Hi Steve;
Can you do two tests on webpagetest.org (with a minimum of 5 runs each, see “advanced settings”);
* one with only W3TC
* one with AO minification + W3TC page cache
and provide me with the test result URL’s so I can compare?
frank
Sure, but how on earth do I get the results over?
Hello?
All done – how do I get the results over?
just copy/paste the URL’s of the result-pages here 🙂
I didn’t save the pages – gimme a minute whilst I do it again
well, I guess I didn’t explain the concept of the 5 runs well enough (you can tell webpagetest to do one test which includes 5 runs, which allows us to see both average and median results, reducing the chance of seeing non-representative results).
but let’s compare the main numbers for the first of each;
1. without AO:
time to first byte: 3.735s
time to start render: 5.518s
time to doc complete: 8.035s
2. with AO:
time to first byte: 1.256s
time to start render: 2.740s
time to doc complete: 5.077s
so it does not look like things are slower with AO, on the contrary.
so I did a 5-run test on both to confirm (disabling AO by adding ?ao_noptimize=1 to the URL) and the median results are;
1. without AO:
TTFB 1.100s
TTSR 3.050s
TTDC 4.470s
2. with AO;
TTFB: 0.236s
TTSR: 1.415s
TTDC: 4.260s
so conclusion: your site is faster with AO (well, at least according to these tests).
which begs the question; what made you conclude the site was slow with AO?
I agree, the site should be faster with Autoptimize.
The problem is, Autoptimize is lagging and making everything slower.
The concept is there, fair enough, but the website lag is horrendous.
That’s why I asked if there was a problem.
The figures from above – fair enough – show that Autoptimize is doing it’s stuff.
What’s not mentioned, is that the 5.whatever seconds to load, hit first byte or whatever, is the same amount of time that it takes the website to load fully. This is probably why the site using AO looks slower. In any case, I’d rather have a site thats bombing along instead of the one that loads all the resources up before loading.
-
This reply was modified 9 years, 6 months ago by
steveraven.
I don’t understand steveraven; can you somehow show me that the site is slower with AO, because when looking at all the data we collected just now, I don’t see it?
Now assuming there is something not right, did you try disabling JS or CSS optimization to single out the culprit? Maybe try to force JS in head (which makes it load sooner)? You could even force AO not to defer the JS-execution (I can tell you how)?
frank
a visual comparison of the site loading without and with AO, based on our webpagetest-tests can be seen here by the way.
As I already said, the problem is just that Autoptimize is causing a lag.
I’m not saying that AO is not doing it’s work – it obviously is. By mistake last night, I disabled AO and rechecked the speed test (without refreshing the cache) and my speed score rocketed! Then I realised why and went back to normality.
All I’m saying is, that at the moment, AO is working but NOT until the page is fully loaded – by which time a W3TC page has already started working.
Update –
I reactivated AO and found that the javascript is now not working, so I turned it off.
It runs excellently without the javascript!
Finally decided to go without AO because of the lag.
Sure, the physical and theoretical results look ok, but the lagging says different.
Thanks anyway!
That’s a pity Steve, esp. since you confirmed the “lag” wasn’t there any more once you disabled JS optimization.
If ever you want to pick this up again let me know. First thing though, would be for me to understand what you mean that your AO’d pages lag, because we do seem to be talking a different language here 😉
