forked from olearczuk/gpu-louvain
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathutils.cu
186 lines (171 loc) · 8.11 KB
/
utils.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
#include "utils.cuh"
#include <vector>
#include <iostream>
#include <thrust/partition.h>
#include <fstream>
#include <getopt.h>
#include <sstream>
host_structures readInputData(char *fileName) {
std::fstream file;
file.open(fileName);
int V, E;
std::string s;
do {
std::getline(file, s);
} while (s[0] == '%');
std::istringstream stream(s);
stream >> V >> E;
int v1, v2;
host_structures hostStructures;
hostStructures.originalV = V;
hostStructures.V = V;
HANDLE_ERROR(cudaHostAlloc((void**)&hostStructures.vertexCommunity, V * sizeof(int), cudaHostAllocDefault));
HANDLE_ERROR(cudaHostAlloc((void**)&hostStructures.communityWeight, V * sizeof(float), cudaHostAllocDefault));
HANDLE_ERROR(cudaHostAlloc((void**)&hostStructures.edgesIndex, (V + 1) * sizeof(int), cudaHostAllocDefault));
HANDLE_ERROR(cudaHostAlloc((void**)&hostStructures.originalToCommunity, V * sizeof(int), cudaHostAllocDefault));
std::vector<std::vector<std::pair<int, float>>> neighbours(V);
// TODO: here is assumption that graph is undirected
int aux = E;
for (int i = 0; i < aux; i++) {
file >> v1 >> v2;
const float w = 1.0;
//v1--;
//v2--;
hostStructures.communityWeight[v1] += w;
neighbours[v1].emplace_back(v2, w);
if (v1 != v2) {
E++;
hostStructures.communityWeight[v2] += w;
neighbours[v2].emplace_back(v1, w);
hostStructures.M += w;
}
hostStructures.M += w;
}
hostStructures.M /= 2;
HANDLE_ERROR(cudaHostAlloc((void**)&hostStructures.edges, E * sizeof(int), cudaHostAllocDefault));
HANDLE_ERROR(cudaHostAlloc((void**)&hostStructures.weights, E * sizeof(float), cudaHostAllocDefault));
hostStructures.E = E;
int index = 0;
for (int v = 0; v < V; v++) {
hostStructures.edgesIndex[v] = index;
for (auto & it : neighbours[v]) {
hostStructures.edges[index] = it.first;
hostStructures.weights[index] = it.second;
index++;
}
}
hostStructures.edgesIndex[V] = E;
file.close();
return hostStructures;
}
void copyStructures(host_structures& hostStructures, device_structures& deviceStructures,
aggregation_phase_structures& aggregationPhaseStructures) {
// copying from deviceStructures to hostStructures
int V = hostStructures.V, E = hostStructures.E;
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.vertexCommunity, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.communityWeight, V * sizeof(float)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.edges, E * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.weights, E * sizeof(float)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.edgesIndex, (V + 1) * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.originalToCommunity, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.vertexEdgesSum, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.newVertexCommunity, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.V, sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.E, sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.originalV, sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.communitySize, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.partition, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&deviceStructures.toOwnCommunity, V * sizeof(int)));
thrust::fill(thrust::device, deviceStructures.communitySize, deviceStructures.communitySize + V, 1);
thrust::sequence(thrust::device, deviceStructures.vertexCommunity, deviceStructures.vertexCommunity + V, 0);
thrust::sequence(thrust::device, deviceStructures.newVertexCommunity, deviceStructures.newVertexCommunity + V, 0);
thrust::sequence(thrust::device, deviceStructures.originalToCommunity, deviceStructures.originalToCommunity + V, 0);
HANDLE_ERROR(cudaMemcpy(deviceStructures.communityWeight, hostStructures.communityWeight, V * sizeof(float), cudaMemcpyHostToDevice));
HANDLE_ERROR(cudaMemcpy(deviceStructures.edges, hostStructures.edges, E * sizeof(int), cudaMemcpyHostToDevice));
HANDLE_ERROR(cudaMemcpy(deviceStructures.weights, hostStructures.weights, E * sizeof(float), cudaMemcpyHostToDevice));
HANDLE_ERROR(cudaMemcpy(deviceStructures.edgesIndex, hostStructures.edgesIndex, (V + 1) * sizeof(int), cudaMemcpyHostToDevice));
HANDLE_ERROR(cudaMemcpy(deviceStructures.V, &hostStructures.V, sizeof(int), cudaMemcpyHostToDevice));
HANDLE_ERROR(cudaMemcpy(deviceStructures.E, &hostStructures.E, sizeof(int), cudaMemcpyHostToDevice));
HANDLE_ERROR(cudaMemcpy(deviceStructures.originalV, &hostStructures.originalV, sizeof(int), cudaMemcpyHostToDevice));
// preparing aggregationPhaseStructures
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.communityDegree, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.newID, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.edgePos, V * sizeof(int)));;
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.vertexStart, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.orderedVertices, V * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.edgeIndexToCurPos, E * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.newEdges, E * sizeof(int)));
HANDLE_ERROR(cudaMalloc((void**)&aggregationPhaseStructures.newWeights, E * sizeof(float)));
}
void deleteStructures(host_structures& hostStructures, device_structures& deviceStructures,
aggregation_phase_structures& aggregationPhaseStructures) {
HANDLE_ERROR(cudaFreeHost(hostStructures.vertexCommunity));
HANDLE_ERROR(cudaFreeHost(hostStructures.communityWeight));
HANDLE_ERROR(cudaFreeHost(hostStructures.edges));
HANDLE_ERROR(cudaFreeHost(hostStructures.weights));
HANDLE_ERROR(cudaFreeHost(hostStructures.edgesIndex));
HANDLE_ERROR(cudaFreeHost(hostStructures.originalToCommunity));
HANDLE_ERROR(cudaFree(deviceStructures.originalV));
HANDLE_ERROR(cudaFree(deviceStructures.vertexCommunity));
HANDLE_ERROR(cudaFree(deviceStructures.communityWeight));
HANDLE_ERROR(cudaFree(deviceStructures.edges));
HANDLE_ERROR(cudaFree(deviceStructures.weights));
HANDLE_ERROR(cudaFree(deviceStructures.edgesIndex));
HANDLE_ERROR(cudaFree(deviceStructures.originalToCommunity));
HANDLE_ERROR(cudaFree(deviceStructures.vertexEdgesSum));
HANDLE_ERROR(cudaFree(deviceStructures.newVertexCommunity));
HANDLE_ERROR(cudaFree(deviceStructures.E));
HANDLE_ERROR(cudaFree(deviceStructures.V));
HANDLE_ERROR(cudaFree(deviceStructures.communitySize));
HANDLE_ERROR(cudaFree(deviceStructures.partition));
HANDLE_ERROR(cudaFree(deviceStructures.toOwnCommunity));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.communityDegree));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.newID));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.edgePos));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.vertexStart));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.orderedVertices));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.edgeIndexToCurPos));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.newEdges));
HANDLE_ERROR(cudaFree(aggregationPhaseStructures.newWeights));
}
int blocksNumber(int V, int threadsPerVertex) {
return (V * threadsPerVertex + THREADS_PER_BLOCK - 1) / THREADS_PER_BLOCK;
}
bool isPrime(int n) {
for (int i = 2; i < sqrt(n) + 1; i++)
if (n % i == 0)
return false;
return true;
}
int getPrime(int n) {
do {
n++;
} while(!isPrime(n));
return n;
}
void parseCommandLineArgs(int argc, char *argv[], float *minGain, bool *isVerbose, char **fileName) {
bool isF, isG;
char opt;
while ((opt = getopt(argc, argv, "f:g:v")) != -1) {
switch (opt) {
case 'g':
isG = true;
*minGain = strtof(optarg, NULL);
break;
case 'v':
*isVerbose = true;
break;
case 'f':
isF = true;
*fileName = optarg;
break;
default:
printf("Usage: ./gpulouvain -f mtx-matrix-file -g min-gain [-v]\n");
exit(1);
}
}
if (!isF || !isG) {
printf("Usage: ./gpulouvain -f mtx-matrix-file -g min-gain [-v]\n");
exit(1);
}
}