Code inside omp critical directive is not executed by a single thread at a time in the parallelized block of code

Question

I want to split a big task into 8 small tasks and execute each task by a single thread, however there should exist a better way to make it faster. The goal is to make the big ta