I heard asio has performance issue on OS which use epoll. In summary, when app send huge amount small package in high frequency, there are too much mutex lock unlock.
I want ask:
- Does newest version asio patch this?
- If not, does one io_service per-thread alleviate this issue?