The pragma "master" OpenMP should not be included in the "prime" for parallelism "

Question

Why does the Intel compiler not allow me to indicate that some actions in the openmp parallel for block should only be performed using the main thread?

And how can I do what I am trying to achieve without such functionality?

What I'm trying to do is update the progress bar via a callback in parallel for:

 long num_items_computed = 0; #pragma omp parallel for schedule (guided) for (...a range of items...) { //update item count #pragma omp atomic num_items_computed++; //update progress bar with number of items computed //master thread only due to com marshalling #pragma omp master set_progressor_callback(num_items_computed); //actual computation goes here ...blah... } 

I want only the main thread to call the callback, because if I don't use it (say using omp critical instead, to ensure that only one thread uses the callback right away), I get the following runtime exception:

 The application called an interface that was marshalled for a different thread. 

... therefore, the desire to keep all callbacks in the main thread.

Thanks in advance.

Answer 1

 #include <omp.h> void f(){} int main() { #pragma omp parallel for schedule (guided) for (int i = 0; i < 100; ++i) { #pragma omp master f(); } return 0; } 

C3034 compiler error OpenMP "master" directive cannot be directly embedded in the "parallel for" directive Visual Studio 2010 OpenMP 2.0

Maybe so:

 long num_items_computed = 0; #pragma omp parallel for schedule (guided) for (...a range of items...) { //update item count #pragma omp atomic num_items_computed++; //update progress bar with number of items computed //master thread only due to com marshalling //#pragma omp master it is error //#pragma omp critical it is right if (omp_get_thread_num() == 0) // may be good set_progressor_callback(num_items_computed); //actual computation goes here ...blah... } 

Answer 2

The reason you get the error is because the main thread does not exist in most cases when the code reaches the line #pragma omp master . For example, take the code from Artyom:

 #include <omp.h> void f(){} int main() { #pragma omp parallel for schedule (guided) for (int i = 0; i < 100; ++i) { #pragma omp master f(); } return 0; } 

If the code is compiled, the following may happen:

Suppose thread 0 starts (main thread). He reaches the pragma, which actually says "Master, do the following piece of code." The master function can perform a function. However, what happens when thread 1 or 2 or 3, etc. Reaches this part of the code?

The main directive tells the present / listen command that the main thread must execute f() . But the team is the only flow, and there is no presence of the master. The program did not know what to do in the past.

And therefore, I think the master cannot be inside the for loop.

Substituting master directive with if (omp_get_thread_num() == 0) works, because now the program says: "If you are a master, do it, otherwise ignore it."