1665720 - lffl.org article page takes 2x as much time to load on Moto G

Status: REOPENED

(Web Compatibility :: Site Reports, defect)

Description

[Asif Youssuff]

Saw some comments about Fenix being slower than Chrome, got this URL and profiled.

Basic information

Steps to Reproduce:

1. Enable warp (makes this about 2 seconds faster in my brief testing)
2. Restart device
3. Type https://www.lffl.org/2020/09/gnome-40-nuova-numerazione.html into a notepad
4. Tap the URL and open in Fenix nightly

Expected Results:

As fast as Chrome - around 5-6 seconds to load.

Actual Results:

12 seconds or so to load page.

---

More information

Profile URL: https://share.firefox.dev/32GCUWN

Basic systems configuration:

OS version: Android 5.1.1

GPU model: Adreno 305

Number of cores: Quad-core 1.2 GHz Cortex-A7

Amount of memory (RAM): 1GB

Thanks so much for your help.

Comment 1

[Sean Feng [:sefeng]]

I found this site delivers different content to Firefox compare to Chrome.

And switching to Chrome's user agent makes the site loads much faster to me. I used a Moto G5 to test it.

Asif, can you ask the reporter to use give changing user agent a try?

Thanks for filing all of these bugs!

Comment 2

[Asif Youssuff]

(In reply to Sean Feng [:sefeng] from comment #1)

And switching to Chrome's user agent makes the site loads much faster to me. I used a Moto G5 to test it.

Asif, can you ask the reporter to use give changing user agent a try?

I can confirm the increased speed with a Chrome UA, and the original person I saw the comments from was also able to confirm this.

With the user agent it goes fast with and without the https addon. Similar speed to the other browser. With the addon I noticed a slightly slower charging of the pics, but not so relevant.

(In reply to Sean Feng [:sefeng] from comment #1)

Thanks for filing all of these bugs!

You are welcome!

Comment 3

[Sean Feng [:sefeng]]

Moving this bug to Web Compatibility.

Comment 4

[Dennis Schubert [:denschub]]

I sent them a message today. We can ship a temporary user agent override if they don't respond, which will be tracked in bug 1663967.

Comment 5

[Oana Arbuzov [:oanaarbuzov]]

I don't have a Moto G device, but I've checked on my devices and I was not able to reproduce the issue. For me the site loads fast below 4-5 seconds even if I disable "lffl.org" UA Override from about:compat.

Tested with:
Browser / Version: Firefox Nightly 210315 (🦎 88.0a1-20210313094300)
Operating System: Samsung Galaxy S8 (Android 9) - 1440 x 2960 pixels, 18.5:9 ratio (~570 ppi density), Google Pixel 5 (Android 11) - 1080 x 2340 pixels, 19.5:9 ratio (~432 ppi density)

Asif Youssuff do you still reproduce it?

Comment 6

[Asif Youssuff]

Tried the same page on the same device, still saw a significant difference between Firefox and Chrome on the same device. Chrome took a little less than 3 seconds to load the page, while Firefox took around 6.

Since Fenix doesn't allow me to capture profiles in Android 5 anymore, I grabbed a new profile on Reference Browser: https://share.firefox.dev/31ePta8

Comment 7

[Dennis Schubert [:denschub]]

Is this profile from a debug build, or a release/optimized build? If it's compiled with ac_add_options --enable-debug, could you capture a profile again without the debug flag? Also, it looks like there are two tabs with that page open, which might skew things a bit (or maybe I'm missing something). If there are two tabs, could you capture a profile with only one? :)

I'm a bit confused by the profile, because everything takes a lot longer than expected. There's jank everywhere, but I can't even pin that down to specific things (except maybe the surprisingly long paints/reflows/flushes). I also notice that the doubleclick.net script is eating a bit. In Fenix, could you give using an Adblocker a try to see if everything is suddenly fast? Knowing that would make debugging that a bit easier.

Comment 8

[Asif Youssuff]

(In reply to Dennis Schubert [:denschub] from comment #7)

Is this profile from a debug build, or a release/optimized build? If it's compiled with ac_add_options --enable-debug, could you capture a profile again without the debug flag? Also, it looks like there are two tabs with that page open, which might skew things a bit (or maybe I'm missing something). If there are two tabs, could you capture a profile with only one? :)

Dennis, I'm not entirely sure if it is a debug build - I am using the build posted on Google Play. In any case, to try to eliminate confounding variables, I removed and reinstalled the app. I also started the app and opened example.com before profiling (and making sure I only had one tab open) to ensure that content processes were already open and ready to serve requests.

I'm a bit confused by the profile, because everything takes a lot longer than expected. There's jank everywhere, but I can't even pin that down to specific things (except maybe the surprisingly long paints/reflows/flushes). I also notice that the doubleclick.net script is eating a bit. In Fenix, could you give using an Adblocker a try to see if everything is suddenly fast? Knowing that would make debugging that a bit easier.

I grabbed two profiles - the profile with uBlock Origin is a lot faster - much more like Chrome on this device. The un-adblocked one is unfortunately still slow:

uBlock Origin: https://share.firefox.dev/3smTSnp
No ad blocker: https://share.firefox.dev/39dd5Aw

Comment 9

[Dennis Schubert [:denschub]]

(In reply to Asif Youssuff from comment #8)

Dennis, I'm not entirely sure if it is a debug build - I am using the build posted on Google Play.

Oh I'm sorry. I somehow thought you compiled the Reference Browser yourself. Since you didn't, that comment is irrelevant.

Thanks for the new profiles, that helps a lot. I will look into that advertising script to see if it's doing something weird, but that's only part of the issue. Another part is... rendering just being super slow. Usually, I'd suggest you try out WebRender, but I'm aware bug 1630356 exists. Let's hope that that WebRender issue is getting resolved soon, so we can re-visit this and see if the rendering performance improves. :)

Comment 10

[Karl Dubost💡 :karlcow]

Needs Triage.

Comment 11

[Raul Bucata]

We appreciate your report. I was not able to reproduce the issue. The page loads approx in the same time frame as using Chrome. This might be device-dependent.

Tested with:
Browser / Version: Firefox Nightly 92.0a1 (2015823243 -🦎92.0a1-20210718090829🦎)
Operating System: Samsung A51 (Android 11) -1080 × 2400 pixels 20:9 aspect ratio (~405 ppi density)

Suggestion: Try clearing cache/data/cookies, disable Ad-blocker (if available), or use a clean profile, and check again? If there are any changes made to the default settings of the browser (e.g. in about: config) please revert to the default settings and try again. Also, have the required cookies been accepted for this page?

Comment 12

[Oana Arbuzov [:oanaarbuzov]]

I've also checked on 2 devices but was not able to reproduce it.

Tested with:
Browser / Version: Firefox Nightly 100.0a1 (🦎 100.0a1-20220321065848)
Operating System: Google Pixel 5 (Android 12) - 1080 x 2340 pixels, 19.5:9 ratio (~432 ppi density), Samsung Galaxy S8 (Android 9) - 1440 x 2960 pixels, 18.5:9 ratio (~570 ppi density)

Asif can you still reproduce it on your Moto G device?

Comment 13

[Oana Arbuzov [:oanaarbuzov]]

Being unable to reproduce the issue on our devices, Asif can you check on your side if the issue still occurs, if you disable the UA OVerride (about:compat> lffl.org)?

Comment 14

[Calin Tanase]

I'm unable to reproduce this issue on my OnePlus 6 A6000 (Android 11). Asif, can you check on the latest Nightly if you toggle to false "extension.webcompat.perform_ua_overrides" from about:config?

Comment 15

[Dennis Schubert [:denschub]]

Given comment 14, I'll close this as WORKSFORME. Please file a new bug if this is still an issue.

Comment 16

[Thomas Wisniewski [:twisniewski]]

Reopening this one, since we're now tracking the status of webcompat issues with interventions differently (and since we never were able to confirm if the Moto G still has the issue, so the intervention is still active).