Month: September 2018

In a previous speed test on Ancient Greek Keyboard, Pingdom.com graded W3 Total Cache poorly for JavaScript combination & minification. Because of this, a different caching plugin (WP-Rocket, with default settings) was tested against a control (Siteground’s Static Cache running in the background).

Methods

Each test was performed using Pingdom.com’s Website Speed Test. The San Jose, California location option was used initially. (For the second half of the tests, Pingdom.com changed its system & the San Francisco, California location was the only similar option). Each test was conducted with the HTTP URL rather than the HTTPS URL.

In all, 9 tests were run each day for 4 days. Load time results were recorded, and averaged at the end of the study.

Results
The following HTML tables were generated by the Tableizer Excel to HTML conversion site:

Page Load Speed with WP-Rocket

Date	Time	Load Time of Control (sec)	Load Time with WP-Rocket (sec)
9/16/2018	16:45	1.230	0.873
		1.490	0.914
		1.240	0.970
	18:30	1.240	0.907
		1.510	1.150
		1.240	1.210
	20:52	1.420	1.120
		1.220	0.910
		1.190	1.110
9/17/2018	15:45	1.200	1.140
		1.440	1.150
		1.480	0.886
	17:50	1.240	0.931
		1.490	0.932
		1.190	1.130
	20:06	1.500	1.100
		1.490	1.200
		1.440	1.010
9/18/2018	10:45	1.480	1.360
		1.520	1.250
		1.410	1.360
	13:46	1.410	1.260
		1.520	1.330
		1.630	1.250
	16:12	1.610	1.320
		1.490	1.420
		1.390	1.270
9/19/2018	12:01	1.680	1.330
		1.390	1.430
		1.450	1.370
	14:18	2.160	1.220
		1.630	1.580
		1.780	1.480
	17:52	1.460	1.240
		1.530	1.340
		1.530	1.390
n		36	36
Average		1.453	1.190
Std Dev		0.191	0.189

Results

WP-Rocket improved page load speed by 0.263 seconds over the control (1.190 sec vs. 1.453 sec).

Conclusions

An alternative to W3 Total Cache was sought because this plugin did not minify Javascript well. WP-Rocket was therefore chosen and speed tested with its default settings, yielding a page load speed of 1.190 seconds.

This result seems unimpressive, since W3 Total Cache was previously found to increase page load speed to 1.046 seconds. However, it should be noted that the two speeds are not really comparable: W3 Total Cache’s speed was recorded under conditions in which it broke CSS styles. (The two plugin speeds can only be compared under conditions in which they both leave CSS styles unbroken).

WP-Rocket speeds will be further tested over the next few weeks with settings adjusted beyond the default settings.

---

Other posts in this series:

1. Siteground Speed Test (With & Without Static Caching)
2. Speed Test of Swift Performance Lite plugin
3. Caching Plugin Speed Tests: Swift Performance Lite vs. W3 Total Cache
4. Speed Test: W3 Total Cache With Autoptimize
5. Speed Test of WP-Rocket plugin (default settings)
6. Speed Test of WP-Rocket plugin (additional settings)
7. WP Rocket Speed Test (with & without SiteGround Static Caching)

In previous tests, Pingdom.com rated W3 Total Cache rather poorly for combining JavaScript files. Because of this, an effort was made to run W3 Total Cache alongside Autoptimize (the latter was being used to combine & minify JavaScript).

Initial results appeared very encouraging, with page load speeds around 0.5 – 0.6 seconds. However, it was later discovered that these numbers were completely wrong: Autoptimize & W3 Total Cache were interfering with each other and causing a 500 Internal Server Error on Ancient Greek Keyboards.

This 500 error occurred under the following conditions:

1. W3 Total Cache (JS Minify enabled) plus Autoptimize (Optimize JavaScript Code enabled)
2. W3 Total Cache (JS Minify disabled) plus Autoptimize (Optimize JavaScript Code enabled)

This incompatibility means that Autoptimize cannot be used to optimize JavaScript alongside W3 Total Cache (ie: Autoptimize can only be used to Combine & minify CSS alongside W3 Total Cache).

---

Other posts in this series:

1. Siteground Speed Test (With & Without Static Caching)
2. Speed Test of Swift Performance Lite plugin
3. Caching Plugin Speed Tests: Swift Performance Lite vs. W3 Total Cache
4. Speed Test: W3 Total Cache With Autoptimize
5. Speed Test of WP-Rocket plugin (default settings)
6. Speed Test of WP-Rocket plugin (additional settings)
7. WP Rocket Speed Test (with & without SiteGround Static Caching)

A problem was experienced with the Swift Performance Lite caching plugin over the past week. Whenever a plugin was activated or deactivated, the Swift Performance Lite plugin would crawl the entire site, cluttering up the statistics in the security plugin. This also occurred at night, when the site was unattended.

In all, it appeared as though Swift Performance Lite crawled the site about a hundred times a day.

(Initially, this behavior was mistaken for an external bot coming from Siteground Hosting since its hostname was given as XXX.YYYYY.siteground.us. However, blocking this hostname resulted in frequent crawling attempts from WP-CRON.PHP — a cron job most likely caused by a plugin.)

Although it was always my intention to compare page load speeds of Swift Performance Lite with W3 Total Cache, the discovery of this issue gave this comparison test a high priority.

The two caching plugins were therefore tested 36 times to gauge their effect on site speed. (Both were tested with Siteground’s Static Cache running in the background.)

Methods

Each test was performed using Pingdom.com’s Website Speed Test, using the San Jose, California location option. Each test was conducted with the HTTP URL rather than the HTTPS URL.

In all, 9 tests were run each day for 4 days. Load time results were recorded, and averaged at the end of the study.

Results

The following HTML tables were generated by the Tableizer Excel to HTML conversion site:

Table 1. Page Load Speed with Swift Performance Lite vs. W3 Total Cache

Date	Time	Load Time with Swift Performance Lite (sec)	Load Time with W3 Total Cache (sec)
9/5/2018	15:04	1.070	0.861
		0.904	1.040
		1.030	1.060
	17:09	1.050	1.130
		1.060	1.190
		1.890	1.090
	19:40	1.270	1.770
		1.250	1.220
		1.090	0.975
9/6/2018	13:46	1.140	1.150
		1.150	1.160
		1.070	0.888
	16:57	0.927	1.150
		0.967	0.860
		1.090	1.150
	19:14	0.959	0.849
		1.020	0.994
		1.130	0.994
9/7/2018	17:16	0.962	0.873
		0.945	0.899
		1.060	0.906
	18:58	0.958	1.100
		1.080	0.873
		0.976	1.120
	20:44	1.110	1.140
		1.090	1.120
		0.902	0.866
9/8/2018	10:21	0.958	0.910
		1.110	1.210
		1.070	0.951
	13:36	0.911	1.080
		0.994	0.887
		0.954	1.110
	16:02	0.967	0.872
		1.100	1.070
		1.050	1.130
n		36	36
Average		1.063	1.046
Std Dev		0.168	0.173

Table 2. Pingdom Grades for Swift Performance Lite vs. W3 Total Cache

Performance Factors	Swift Performance Lite	W3 Total Cache
Overall Grade	A (91%)	B (88%)
Combine External JavaScript	D (61%)	F (44%)
Combine External CSS	B (80%)	B (86%)
Minimize DNS Lookups	B (87%)	B (88%)
Leverage Browser Caching	A (92%)	B (80%)
Remove Query Strings	A (92%)	B (87%)
Specify A Cache Validator	A (100%)	A (95%)
Serve Static Content fr. Cookieless Domain	A (100%)	A (97%)

Table 1 shows that the Swift Performance Lite caching plugin yielded slightly slower page load speeds than W3 Total Cache (1.063 sec vs. 1.046 sec). Although the 0.017 second difference may not be significant, at the very least it appears that the two caching plugins are at least comparable in their effect on page load speeds.

Table 2 shows that Pingdom.com gave Swift Performance Lite slightly higher letter & numerical grades than W3 Total Cache (A vs. B; 91% vs 88%). It also granted Swift Performance Lite higher grades for a variety of scoring factors.

Despite this apparent advantage, Swift Performance Lite was still still slightly slower than W3 Total Cache.

Conclusions

An alternative to Swift Performance Lite was sought because this plugin tended to clutter up the statistics by excessively crawling the site during cron jobs. Since page load speeds with W3 Total Cache were determined to at least as fast as Swift Performance Lite (and possibly faster), W3 Total Cache will be used as the caching plugin on Ancient Greek Keyboard from now on.

It should be mentioned that W3 Total Cache sometimes breaks CSS styles, so it is anticipated that some of the site’s styles will have to be given greater priority (either through use of the !IMPORTANT property or by using greater specificity in CSS rules).

---

UPDATE (Sep 10, 2018): W3 Total Cache activated on site. As expected, this plugin broke link and header styles. This could not be fixed with the !IMPORTANT property, so the CSS Minify setting was disabled.

After this was done, Swift Performance Lite was deleted from the site.

---

Other posts in this series:

1. Siteground Speed Test (With & Without Static Caching)
2. Speed Test of Swift Performance Lite plugin
3. Caching Plugin Speed Tests: Swift Performance Lite vs. W3 Total Cache
4. Speed Test: W3 Total Cache With Autoptimize
5. Speed Test of WP-Rocket plugin (default settings)
6. Speed Test of WP-Rocket plugin (additional settings)
7. WP Rocket Speed Test (with & without SiteGround Static Caching)