-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathomp.c
More file actions
435 lines (332 loc) · 10.2 KB
/
omp.c
File metadata and controls
435 lines (332 loc) · 10.2 KB
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
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
/* !!!Important!!! Follow the instruction on line 92 to run the code on a mtx file without values column*/
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <omp.h>
#include <time.h>
#include <sys/time.h>
#include "Matrix_Market/mmio.h"
#include <stdbool.h>
#define NUM_THREADS 4
/*********************My structs**************************/
struct COO{
int* rows;
int* cols;
int nodes;
int edges;
};
struct CSC_CSR{
int* pointers;
int* indices;
int nodes;
int nz;
};
/**********************************************************/
/*********************My functions***********************/
int trivial_scc(struct CSC_CSR csc, struct CSC_CSR csr, bool* used_nodes);
int* coloring(struct CSC_CSR csc, bool* used_nodes);
int* unique(int* colors, bool* used_nodes, struct CSC_CSR csc);
void pred_scc(int* colors, struct CSC_CSR csc, bool* used_nodes, int index);
struct CSC_CSR coo_to_csc(struct COO coo);
struct COO create_coo(struct COO coo, int* I, int* J, int M, int nz);
struct CSC_CSR coo_to_csr(struct COO coo);
/********************************************************/
int main(int argc, char** argv){
int ret_code;
MM_typecode matcode;
FILE *f;
int M, N, nz;
int i, *I, *J;
double *val;
if (argc < 2)
{
fprintf(stderr, "Usage: %s [martix-market-filename]\n", argv[0]);
exit(1);
}
else
{
if ((f = fopen(argv[1], "r")) == NULL)
exit(1);
}
if (mm_read_banner(f, &matcode) != 0)
{
printf("Could not process Matrix Market banner.\n");
exit(1);
}
if (mm_is_complex(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) )
{
printf("Sorry, this application does not support ");
printf("Market Market type: [%s]\n", mm_typecode_to_str(matcode));
exit(1);
}
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &nz)) !=0)
exit(1);
I = (int *) malloc(nz * sizeof(int));
J = (int *) malloc(nz * sizeof(int));
val = (double *) malloc(nz * sizeof(double));
for (i=0; i<nz; i++)
{
fscanf(f, "%d %d %lg\n", &I[i], &J[i], &val[i]);
I[i]--; /* adjust the fscanf with fscanf(f, "%d %d\n", &I[i], &J[i]); */
J[i]--;
}
free(val);
if (f !=stdin) fclose(f);
/************************/
/* now write out matrix */
/************************/
mm_write_banner(stdout, matcode);
mm_write_mtx_crd_size(stdout, M, N, nz);
/******************Read matrix********************************/
double start, end;
start = omp_get_wtime();
struct COO coo = create_coo(coo, I, J, M, nz);
struct CSC_CSR csc = coo_to_csc(coo);
struct CSC_CSR csr = coo_to_csr(coo);
free(I);
free(J);
/***********************************************************/
int global_scc, used_nodes_num = 0;
bool* used_nodes = calloc(csc.nodes, sizeof(bool));
int trivial = trivial_scc(csc, csr, used_nodes);
global_scc = trivial;
free(csr.pointers);
free(csr.indices);
printf("trivial_scc: %d\n", global_scc);
int *colors, *unique_colors;
double bfs_time_start, bfs_time_end;
bool done = false;
while(!done){
done = true;
colors = coloring(csc, used_nodes);
unique_colors = unique(colors, used_nodes, csc);
bfs_time_start = omp_get_wtime();
for(int i = 0; i < csc.nodes; i++){
if(unique_colors[i] != -1){
pred_scc(colors, csc, used_nodes, unique_colors[i]);
global_scc++;
}
else
break;
}
bfs_time_end = omp_get_wtime();
for(int i = 0; i < csc.nodes; i++){
if(used_nodes[i] == false){
done = false;
break;
}
}
free(colors);
free(unique_colors);
}
printf("global_scc: %d\n", global_scc);
free(used_nodes);
free(csc.pointers);
free(csc.indices);
end = omp_get_wtime();
printf("time: %lf\n", end - start);
return 0;
}
void pred_scc(int* colors, struct CSC_CSR csc, bool* used_nodes, int index){
double start_time, end_time;
start_time = omp_get_wtime();
bool check = true;
int id, start, end;
bool* queue = calloc(csc.nodes, sizeof(bool));
queue[index] = 1;
used_nodes[index] = true;
double temp_time_start, temp_time_end;
while(check){
check = false;
#pragma omp parallel for num_threads(NUM_THREADS) private(start, end, id)
for(int i = 0; i < csc.nodes; i++){
if(queue[i] == 1){
queue[i] = 0;
start = csc.pointers[i];
end = csc.pointers[i + 1];
for(int j = start; j < end; j++){
id = csc.indices[j];
if(colors[index] == colors[id] && used_nodes[id] == false){
queue[id] = 1;
check = true;
used_nodes[id] = true;
}
}
}
}
}
end_time = omp_get_wtime();
free(queue);
}
int* unique(int* colors, bool* used_nodes, struct CSC_CSR csc){
double start_time, end_time;
start_time = omp_get_wtime();
int* unique_colors = malloc(csc.nodes * sizeof(int));
int id = 0;
for(int i = 0; i < csc.nodes; i++)
unique_colors[i] = -1;
for(int i = 0; i < csc.nodes; i++){
if(used_nodes[i] == false && colors[i] == i){
unique_colors[id] = i;
id++;
}
}
end_time = omp_get_wtime();
return unique_colors;
}
int* coloring(struct CSC_CSR csc, bool* used_nodes){
double start_time, end_time;
start_time = omp_get_wtime();
bool check = true;
int start, end, index, min;
int* colors = malloc(csc.nodes * sizeof(int));
#pragma omp parallel for num_threads(NUM_THREADS)
for(int i = 0; i < csc.nodes; i++)
colors[i] = i;
while (check){
check = false;
#pragma omp parallel for num_threads(NUM_THREADS) private(start, end, min, index)
for(int i = 0; i < csc.nodes; i++){
start = csc.pointers[i];
end = csc.pointers[i + 1];
min = colors[i];
if(used_nodes[i] == false){
for(int j = start; j < end; j++){
index = csc.indices[j];
if(used_nodes[index] == false){
if(min > colors[index]){
min = colors[index];
colors[i] = min;
check = true;
}
}
}
}
}
}
end_time = omp_get_wtime();
return colors;
}
int trivial_scc(struct CSC_CSR csc, struct CSC_CSR csr, bool* used_nodes){
double start_time, end_time;
start_time = omp_get_wtime();
int trivial = 0;
#pragma omp parallel num_threads(NUM_THREADS)
{
#pragma omp for reduction(+:trivial)
for(int j = 0; j < csc.nodes; j++){
if(csc.pointers[j + 1] - csc.pointers[j] == 0 || csr.pointers[j + 1] - csr.pointers[j] == 0){
used_nodes[j] = true;
trivial = trivial + 1;
}
}
#pragma omp for reduction(+:trivial)
for(int j = 0; j < csc.nodes; j++){
if(used_nodes[j] == false){
if (csr.pointers[j + 1] - csr.pointers[j] == 1 && used_nodes[csr.indices[csr.pointers[j]]] == true){
used_nodes[j] = true;
trivial = trivial + 1;
}
else if(csc.pointers[j + 1] - csc.pointers[j] == 1 && used_nodes[csc.indices[csc.pointers[j]]] == true){
used_nodes[j] = true;
trivial = trivial + 1;
}
}
}
}
end_time = omp_get_wtime();
return trivial;
}
struct CSC_CSR coo_to_csr(struct COO coo){
struct CSC_CSR csr;
csr.nodes = coo.nodes;
csr.nz = coo.edges;
csr.pointers = malloc((csr.nodes + 1) * sizeof(int));
csr.indices = malloc(csr.nz * sizeof(int));
#pragma omp parallel num_threads(NUM_THREADS)
{
#pragma omp for
for(int i = 0; i < csr.nodes; i++)
csr.pointers[i] = 0;
#pragma omp for
for(int i = 0; i < csr.nz; i++)
csr.indices[i] = 0;
}
for(int i = 0; i < csr.nz; i++)
csr.pointers[coo.rows[i]]++;
int temp = 0;
int sum = 0;
for(int i = 0; i < csr.nodes; i++){
temp = csr.pointers[i];
csr.pointers[i] = sum;
sum += temp;
}
csr.pointers[csr.nodes] = csr.nz;
int row = 0;
int dest = 0;
for(int i = 0; i < csr.nz; i++){
row = coo.rows[i];
dest = csr.pointers[row];
csr.indices[dest] = coo.cols[i];
csr.pointers[row]++;
}
temp = 0;
int last = 0;
for(int i = 0; i < csr.nodes + 1; i++){
temp = csr.pointers[i];
csr.pointers[i] = last;
last = temp;
}
return csr;
}
struct CSC_CSR coo_to_csc(struct COO coo){
struct CSC_CSR csc;
csc.nodes = coo.nodes;
csc.nz = coo.edges;
csc.pointers = malloc((csc.nodes + 1) * sizeof(int));
csc.indices = malloc(csc.nz * sizeof(int));
#pragma omp parallel num_threads(NUM_THREADS)
{
#pragma omp for
for(int i = 0; i < csc.nodes; i++)
csc.pointers[i] = 0;
#pragma omp for
for(int i = 0; i < csc.nz; i++)
csc.indices[i] = 0;
}
for(int i = 0; i < csc.nz; i++)
csc.pointers[coo.cols[i]]++;
int temp = 0;
int sum = 0;
for(int i = 0; i < csc.nodes; i++){
temp = csc.pointers[i];
csc.pointers[i] = sum;
sum += temp;
}
csc.pointers[csc.nodes] = csc.nz;
int row = 0;
int dest = 0;
for(int i = 0; i < csc.nz; i++){
row = coo.cols[i];
dest = csc.pointers[row];
csc.indices[dest] = coo.rows[i];
csc.pointers[row]++;
}
temp = 0;
int last = 0;
for(int i = 0; i < csc.nodes + 1; i++){
temp = csc.pointers[i];
csc.pointers[i] = last;
last = temp;
}
return csc;
}
struct COO create_coo(struct COO coo, int* I, int* J, int M, int nz){
coo.nodes = M;
coo.edges = nz;
coo.cols = J;
coo.rows = I;
return coo;
}