Measuring theme performance impact
Categories
(Data Science :: Investigation, task)
Tracking
(Not tracked)
People
(Reporter: amyt, Assigned: esmyth)
Details
Brief Description of the request (required):
A few years ago, we applied some optimization code to animated lightweight themes because the files were huge and impacted startup performance. The result was animated themes no longer animated. When we migrated to static themes, that optimization code wasn't ported, and animated themes started moving again.
Business purpose for this request (required):
Our engineers came upon this optimization code recently and want to know if we should apply them again, but we want to first find out if themes are actually still impacting performance, because we don't want to take something away that people have had for the past 1.5 years without a good reason to. We think there's a chance performance is fine because static themes use CSS and smaller image files.
Requested timelines for the request or how this fits into roadmaps or critical decisions (required):
We need to know if we need to re-apply optimization code to animated themes. If there is a significant performance impact on Firefox, we need to fix it asap.
Links to any assets (e.g Start of a PHD, BRD; any document that helps describe the project):
Would it be possible for you to tell us the following:
- Impact of enabled theme on startup
- Impact of enabled animated theme on startup performance.
- Impact of enabled theme on general performance (particularly CPU usage and jank on main thread).
- Impact of enabled animated theme on general performance (particularly CPU usage and jank on main thread).
- Here are some popular ones to use as samples: https://addons.mozilla.org/en-US/firefox/search/?platform=mac&q=animated&sort=users&type=statictheme
Name of Data Scientist (If Applicable):
Ben Miroglio
Please note if it is found that not enough information has been given this will delay the triage of this request.
|
||
Comment 1•7 years ago
|
||
Hi Amy...is this something that can be done by Performance or QA? It's not clear what the Data Science ask is here.
|
|
||
Updated•7 years ago
|
|
|
Reporter | |
Comment 2•7 years ago
|
||
This is normally something Ben would look into for us, but if there is a new team or process I am not sure what it is.
|
|
||
Comment 3•7 years ago
|
||
As stated, we would need to get priority from Reese over the other longer term data science work we are doing, which is why I suggested asking QA if you wanted this done sooner...I would also want to get priority from Eric (performance PM) for what kind of performance measurements he would want to see.
|
|
||
Updated•7 years ago
|
As discussed yesterday, the front end performance team will conduct one-off startup testing to assess the impact of static and animated themes. Results expected this week.
Separately, 1541239 tracks the need for CI-integrated tests for themes.
The test ran four times using the March 28 build of Firefox Nightly (v68.0a). The results below are the geometric mmean of the four runs.
Time to first blank in ms:
Default: 5067.34
Static: 8718.03 (72% longer than default)
Animated: 8769.70 (73% longer than default)
Time to settled about:newtab in ms:
Default: 15807.88
Static: 15991.00 (1.2% longer than default)
Animated: 16533.47 (4.6% longer than default)
|
||
Comment 6•7 years ago
|
||
Do you know how this compares to having an extension installed? Does having any kind of add-on installed have a similar startup penalty, or is loading the theme contributing most of it?
|
|
||
Comment 7•7 years ago
|
||
There's no such thing as not having an extension installed anymore.
Received reference add-ons and themes from Jorge. Verifying this issue and tracking variations of performance tests with add-ons in the perf team backlog
Description
•