Open Bug 1616520 Opened 6 years ago Updated 3 years ago

speedtest.net downstream measurement fluctuates more during the test than in Chrome

Categories

(Core :: Networking, defect, P3)

73 Branch
defect

Tracking

()

People

(Reporter: emails44me2+forums, Unassigned)

References

(Blocks 1 open bug)

Details

(Whiteboard: [necko-triaged])

Attachments

(2 files)

1.06 MB, application/x-zip-compressed
Details
466.67 KB, application/x-zip-compressed
Details
Attached file Videos.zip

User Agent: Mozilla/5.0 (Windows NT 6.3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.106 Safari/537.36

Steps to reproduce:

I've ALWAYS had this issue with my Firefox....even before Quantum.
I have waited year for this issue to be resolved, but it jut never does get solved, so I thought I'd write in and show you guys the issue.

Its this issue that keeps me from making Firefox my browser of choice, even though I so badly do want to do so.
Please fix this issue, and I can't wait to switch.

When I run speed tests on https://www.speedtest.net, there results I get from Firefox are ALWAYS more sluggish and slower compared to what I get using Google Chrome. The issue is not with the actual connection speeds, but the way the tests are shown in Firefox.

Chrome displays the tests so smoothly, and the speeds don't jump up and down like they do on Firefox.
This is the main issue.

I've made two videos of the speed tests.

When you look at the videos, you'll see exactly what I mean...
Again...remember...the connection speeds do not matter at all.
What matters is the way Chrome is so smooth and Firefox jumps up and down.

Why is this the case?

See the results for yourself....

Firefox video: https://ufile.io/0e7dknpe
Chrome video: https://ufile.io/iy5qxv99

Did you guys see the issue that I'm talking about?

Try these tests yourself and you'll see exactly what I'm talking about.

PS: I've tried these tests on my friends, parents, sisters, and work computers, and all the results are the same.
And obviously all these systems are indeed very different.
The only things remaining the same are the two browser versions....

=====================
Windows 8.1 Pro (32-bit)
4GB RAM
Google Chrome 80.0.3987.106 (32-bit)
Mozilla Firefox Quantum 73.0.1 (32-bit)

PS: I have also uploaded the video files mentioned above.

Actual results:

The way the speed tests were jumping around in Firefox makes it look so bad compared with the constant and fluid way the page is displayed in Google Chrome.

Firefox video: https://ufile.io/0e7dknpe
Chrome video: https://ufile.io/iy5qxv99

Again...remember...the connection speeds do not matter at all.
What matters is the way Chrome is so smooth and Firefox jumps up and down.

PS: I have also uploaded the video files mentioned.

Expected results:

The way Chrome is able to show the tests and have a smooth speed and speedometer movement, instead of the way Firefox jumps up and down.

PS: I have also uploaded the video files mentioned.

I can see a small difference on Windows 10 x64 between the latest Chrome and latest Nightly. It is a bit flowier on Chrome.

Do you mean to say that the animation of the Ookla speed test is stuttering on Firefox and it's more flowy on Chrome?

Component: Untriaged → DOM: Events
Flags: needinfo?(w0ouw)
Product: Firefox → Core

n Chrome, it reaches the top speed and stays there.
In Firefox, it reaches top speed, then jumps around to 2-3 Mbps less and then more, etc....

Its a bit hard to explain.

The speeds itself don't matter, as in how high they go...but the way Firefox jumps up and down in speeds like its stuttering or something....

Flags: needinfo?(w0ouw)

AFAICT, this isn't an animation issue but downstream speed measurement itself appears to fluctuate more during the test in Firefox than in Chrome.

(Also, the animation stops anyway and you need to look at the number if the downstream speed exceeds 100 Mb/s.)

Dev tools indicate that in Firefox, the downstream test is via Web Socket and the upload is via XHR. In Chrome, both are via XHR.

Chances are that this is simply a matter of a Web Socket measurement being more sensitive than an XHR measurement, but sending over to the networking component for confirmation.

Status: UNCONFIRMED → NEW
Component: DOM: Events → Networking
Ever confirmed: true
Summary: Why the sluggish performance compared to Google Chrome? → speedtest.net downstream measurement fluctuates more during the test than in Chrome
Priority: -- → P2
Whiteboard: [necko-triaged]

What has been happening with this bug?

I would have thought that this issue would have been fixed by now.
Why has there been such a hold up of this issue?

I have updated Firefox several times, but the issue still occurs 100% of the time.
I have just made new videos that you can easily see the issues on.
Video (Results.zip) can be downloaded here: https://ufile.io/ziuh2663

Just a reminder:
Chrome download speeds are so smooth, yet Firefox download speeds jump up and down as shown on the animation.
The upload speed test of Firefox is so much better than the download speed test. You can see the stability difference.

I just reproduced the same issue on the latest Firefox Browser.
My system specs have changed a little bit from above, as I added more RAM and moved onto 64 bit Windows.

My system specs are:
Windows 8.1 Pro (64-bit)
12 GB RAM
Google Chrome 85.0.4183.102 (64-bit)
Firefox Browser 80.0.1 (64-bit)

Attached file Results.zip

(In reply to Henri Sivonen (:hsivonen) from comment #3)

AFAICT, this isn't an animation issue but downstream speed measurement itself appears to fluctuate more during the test in Firefox than in Chrome.

(Also, the animation stops anyway and you need to look at the number if the downstream speed exceeds 100 Mb/s.)

Dev tools indicate that in Firefox, the downstream test is via Web Socket and the upload is via XHR. In Chrome, both are via XHR.

Chances are that this is simply a matter of a Web Socket measurement being more sensitive than an XHR measurement, but sending over to the networking component for confirmation.

I can only confirm that the downstream test in Firefox is via websocket and chrome is via XHR.
Given the truth that Firefox and Chrome use different ways to do downstream test, I think it doesn't make sense to figure out why Firefox is not as "smooth" as Chrome.

Blocks: necko-perf
Priority: P2 → P3
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: