Open Bug 1320745 Opened 8 years ago Updated 2 months ago

Download speeds limited on gigabit internet connection

Categories

(Core :: Networking, defect, P3)

52 Branch
defect

Tracking

()

UNCONFIRMED
Performance Impact medium

People

(Reporter: mozilla, Unassigned)

References

(Blocks 1 open bug)

Details

(Keywords: perf, Whiteboard: [necko-triaged])

Attachments

(1 file)

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.11; rv:52.0) Gecko/20100101 Firefox/52.0
Build ID: 20161128004004

Steps to reproduce:

Either visit a speed test site like fast.com or download a large file ( e.g. Ubunti ISO ). In Firefox Developer Edition and consumer Firefox 52 speeds are limited to about 300Mbps whereas in Chrome, Safari and Opera speeds reach 850-900 Mbps. Tests done via gigabit ethernet and via Wifi with same results. Confirmed across multiple machines and connections.


Actual results:

Large files take 2-3 times longer to download in Firefox than in Chrome/Opera/Safari. The only test where I saw gigabit speeds in Firefox was in http://beta.speedtest.net/result/5834698839. But on fast.com and in real world download tests Firefox was 2 to 3 times slower and did not approach gigabit speeds.


Expected results:

Files should take approximately the same amount of time to download and fast.com should report similar speeds. Speeds should approach gigabit speeds 900+ Mbps but instead see about 300 Mbps in Firefox while other browsers download at gigabit speeds.
Component: Untriaged → Downloads API
Product: Firefox → Toolkit
Component: Downloads API → Networking
Product: Toolkit → Core
Which OS are you using?

Can you make wireshark cap? It would help me figure out if tcp is limited. (you can also send it via e-mail)

Http logging would be good as well. how to obtain it is described at:
https://developer.mozilla.org/en-US/docs/Mozilla/Debugging/HTTP_logging
and please replace line:
export MOZ_LOG=timestamp,rotate:200,nsHttp:5,nsSocketTransport:5,nsStreamPump:5,nsHostResolver:5
with
export MOZ_LOG=timestamp,nsHttp:5,nsSocketTransport:5,nsStreamPump:5,nsHostResolver:5

Thank you!
Flags: needinfo?(mozilla)
Whiteboard: [necko-backlog]
See Also: → 1329486
Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258
Priority: -- → P1
Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258
Priority: P1 → P3
here is a similar issue, can anyone repro or confirm it?   bug 1444681
Attached file Speedtest.7z

Ookla Speedtest.net results comparison for Firefox 70 on both, Windows 10 1903, and Linux Mint 19.2 with multiple browsers & desktop and CLI apps.

Comment on attachment 9105959 [details]
Speedtest.7z

Ookla speedtest results comparison for Firefox 70 on both, Windows 10 1903, and Linux Mint 19.2 with multiple browsers & desktop and CLI apps.

I can confirm that I'm experiencing the same issue in Windows 10 1903 (64-bit) and Linux Mint 19.2 64-bit Cinnamon. My download speed seems to be limited to around 300 Mbps using Firefox 70.0.1 (64-bit) on Windows 10. The situations is just a little bit better on Linux Mint. I've noticed that all browsers seems to suffer some performance penalty vs the Ookla desktop app and CLI tool. The straight desktop app seems to get the best performance. Chrome and Brave get about the same, which is slightly less. Firefox's speed is pitiful vs its competition. Obviously, this will need to be rectified as we're about to enter the 10Gbase-T era. I don't see how Firefox can expect to continue without a major overhaul and speed boost.

Same problem here Windows 7 64-bit. Firefox is limited to 300 Mbps. Chrome reaches ~920 Mbps.

Also this problem seems only to appear over HTTPS, over HTTP the regular speed of 920 ish is achieved.

(In reply to Loreno Heer from comment #9)

Also this problem seems only to appear over HTTPS, over HTTP the regular speed of 920 ish is achieved.

I can confirm this speed limit issue. Firefox's download speed is throttled at 250mbps for me using the latest Firefox 64-bit. And as Loreno Heer said above, it appears to only affect HTTPS sites, not HTTP. This bug has been submitted over four years ago and still not resolved. Please see the support thread I made about this bug for more info: https://support.mozilla.org/en-US/questions/1327222

I have the same issue here.

O upgraded my internet plan to 600 Mbps (download) / 300 Mbps (upload) and Firefox is the only browser that can't reach my full speed.

Here my tests with Ookla Speedtest:

Mozilla Firefox v99.0 (64-bits)
https://www.speedtest.net/result/12994244380
(295 Mbps Down / 279 Mbps Up)

Google Chrome v100.0.4896.75 (64-bits)
https://www.speedtest.net/result/12994253267
(643 Mbps Down / 337 Mbps Up)

Microsoft Edge v100.0.1185.29 (64-bits)
https://www.speedtest.net/result/12994258092
(666 Mbps Down / 333 Mbps Up)

Ookla Speedtest App for Windows v1.15.163.0
https://www.speedtest.net/my-result/d/4ac2429b-fe40-448c-9c38-39aecf1c5952
(741 Mbps Down / 374 Mbps Up)

I have a wired connection to internet with my desktop computer.

OS: Windows 10 Pro (64-bit) 21H2 Build 19044.1620

Me too. Windows 10, two different network cards. Chrome, Edge >900mbit. Firefox 64bit <350mbit.
A lot speed tests. I can confirm that it does not occur on http speedtests (!) for example on http://kabelspeed.telekom-dienste.de/

First I thought it was a setting of network card.
https://www.glasfaserforum.de/forum/thread/1240-nur-300mbit-von-1gbit-auf-einem-rechner-geloest/?postID=10990#post10990

If I can help in any way, providing more data.. let me know.

What's wrong with your SSL Firefox?

Severity: normal → S3

Redirect a needinfo that is pending on an inactive user to the triage owner.
:kershaw, since the bug has recent activity, could you have a look please?

For more information, please visit auto_nag documentation.

Flags: needinfo?(mozilla) → needinfo?(kershaw)
Blocks: necko-perf
Flags: needinfo?(kershaw)
Whiteboard: [necko-backlog] → [necko-triaged]

Just to report that the issue has been resolved in some version of Firefox between 99.0.1 and 106.0.1.

I have meanwhile Firefox 106.0.2 and problem still persists.

https://www.wieistmeineip.de
Chrome: 692mbit / 184mbit, Ping 12ms
Firefox: 224mbit/ 158mbit, Ping 12ms
Edge: 873 mbit / 163 mbit, Ping 10ms

https://www.deutsche-glasfaser.de/internet/speedtest
Chrome: 919mbit /515mbit, Ping 10ms
Firefox: 219 mbit / 226 mbit, Ping 11ms
Edge: 911 mbit / 431 mbit , Ping 11ms

https://www.speedtest.net/de
Chrome: 583 mbit / 516 mbit, Ping 8ms
Firefox: 209 mbit / 226 mbit, Ping 8ms
Edge: 891 mbit / 484 mbit, Ping 16ms

Adam, Loreno,
Do you still see this problem?

If you do, please create a performance profile https://profiler.firefox.com/

Flags: needinfo?(mozilla)
Flags: needinfo?(helohe)

No change in internet speed in latest version.
Started Profiling went to speedtest.net did a test:
https://share.firefox.dev/3kBZds3

Flags: needinfo?(helohe)

Redirect a needinfo that is pending on an inactive user to the triage owner.
:smayya, since the bug has recent activity, could you have a look please?

For more information, please visit auto_nag documentation.

Flags: needinfo?(mozilla) → needinfo?(smayya)
Flags: needinfo?(smayya)
Performance Impact: --- → medium
See Also: → 1716156

Tested today with Firefox 106.02/Chrome 121 on Mac and unable to reproduce the reported issue with two referenced speed tests.

https://www.wieistmeineip.de
Chrome: 382mbit / 244mbit, Ping 108ms
Firefox: 336mbit/ 247.mbit, Ping 118ms

https://www.speedtest.net/de
Chrome: 654mbit / 771 mbit, Ping 6ms
Firefox: 680 mbit / 772 mbit, Ping 5ms

And https://www.speedtest.net
Chrome: 670mbit / 767mbit, Ping 4ms
Firefox: 674 mbit / 759mbit, Ping 5ms

You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: