As promised in my earlier Ubuntu 16.10 review, I have come up with an Ubuntu 16.10 flavors comparison as well, although, I was planning on coming up with this comparison much sooner (but hey, it’s here!)
Unlike in my Ubuntu 16.04 LTS flavors comparison which only included two main Ubuntu flavors (Ubuntu GNOME & Kubuntu), this time, I’ve also added Xubuntu 16.10 to the comparison because it was requested by a couple of my readers. The ISO disc image sizes are as follows: Ubuntu 16.10 (1.6 GB), Ubuntu GNOME 16.10 (1.5 GB), Kubuntu 16.10 (1.6 GB) & Xubuntu 16.10 (1.3 GB). And also, I only chose the 64-bit versions of the disc images for the flavors review as well.
And in this comparison, I’ll only be comparing the performance related data, the stability and hardware recognition of each flavor. I’ll skip new features and whatnot, because you can find information about those features elsewhere, quite easily.
Anyhow, Ubuntu 16.10 and the other 3 flavors were installed into the same partition of the same drive. And once I was finished installing a flavor, I made sure to boot into the newly installed operating system for at least 5-6 times for letting things settle down. I used each at least for 2 days for evaluating the stability as well. Once I was finished testing one OS, before installing another, I securely wiped out all the partitions that were used by the previous one (to make sure that each OS was installed under ‘pristine’ conditions which in turn increases the accuracy of the data).
Except for Xubuntu 16.10, I also made sure not to make any tweaks or changes to the each operating system yet again to preserve the accuracy of the data (Boot-Up Times, Memory Usage, Power Usage, System Responsiveness and Shutdown Delay mainly), except that for measuring the memory usage I added a shortcut for System Monitor (of each desktop environment) to the taskbar, so that I can avoid opening it through the System Menu which could’ve otherwise increased the memory usage thus negatively effecting the accuracy of the memory usage reading. Whenever possible, I also made sure to disable the update manager from searching for updates for the same reason I just explained. And, except under the Power Usage and System Responsiveness tests, I took 5 samples of all the other tests for coming up with the average values as well.
What happened with Xubuntu 16.10 was that its system moniter application displays the memory usage as a percentage where it does not register small to larger (10 MiB -- 100 MiB I would say) memory usage fluctuationes. Therefore, I was forced to run
apt-get and install ‘Htop’ system monitor (a command-line based, very ligthweight system monitor). Of course this wouldn’t have made any serious negative effects, but I just wanted my readers to know it happened 🙂 .
Before I begin the Ubuntu 16.10 flavors comparison review, below is the hardware details of the laptop that I used to test each operating system:
Intel Core i7-5500U, Hybrid GPU Setup (Intel Broadwell HD Graphics 5500, Nvidia 920M), 4GB RAM DDR3, Hybrid Permanent Storage Setup (Seagate 5400 RPM, 500 GB rotational disk and a Kingston 24 GB SSD), Qualcomm Atheros AR9565 Wireless Adapter, Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller, Realtek ALC3236 Sound Card, LED Display (1366 x 768 resolution, 60 FPS/HZ). It's an Asus laptop (F302LJ-FN024H).
In my previous Ubuntu 16.04 LTS flavors comparison I complained that Kubuntu 16.04 LTS’s installer crashed while trying to setup the keyboard and the layout (the installer used by Kubuntu which is also used in many other GNU/Linux distributions, has done this in few other distributions as well). This time however, I encountered no such issue in Kubuntu 16.10. Other two flavors use Ubuntu’s installer, therefore, just like Ubuntu 16.10, they didn’t come up with any issues whatsoever either.
All in all, I was quite happy.
I use my Android Phone’s ‘Stopwatch’ app to measure the Boot-Up Speed which I ‘define’ as; from the moment I hit the Enter key at GRUB (boot-loader) menu, till each operating system loads the desktop (I always enable ‘Auto User-Login’). And I do not necessarily wait till all of the startup applications are finished loading. If I can see that the desktop is usable (file manager is usable or the system is ready to open an app for instance), then I consider the OS has finished its boot-up process.
This maybe a little controversial for some readers, but this is how I’ve run this test all these years and the reason is because when taking measurements, I always tend to be more practical (from the average end-user’s perspective) than being ‘technically correct’. That said, 90% of the time, by the time I stop measuring, operating systems are fully finished loading the desktops anyway. So anyhow, below is the graph that I created using the gathered data:
As you can see, compared to Ubuntu 16.10, Xubuntu 16.10 was the fastest (13%), Ubuntu GNOME 16.10 was the second fastest (8.2%), and Kubuntu 16.10 was the slowest (12.3%) of all the 4. As far as Kubuntu is concerned, this is not surprising at all because KDE based distributions for some reason have a tendency to take a bit of time while booting.
It is also worth noting that, once every while Ubuntu GNOME 16.10 registered a ‘mammoth’ 53 seconds+ boot-up times because Plymouth (the utility that displays the logo during boot-up and shutdown) for some reason crashed. I was gracious enough to skip those readings when calculating its boot-up times. I came up with this exact same issue while comparing the Ubuntu 16.10 LTS flavors as well.
Memory Usage Upon Desktop Loading…
As you can see, Ubuntu 16.10 recorded the highest memory usage. And compared to it, Ubuntu GNOME 16.10 consumed about 5.7% less memory. The gap is a small one, and it is not surprising either, because both these desktops (Unity and GNOME 3) share so much in common. But the most impressive of all is Kubuntu 16.10. Here we have a very modern (both in looks and in functionality) KDE desktop that has consumed (55.4% decrease compared to Ubuntu 16.10) a fraction less memory than a desktop which most people consider to be a ‘lightweight’ desktop environment (XFce)! It is seriously intriguing why Unity & GNOME, which are also two modern desktops, in relative to KDE, have consumed more than twice that amount!
Kudos to KDE developers!
CPU Usage at Idle…
When let to idle (zero user engagement with the operating system), all the 4 operating systems were able to register between 1% to 0% CPU usages for long periods. No complaints here, well done!
Power Usage at Idle…
When measuring power (I use an awesome tool called ‘powerstat‘ which is built for this exact purpose. It’s highly accurate) I made sure all the display screen related power options such as Screensavers, Dimming, Turning OFF, Locking etc were disabled. WiFi was turned ON (connected to my wireless router) and Blutooth turned OFF, and then I let the OS to idle.
Based on the data I gathered, below is the graph I created, and below that are the screenshots containing the actual outputs of the power usage measuring tool used (to prove that I’m not making any of this up) under Ubuntu 16.10, Ubuntu GNOME 16.10, Kubuntu 16.10 and Xubuntu 16.10, respectfully.
Here Ubuntu 16.10 consumed roughly 9.7% less power compared other 3. 9.7% less energy consumption can make a difference when spanned over a couple of hours.
Under all these 4 operating systems, my Focaltech touchpad failed to work when I turned ON the laptop where I had put it to Sleep mode, previously. This is a known problem with the Kernel.
Here & there Xubuntu failed to restore the previous instance of Bluetooth adapter while loading the desktop where it tuns it ON automatically (other 3 operating systems didn’t suffer from this issue). And also, for some unknown reason, Kubuntu 16.10 and Xubuntu 16.10 somehow mishandle the keyboard inputs.
What happens is that whenever I type something, quite frequently the cursor moves to random places and thus partially written letters get scattered all over the place. This is extremely annoying and my guess is that sometimes my palms gently rest on the two top edges of the touchpad and these sometimes are registered as ‘clicks’ (and ‘scrolls’ as well). If I’m very careful not to let my palms touch the touchpad, this doesn’t occur. So it’s probably related to some mishandling of the touchpad inputs. However, interestingly, this does not occur in Ubuntu 16.10 or Ubuntu GNOME 16.10 where both operating systems pretty much share the GNOME’s technologies.
Kubuntu 16.10 was unable to install proprietary drivers for the Nvidia GPU (920M). Actually, when I opened the ‘Driver Manager’, it said ‘Collecting information about your system’ and it was never able to ‘finish collecting’ that information.
However, Ubuntu and its flavors do have a command-line utility that provides the same functionality. It’s called
ubuntu-drivers. So I opened it up and entered the below command:
When you run this command, this utility will automatically scan for hardware devices that are in need of proprietary drivers, and so it did in this instances. Then I issued the below command which automatically installed the driver:
That’s it, job done! All the operating systems functioned well after installing the proprietary GPU and CPU microcode driver. I also tested them while switching to the more powerful Nvidia GPU as well.
I must say that except for the touchpad related issue, Ubuntu & Ubuntu GNOME did extremely well as far as recognizing and configuring hardware was concerned because that was the only issue I came up with. Despite the driver manager related issue, I would’ve given Kubuntu and Xubuntu the same amount of respect in this regard, but that mishandling of the keyboard inputs (which again is probably related to the touchpad handling) is quite irritating I must say.
I also tested Adobe Flash playback on all these 4 operating systems without any major issues as well.
As mentioned in the beginning, I tested each OS for about 2 days for getting a sense of how stable they are. And, while in all the 4 operating systems I saw an application crash once or twice, but those didn’t affect the running system at all. In Kubuntu 16.10, when trying to switch to the Nvidia GPU where you have to log out of the desktop session (you have to follow this procedure in all the desktops), the KDE session manager crashed. The system appeared to be stuck for 3-4 seconds, but I was able to successfully log out of the system and activate the Nvidia GPU at the end. All in all, I’m quite satisfied.
It matters little whether it’s a rotational disk or a much faster SSD, because compared to other major components that are closely associated with the performance of a computer, the main storage device is quite slow. Therefore, provided that the main storage device is in good condition (because there nothing the OS can do if it’s faulty one), one of the ways to measure how well tuned an operating system is to put the main storage device under heavy stress and observe how well the OS manages the situation.
So what I do in this test is quite simple. I copy a file (about 1.5 GB usually) within two locations under the currently logged in user’s ‘Home’ folder. And as soon as the file copy starts, I try to open a multimedia file through VLC (yes, I have to manually install it on most occasions because most GNU/Linux distributions don’t include proprietary multimedia codec support), then I try to open a couple of applications through the main menu, some by searching in the main menu (if such functionality is available) to add more stress to the storage device (because searching for apps means more work for the hard disk), and then I also try to view a folder in the file manager that is filled with somewhat a large number of files & folders as well (
/usr/bin). When all this is happening, I also try to notice the sensitivity of the cursor because when an OS becomes unresponsive, on most occasions, the cursor has a tendency to be stuck as well.
So I performed this test in Ubuntu 16.10, Ubuntu GNOME 16.10, Kubuntu 16.10 and Xubuntu 16.10. How did it go then? (below images are all illustrations)
Well, unsurprisingly, both Ubuntu 16.10 and Ubuntu GNOME 16.10 performed quite identically (because again, they both share so much in common) and they both performed really well. On both these operating systems, VLC was interrupted twice but each time only for about half a second (it was nothing really), while the file manager struggled a bit while opening location, almost all the applications were opened before the file copy job was finished. And the cursor didn’t lose its sensitivity either.
Unlike Ubuntu and Ubuntu GNOME, Kubuntu 16.10 uses the ‘CFQ’ I/O scheduler (this is utility that controls read & write requests to the main storage device) and under it the responsiveness was horrible! Twice the system completely got stuck for about 3-4 seconds and most of ‘heavyweight’ applications were only opened after the file copy finished. There was also a big delay for VLC to open the multimedia file too.
However, then I changed the I/O scheduler to ‘deadline’ (the Kernel comes with 3 actually), one that’s used by Ubuntu & Ubuntu GNOME by default, rebooted the computer and ran the test again. And after the ‘switch’ the responsiveness was so impressive that I would say that it was even slightly better than what I observed in Ubuntu & Ubuntu GNOME!
Xubuntu 16.10 also did very well. The cursor got stuck for about a second, and VLC playback was interrupted twice (for about 1 second on the first instance and 0.5 second on the second instance, I would say) and all of the apps were opened up before the file copying finished.
So all in all, I would say Ubuntu 16.10, Ubuntu GNOME 16.10 and Xubuntu 16.10 were on the same performance level. If I was forced give it a score, I’d say 8.5 points out of 10. After changing the I/O scheduler, Kubuntu 16.10 was the most impressive of them all. On a scoring level, I’d give it 9 points out of 10.
Compared to Ubuntu 16.10, Kubuntu 16.10 was the most slowest (165%). Both Ubuntu 16.10 and Xubuntu 16.10 were marginally faster (8.7%). Just like I said under Boot-Up Times, the shutdown delay of Kubuntu 16.10 is also not surprising. Kubuntu (most KDE distributions in general) for some reason does take it time for shutting down.
I gave you the numbers, and even though I only used each operating system for about 2 days (some such as Ubuntu 16.10 & Xubuntu 16.10 which is what I’m using to write this review, yes that touchpad & key input issue is killing me!, I used for 4-5 days), I shared with you the level of stability, and I also showed you the type of responsiveness that I observed in Ubuntu 16.10 and 3 of its flavors. As always the choice is yours to make, but remember, every opinion is personal, not an abstract one. Therefore, the type of performance that you saw here you may not see under different hardware configurations (for instance, you should see much faster Boot-Up Times under an SSD), or even stability & responsiveness for that matter. So take that into account, and before downloading, make sure to read the release notes as well.
If interested, you can download Ubuntu 16.10 from here (release notes from here), Ubuntu GNOME 16.10 from here (here’s the release notes), Kubuntu 16.10 from here (read the release notes from here) and Xubuntu 16.10 from here (and here’s its release notes). Good luck and thank you for reading!
38 thoughts on “Ubuntu 16.10 Flavors Comparison: Ubuntu vs Ubuntu GNOME vs Kubuntu vs Xubuntu”
Thank you for all the work done! Great review!
You’re welcome 🙂 .
thanks a lot, very useful
I didn’t imagined that KDE used so little memory!
I know! 🙂 .
How can I change the I/O scheduler of kernel to ‘deadline’?
Follow the instructions laid out in the below link:
A very nice review and well thought out. Thank you very much for your effort…..well done.
Thank you! 🙂 .
Sorry, you are incorrect.
Kubuntu 16.10 (as with all other ubuntu variants) ship using the deadline scheduler as default.
However if you have a SATA drive It uses CFQ on that disk (as does ubuntu and others)
i.e – this is on a brand new install of kubuntu 16.10
dmesg |grep sched
[ 0.899430] io scheduler noop registered
[ 0.899430] io scheduler deadline registered (default)
[ 0.899435] io scheduler cfq registered
On my SATA disk it uses CFQ. But Ubuntu, Ubuntu GNOME and Xubuntu all uses deadline on the same disk.
I thought that was the default in all distros… It is in RHEL and Solus I believe.
CFQ is meant to be better for SATA I thought…
I know that CFQ is supposed to deliver better performance under SATA and it has done that in certain distributions. But on the same hardware (I’m talking about my old laptop because I haven’t tested a lot distributions on the new one) under a different distribution, CFQ doesn’t perform that well (yes, on such occasions, I run the test 3-4 times just to make sure it’s not an anomaly). It could be because the developers of different distributions have had altered the original settings of the I/O scheduler. Just a guess.
How do you it seems, why Kubuntu developers uses CFQ i/o scheduller?
I don’t know my friend. A personal preference I guess. That said, on some distributions, CFQ perform better than deadline under the same hardware. Why is that? I honestly don’t know. And trust me, on such occasions, I’ve run the test 3-4 times just to make sure that it wasn’t due to some sort of an ‘anomaly’. It’s what it is… my guess on such occasions is that maybe the developers behind a certain distribution may have had tweaked the I/O scheduler, thus altering the performance. Therefore even though name-wise, on the same I/O scheduler (but performance-wise their ‘behavior’ is now changed), the OS delivers different responsiveness.
Thanks for answer.
Maybe there are some keys (for detailed tuning behaviour disk i/o scheduler), which we (end users) to give to the kernel (for example, via GRUB conf)?
Certainly. However, the best way is to Google by using keywords like ‘tuning Linux I/O schedulers’. Below is a link to OpenSUSE documentation that that describes some of the tweaks:
I’m going to install kubuntu 16.10, try it for few days. Your ‘tuning Linux I/O schedulers’ gonna be help. Thanks!
You’re welcome 🙂 .
Its slightly better than ubuntu, only needs 100mb than bodhilinux, i dunno why superblock happened to kubuntu, last time i installed ubuntu 16.04.1 superblock didn’t happen.
Gayan, thank you.
nice analysis. xubuntu always take lead in *untu family. I have one suggestion. Instead of putting screen shots, its good to see your test results in nice infographics.
You’re welcome. I’ve actually never worked with infographics, but that’s an interesting idea. Can you give me some suggestions?
Yes. Infographics is another raising field. Its catchy esp reviews / test results / researches, etc. I’m also not familiarize with infographics. but as far as i know, there are lots of free tools available. Just google. If you put your results in infographic, lots of space / time would be saved and shortened the reading time.
Thank you… I’ll look into it.
Thank you for all your very interesting and instructive analyses about the various Ubuntu releases.
I am just a little surprised that you never (correct me if I’m wrong) tested Lubuntu operating system. Is it something that you are planning in a near future?
In fact, I am a quite new user of Ubuntu system and I am currently using the 14.04 LTS release on an antique 12 years old laptop (powered by a Pentium M processor @1.73GHz ) which is running very well. I would like to migrate this OS to the 16.04 LTS release, but I’ve read that the minimum system requirement for this OS is a 2GHz processor. So I heard about the Lubuntu system which is apparently less hardware consuming, so I would like to get your opinion about this Linux distribution.
Thanks in advance for your feedback.
First of all, thank you for the appreciation 🙂 . Coming back to the question, I have used Lubuntu in the past, but have not reviewed it on my website. The main reason is because as you’ve said, it’s there for those who’re looking for a more ‘leaner’ operating system which they can use on older hardware, and I don’t have hardware which are 8-10 years old to test it. And I don’t see the point of testing it on newer hardware because people who’re interested in reading such a review ultimately expects an opinion after testing it on older hardware.
That said, I wouldn’t worry too much on the CPU requirements of Lubuntu 16.04 LTS , I’d rather worry about the requirement of the RAM mainly. If it’s (RAM) too high then you may not get the best performance out of it.
Thank you for your answer. The RAM requirement should not be a problem, my old laptop has 2GB of RAM, so it should be enough for running Lubuntu 16.04 LTS smoothly.
I think I will try it and let you know the first results soon.
You’re welcome, and good luck.
As announced, I installed Lubuntu 16.04 LTS on my old laptop. My first impression is that this system is consuming very few machine resources as the reactiveness of this old computer is very good running this OS. The system is booting very fast and all the applications can be launched very fast too. I’ve the impression to use a brand new PC running the latest version of Windows.
That said, I have to admit that the layout of the LXDE interface is not as ‘sexy’ as the one of Ubuntu (Unity). It looks like the good old Windows NT 4.0. But anyway, my objective was to know whether this system could bring a 12 years old PC to life, and this objective is clearly reached.
I faced however some difficulties with some applications, e.g.:
– Impossible to configure my gmail address on Sylpheed wich is the default application for emails management. I had to download and install Mozilla Thunderbird.
– Impossible to use the text application “Abiword”: it is totally unstable! I had to download and install LibreOffice.
– CUPS printing server is not installed by default. I had to install it manually.
– The driver for my HP OfficeJet Pro printer is not working well, though I did not face any problem under Ubuntu with this printer. I had to download another driver which is not available on the Lubuntu disc image.
But anyway, I am quite happy with this system which allows me to keep on using my old laptop.
I am looking forward to hearing from you soon about the next version (i.e. 17.04) of Ubuntu.
Thank you very much for the update, appreciate it (yes, I’ll review the 17.04 when it’s officially released).
Thank you. Very helpful.
Perhaps a review of Black Lab Linux or even the NetOS project of the company?
There are different versions and maybe your review could help people with this operating system.
I think it might be based on Debian and Ubuntu so you would find this interesting, maybe even Steam gaming with this system.
Previous OS/4 OpenLinux.
The test on the memory used after booting is relatively useless. The people tuning the system can get almost any value they prefer. You simply trade off between memory usage, page cache size, preload, zswap and how much you swap out to HDD/SSD during that boot process. You can also delay the loading of services till they are used for the first time. Personally I trimmed Ubuntu 16.04 to around 400MB just by blocking some services to start at boot time.