3G download speed is very slow compared to Hardy on elderly PIII computers or Microsoft Windows OSs
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Linux |
Invalid
|
Undecided
|
Unassigned | ||
linux (Ubuntu) |
Invalid
|
Undecided
|
Unassigned |
Bug Description
Binary package hint: ppp
A mobile broadband download is very slow (only 10–20 % of what it is supposed to be) in Jaunty, Karmic and Lucid Lynx Alpha 3 (and also in Hardy used in more modern computers, ie. P4->) 0.4–0.6 Mb/s, when the theoretical max. speed of the affected ISP's plan is 3.6 Mb/s (in practise the speed is about 3 Mb/s in Windows Vista).
To reproduce:
Download an Ubuntu iso-image from a nearest mirror. In Hardy used with elderly PIII laptop (see Comment #2 for lshw) download speed is 250–300 kB/s, which is a good five times faster than in Jaunty (about 60 kB/s).
This is a regression and performance issue related to download speed, which is at worst ten times slower on average compared to Windows OS's (40 kB/s vs. 400 kB/s). The extremely slow speed seems to be USB/3G modem independent (tested in Nokia CS-15, Huawei e1552, Huawei E230, Nokia cell phone 6220 classic, ...). Currently there is a Finnish discussions going on at
http://
http://
Karmic:
http://
Interesting is that, when you download for example four Ubuntu distros in the same time, your download speed is approximately: 60 kB/s + 60 kB/s + 60 kB/s + 60 kB/s = 240 kB/s. When you cancel three of those downloads, you will continue with the "normal" Jaunty download speed that is surprisingly 60 kB/s. Some other day the individual download speed could be only about 40 kB/s. On the other hand, torrent download (several seeders) is quite fast: about 300 kB/s.
Tested both near the capital city centre of Helsinki and also at the very edge of the 3G field at countryside. There is no diffenrence between locations.
Tested with: NetworkManager and Wvdial (and by other people with umtsmon and vodavone-
Most probably these speed / network bandwidth issues are related only to one internet service provider called Sonera (TeliaSonera) in Finland. Nevertheless, last autumn the "HelpDesk" at Sonera did not comment this issue.
ProblemType: Bug
Architecture: i386
DistroRelease: Ubuntu 9.04
ExecutablePath: /usr/sbin/pppd
Package: ppp 2.4.5~git200811
ProcAttrCurrent: unconfined
ProcEnviron: PATH=(custom, no user)
SourcePackage: ppp
Uname: Linux 2.6.28-18-generic i686
tags: | added: lucid |
Changed in linux: | |
assignee: | nobody → eetu.piki@gmail.com (eetu-piki) |
status: | New → Confirmed |
assignee: | eetu.piki@gmail.com (eetu-piki) → nobody |
description: | updated |
Changed in linux (Ubuntu): | |
assignee: | Pertti Nieminen (julaos) → nobody |
summary: |
- 3G download speed is very slow compared to Hardy or Microsoft Windows - OSs + 3G download speed is very slow compared to Hardy on elderly PIII laptop + or Microsoft Windows OSs |
Changed in linux: | |
status: | Confirmed → Fix Committed |
status: | Fix Committed → Fix Released |
Changed in linux (Ubuntu): | |
status: | Confirmed → Fix Released |
Very interesting findings: I found a hardware dependency!
The bandwidth is quite fast 270–290 kB/s (almost as fast as it is supposed to be in practice – eg. in M$ Vista even 400 kB/s. The nominal speed used in a bit false marketing is 3,6 Mb/s) using Hardy Live-CD with an old IBM laptop with Pentium III (800 MHz). In the same time, using Hardy Live-CD with a more recent IBM desktop with Pentium 4 (3.20 GHz) the speed is slow 80 kB/s (almost as slow as in Jaunty, ie. ~40–55 kB/s).
I did several download tests from Ubuntu servers during about two hours today. I Have also earlier found that Hardy Live-CD gives a bit faster download compared to Jaunty. Once, in the otherwise fast IBM laptop, the download started slower (about 120–150 kB/s), but that could have been a network or server "artefact", because when I immediately started a new download from other place the speed was again 270–290 kB/s.
Jaunty (others have tested Karmic, check the links on my earlier post) is slow (40–55 kB/s) also in Acer Aspire 3623wxci laptop with Intel Celeron M 370 (1,5 GHz) and HP Compaq Business Desktop dc7100 P4 540 (3,2 GHz Prescott Socket T).
Who can make conclusion? Should this bug report be extended to affect several packages?