I am applying kqueue() to Nimf.

Wed, Jan 24 2024 16:50:34 KST

In the case of poll(), there is a problem that if n fds are added, the loop to check the event must repeat n times. However, since kevent() returns a list of events that have occurred, the loop only needs to iterate as many times as the number of events that have occurred. So I’m expecting that using kqueue() will improve the performance of Nimf.

Are there any other ways to improve performance?

• When I have time later, I would like to add padding bytes to structures and packets to prevent performance degradation.
• The performance of my self-implemented hashmap is not very good. So, I am thinking about using the hash map in cpp instead of the hash map I implemented myself.