Hi, it is not exactly part of the plugin, but Pro version has two tools used for it: Tracker and Debugger. Tracker can log in many different things and show them in debugger, measure performance, executes queries and all sorts of other things.

Thanks!

Yes, it could be using much more, but there are limitation to what I can measure, and some plugins can get better results than they actually deserve. Testing such plugins is a problem, and in the future methodology and format used for testing will change. This is not perfect testing, and considering how plugins are loaded, there is no way to make it 100% accurate.

There’s no way that it uses just 5 hooks. It uses hundreds. I know that for fact, because we hook to several dozens of them as well.

Any chance that you didn’t configure the e-commerce functionality or are not using a theme that loads products?

As far as e-commerce plugins are involved, I think that it’s a wide enough topic to warrant a complete side-by-side analysis of fully loaded e-commerce sites.

Thanks for stopping by Carl. I don’t compare any of these plugins to each other in that regard, table only states facts in memory use and other things, and it is not comparison. Contact Form 7 is not even close to power of Gravity Forms, even if it uses less memory. Gravity Forms is amazing plugin and considering features in it, I don’t expect that it can use 1MB of memory. My GD Press Tools Pro is highly optimized plugin, yet it uses 2.9MB/2MB of memory. But if it was not optimized, GD Press Tools Pro would use around 6MB.

And that is the point of all this, not how much memory plugin uses, but how much it can use if it was optimized.

Comparing a plugin like Contact Form 7 with Gravity Forms is simply not a valid comparison. The functionality, capabilities and feature set of these 2 plugins is vastly different. Comparing simple plugins which do not contain very much functionality with fully functioning APPLICATIONS in their own right like Gravity Forms, WooCommerce, etc. is like comparing apples and oranges.

The larger and more complex the application, the more code is involved and the more resources will be used. This isn’t a reflection on how good or bad the application is as far as code quality, it’s simply going to require more server resources.

I think comparing such a large number of WordPress plugins to one another as far as load time impact and memory utilization is simply an unfair comparison when it comes to the larger, more complex and feature rich plugins.

The truth is “plugins” like Gravity Forms, WooCommerce, etc. aren’t really just a “plugin”, i’ve said it many times that they are full blown applications that run on top of WordPress just like iPhone and iPad apps run on iOS.

Excellent idea! I will try to add PHP warnings and WP deprecated warnings next time.

DivaVocals,

Of course I’m going to defend my plugin! I’m quite open to criticism if I believe it’s fair criticism.

The purpose of this article is about assessing the load time of plugins as assessed by RAM usage, SQL queries, hooks used and time taken. All of this was done using another plugin – so very objective. I feel that Subscribe2 scored comparatively well against other plugins.

BUT, it was given a rating of “Bad optimization results, must be improved” purely based on the fact that it is one file with one class.

No, forgive my defensive stance but that’s not very objective now is it? To define a series of tests, get your results and then essentially ignore them because subjectively you feel the plugin code could be arranged a little better.

Then, to top it all since I’ve started looking at splitting subscribe2 into several sub-files I have realised I would have to buy the benchmarking plugin to see if it makes any difference. If this really if for plugin developer benefit, how about giving free access to the benchmarking code?

Well, now you know! 🙂

I’m not a coder, and can only “script” in PHP (in my own opinion), but this stuff is something I had to learn. Heck, the basic info is in the Wikipedia article…

Opcode caches and such may require some knowledge of the underlying stuff, and those are often needed when the sites and CMS get bigger. Plugin optimization and code quality are an issue for a site builder like me.

Matt, do not take this the wrong way, but while everyone appreciates the time it takes to develop plugins for free on one’s own time, I think you are missing hte point point here.. You are so busy being defensive about this being a hobby, about you not having training in coding an PHP that you are missing the fact that NONE of this matters.. The point of this article was to bring to the attention of those interested (site owners, site developers, and even plugin authors) that there is room for improvement in some plugins commonly used. This doesn’t appear to be intended as a slight against you (as you seem to have interpreted it as based on your two responses) it was simply meant to be helpful information.. Instead of taking it as information you could use (since by your OWN words, you are not a trained coder) you seem to be a tad offended by it.. **shrugs**

Well, I can understand that, but don’t blame the results because of that. Based on all reactions for the optimization articles, all the emails I got, a lot of WordPress users care how the plugins they use are working and they want them optimized. And that’s why I have written optimization tutorial also.

I expect that all PHP/WordPress developers should be informed about the way PHP works.

That’s maybe because you expect that PHP/WordPress developers have some level of training in coding and/or PHP.

I have none. This is a hobby for me and the last time I formally learned anything about computer languages was dBase3+. I spend enough time remaining up to date for my day job in healthcare as this pays my mortgage. As I said before, I need to keep it as simple as I can for me as from my perspective this is supposed to be something that’s fun and that I enjoy.

I expect that all PHP/WordPress developers should be informed about the way PHP works. If your file is 200KB in size, it doesn’t mean that it will take 200KB of PHP memory. When loaded, file is interpreted into the operation code that is actually executed, source code is not something PHP can run directly. How much memory this code requires depends on data structures and code size. So, the interpreted code would need from 3 to 10 times more memory than the size of the source files. WordPress has about 3MB of PHP source code, yet when loaded it needs from 15 to 20MB of memory to run.

If you have parts of code needed for both front and back end (as any plugin would need), you put that code in a file loaded on both sides. You need one file for admin only code and one for front end only code. And main file to load these 3 files based on the location. This is simplified way to make the code optimized, depending on plugin size and features it is usually fragmented further.

You are basically saying that Subscribe2 uses too much RAM as it loads as a single class – despite the evidence you also present that indicates that this class only hooks where needed, makes few unnecessary SQL requests and does not massively add to page load times.

Maybe I’m being overly defensive but that seems a bit harsh to me. Personally, I find the code much easier to handle in one file containing all the functions rather than trying to split it out. You see a great number of the functions I have are needed on the front end and the admin area. Splitting into 3 or 4 files would just results in always including 2 or 3 of those files with the others being essentially empty.

I will however test your theory and see if splitting the files out does reduce RAM use. I still cannot understand how a plugin script of less that 200kilobytes in size needs 1.4megabytes of RAM to run though!

If there is a legitimate reason for plugin to use the memory it uses all the time, that is OK. Problem is if the plugin loads things it doesn’t need all the time, and that’s why there is a difference in color or optimization grade. WP Download Monitor uses small main file that determines where the user is (admin or front) and loads different parts of the plugin, or attached different hooks.

Subscribe2 has huge 4000 lines long main file that includes almost every piece of code used by the plugin and loads it always, regardless of the need for it. It doesn’t matter how much memory is used, all it matters how much memory can be used if optimized properly, and Subscribe2 is not. This main file should be split into 3-4 smaller files that would be loaded if they are needed, and that will lower memory usage and make it load faster. Once Subscribe2 is optimized this way, I am sure it would use under 1M on both admin side and front.

WordPress SEO by Yoast is plugin with great loading optimization, and it is 3 times the size of Subscribe2 (amount of PHP code),and yet it uses 15% less on admin side and 50% less on front end than Subscribe2, but its code is split into many small files loaded only when actually needed or used.

Please can you explain the results table above for me. In particular, how have you arrived at your colour rankings?

GravityForms 2.9M 0 0.025 4 2.8M 0 0.030 4
Subscribe2 1.4M 0 0.013 5 1.4M 0 0.016 5
WordPress Download Monitor 1.5M 1 0.036 7 1M 1 0.033 7

Picking out these 3 plugins (Subscribe2 is the one I write) as comparators. GravityForms uses more RAM, takes longer and uses 1 less hook and is ranked ‘should be improved’
WP Download Monitor uses about the same amount of RAM takes longer and uses more hooks and gets an ‘Excellent’ rating. Yet Subscribe2 which performs better in terms of speed than both of these, uses similar numbers of hooks, SQL queries and similar or less RAM gets a ‘must be optimized’ result.

The rating seems completely arbitrary to me. While I understand and accept all you say about optimizing code you seem to have taken no account of the complexity of what the plugin is actually deliver either. You cannot expect Microsoft Word to use the same RAM and resources as NotePad when it offers so much more.

Thanks for the comment. I know that WooCommerce team is on this and improvements can be expected. Either way, I too use bbPress and GravityForms, and these results don’t mean much if you like the plugins you are using, and you use cache and some quality hosting.

I’ve build a complex site that uses WooCommerce and Gravity Forms (among others). It feels somewhat slow (without cache, on an “empty” dev server). Seeing this memory usage chart confirms my gut feeling.

I’m really happy that my customer opted for a robust hosting platform. 🙂

Hopefully WooCommerce team sees this, and makes some optimizations.

whilst motivation to speed up was always here for me, it is a good motivator anyway. as you say atm we’re back onto adding features 🙂

the only fault I find with comparing EM to a few others on here is that you can have 1000s of events (I test with 1000s when possible) and that becomes a different ball-game.

Yes, well that is very important, to see how the plugins are doing over time. I understand that sometimes features for the plugin are more important, but when you get that right, you have more time to improve other things. So, good plugins evolve over time and these tests will hopefully show that. Few plugins in the next month tests will show improvements based on versions released after January tests and I am pleased that these tests are having an effect.

For the performance testing I use plugins set by default, but for plugins like this one I took time to make some test data, but it doesn’t show any measurable difference in this case.

P3 plugin from GoDaddy is interesting plugin and I was excited to see it, but I was very disappointed when I tried it. This plugin tries to act as a profiler but it doesn’t work as profiler at all. Profiler is hooking into existing code and works from the inside, this P3 plugin is not doing that. I have run hundreds of tests with it, and on more than half of those plugin that had worst performance was: P3 plugin from GoDaddy. Results are completely unreliable, since I made a dummy plugin that is made to use large amount of memory and to take very long time to load, yet P3 in most tests didn’t even registered it. Also, during 6 hours I tried it, I had to restart my server 6 times because P3 crashed it. Right now P3 plugin from GoDaddy is a good idea, but for practical use I can’t see a way to get reliable results from it.

All these test information should be much more useful to developers. As I said many times, if you like a plugin you are using, it is important that it plays well with other plugin, and if it uses more memory is not that important, because you like it and it is working for what you need. I use bbPress despite it results here, it is a great plugin.

Someone just showed me this, so I’ll be keeping up with your future posts from now on 🙂 It’s very useful to see, and great that you update the results regularly!

Out of curiosity, how many events do you test with?

Thanks for the comment Marcus. Yes, Events Manager 5.0.1 is tested in monthly testing in January, and while it uses a bit less memory, it is still not there. But, for the next test in February, I already started choosing new plugins and updating old ones, and EventsManager is getting better (I used 5.0.3). I am looking forward to this, and I was using the plugin for testing and bit more, and I like really like it.

Worth noting that you probably used Events Manager version 4, version 5 came out late December and has had some major performance improvements (and more on the way as we go).

I am a little surprised you didn’t mention the the “P3 plugin from GoDaddy” in this context, because, of course, it too creates a profile of your WordPress site’s plugins’ performance by measuring their impact on your site’s load time.

You just click “Start Scan” to run an automated scan of your site. The scanner generates some traffic on your site and monitors your site’s performance on the server, then shows you the results.

How accurate are these results? The results have an inherent margin of error because of the nature of the tool and its multi-layered design, but the plugin changes the environment to measure it, and that makes it impossible to get completely accurate results.

It gets really close, though! The “margin of error” on the Advanced Metrics page displays the discrepancy between the measured results (the time for your site’s PHP code to completely run) and the expected results (sum of the plugins, core, theme, profile load times) to show you the plugin’s accuracy.

If you want more accurate results, you’ll need to resort to a different profiler like xdebug, but this will not break down results by plugin.

With this information, you can decide what action to take, but of course, the big question then is – if you’re not the developer, exactly what do I do with these results?

But I guess, at the very least, you could ask the authors to improve them, or if they won’t/can’t, stop using the worst plugins and try to find something better. I did that recently and studied the effect of various “Lightbox” type plugins and was very surprised to see just how much Fancybox for WP is so much better than all the rest (as far as share of site load-speed goes, that is).

Terence.

Thanks for stopping by Dan. Many users wanted to see how currently best eCommerce solutions are doing in terms of optimization. In terms of usability it comes down to personal preference or some specific features, but it is good to know how much resources plugin will use. I will contact you via email next week, in connection to my current project, and we can discuss how can we work together.

bbPress will get there, even now it is not so bad considering that it is the best forum for WP I have tried, and optimization will come with next few versions. I have some ideas about it, and I am sure that JJJ and other developers will think of something to make it even better.

It is awesome to see that the WP e-Commerce Plugin does better then the other e-Commerce Plugins.

We too are interested in working with you to ensure that WPEC is as optimized as it can be for now and for future versions.

I’m also a little sad that bbPress isnt there just yet in terms of optimization. We use it and love it and we just hope that Automattic leave JJJ on the project so that it becomes all that it can be. We’ll never stop using it though… we’re WordPress guys 😛

Thank you Ted!

This is an investigation that has been sorely needed. That you describe the results (and limitations!) carefully, and provide a separate tutorial on the topic, is very generous.

Thank you!

I don’t have that list anymore. But, I am pretty sure that list would include over 100 or maybe more plugins.

“My GD Press Tools Pro and GD Custom Posts And Taxonomies Tools Pro can replace great many other plugins, and that alone can save you a lot of resources.”

I am pretty shure i saw a list of plugins that could potentially replaced by gd press tools (which i bought already) but i cant find it anymore. Can you point me out for it?

Thanks
Mike