t/genzipf.c

   1 /*
   2  * Generate/analyze pareto/zipf distributions to better understand
   3  * what an access pattern would look like.
   4  *
   5  * For instance, the following would generate a zipf distribution
   6  * with theta 1.2, using 262144 (1 GB / 4096) values and split the reporting into
   7  * 20 buckets:
   8  *
   9  *      ./t/fio-genzipf -t zipf -i 1.2 -g 1 -b 4096 -o 20
  10  *
  11  * Only the distribution type (zipf or pareto) and spread input need
  12  * to be given, if not given defaults are used.
  13  *
  14  */
  15 #include <stdio.h>
  16 #include <stdlib.h>
  17 #include <fcntl.h>
  18 #include <string.h>
  19 #include <unistd.h>
  20
  21 #include "../lib/zipf.h"
  22 #include "../flist.h"
  23 #include "../hash.h"
  24
  25 #define DEF_NR_OUTPUT   20
  26
  27 struct node {
  28         struct flist_head list;
  29         unsigned long long val;
  30         unsigned long hits;
  31 };
  32
  33 static struct flist_head *hash;
  34 static unsigned long hash_bits = 24;
  35 static unsigned long hash_size = 1 << 24;
  36
  37 enum {
  38         TYPE_NONE = 0,
  39         TYPE_ZIPF,
  40         TYPE_PARETO,
  41 };
  42 static const char *dist_types[] = { "None", "Zipf", "Pareto" };
  43
  44 static int dist_type = TYPE_ZIPF;
  45 static unsigned long gb_size = 500;
  46 static unsigned long block_size = 4096;
  47 static unsigned long output_nranges = DEF_NR_OUTPUT;
  48 static double percentage;
  49 static double dist_val;
  50 static int output_csv = 0;
  51
  52 #define DEF_ZIPF_VAL    1.2
  53 #define DEF_PARETO_VAL  0.3
  54
  55 static struct node *hash_lookup(unsigned long long val)
  56 {
  57         struct flist_head *l = &hash[hash_long(val, hash_bits)];
  58         struct flist_head *entry;
  59         struct node *n;
  60
  61         flist_for_each(entry, l) {
  62                 n = flist_entry(entry, struct node, list);
  63                 if (n->val == val)
  64                         return n;
  65         }
  66
  67         return NULL;
  68 }
  69
  70 static struct node *hash_insert(struct node *n, unsigned long long val)
  71 {
  72         struct flist_head *l = &hash[hash_long(val, hash_bits)];
  73
  74         n->val = val;
  75         n->hits = 1;
  76         flist_add_tail(&n->list, l);
  77         return n;
  78 }
  79
  80 static void usage(void)
  81 {
  82         printf("genzipf: test zipf/pareto values for fio input\n");
  83         printf("\t-h\tThis help screen\n");
  84         printf("\t-p\tGenerate size of data set that are hit by this percentage\n");
  85         printf("\t-t\tDistribution type (zipf or pareto)\n");
  86         printf("\t-i\tDistribution algorithm input (zipf theta or pareto power)\n");
  87         printf("\t-b\tBlock size of a given range (in bytes)\n");
  88         printf("\t-g\tSize of data set (in gigabytes)\n");
  89         printf("\t-o\tNumber of output rows\n");
  90         printf("\t-c\tOutput ranges in CSV format\n");
  91 }
  92
  93 static int parse_options(int argc, char *argv[])
  94 {
  95         const char *optstring = "t:g:i:o:b:p:ch";
  96         int c, dist_val_set = 0;
  97
  98         while ((c = getopt(argc, argv, optstring)) != -1) {
  99                 switch (c) {
 100                 case 'h':
 101                         usage();
 102                         return 1;
 103                 case 'p':
 104                         percentage = atof(optarg);
 105                         break;
 106                 case 'b':
 107                         block_size = strtoul(optarg, NULL, 10);
 108                         break;
 109                 case 't':
 110                         if (!strncmp(optarg, "zipf", 4))
 111                                 dist_type = TYPE_ZIPF;
 112                         else if (!strncmp(optarg, "pareto", 6))
 113                                 dist_type = TYPE_PARETO;
 114                         else {
 115                                 printf("wrong dist type: %s\n", optarg);
 116                                 return 1;
 117                         }
 118                         break;
 119                 case 'g':
 120                         gb_size = strtoul(optarg, NULL, 10);
 121                         break;
 122                 case 'i':
 123                         dist_val = atof(optarg);
 124                         dist_val_set = 1;
 125                         break;
 126                 case 'o':
 127                         output_nranges = strtoul(optarg, NULL, 10);
 128                         break;
 129                 case 'c':
 130                         output_csv = 1;
 131                         break;
 132                 default:
 133                         printf("bad option %c\n", c);
 134                         return 1;
 135                 }
 136         }
 137
 138         if (dist_type == TYPE_PARETO) {
 139                 if ((dist_val >= 1.00 || dist_val < 0.00)) {
 140                         printf("pareto input must be > 0.00 and < 1.00\n");
 141                         return 1;
 142                 }
 143                 if (!dist_val_set)
 144                         dist_val = DEF_PARETO_VAL;
 145         } else if (dist_type == TYPE_ZIPF) {
 146                 if (dist_val == 1.0) {
 147                         printf("zipf input must be different than 1.0\n");
 148                         return 1;
 149                 }
 150                 if (!dist_val_set)
 151                         dist_val = DEF_ZIPF_VAL;
 152         }
 153
 154         return 0;
 155 }
 156
 157 struct output_sum {
 158         double output;
 159         unsigned int nranges;
 160 };
 161
 162 static int node_cmp(const void *p1, const void *p2)
 163 {
 164         const struct node *n1 = p1;
 165         const struct node *n2 = p2;
 166
 167         return n2->hits - n1->hits;
 168 }
 169
 170 int main(int argc, char *argv[])
 171 {
 172         unsigned long offset;
 173         unsigned long i, j, k, nr_vals, cur_vals, interval_step, next_interval, total_vals, nnodes;
 174         unsigned long long nranges;
 175         struct output_sum *output_sums;
 176         struct node *nodes;
 177         double perc, perc_i;
 178         struct zipf_state zs;
 179
 180         if (parse_options(argc, argv))
 181                 return 1;
 182
 183         if( !output_csv )
 184                 printf("Generating %s distribution with %f input and %lu GB size and %lu block_size.\n", dist_types[dist_type], dist_val, gb_size, block_size);
 185
 186         nranges = gb_size * 1024 * 1024 * 1024ULL;
 187         nranges /= block_size;
 188
 189         if (dist_type == TYPE_ZIPF)
 190                 zipf_init(&zs, nranges, dist_val, 1);
 191         else
 192                 pareto_init(&zs, nranges, dist_val, 1);
 193
 194         hash_bits = 0;
 195         hash_size = nranges;
 196         while ((hash_size >>= 1) != 0)
 197                 hash_bits++;
 198
 199         hash_size = 1 << hash_bits;
 200
 201         hash = malloc(hash_size * sizeof(struct flist_head));
 202         for (i = 0; i < hash_size; i++)
 203                 INIT_FLIST_HEAD(&hash[i]);
 204
 205         nodes = malloc(nranges * sizeof(struct node));
 206
 207         for (i = j = 0; i < nranges; i++) {
 208                 struct node *n;
 209
 210                 if (dist_type == TYPE_ZIPF)
 211                         offset = zipf_next(&zs);
 212                 else
 213                         offset = pareto_next(&zs);
 214
 215                 n = hash_lookup(offset);
 216                 if (n)
 217                         n->hits++;
 218                 else {
 219                         hash_insert(&nodes[j], offset);
 220                         j++;
 221                 }
 222
 223         }
 224
 225         qsort(nodes, j, sizeof(struct node), node_cmp);
 226         nnodes = j;
 227         nr_vals = nnodes;
 228
 229         if (output_csv) {
 230                 printf("rank, count\n");
 231                 for (k = 0; k < nnodes; k++) {
 232                         printf("%lu, %lu\n", k, nodes[k].hits);
 233                 }
 234         } else {
 235                 unsigned long blocks = percentage * nranges / 100;
 236                 double hit_percent_sum = 0;
 237                 unsigned long long hit_sum = 0;
 238                 interval_step = (nr_vals - 1) / output_nranges + 1;
 239                 next_interval = interval_step;
 240                 output_sums = malloc(output_nranges * sizeof(struct output_sum));
 241                 for (i = 0; i < output_nranges; i++) {
 242                         output_sums[i].output = 0.0;
 243                         output_sums[i].nranges = 0;
 244                 }
 245
 246                 j = total_vals = cur_vals = 0;
 247
 248                 for (k = 0; k < nnodes; k++) {
 249                         struct output_sum *os = &output_sums[j];
 250                         struct node *node = &nodes[k];
 251                         cur_vals += node->hits;
 252                         total_vals += node->hits;
 253                         os->nranges += node->hits;
 254                         if (k == (next_interval)  -1 || k == (nnodes - 1)) {
 255                                 os->output = (double)(cur_vals) / (double)nranges;
 256                                 os->output *= 100.0;
 257                                 cur_vals = 0;
 258                                 next_interval += interval_step;
 259                                 j++;
 260                         }
 261
 262                         if (percentage) {
 263                                 if (total_vals >= blocks) {
 264                                         double cs = k * block_size / (1024 * 1024);
 265                                         char p = 'M';
 266
 267                                         if (cs > 1024.0) {
 268                                                 cs /= 1024.0;
 269                                                 p = 'G';
 270                                         }
 271                                         if (cs > 1024.0) {
 272                                                 cs /= 1024.0;
 273                                                 p = 'T';
 274                                         }
 275
 276                                         printf("%.2f%% of hits satisfied in %.3f%cB of cache\n", percentage, cs, p);
 277                                         percentage = 0.0;
 278                                 }
 279                         }
 280                 }
 281
 282                 perc_i = 100.0 / (double)output_nranges;
 283                 perc = 0.0;
 284
 285                 printf("\n   Rows           Hits %%         No Hits         Size\n");
 286                 printf("--------------------------------------------------------\n");
 287                 for (i = 0; i < output_nranges; i++) {
 288                         struct output_sum *os = &output_sums[i];
 289                         double gb = (double)os->nranges * block_size / 1024.0;
 290                         char p = 'K';
 291
 292                         if (gb > 1024.0) {
 293                                 p = 'M';
 294                                 gb /= 1024.0;
 295                         }
 296                         if (gb > 1024.0) {
 297                                 p = 'G';
 298                                 gb /= 1024.0;
 299                         }
 300
 301                         perc += perc_i;
 302                         hit_percent_sum += os->output;
 303                         hit_sum += os->nranges;
 304                         printf("%s %6.2f%%\t%6.2f%%\t\t%8u\t%6.2f%c\n",
 305                                i ? "|->" : "Top", perc, os->output, os->nranges,
 306                                gb, p);
 307                 }
 308                 printf("--------------------------------------------------------\n");
 309                 printf("Total\t\t%6.2f%%\t\t%8llu\n", hit_percent_sum, hit_sum);
 310                 free(output_sums);
 311         }
 312
 313         free(hash);
 314         free(nodes);
 315         return 0;
 316 }