Unexpected output in OpenMP parallel Block

The following program, if my understanding is correct, should create at most 10 threads and run the omp parallel block :

#include <iostream>
#include <omp.h>
using namespace std;

int main()
{
    omp_set_num_threads(10);
#pragma omp parallel
    {
        int id = omp_get_thread_num();

        printf("%d %d %d\n", id, omp_get_num_threads(), omp_get_max_threads());
    }
    return 0;
}

The output shows that it was run 12 times. But omp_get_num_threads() = 1 in all of the threads, and somehow omp_get_max_threads = 12 except for the first thread.

output :

0 1 10
0 1 12
0 1 12
0 1 12
0 1 12
0 1 12
0 1 12
0 1 12
0 1 12
0 1 12
0 1 12
0 1 12

I have compiled the exact same code with cmd line : g++ -fopenmp .\openMp.c -o prog in VS code terminal, the output was expected and is correct :

1 10 10
2 10 10
3 10 10
4 10 10
0 10 10
5 10 10
6 10 10
7 10 10
8 10 10
9 10 10

• I have added libomp.lib to : Configuration Properties > Linker > Input > Additional Dependencies
• -fopenmp -Xclang flags added to C/C++ > Additional Options
• C/C++ > All Options > Open MP Support > Yes
• Not sure if there is any libomp.dll, can't find it in MSVC intallation directory.

Note : I am using a 6 cores CPU with 12 logical processors.

the only thing you need to do in MSVC to enable openmp is to enable it with

• C/C++ > Language > Open MP Support > Yes

this step is equivalent to adding a command line argument /openmp

the issue you see is caused by manually linking libomp.lib, you should remove that. you also shouln't pass -fopenmp -Xclang, as those are clang arguments, not MSVC arguments.

If you are working with multiple compilers and platforms, I recommend you use a build system to automatically set-up your toolchain, CMake has those flags built-in How to set linker flags for OpenMP in CMake's try_compile function, and most IDEs including VSCode and Visual Studio have good support for CMake, and you don't need to learn the quirks of each compiler.