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
|
#include <mpi.h>
#include <sys/time.h>
#include <iostream>
#include <functional>
#include <algorithm>
#include <vector>
#include <sstream>
#include "CombBLAS/CombBLAS.h"
using namespace std;
using namespace combblas;
#define ITERATIONS 10
template <class T>
bool from_string(T & t, const string& s, std::ios_base& (*f)(std::ios_base&))
{
istringstream iss(s);
return !(iss >> f >> t).fail();
}
int main(int argc, char* argv[])
{
int nprocs, myrank;
MPI_Init(&argc, &argv);
MPI_Comm_size(MPI_COMM_WORLD,&nprocs);
MPI_Comm_rank(MPI_COMM_WORLD,&myrank);
if(argc < 3)
{
if(myrank == 0)
{
cout << "Usage: ./IndexingTiming Input/Force/Binary <Inputfile>/<Scale>/<BinaryFile>" << endl;
}
MPI_Finalize();
return -1;
}
{
typedef SpParMat <int, double, SpDCCols<int,double> > PARDBMAT;
PARDBMAT * A; // declare objects
if(string(argv[1]) == string("Input"))
{
A->ReadDistribute(argv[2], 0);
}
else if(string(argv[1]) == string("Binary"))
{
uint64_t n, m;
from_string(n,string(argv[3]),std::dec);
from_string(m,string(argv[4]),std::dec);
ostringstream outs;
outs << "Reading " << argv[2] << " with " << n << " vertices and " << m << " edges" << endl;
SpParHelper::Print(outs.str());
DistEdgeList<int64_t> * DEL = new DistEdgeList<int64_t>(argv[2], n, m);
SpParHelper::Print("Read binary input to distributed edge list\n");
PermEdges(*DEL);
SpParHelper::Print("Permuted Edges\n");
RenameVertices(*DEL);
SpParHelper::Print("Renamed Vertices\n");
A = new PARDBMAT(*DEL, false);
delete DEL; // free memory before symmetricizing
SpParHelper::Print("Created double Sparse Matrix\n");
}
else if(string(argv[1]) == string("Force"))
{
double initiator[4] = {.6, .4/3, .4/3, .4/3};
DistEdgeList<int64_t> * DEL = new DistEdgeList<int64_t>();
int scale = static_cast<unsigned>(atoi(argv[2]));
ostringstream outs;
outs << "Forcing scale to : " << scale << endl;
SpParHelper::Print(outs.str());
DEL->GenGraph500Data(initiator, scale, 8 * ((int64_t) std::pow(2.0, (double) scale)) / nprocs );
SpParHelper::Print("Generated local RMAT matrices\n");
PermEdges(*DEL);
SpParHelper::Print("Permuted Edges\n");
RenameVertices(*DEL);
SpParHelper::Print("Renamed Vertices\n");
// conversion from distributed edge list, keeps self-loops, sums duplicates
A = new PARDBMAT(*DEL, false);
delete DEL; // free memory before symmetricizing
SpParHelper::Print("Created double Sparse Matrix\n");
}
A->PrintInfo();
FullyDistVec<int,int> p;
p.iota(A->getnrow(), 0);
p.RandPerm();
SpParHelper::Print("Permutation Generated\n");
PARDBMAT B = (*A)(p,p);
B.PrintInfo();
float oldbalance = A->LoadImbalance();
float newbalance = B.LoadImbalance();
ostringstream outs;
outs << "Running on " << nprocs << " cores" << endl;
outs << "Old balance: " << oldbalance << endl;
outs << "New balance: " << newbalance << endl;
SpParHelper::Print(outs.str());
MPI_Barrier(MPI_COMM_WORLD);
double t1 = MPI_Wtime(); // initilize (wall-clock) timer
for(int i=0; i<ITERATIONS; i++)
{
B = (*A)(p,p);
}
MPI_Barrier(MPI_COMM_WORLD);
double t2 = MPI_Wtime();
if(myrank == 0)
{
cout<<"Indexing Iterations finished"<<endl;
printf("%.6lf seconds elapsed per iteration\n", (t2-t1)/(double)ITERATIONS);
}
// Test #2
int nclust = 10;
vector< FullyDistVec<int,int> > clusters(nclust);
int nperclus = A->getnrow() / nclust;
for(int i = 0; i< nclust; i++)
{
int k = std::min(nperclus, A->getnrow() - nperclus * i);
clusters[i].iota(k, nperclus * i);
clusters[i] = p(clusters[i]);
}
for(int i=0; i< nclust; i++)
{
B = (*A)(clusters[i], clusters[i]);
B.PrintInfo();
}
MPI_Barrier(MPI_COMM_WORLD);
t1 = MPI_Wtime(); // initilize (wall-clock) timer
for(int i=0; i< nclust; i++)
{
for(int j=0; j < ITERATIONS; j++)
B = (*A)(clusters[i], clusters[i]);
}
MPI_Barrier(MPI_COMM_WORLD);
t2 = MPI_Wtime();
if(myrank == 0)
{
cout<<"Indexing Iterations finished"<<endl;
printf("%.6lf seconds elapsed per iteration\n", (t2-t1)/(double)ITERATIONS);
}
// Test #3: Pruning for SpAsgn
for(int i=0; i< nclust; i++)
{
PARDBMAT C = *A;
C.Prune(clusters[i], clusters[i]);
C.PrintInfo();
B = (*A)(clusters[i], clusters[i]);
C.SpAsgn(clusters[i], clusters[i], B);
// We should get the original A back.
if( C == (*A))
{
SpParHelper::Print("Pruning and SpAsgn seem to be working\n");
}
else
{
SpParHelper::Print("A and C don't match, below is C's info followed by B's info\n");
C.PrintInfo();
B.PrintInfo();
}
}
delete A;
}
MPI_Finalize();
return 0;
}
|
