如何将GDB与openMP和并行编程一起使用？

Question

I was given a text file as input with 1,000,000 entries between 0 and 9. Each entry corresponds to a vote for a candidate made by a citizen of a country. I was asked to read the file into a sequence and then determine who won the race. I was asked to do this with both parallel and non-parallel programming. My parallel function (pCounter) does not work and I am trying to understand why, but I can't get GDB to run in openMP. It just "continues" over the function when I want it to "step" through. Does anyone know how I can use GDB for openMP and parallel programming?

EDIT When I try to step through a parallel function, I get the following result:

[New Thread 0x7ffff619e700 (LWP 11152)]
[New Thread 0x7ffff599d700 (LWP 11153)]
[New Thread 0x7ffff519c700 (LWP 11154)]

Compiler line: g++ -g homework_5_1.cpp -o hw51 -std=c++11 -fopenmp

Code:

void sCounter(myint *x, myint len){
  myint candidates[10];
  for(int i=0; i<10; ++i){
    candidates[i]=0;
  }
  for(int i=0; i<len; ++i){
    candidates[x[i]]+=1;
  }
  int winner=0;
  for(int i=0; i<10; ++i){
    if(i==winner)
      ++i;
    if(candidates[i]>candidates[winner]){
      winner=i;
      i=0;
    }
  }
  cout<<"The winner is candidate "<<winner<<endl;
  for(int i=0; i<10; i++){
    cout<<candidates[i]<<" ";
  }
  cout<<endl;
  int sum=0;
  for(int i=0; i<10; ++i){
    sum+=candidates[i];
  }
  cout<<sum<<endl;
}

void pCounter(myint *x, myint len){
  myint numberOfThreads;
  myint candidates[10];
  for(int i=0; i<10; ++i){
    candidates[i]=0;
  }
  int winner=0;
  #pragma omp parallel
  {
    #pragma omp for
    for(int i=0; i<len; ++i){
      candidates[x[i]]+=1;
    }
    #pragma omp for
    for(int i=0; i<10; ++i){
      if(i==winner)
        ++i;
      if(candidates[i]>candidates[winner]){
        winner=i;
        i=0;
      }
    }
  }
  cout<<"The winner is candidate "<<winner<<endl;
  for(int i=0; i<10; i++){
    cout<<candidates[i]<<" ";
  }
  cout<<endl;
  int sum=0;
  for(int i=0; i<10; ++i){
    sum+=candidates[i];
  }
  cout<<sum<<endl;
}

Answer 1

To step through the parallel part of your code use the breakpoint command to reach the start of the parallel region.

The syntax is as follows:

b <filename>:<linenumber>

When you've started your gdb session enter the command above, then enter "y" when prompted with the following question:

No source file named <filename>.
Make breakpoint pending on future shared library load? (y or [n])

Now run your code by entering "r".

Once your code reaches the breakpoint it should start with the following line:

Thread 8 hit Breakpoint 1.

Now you can step through it like you did in the single-threaded region.