I was about to point the same issue. A better testing would be preparing the same timeouts for both cases, and for eg storing them in an array or file for future use. Therefore you'll have truly comparable results.

Besides that, I like the idea of a promise pool.