I have this code, parallel odd-even merge sort. I am now trying to optimize it more, and I think a good idea is to parallelize the for loop that is commented, as that comparations are independent. But, when I parallelize that loop, only worse performance is achieved. I printed the thread ID inside the loop, and only one thread is looping, even if I write num_threads(2)

I am parallelizing odd-even Merge Sort. But I only get worst results with the parallel approach than the sequential one.