Apache itself is fine. Just need to set MaxClients/MaxRequestWorkers equal to StartServers and MaxSpareThreads (assuming prefork), so that all the forking happens at start up. Similar things for other mpms, I assume.