0;
setIterationCount<GPU>(iteration);
gpu_timer.Start();
while( engine.countActive() )
{
//run apply without gather
engine.gatherApply(false);
engine.scatterActivate(false);
engine.nextIter();
setIterationCount<GPU>(++iteration);
}
engine.getResults();
gpu_timer.Stop();
elapsed += gpu_timer.ElapsedMillis();
}
elapsed /= 10;
// printf("Took %f ms\n", elapsed);
printf("search depth (number of iterations): %d\n", iteration);
return elapsed;
}
void outputDepths(int nVertices, BFS::VertexData* vertexData, FILE *f = stdout)
{
for( int i = 0; i < nVertices; ++i )
fprintf(f, "%d %d\n", i, vertexData[i].depth);
}
int main(int argc, char** argv)
{
char *inputFilename;
char *outputFilename = 0;
int sourceVertex;
bool runTest;
bool dumpResults;
bool useMaxOutDegreeStart;
if(!parseCmdLineSimple(argc, argv, "si-t-d-m|s", &inputFilename, &sourceVertex
, &runTest, &dumpResults, &useMaxOutDegreeStart, &outputFilename) )
{
printf("Usage: bfs [-t] [-d] [-m] inputfile source [outputFilename]\n");
exit(1);
}
//load the graph
int nVertices;
std::vector<int> srcs;
std::vector<int> dsts;
loadGraph(inputFilename, nVertices, srcs, dsts);
//initialize vertex data
std::vector<BFS::VertexData> vertexData(nVertices);
for( int i = 0; i < nVertices; ++i )
vertexData[i].depth = -1;
if( useMaxOutDegreeStart )
{
//convert to CSR layout to find source vertex
std::vector<int> srcOffsets(nVertices + 1);
std::vector<int> csrSrcs(srcs.size());
edgeListToCSR<int>(
nVertices, srcs.size(), &srcs[0], &dsts[0], &srcOffsets[0], 0, 0);
int maxDegree = -1;
sourceVertex = -1;
for(int i = 0; i < nVertices; ++i)
{
int outDegree = srcOffsets[i + 1] - srcOffsets[i];
if( outDegree > maxDegree )
{
maxDegree = outDegree;
sourceVertex = i;
}
}
printf(
"using vertex %d with degree %d as source\n", sourceVertex, maxDegree);
}
std::vector<BFS::VertexData> refVertexData;
if( runTest )
{
refVertexData = vertexData;
float elapsed = run<GASEngineRef<BFS>, false>(nVertices
, &refVertexData[0], (int)srcs.size(), &srcs[0], &dsts[0], sourceVertex);
if( dumpResults )
{
printf("Reference:\n");
outputDepths(nVertices, &refVertexData[0]);
}
}
float elapsed = run<GASEngineGPU<BFS>, true>(nVertices, &vertexData[0]
, (int) srcs.size(), &srcs[0], &dsts[0], sourceVertex);
// compute stats
int nodes_visited = 0;
int edges_visited = 0;
std::vector<int> srcOffsets(nVertices + 1);
std::vector<int> csrSrcs(srcs.size());
edgeListToCSR<int>(
nVertices, srcs.size(), &srcs[0], &dsts[0], &srcOffsets[0], 0, 0);
for (int itr = 0; itr < nVertices; ++itr)
{
if (vertexData[itr].depth > -1)
{
nodes_visited += 1;
edges_visited += srcOffsets.at(itr+1) - srcOffsets.at(itr);
}
}
printf("nodes visited: %d edges visited: %d\n", nodes_visited, edges_visited);
float m_teps = (float) edges_visited / (elapsed * 1000);
printf("elapsed: %.4f ms, MTEPS: %.4f MiEdges/s\n", elapsed, m_teps);
if( dumpResults )
{
printf("GPU:\n");
outputDepths(nVertices, &vertexData[0]);
}
if( runTest )
{
bool diff = false;
for( int i = 0; i < nVertices; ++i )
{
if( vertexData[i].depth != refVertexData[i].depth )
{
printf("%d %d %d\n", i, refVertexData[i].depth, vertexData[i].depth);
diff = true;
}
}
if( diff )
return 1;
else
printf("No differences found\n");
}
if( outputFilename )
{
printf("writing results to %s\n", outputFilename);
FILE* f = fopen(outputFilename, "w");
outputDepths(nVertices, &vertexData[0], f);
fclose(f);
}
free(inputFilename);
free(outputFilename);
return 0;
}