I am very much confused with respect to TBB
. I am trying to insert <key, value>
pair in an unordered bimap, where the key
is of type uint64_t
and value
is of type string
. I made an attempt to create the loop object
, which is in TBB.h
file, and looks like
void ParallelIndex(uint64_t &kmer_len, std::string &split_kmer, std::string &sequence_content,
uint64_t &sequence_length, size_t &kmer_position ) {
parallel_for( blocked_range<size_t>(0, sequence_length),
[&](const blocked_range<size_t>& r) {
for(i = r.begin(); i < r.end(); ++i ){
split_kmer = sequence_content.substr(i, kmer_len);
reference_index_vector.insert(position(kmer_position, split_kmer));
i += kmer_len-1;
kmer_position += kmer_len;
}
}
);
}
and in the main
function, which is in TBB.cpp
, I tried to invoke the function
index.ParallelIndex(index.kmer_len, index.split_kmer, index.sequence_content,
index.sequence_length, index.kmer_position );
It starts with 0 and goes through n. I am trying to increment i
by 5 (first ++i
and i +=kmer_len-1
, like i = 0, 5, 10, 15, ...
till n), but i
gets incremented by only 1.
Complete code is shown below:
TBB.cpp
#include <iostream>
#include <string>
#include <algorithm>
#include <boost/bimap.hpp>
#include <boost/bimap/unordered_set_of.hpp>
#include <boost/bimap/unordered_multiset_of.hpp>
#include "tbb/parallel_for.h"
#include "TBB.h"
#include "tbb/tbb.h"
#include <tbb/blocked_range.h>
using namespace tbb;
namespace bimaps = boost::bimaps;
typedef boost::bimap<bimaps::unordered_set_of<uint64_t>,
bimaps::unordered_multiset_of<std::string> > bimap_reference;
typedef bimap_reference::value_type position;
bimap_reference reference_index_vector;
uint64_t thread_test::create_index::kmer_len = 5;
std::string thread_test::create_index::split_kmer = "";
std::string thread_test::create_index::sequence_content = "ABCDDBACDDDCBBAAACBDAADCBDAAADCBDADADACBDDCBBBCDCBCDAADCBBCDAAAD";
uint64_t thread_test::create_index::sequence_length = 0;
size_t thread_test::create_index::kmer_position = 0;
size_t thread_test::create_index::i = 0;
int main(){
thread_test::create_index index;
index.sequence_length = index.sequence_content.length();
index.ParallelIndex(index.kmer_len, index.split_kmer, index.sequence_content,
index.sequence_length, index.kmer_position );
for( bimap_reference::const_iterator iter = reference_index_vector.begin(), iend = reference_index_vector.end();
iter != iend; ++iter ) {
std::cout << iter->left << " <--> "<< iter->right <<std::endl;
}
}
TBB.h
#ifndef TBB_H_
#define TBB_H_
#include<iostream>
#include <algorithm>
#include <utility>
#include <boost/bimap.hpp>
#include <boost/bimap/unordered_set_of.hpp>
#include <boost/bimap/unordered_multiset_of.hpp>
#include "tbb/tbb.h"
#include <tbb/parallel_for.h>
#include <tbb/blocked_range.h>
#include <tbb/mutex.h>
using namespace tbb;
namespace bimaps = boost::bimaps;
typedef boost::bimap<bimaps::unordered_set_of<uint64_t>,
bimaps::unordered_multiset_of<std::string > > bimap_reference;
typedef bimap_reference::value_type position;
extern bimap_reference reference_index_vector;
namespace thread_test{
class create_index{
public:
static uint64_t kmer_len;
static std::string split_kmer;
static std::string sequence_content;
static uint64_t sequence_length;
static size_t kmer_position;
static size_t i;
// using lambda function
void ParallelIndex(uint64_t &kmer_len, std::string &split_kmer, std::string &sequence_content,
uint64_t &sequence_length, size_t &kmer_position ) {
parallel_for( blocked_range<size_t>(0, sequence_length),
[&](const blocked_range<size_t>& r) {
for(i = r.begin(); i < r.end(); ++i ){
split_kmer = sequence_content.substr(i, kmer_len);
reference_index_vector.insert(position(kmer_position, split_kmer));
i += kmer_len-1;
kmer_position += kmer_len;
}
}
);
}
};
}
#endif /* TBB_H_ */
problem is i
gets incremented by 1 but I try to increment it by 5 as mentioned above.
Change for(i = r.begin(); i < r.end(); ++i )
to for(; i < r.end(); ++i )
. In the first case you are reinitializing i = r.begin()
.