Best configurations for speeding up standalone discourse

The queuing means you can do with more web workers

Additionally your median times look pretty high, can you get this on a faster CPU?