- W3 Total cache
- WP Super-Cache
- No Plugin (APC only)
- Default install (no APC)
All tests are done with APC enabled except for the “default install”.
This test is not meant to compare static file caching vs APC! Since static caching and “PHP” caching are not the same beast, results will vary depending on your setup! The CPU load was not directly tested which of course is one of the main reasons to be using APC.
There are noticeable differences when you start to drill down into the server and plugin settings and this benchmark did none of that. Please do not take these results as facts, I wanted to emulate a basic setup by just pressing “enable” with zero configuration, aka lazy optimization for people who don’t have time.
This is a “rough” test, I did not check for initial priming, pour over analysis or look for sweet spots. I just threw shit at WordPress with ABench. These results are the average of a small data-set, only 4 tests were run and averaged out for each test case. The benchmark is based on total connections and simultaneous connections (concurrency) a well as failed requests.
- CPU: 4GB 2 Duo 3.17 GHz 64-bit
- OS: Xampp (yuck I know)
- Apache 2.2.21, PHP 5.3.8, APC 3.1.9, ABench 2.3
- WordPress: 3.5.1
- Theme: Twenty Twelve
- URL: post with 5 large images and a bunch of widgets
The higher the better, plugins start to even out with more concurrency
This only tests the 2 higher concurrency data-sets concurrency below 400 typically resulted in zero failed requests. APC -only blew up on the last test and the default blew up on both. The lower the better.
Conclusion: I like batcache.
Though the other plugins did perform very well, and Total Cache has some pretty amazing options, using batcache and the APC drop-in results in a whopping 2 files. Total Cache in comparison is 451 files and 70 folders, at this point it is 1/2 the size of WordPress! I’m not saying it does a bad job, all those features have to go somewhere, but I just like the ease of not really having to deal with any settings, just drop them in and move on to something else.
ps. Sorry for the ugly ass graphs.