In short: there is no significant performance difference.
As it has already been pointed out in the comments of your blog, the benchmark is useless. First of all you are testing without the trailing forward slash so web server needs to do a 302 redirect first. Secondly, at just a bit more than 1 request per second you are pretty much testing the PHP performance rather than web server's performance.
Even in your static file test it clearly shows an async-IO based webserver (Litespeed) out performs Apache, which I assume is using prefork MPM.
And there is one more thing that you have failed to test -- memory usage. With high concurrency (ab -c50) you can very quickly find out that there are 50 instances of Apache httpd running at 10MB each, verses one single instance of Litespeed at 5-10MB. That's 50x more memory usage, where you can use for disk cache or MySQL cache or memcached to further improve your performance.
Alright, thanks. But I still don't understand what those concurrent connections mean...
Concurrent connection -- how many TCP connections that are currently in ESTABLISHED state between users' web browser (or proxy) to your web server. As HTTP is really a state-less protocol (if you don't look at persistent connections), a connection is hardly correlated to the number of "users online". Someone sends a request to retrieve a page, and after page is retrieved, the connection is closed (if there is no keep alive). In that case the connection count is 0, but someone is certainly online reading your page! Vice versa that someone can issue multiple connections requesting images together with the document, even though there is only one person online.
Real world example -- one of my Drupal site usually reports around 150-180 "users online" (i.e. unique sessions that were active over the last 15 minutes), however Lighttpd that it uses reports less than 40 connections. It will be even less if I haven't extend the HTTP Keep Alive to 30 seconds. Default keep alive for Apache is 15 seconds and for most busy sites it is reduced to 1-2 seconds to reduce the concurrent connections. How many "online users" can 150 concurrent connections support, when you have a sensible HTTP Keep Alive setting? A LOT (for the VPS anyway).