Performance Based, Ubuntu 13.04 Review


The recently released, Ubuntu 13.04, not only brings up-to-date packages (including LibreOffice 4.0), which is pretty usual, but according to its ‘Release Notes’ page, ‘Unity’ desktop 7.0 too brings noticeable improvements, concerning memory consumption & performance.

On a side note, I know this might sound a bit weird, but from all the Ubuntu versions that I’ve tried over the years, the best performing ones (less bugs & solid performance) were the ‘ .04′ numbered versions, the ones that get released in April. Where the ‘.10′ versions are usually buggy.

Anyhow, I decided to write a simple review, by comparing both Ubuntu 12.10 & 13.04, from a performance perspective, including details such as: the Boot-up time, memory usage, initial number of system services, power consumption on idle, CPU usage on idle, ACPI & Hardware issues, system responsiveness, last but not least, the Shutdown delay.

Ubuntu 13.04 desktop

Gnome Terminal emulator displaying Unity 7.0

So in this review, I’ll only concentrate on those, rather than writing about software & their new features etc. Here is a simple description about my hardware.

Intel Core i3-2330M CPU, Intel HD 3000 GPU, 4GB RAM (DDR3), Toshiba 7200 RPM (320GB) SATA HDD, Intel N-1030 Wireless adapter, Realtek network adapter ('RTL8168'), LED display with 1366x768 resolution (60Hz/60FPS). It's a Dell Vostro V-131 notebook.

Also remember that, while measuring all the above mentioned tests, I took 5 ‘samples’ of each test, for getting more conservative readings.

Boot-up Time

After clicking the GRUB entry for Ubuntu 13.04 & 12.10, till it booted to the desktop fully, I measured how many seconds it took, compared the results (including the data that I had for ‘openSUSE 12.3 KDE‘) & came up with the below graph.

12.10 vs 13.04 vs 12.3 (openSUSE) - Graph

As you can see, there’s is only a marginal difference between Ubuntu 12.10 & 13.04 while booting, but Ubuntu in general, boots pretty fast when compared with ‘openSUSE 12.3′ or ‘Linux Mint 14 Nadia‘.

The main reason behind the fast boot-up of Ubuntu is because of ‘Ureadhead’. It’s a tool (developed by Ubuntu) that reduces the ‘seek time’ & ‘seek distance’ of the HDD (‘ureadahead’ also improves performance on SSDs) while loading the data required for booting, into the RAM.

Reduced ‘seek time & distance’ improves the throughput of the drive & thus leads to faster start-up times. Without ‘ureadahead’, Ubuntu (12.10) takes around 25-27 seconds to boot!.

Memory Usage

Anyway, after that I measured the initial memory usage, upon loading the desktop fully & came up with the below graph.

Initial Memory Usage Graph - 13.04, 12.10 & 12.3 (openSUSE)

As you can see, Ubuntu 13.04’s memory usage, when compared to 12.10 roughly shows a 15% decrease!.

Now, it’s very difficult to find what has helped for reducing the memory usage as such. However, I also saw a good decrease of memory consumption of ‘Compiz’, again, upon the desktop log-in.

Initial memory usage of 'Compiz' (graph) - 13.04, 12.10

As you can see, in comparison, ‘Compiz’ in Ubuntu 13.04 has a reduced memory consumption by 22% (roughly), which is also pretty impressive!. You know, I was never that impressed by ‘Compiz’ to be honest, but whoever that did this, I’m sure the users would absolutely appreciate it :).

LightDM has also become lightweight, a lot! …

Another application, in which I saw the most daunting amount of memory reduction was in ‘lightdm’ (display manager) service. I monitored its memory usage upon desktop loading and came up with the below graph.

'LightDM' memory usage drop (graph) - 13.04, 12.10

As you can see, it’s almost a 90% decrease! (89.4% to be precise). However, sometimes, ‘lightdm’ forks another process (with ‘root’ permissions). So just to check, I killed it manually, because if you kill the running processes of your ‘display manager’, one will lose his pretty desktop & the app windows.

But the killing of the ‘fork’ or the ‘non-original’ ‘lightdm’ didn’t do any harm, though its existence is unusual, it’s probably a bug. Now I’m no expert, but perhaps display manager ‘forking’ its own process, with ‘root’ permissions might also lead to some security concerns.

Anyway, there are other changes that have helped. With 13.04, some system services are loaded on demand, rather than running on the background, all the time. Also, some of the ‘Lenses’ that come installed by default, seem to be using a bit less as well.

That said, the ‘Xorg’ process (‘display server’) in Ubuntu 12.10 is around 16MB, but in 13.04, it’s around 31-32MB. But the overall ‘effect’ is a positive one nevertheless.

I also used ‘top’ command in the Terminal to measure how many processes are running. Below is the graph for that (Gnome Terminal’s processes excluded).

Number of processes after fully booting - 13.04, 12.10

Although it seems that Ubuntu 13.04 runs a few more processes, it still manages to cut the memory usage, as mentioned above.

Power Usage at Idle

I also measured the power usage while the computer was idling (Bluetooth OFF, Wi-Fi ON, brightness set to maximum), below are the results.

Power usage at idle (graph) - 13.04, 12.10

There’s isn’t a big difference, with Ubuntu 13.04 being slightly efficient, I’ll take that though :).

CPU Usage at Idle

‘openSUSE 12.3′ (closely followed by ‘12.2‘) is one of the best KDE based GNU/Linux distributions, in which I’ve seen impressively lower CPU usage readings at idle. That said, except for the age-long bug (?) in ‘Gnome System Monitor’ which causes 4-6 CPU cycles per-seconds (on my Core i3) when idle, I found that the rest of the processes left the CPU alone, for longer periods.

CPU usage while Ubuntu 13.04 is at idle

‘Compiz’ too behaved pretty excellently, as in 12.10, it usually consumes 2-4 CPU cycles somewhat frequently.

Window Handling in ‘Compiz’ (‘FPS’ Benchmarks)

‘Compiz’ is in control of ‘drawing’ application windows, moving them around, adding ‘window control buttons’ etc (although most of the work is done by the ‘display server’ & GTK or Qt interface libraries).

Just like in videos, when an OS is running, the screen that it relies on for displaying its graphical output also has a ‘FPS’ (frames per second), or ‘frame rate’. When nothing is changing on your screen (meaning that it stays as it is, like a still image), the ‘FPS’ should be set around zero, as shown below.

Compiz idling on Ubuntu 13.04 desktop

But if a notification comes in, or if the user started to play a video, minimize a window etc, then the ‘FPS’ on the display screen goes up, more or less, depending on changes on the screen. More ‘FPS’ means more work for the ‘GPU’ (video card) & it obviously leads to more power consumption.

Now one thing that I’ve noticed with ‘Compiz’ in Ubuntu 12.10 was that, whenever I opened ‘HUD’, even if I didn’t enter anything to its search-box, and just let the screen to idle, then ‘Compiz’ still renders the screen around a frame rate of 22-23, rather than setting the ‘FPS’ close to zero.

Idle FPS of 'Compiz' with 'HUD' opened - Ubuntu 12.10

Idle FPS with ‘HUD’ opened -- Ubuntu 12.10

So I though, perhaps it has something to do with the transparency, so to test, I clicked on the ‘Dash’ icon, and kept it opened without typing or anything. Because, if it’s a result of the transparent window (highly unlikely), then the same thing should happen after opening ‘Dash’ as well.

That time however, the ‘FPS’ decreased to ‘0.49’, which is pretty close to zero, and it is more than acceptable. So I hoped that the above issue (under ‘HUD’), would be fixed in 13.04.

Idle 'FPS' for 'Compiz' while 'Dash' is opened - Ubuntu 12.10

Idle FPS with ‘Dash’ opened -- Ubuntu 12.10

I ran the same tests in Ubuntu 13.04 and learned that things are actually gotten worse.

I don’t know whether if it’s an issues with the GPU driver, ‘Compiz’ itself or a bug with the ‘Compiz Benchmark’ plugin (it’s known to come with those). Nevertheless, below is a screenshot of the frame rate, while keeping the ‘HUD’ just opened in Ubuntu 13.04.

Idle 'FPS' of 'Compiz' while 'HUD' is opened - Ubuntu 13.04

Idle FPS with ‘HUD’ opened -- Ubuntu 13.04

As one can easily see, things seems to have gotten pretty bad, as ‘Compiz’ is now showing a frame rate of 59.67, almost 60 — the full ‘FPS’ supported by my display device (the default ‘FPS’ ‘ is automatically detected by ‘Compiz’ based on your display device).

Below is the frame rate that I captured while keeping the ‘Dash’ opened.

Idle 'FPS' of 'Compiz' while 'Dash' is opened - Ubuntu 13.04

Idle FPS with ‘Dash’ opened -- Ubuntu 13.04

Here too, the idle ‘FPS’ has gone from ‘0.49’ to ‘16.03’. Again, I don’t know what is the real reason behind it, but it doesn’t look that good.

ACPI & Hardware Issues

All the versions of Ubuntu that I’ve used in my Dell Vostro V-131 have detected & configured the hardware without any major issues. However, the fingerprint reader has never worked, & sometimes, upon rebooting, the Bluetooth adapter gets turned ON & the brightness also resets to full.

This is also the case with Ubuntu 13.04, but luckily, latter two issues are somewhat easily solvable. While resuming from sleep (in 13.04), sometimes both Bluetooth & the screen brightness got reset, this is the first time it has happened. Other than that, I didn’t encounter any major issues.

System Responsiveness

As usual, I also tested the system responsiveness under heavy multitasking.

While a 1.7 GB size file was being copied (within ‘Home’), I opened Firefox, LibreOffice Writer & Calc, Software Center, System Settings window, Text editor, System monitor, Ubuntu One, opened Terminal emulator for running a command, searched in ‘Dash’, opened & started to use ‘HUD’.

So how did it go ?

Multitasking under Ubuntu 13.04

A recreation of what I did (just a fraction :P) …

It went pretty decently, however, the responsiveness was not as good as how it was under Ubuntu 12.10. Because for few seconds the file copy seemed stuck (‘Compiz’ darkens the so called ‘non responding apps), including few others and the mouse movement wasn’t also as good as under 12.10.

However, given the workload I put 13.04 through, the responsiveness was pretty decent. It was certainly better that how it was under ‘openSUSE 12.3 KDE’.

____________________

Note: Now, I’m not so sure about this, so I’m just gonna say it. I also felt a somewhat aggressive memory management in 13.04. The reason I say that is, even when doing all the above, the overall memory usage had a tendency of quickly changing (reducing), when compared to other versions.

I don’t know with the latest Kernel that comes with 13.04 that they’ve tweaked the memory manager or Ubuntu has written an optimizer by themselves, as I recall a similar incident with ‘Fedora Core 5′. If it’s so, then that could be the reasons behind the slight lag in performance (?)

____________________

Shutdown delay 

I’ve had few issues with the shutdown speed under Ubuntu 12.10. Even when it was working without issues, it still took around 9 seconds for shutting down. It was a bit disappointing when comparing with the 5-6 seconds delay Ubuntu 12.04 had.

But as you can see, Ubuntu 13.04 is one of the most fastest to shutdown that I’ve used so far, as on average, it only takes around 3.7 seconds!.

Shutdown delay (graph) - 13.04, 12.10, 12.3 (openSUSE)

However, here & there, while both shutting-down and rebooting, the OS got completed stuck. The only fix that could be used, was to use the power button manually. That said, when its working properly, what can I say, it was pretty impressive.

Final words

Despite few issues, Ubuntu 13.04 has managed to decrease the memory usage dramatically, comes with an improved ‘Unity’ desktop & ‘Compiz’ (though it has worsened the ‘FPS’ issue), boots fast (no visible improvements though) & shuts-down impressively faster.

It also includes some nice other changes that I didn’t mention here as well.

So, as a final word, in most aspects, it’s certainly better than 12.10, according to my experience. If interested, you can download it from here, after all, it only costs $70 (just kidding! :D).

17 thoughts on “Performance Based, Ubuntu 13.04 Review

  1. serdotlinecho

    Really enjoy this review. This is what i want to know, rather than boring new features or “10 things to do after installing” review. The technical aspects of what makes good OS/distro. Quite shocking to know your discoveries with Dash and HUD, though. I hope the developers will patch this, ASAP.

  2. HateUnity

    I absolutely hate Unity, however the Ubuntu base is great and swapping desktops is easy (Kubuntu is a great distro right now)…That is the great thing about Linux in general, you actually have choice.

  3. Ghis

    Same here, good review, well focused and written. Liked that you managed to quantify the performance improvement and not just report that it feels more responsive.

    1. Gayan Post author

      Hi,

      1. The fonts are ‘Yanone Kaffeesatz’ (used in post titles/headers) & the content is displayed using ‘Droid Sans’.

      2. I don’t have the data concerning ‘openSUSE’ Gnome edition. Plus, I thought, adding a KDE distro related info with ‘Unity’ would come in handy for some as well :).

  4. Adi

    “While resuming from sleep (in 13.04), sometimes both Bluetooth & the screen brightness got reset [...] latter two issues are somewhat easily solvable.”
    So, how do you solve these problems?

    1. Gayan Post author

      Hi ‘Adi’,

      Well, that’s not exactly what I said actually. I said:

      ‘… the fingerprint reader has never worked, & sometimes, upon rebooting, the Bluetooth adapter gets turned ON & the brightness also resets to full.

      This is also the case with Ubuntu 13.04, but luckily, latter two issues are somewhat easily solvable.’

      So, if Bluetooth and Brightness gets reset upon rebooting, then you can try applying these two tips – 1 & 2. But I don’t think that they’ll help while resuming from sleep though (it only happened once or twice, I guess I’ve been lucky ever since).

      You can also try adding the ‘vendor’ to the GRUB, though it hasn’t helped me on 12.10 (haven’t tried it on 13.04). Go to this link and scroll a bit for finding the instructions.

      Before applying this however, make sure to undo the changes of the above tip ‘2’, if you had followed that.

  5. BC

    Very nice review, really good!
    So here’s a suggestion: would you do a review like this but also including the lattest LTS version, today is 12.04?
    Because I always try to keep with the LTS versions, I just change if there’s a significant update (as a performance boost) or the softwares are getting too old.

    1. Gayan Post author

      Hi ‘BC’,

      First of all, thank you! :).

      As per your request, it’s really good one!. However, since we’re like 11 months away from the next ‘LTS’ version and it’s been like almost a year since 12.04 came out, I kinda feel a bit lazy about writing about 12.04, sorry about that mate.

      That said, again, this is great suggestion and I’ll definitely keep that in mind while coming up with future reviews. Thanks again.

      1. BC

        Hey Gayan,

        Sorry to take so (soooo, ehehe) long to answer you (actually I had ‘lost’ this post of yours in the middle of my numerous bookmarks and haven’t returned to see your reply…).

        But, very (very, very, very, ehehe) nice you liked the idea. I understand your point, indeed 12.04 is a little bit older, but I really think it would be nice and kind and useful if in the future posts you also would include a “LTS comparison”.
        *
        Finally (finally!) I’m taking a good action specially about this post (already did about your blog): signing the RSS just for the comments and replies of this post, so I’ll be easylly* updated.

        [*How do I write this in English? As you can see, English is not my primary language, which is Portuguese, and from Brazil, so, sorry for misspellings, grammar, vocabulary (etc. etc. etc.) errors... but, in the end of the day, I think we can understand each other, ehehe].

        Cheers and, you know it, good job!

  6. Greg Zeng

    Interested in how the many variants of Ubuntu 13.04 compare: LXDE, XFCE, KDE, …. with Unity. Is that possible – please?

    Chief Information Officer Retired, 1984
    Australian Capital Territory

    1. BC

      Hey Greg (and also Gayan),

      I also think this suggestion very nice (specially about XFCE and LXDE, my two favorites DEs), but I also understand that the work would be huge. And keep it update every six months, and it would have to be done in the same cpu (or at least, with very similar hardware), doesn’t seem a very attractive thing to do… (ehehe).

      But here’s ‘the suggestion #2′: what about try to ‘recruit’ some volunteers to write this kind of comparison about a specific DE (and, of course, in the same cpu, probably the writer’s own)? Maybe some kind of partnership with another(s) blog(s)?

      I would volunteer to write about LXDE, but I have a very old and very (sorry for word, but it is true! ehehe) crappy cpu: a 2006 Brazilian branded Itautec notebook with Pentium M 2.0 GHz, 2 GB of RAM, in the top of a Via VN800 motherboard, with openchrome video driver for the integrated graphics… (which is, as you can imagine, horrible). So all my results will be… horrible, of course. And it would help no one (or very, very few people).

      But I’m kind of ‘scheduled’ to buy a new laptop next year, and this time a good one (AMD A8 or Intel i7), so if we keep in touch till then, I would volunteer again (but someone would have to revision or even rewrite my text, because my horrible English… ehehe).

      Cheers and waiting for replies,

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>