"Last Change" vs "Fully Loaded" - 900msec difference?

Web Performance Discuss Test Results
1

Hey!
Do you have any idea why there is 900msec difference between ‘Fully loaded’ and ‘last change’ (10.79s vs 11.7s)?

Report:
http://www.webpagetest.org/result/120309_8X_3H3J4/5/details/

FilmStrip:
http://www.webpagetest.org/video/compare.php?tests=120309_8X_3H3J4-r%3A5-c%3A0&thumbSize=200&ival=100&end=all

Another test shows 700msec difference (7.86s vs 8.6s):

Report:
http://www.webpagetest.org/result/120309_YC_3H69S/6/details/

Filmstrip:
http://www.webpagetest.org/video/compare.php?tests=120309_YC_3H69S-r%3A6-c%3A0&thumbSize=200&ival=100&end=all

Questions
[list]
[]What’s the difference between Fully Loaded and Last Change?
[]Why does ‘Last Change’ metric appear only on filmstrip?
[*]Why there’s no CPU charts for filmstrip?
[/list]

I’m assuming that:
Fully loaded is a moment when all resources were loaded including these loaded dynamically after onLoad event.

Last Change is a moment after fully loaded event occurred and browser did some processing (can’t confirm this without cpu chart).

Regards.

2

I customized waterfall view by extending Max time to longer period, now I can see some CPU activity that was very likely ‘last change’

Update MaxTime manually in custom view to 13 sec:
http://www.webpagetest.org/customWaterfall.php?width=930&test=120309_8X_3H3J4&run=5&cached=0

Update MaxTime manually in custom view to 10 sec:
http://www.webpagetest.org/customWaterfall.php?width=930&test=120309_YC_3H69S&run=6&cached=0

In 1st case ‘last cpu activity’ match ‘last change’ from filmstrip.
In 2nd case Last change looks more like 9.6s than 8.6s, any ideas why?

I think this cpu activity confirms difference between last change and fully loaded.

Comments?

3

Fully Loaded is a measure based on the network activity and is the point after onload when there was no activity for 2 seconds.

Last change is a completely visual measurement and is the last point in the test when something visually changed on the screen. It could be something as simple as an animated gif or ad even that didn’t really cause much CPU work but changed some pixels on the screen. It is only captured when video is recorded because it depends on the video capture to measure it.

4

Thanks, now it’s clear.

5

First, thanks for WebPagetest. It it extremely helpful.

I am wondering why “Document Complete” and “Fully Loaded” times are often identical with IE, but with Chrome “Fully Loaded” times tend to be about double “Document Complete” times. Is that browser difference due to Chrome changing something on the screen long after “Document Complete?”

6

It shouldn’t. Do you have sample tests? Fully loaded is based entirely on network activity and it should show up in the waterfall (though HTTPS is a blind spot for Chrome right now).

7

I am seeing a similar issue between Document Complete and Fully Loaded on Chrome. I just posted a new thread about the problem in the Bugs/Issues category, but it’s waiting to be moderated.

In short, take a look at the difference between those two metrics on Run 1 vs. Run 2 of this very simple test: http://www.webpagetest.org/result/120501_6C_4676S/

8

Argh, for the life of me I can’t get it to reproduce.

Trying the test with the dev tools timeline captured to see if there is some network activity that isn’t in the Chrome waterfall. We can’t display SSL traffic for Chrome right now which could explain it (top of my list to get finished - most of the code is done).

I’ll take a look at your other post to see if it can be reproduced in one of the other browsers.

9

Thanks for looking Pat – you’re right that it doesn’t happen every time.

Here’s the link to the new thread in case someone stumbles across this in the future: http://www.webpagetest.org/forums/showthread.php?tid=11068