I have completed the second stage of testing different WordPress versions. This time I have tested on some of the blog pages using the same conditions as for the previous tests. But, in this case, I have also tested number of SQL queries executed. And finally, we can see improvements in WP 2.8 performance on the server side.

I had some troubles with WP 2.8 testing and Arthemia theme, because for some reason a lot of time was wasted on waiting for MySQL server to respond. This was not the problem with WP 2.7.1 for some reason. So, I have spent some time tweaking MySQL on the localhost until the response time improved. Why this was problem only with WP 2.8, I don’t know but it’s not much important, since I didn’t noticed similar problem on this website that is hosted on BlueHost servers. So, localhost MySQL server for Windows need to be set right if you want maximal speed.

Also, important thing is to know that some of the results are caused by the theme. Arthemia is very complicated theme, and it uses multiple loops to render posts, featured lists, random posts, several sidebars, pages and categories menus and on top of that is still not optimized to use WordPress 2.8 optimizations for loading JavaScript and CSS files. Because of that I will not talk much about time needed for server to create pages because in all the tests this time was more or less the same in all tests between browsers, and increased number of plugins added only 10-12% to load page time.

Used memory (MB)

Used memory (MB)

Used memory

As with the administration benchmarks, amount of memory used depend on number of plugins you use.

Due to always improving WP engine parts used for widgets, queries, re-writer and other important elements, WP 2.7 and WP 2.8 use more memory than WP 2.6.5, but on the blog pages this is more balanced and doesn’t causes problems as we had with administration. And the difference in memory use from previous versions is not so big on this side. And for few more MB used, you get some real speed and optimization improvements (as you can see later), so this is acceptable.

Page size

Page size (KB)

Page size (KB)

Measuring page size on the blog is actually tricky to do, because most themes are not using WP 2.8 features for optimization of JavaScript and CSS files, loading them in footer and few other things, so WP 2.8 can’t be seen in the best light until we get more themes that actually know how to use it.

Again, results are more or less the same, except for WP 2.7.1 that generated biggest pages on blog front page and archive. WP 2.6.5 and WP 2.8 generated pages of almost the same size. All measurements are without use of cache.

But, blog running on WP 2.8 has better caching, and on page reload browser needed less than from WP 2.6.5 blog, and this is very important if your blog generates pages with 500KB or more. With WP 2.8,  getting data from server is definitely faster.

Number of queries

And now we are coming to really interesting part. For blog pages I also measured the number of queries WP executes. Because Arthemia is complex theme, I used also WP default theme for this test. First 3 columns are from Arthemia theme, and the 3 small bars are from default theme.

Number of queries

Number of queries

As you can see, number of queries for Arthemia is a bit excessive, but as I said this theme renders a lot on a single page, and that costs in terms of performance. But, important thing is a trend this chart shows. WP 2.8 is better than WP 2.6 and WP 2.7 and it needs less queries to do the same thing. Same trend is visible for default theme that only needs 10-15% queries from Arthemia’s number of queries. Rendering of a single post needs a query or two more from WP 2.7.1, but I couldn’t explain this, on all other types of pages WP 2.8 is better, and in some cases number of queries has decreased for 10% or even more.

Conclusion

All this clearly shows that WP 2.8 is very important step in WP development, and that better optimization of database access is clearly visible with less queries needed. Once we start seeing more themes that use loading optimizations this trend will be visible in generated page size and loading times also. Since displaying blog pages is much more important from administration, then the problems from the previous benchmark don’t look so bad. Take into account that WP 2.8.1 is coming soon, with expected improvements in administration page memory use and speed, it is obvious that it is time to upgrade your blogs to WP 2.8 (or at least wait for 2.8.1).

What’s next?

As Matt suggested in one of the comments for previous benchmark post, we now only need to see the speed of rendering on the browser side where we should see more improvements for WP 2.8. JavaScript/Browser test will be completed in the next few days, and you can expect them next week. I will also analyze the impact of some plugins to blog performances,  and post(s) about that will be published soon.

As always, comments and suggestions are welcome.


, , , , , , , , , , , , , , , , , , ,

Leave a Reply

Date Archives