Multiple Readers / Multiple Writers Solution

Asked Mar 27 '13 at 10:19

Active Mar 27 '13 at 10:30

Viewed 521 times

I have the following function that I would like to run by multiple threads:

float * readWrite(int & idx, const float * vector, const int & dimension){
    float * subPart = new float[dimension];
    for(int i=0; i < dimension; i++)
        subPart[i] = *(v+idx*dimension+i);
    idx++;
    return subPart;
}

So Each thread will read the variable idx and also increment it. I would like that at no time should readers and writers be allowed in at the same time. The solution would be that all threads would be able to access to idx and actually create a critical section (boost::mutex) only when idx changes. How could I do that with boost or even for C++11 std::mutex avoiding both the dead lock and slower down the reading access.

Thanks

edited Mar 27 '13 at 10:30

asked Mar 27 '13 at 10:19

saloua

2,433
4
27
37

6

malloc and C++11, nice mix – Denis Ermolin Mar 27 '13 at 10:21
3

There is solution for you http://stackoverflow.com/questions/989795/example-for-boost-shared-mutex-multiple-reads-one-write – Denis Ermolin Mar 27 '13 at 10:22
You can't prevent a little slowdown if you are using synchronization. – bash.d Mar 27 '13 at 10:22
4

why not simply use an `std::atomic_int`? – KillianDS Mar 27 '13 at 10:34
@KillianDS Thank you for your answer. Indeed, I didn't know that before. – saloua Mar 27 '13 at 11:01

Multiple Readers / Multiple Writers Solution

0 Answers0