Leistungsproblem bei der Verwendung von HPX zur Parallelisierung in C++-Code ⇐ C++

[Guest]

Ich versuche, meinen Code mithilfe von HPX zu parallelisieren, um die Leistung zu verbessern. Unten ist der Originalcode und mein Versuch, ihn mit HPX umzugestalten.
Originalcode:

Code: Select all

std::vector find_mem(std::vector data){
std::string output = "";
Timer timer;
uint_t n = 0;
unsigned char* concat_data = concat_strings(data, n);
uint_t *SA = NULL;
SA = (uint_t*) malloc(n*sizeof(uint_t));
int_t *LCP = NULL;
LCP = (int_t*) malloc(n*sizeof(int_t));
int32_t *DA = NULL;
DA = (int32_t*) malloc(n*sizeof(int32_t));

timer.reset();
gsacak((unsigned char *)concat_data, (uint_t*)SA, LCP, DA, n);
double suffix_construction_time = timer.elapsed_time();

timer.reset();
int_t min_mem_length = global_args.min_mem_length;
int_t min_cross_sequence = ceil(global_args.min_seq_coverage * data.size());
std::vector joined_sequence_bound;
uint_t total_length = 0;
for (uint_t i = 0; i < data.size(); i++) {
joined_sequence_bound.push_back(total_length);
total_length += data[i].length() + 1;
}
// Find all intervals with an LCP >= min_mem_length and