1👍
If you are using the machine for the application server only, usually a good approach to start is using the double of processor cores in the machine, and increase gradually watching the CPU and RAM usage, until you find a optimal number.
If there is more services in the same machine (memcached, or even your database), start with a fewer processes, usually the number of cores + 1.
I think the best approach is to watch the resource consumption and increase the number of processes/threads gradually… Many processes = high CPU wait times. Few processes = idle resources (requests will queue up and wait)
Source:stackexchange.com