Nitro enclave and use of python multiprocessing

Hello, I'm running a computation in Python using pool.map() of the multiprocessing library. On the EC2 parent instance, when I run the process first using only one thread and then 2 threads using pool.map the processing time is approximately divided by two as I would expect. However, in the Nitro Enclave (with 2 cpus) the computation time is the same. Any idea what could be wrong? Thanks