Increasing the buffer size will likely just ensure messages aren't dropped and increase your delay. We're doing between 17 and 22 million messages a day on a single server with 80+ rules, lots of logging and scripting. The server is a VM with 4 CPU and 4gb of RAM. Something definitely isn't right with the performance you're seeing. We dont start to hit the buffer until we get over 2 million messages an hour.
With the F5 are you using UDP or TCP syslog connections?
A couple of things to look at to optimize your installation:
Order the rules so the heaviest used is first.
If the message isn't going to be processed by any further rules in the list(they run sequentially) add a "Stop Processing" action at the end of the rule. This will keep the rule engine from evaluating the message against all rules.
Minimize logging to disk from the rules.