1939152 - Poor throughput performance with OpenSpeedTest

Status: UNCONFIRMED

(Core :: Networking, defect, P2)

Description

[Eric Nix]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:133.0) Gecko/20100101 Firefox/133.0

Steps to reproduce:

I host OpenSpeedTest server on a Synology NAS (RS1221+).

Actual results:

I am only able to get 6G throughput with Firefox (6G down, 3G up), but with Chrome I am able to get 9.7G throughput (9.7G down, 9G up).

Expected results:

Would expect Firefox to match Chrome in performance measures.

Comment 1

[Eric Nix]

Attachment: firefox_throughput.webp

Comment 2

[Eric Nix]

Attachment: chrome_throughput.webp

Comment 3

[Eric Nix]

Attachment: chrome macos activity monitor

Comment 4

[Eric Nix]

Attachment: firefox macos activity monitor

Comment 5

[Mayank Bansal]

Thanks for reporting! To help diagnose this issue better, please can you:

1. Type "about:support" in Firefox and copy-paste its contents here
2. Capture a log when the issue occurs:

• In Firefox, ideally freshly started with no other tabs, go to about:logging in a new tab
  - Select the "HTTP/3" preset
• Enable "stack traces for log messages"
• Click on "Set Log Modules"
• Click Start Logging
• Reproduce the bug
• Back on about:logging, click Stop Logging
• In the new tab that appears with the Firefox Profiler web application, in the top right click the button to upload the profile
• Make sure hidden threads are included and upload, then share the link here or send privately to a Mozilla developer
• For reference, please refer to this video: https://paul.cx/public/about-logging-presentation.webm

Comment 6

[Eric Nix]

Attachment: about:support

https://share.firefox.dev/4iQKvHm

Comment 7

[Eric Nix]

https://share.firefox.dev/4iQKvHm

Comment 8

[Eric Nix]

Sorry, I accidentally posted that link twice.

Comment 9

[Eric Nix]

https://share.firefox.dev/49RP6EY

I tried it again because without the logger, I get 6G down but with it running I'm only getting 2.5G up/down.

Comment 10

[Sunil Mayya]

Thank you for reporting this bug.

We believe the upload issue, particularly on high-bandwidth connections, is already a known problem, and we’re actively working to improve it.

Kershaw, does this bug reveal any new insights that could contribute to our ongoing work on the performance issue?

Comment 11

[Valentin Gosu [:valentin]]

I think we can self-host https://openspeedtest.com/selfhosted-speedtest ourselves.
I don't have a 10G connection at home, but hopefully we can reproduce on localhost.

Comment 12

[Kershaw Chang [:kershaw]]

(In reply to Sunil Mayya from comment #10)

Thank you for reporting this bug.

We believe the upload issue, particularly on high-bandwidth connections, is already a known problem, and we’re actively working to improve it.

Kershaw, does this bug reveal any new insights that could contribute to our ongoing work on the performance issue?

I think this is a dup, but I'll try to reproduce this locally.

Comment 13

[Shonk]

I can add to this

i9-9900KS 128GB DDR4 3600 + AQC113C as Open Speedtest Server
i9-14900KS 96GB DDR5 6400 + RTL8126 Client Firefox 5Gbps Down 3.5Gbps Up
i9-14900KS 96GB DDR5 6400 + RTL8126 Client Chrome 5Gbps Down 5Gbps Up
i9-12900K + 64GB DDR4 3900 AQC113 Client Firefox 10Gbps Down 3.5Gbps Up
i9-12900K + 64GB DDR4 3900 AQC113 Client Chrome 10Gbps Down 10Gbps Up
i9-12900KS + 64GB DDR5 4800 AQC113C Client Firefox 10Gbps Down 3.5Gbps Up
i9-12900KS + 64GB DDR5 4800 AQC113C Client Chrome 10Gbps Down 10Gbps Up
All connected via 12 Port 10 Gbps Switch (RTL9313+RTL8264B+RTL8264B)

There is a hard wall of around 3.5Gbps one way

Comment 14

[Shonk]

Here's a brief Recording on two pc's comparing Firefox to chrome
i can do it on more pc's if required but two pc's is ample to confirm this is firefox

Take note of CPU power use
3.5Gbps Up on Firefox over 100W
10Gbps Up on Chrome around 50W

https://youtu.be/OePjZ5rtStQ

Comment 15

[Valentin Gosu [:valentin]]

(In reply to Shonk from comment #14)

Take note of CPU power use
3.5Gbps Up on Firefox over 100W
10Gbps Up on Chrome around 50W

Comment 16

[Eric Nix]

Sorry, I never followed up with this.

The latest builds fixed the download issue. I am able to download at 9.7G. The most I can get upstream is 3.5-4G.

Comment 17

[Shonk]

Nothing has changed in the last 7 months with regard to firefox
any extra download speed is something that has resolved its self on your pc

Comment 18

[Eric Nix]

Would be weird for that to happen considering I'm testing with the same hardware and Chrome hasn't had this issue. Whether intentional or not, I'm still suspicious something changed with Firefox to increase download performance.

Comment 19

[Itiel]

(In reply to Eric Nix from comment #18)

Would be weird for that to happen considering I'm testing with the same hardware and Chrome hasn't had this issue. Whether intentional or not, I'm still suspicious something changed with Firefox to increase download performance.

If possible, I think the devs would appreciate it if you could bisect the builds using mozregression to see what was the change that allegedly fixed your download speeds:
https://mozilla.github.io/mozregression/