static bool init(unsigned char *vertex_state,
unsigned long vertex_index,
unsigned long bsp_phase,
per_processor_data *cpu_state)
{
struct hyperanf_vertex *vstate = (struct hyperanf_vertex *)vertex_state;
if(bsp_phase == 0) {
hll_init(vstate->hll_ctr, &hll_param);
unsigned long hash = jenkins(vertex_index, 0xdeadbeef);
add_hll_counter(&hll_param, vstate->hll_ctr, hash);
vstate->changed = true;
return true;
}
else {
// rollup counters
hyperanf_pp_data * pp_data =
static_cast<hyperanf_pp_data *>(cpu_state);
if(vstate->changed) {
vstate->count = count_hll_counter(&hll_param, vstate->hll_ctr);
vstate->changed = false;
}
pp_data->local_neighbour_cnt += vstate->count;
if(vstate->count > pp_data->local_seed_reach) {
pp_data->local_seed_reach = vstate->count;
pp_data->local_seed = vertex_index;
}
return false;
}
}
END_TEST
START_TEST(test_hll_init_and_destroy)
{
hll_t h;
fail_unless(hll_init(10, &h) == 0);
fail_unless(hll_destroy(&h) == 0);
}
CAMLprim value
caml_hll_merge(value hll_a, value hll_b) {
hll_t *_hll_a= (hll_t*)hll_a;
hll_t *_hll_b= (hll_t*)hll_b;
hll_t *merged_hll = malloc(sizeof(hll_t));
int _init_res = hll_init(_hll_a->precision, merged_hll);
int _merge_res = hll_merge(_hll_a, _hll_b, merged_hll);
return (value)merged_hll;
}
END_TEST
START_TEST(test_hll_size)
{
hll_t h;
fail_unless(hll_init(10, &h) == 0);
double s = hll_size(&h);
fail_unless(s == 0);
fail_unless(hll_destroy(&h) == 0);
}
int main() {
hll_t *hll = malloc(sizeof(hll_t));
int res = hll_init(18, hll);
char line[MAXLINELEN];
double after;
while (fgets(line, MAXLINELEN, stdin)) {
hll_add(hll, line);
}
after = hll_size(hll);
printf("%.1f\n", after);
hll_destroy(hll);
return 0;
}
END_TEST
START_TEST(test_hll_add_hash)
{
hll_t h;
fail_unless(hll_init(10, &h) == 0);
char buf[100];
for (uint64_t i=0; i < 100; i++) {
hll_add_hash(&h, i ^ rand());
}
fail_unless(hll_destroy(&h) == 0);
}
END_TEST
START_TEST(test_hll_add)
{
hll_t h;
fail_unless(hll_init(10, &h) == 0);
char buf[100];
for (int i=0; i < 100; i++) {
fail_unless(sprintf((char*)&buf, "test%d", i));
hll_add(&h, (char*)&buf);
}
fail_unless(hll_destroy(&h) == 0);
}
/**
* Converts a full exact set to an approximate HLL set.
*/
static void convert_exact_to_approx(set_t *s) {
// Store the hashes, as HLL initialization
// will step on the pointer
uint64_t *hashes = s->store.s.hashes;
// Initialize the HLL
s->type = APPROX;
hll_init(s->store.s.precision, &s->store.h);
// Add each hash to the HLL
for (int i=0; i < SET_MAX_EXACT; i++) {
hll_add_hash(&s->store.h, hashes[i]);
}
// Free the array of hashes
free(hashes);
}
END_TEST
START_TEST(test_hll_add_size)
{
hll_t h;
fail_unless(hll_init(10, &h) == 0);
char buf[100];
for (int i=0; i < 100; i++) {
fail_unless(sprintf((char*)&buf, "test%d", i));
hll_add(&h, (char*)&buf);
}
double s = hll_size(&h);
fail_unless(s > 95 && s < 105);
fail_unless(hll_destroy(&h) == 0);
}
END_TEST
START_TEST(test_hll_error_bound)
{
// Precision 14 -> variance of 1%
hll_t h;
fail_unless(hll_init(14, &h) == 0);
char buf[100];
for (int i=0; i < 10000; i++) {
fail_unless(sprintf((char*)&buf, "test%d", i));
hll_add(&h, (char*)&buf);
}
// Should be within 1%
double s = hll_size(&h);
fail_unless(s > 9900 && s < 10100);
fail_unless(hll_destroy(&h) == 0);
}
int main(int argc, char *argv[]) {
long i;
struct HLL hll;
if(hll_init(&hll, 16) == -1) {
perror("hll_init");
exit(1);
}
for(i = 0; i < 100000000; i++) {
long r = random() % 1000000;
hll_add(&hll, &r, sizeof(r));
}
printf("Estimate: %f\n", hll_count(&hll));
hll_destroy(&hll);
return 0;
}
CAMLprim value
caml_hll_init(value precision) {
hll_t *hll = malloc(sizeof(hll_t));
int res = hll_init(Int_val(precision), hll);
return (value)hll;
}
