Make several threads with OpenMP and print "Hello world" as many as a number of threads.
Assign for statement tasks to each thread.
Assign different tasks belong to each section to different threads.
Make private variables which uses different memory space despite of same name.
Gather results of each thread operation, and transfer to host memory region. Use with operator and variable name.
Use OpenMP with CUDA. Assign matrix multiplication CUDA kernel to threads which generated by OpenMP.