Take the 2-minute tour ×
Ask Ubuntu is a question and answer site for Ubuntu users and developers. It's 100% free, no registration required.

I know that Wubi is slower in disk IO, but I can't find anything about it online other than vague statements such as "slightly slower disk seeks". Has anyone done any benchmarks to quantify the speed difference?

share|improve this question

3 Answers 3

up vote 16 down vote accepted

You've inspired an article on Phoronix, entitled The Performance Impact Of Ubuntu's Wubi Windows Installer, which on the 2nd, 3rd and 4th pages has a series of benchmarks that show quite clearly the performance impact, see an example below:

Gzip Compression benchmark in Ubuntu 10.10 and Ubuntu 10.10 with Wubi

share|improve this answer
2  
I recommend all users actually follow the link to the article. The representation here is shamelessly picking the worst benchmark and ignoring the others, leaving a completely biased view that is then used to mislead other users. e.g. here askubuntu.com/a/258189/14916 (i.e. simply blaming Wubi rather than trying to help users with the real underlying problem) –  bcbc Feb 19 '13 at 17:32
    
@bcbc feel free to edit it to make it a better answer :) –  8128 Feb 21 '13 at 0:22
    
Then how could I express my righteous indignation? ;) Sigh, okay, I'll add some more graphs. –  bcbc Feb 21 '13 at 3:54
    
Actually, I think we need written permission to copy the graphics: "All information displayed on Phoronix Media web-sites are the property of Phoronix Media and is protected by copyright. Articles may not be copied or distributed without the prior written consent of Phoronix Media." –  bcbc Feb 21 '13 at 4:18

I found this on the forums, You can read it in it's original post here

  1. Disk performance is slightly slower. You should not be able to notice any difference under normal circumstances. The problem becomes more evident if you have little memory and use swap a lot or if your windows partition is very fragmented. It should be still far faster than under a typical VM or LiveCD.

  2. The filesystem is less reliable than in a real installation. This is because you have 2 filesystems nested within each other which makes it more vulnerable than ntfs or ext3 taken individually. This can be an issue if you hard reboot (unplugging the power). Hard reboots are never ever a good idea, but even less so in wubi. We have taken some measures to minimize the risk (and will add a couple of tricks to next build), so that hard reboots can be better tolerated, but the rule of thumb is: do not hard reboot. In linux there are alt+sysrq key combinations if you get stack for any reason.

  3. Hibernation / Suspend does not work properly. We are looking into this with the help of some ubuntu kernel devs. Either we'll fix that or we will disable hibernation/suspend.

A side issue is that with wubi people tend to allocate less disk space than they would with a normal installation (since they take it as a trial and then keep using it). Of course if the free space is over, it is going to create issues, but that is hardly our fault. We are looking into ways to expand virtual disks, but that would be a separate app anyway. That said in linux you can create a link from a folder within a real partition to a folder within a virtual disk, thus alleviating the pain.

In short: allocate enough space, do not hard reboot and do not suspend/hibernate. Other than that it should be the full monty: same speed (other than for #1), same hardware support/detection, same behaviour, same software. A small trade-off considering that we* provide what can possibly be considered the easiest OS installer ever created, whatever the OS.

As for long term use, I would consider wubi as a mid-term solution. You can use it happily for weeks and months, but because of the 3 issues above, if you find yourself using Ubuntu quite heavily, you might want to do a full installation later on. That said we have a tool to migrate virtual disks to a real partition (LVPM by tuxcantfly). So migration should be quite smooth (that should result in an installation which is 100% identical to a standard one while keeping your data and settings).

If you have a free partition or a spare hard disk and are confident about partitioning and ISO burning there is no much reason to use Wubi, just go straight for the full installation via live CD. But for people that do not know what a partition is, wubi is probably the best solution to date, particularly once tools such as LVPM reach final status. I hope that wubi will bring a small revolution to Linux installers and hence to Linux adoption, similar to what Knoppix/LiveCD did a few years ago', and I would not be too surprised to see Wubi clones implemented by other *nix distros in the near future.

FYI: Wubi will hopefully become an official installation method by next Ubuntu release and we will "merge" with Ubuntu. Even though the recommended long term installation will still be based on the current LiveCD installer (with partitioning), Wubi installer will also be available on the official CD and as a separate download. By Gutsy we will use the LiveCD ISO as opposed to the Alternate ISO as a source of packages and the installer will be fully "graphical".

  • A big chunk of the merit goes to the people that gave us the possibility to boot linux from within windows (grub4dos) and provided us with r/w access to ntfs (ntfs-3g), without such technologies wubi would not be possible today.
share|improve this answer

Well, two things. One, EXT4 is faster than NTFS, here is a proof : http://www.phoronix.com/scan.php?page=news_item&px=ODIxNw

Two, with NTFS there is fragmentation and even if it's difficult to determine how much it will slower the system you know that it will eventually slow it down.

Three (because there is always a three), don't install Ubuntu on Windows, you don't want to use non free software to start your system that will eventually get an update from Microsoft one day and you won't be able to start Ubuntu again.

share|improve this answer
    
I don't doubt it's slower, I'm just curious as to how much slower. You make a good point about the fragmentation. Unless the benchmarks explicitly take it into account, it's something that I'll have to keep in mind when looking at the results. –  Colin Oct 20 '10 at 21:52

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.