So far, I have used two Manjaro releases. The first one was the XFCE desktop based release (Manjaro 0.8.7.1 Ascella) and the other one was the KDE based version (Manjaro 0.8.9). I liked them both, but the XFCE version was more impressive, as far as the performance was concerned (at least).
To this day, Majaro 0.8.7.1 Ascella was one of the GNU/Linux distributions that consumed the least amount of power at idle (it’s worth mentioning that it came with an aggressive hard disk spin down setting that should’ve had contributed to the low power usage readings). It was also impressively responsive when the hard disk was put under pressure, the KDE version, though, was horrible. This is a well known characteristic observation of the KDE desktop, at least in my experience (there are exceptions, though, which makes it all very interesting. Some blame it on the compositing backend of the KDE’s window manager — ‘KWin’).
So anyhow, the last time I tried Manjaro was in the early 2014, and since Manjaro just released its latest KDE version, Manjaro 15.12 Capella (the version naming method is now simply based on the year & the month of the release time), a yet another gorgeous looking KDE desktop (Manjaro developers know a thing or two about artwork 😀 ), I decided to review it.
Compared to my previous Manjaro KDE review the major change here is that this version of Manjaro is based on the KDE Plasma 5.5.1 where as the previous one was based on the KDE 4.12.2. The disc image is also slightly bigger (1.8 GiB) and the default set of applications might also be changed slightly, don’t know for sure though.
For this review, I downloaded the 64-bit version (I highly recommend the BitTorrent based download method as opposed to the direct download as it guarantees that you can always resume, if interrupted) and it uses 4.1.0 LTS kernel by default.
Since in my reviews I quite empathize the performance aspects of the operating system that I’m reviewing, and because these two Manjaro KDE releases differ quite a bit due to the KDE Plasma desktop versions they’re based upon, I didn’t think a mere performance comparison between the two would make way to fair judgments. So I decided to use the Kubuntu 15.10’s data from my Ubuntu flavors comparison review since it too uses the KDE Plasma version 5 (5.4.0 actually).
Before I begin my review, here is a brief information about the hardware that I used to test it (it’s the same system I’ve used to test all these distributions so far):
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.
Manjaro 15.12 KDE comes with three installers. The default one is ‘Calamares’, there’s a command-line based one and also another installer GUI called ‘Thus’ which is beta software.
Out of curiosity I tried to use Thus, but being a beta version, it thusly gave a me a warning ( 😉 ) saying that it’s possible to lose data since it’s still being developed. So I changed my mind and switched over to Calamares, and that was that, as they say.
Anyhow, compared to my previous confrontation, Calamares has been improved slightly (Kubuntu 15.10 also uses the same installer). In the recent past Calamares included a notorious bug that made it to crash at keyboard setup, but I came up with a trick as a temporary fix to this, as mentioned in my Netrunner 14 Frontier review, back in mid 2014, but interestingly this bug was not present in the old Manjaro 0.8.9 back in early 2014.
Luckily, it has all been fixed recently and I came up with no new bugs and Calamares carried out the rest of the installation without any issues. And as I previously did, I admired its intuitive design.
As far as the features are concerned, Calamares doesn’t yet support LVM, which might concern some users. And as a result, although it correctly detected and added an entry to my Ubuntu 15.10 OS (my flagship OS these days. I was forced to let go of Fedora, after the version 23. As they say, ‘When presented, one must always choose the lesser weevil‘.), it failed to recognize & show the two logical volumes Ubuntu 15.10 was using.
First Boot-Up (GRUB & the Boot Logo)…
Manjaro uses a simple GRUB theme that goes nicely with its main dark green theme, and the boot-up logo also follows the same color protocol, as shown below (those small dots don’t represent the actual progress of the boot-up process though).
I’ve always enjoyed Manjaro’s default set of colors and the theme implementation. Simply said, I like how things look in Manjaro 15.12 even more. Application Window colors (default wallpaper included) are pleasing to the eye, icons look gorgeous and KDE Plasma 5.5 desktop has a minimalist appearance. The new theme of Manjaro for the KDE Plasma 5 is called Maia and you can quickly switch between four themes through System Settings window as well.
Apart from the default set of wallpapers of KDE Plasma 5.5, Manjaro too includes a couple of beautiful wallpapers.
Other than that, I won’t go into the details of what new on the new KDE release, mostly because I’m not a heavy KDE user and I have no idea what these subtle changes are. Secondly, with each of its release, KDE does provide a colorful look of what’s new etc anyway.
If you have two minutes to spare, then view the below video which should update you about the changes made to the KDE 5.5 release, or you can also read this bloody thing as well 😛 :
Manjaro comes with a couple of its own tools and these tools are now fully integrated to the KDE’s System Settings window. Nice.
These tools include a hardware device driver installer (which makes the installation of proprietary drivers such as GPU drivers a breeze), a language package manager and another tool for installing/unistalling/configuring various kernels (very useful tool for advanced users) supported by the Manjaro distribution.
Another key feature of Manjaro is that it supports the playback of proprietary multimedia codecs by default. It ships with two video players (VLC 2.2.1 and Xine 0.99.9) and Cantata 1.5.2 is the music manager. Firefox (43.0.1) is the default web browser which also includes the outdated Adobe Flash Plugin. LibreOffice 18.104.22.168 is the office suite and Yakuke 2.9.9 drop down terminal is also included.
It also includes a HDR image manipulator called Exposure Blending Tool (4.14), Karbon 2.9.10, showfoto and digikam image editors which are also from the old KDE 4.14 release. Manjaro 15.12 also comes with a big list of other KDE based applications (such a QBittorent, Kget, Akregator etc) that I really am not going to type into because I’m already board 😛 .
Let’s talk a little about its performance.
I measured these data first, before ‘touching’ anything on the newly installed system. I had to though add the system monitor application icon to the KDE taskbar for opening when measuring the memory usage. Because if I had to navigate the start menu to open the system monitor, every time I measured the memory usage (I took five samples of each of these tests -- boot speed, memory usage & shutdown speed, for coming up with the average results), it could’ve unnecessarily increased the memory usage. It’s the only change I made.
As you can see, there is quite a margin between the old Manjaro 0.8.9 and the new 15.12 releases (Manjaro 15.12 being 39% slower to boot). But as mentioned in the beginning, these releases differ quite a bit, so one shouldn’t be too harsh on Manjaro 15.12, by merely being based on these readings, which is further proved by the speed similarities shared by the recently released Kubuntu 15.10 and Manjaro 15.12, where Manjaro is only marginally slower to boot (3.7% roughly).
Also, as I’ve mentioned in my other reviews (in Fedora 22 for instance), new distributions that use systemd (such as Manjaro 15.12 and Kubuntu 15.10) will also boot-up slowly on rotating disks due to systemd developers abandoning a certain utility (systemd-readahead). Had it been present, Manjaro 15.12 could’ve been able to deliver much faster start-up times.
That said, I don’t know if it was enabled in Manjaro 0.8.9 KDE, and I don’t have its disc ISO anymore to test it out either. Sorry about that.
Memory Usage Upon Desktop Loading…
Here too there’s a big gap between Manjaro 15.12 and 0.8.9 (16.7% roughly) where the gap is further reduced compared to Kubuntu 15.10 (about 5%). As for the reasons, I’m clueless here.
Power Usage at Idle…
When measuring the power usage I use a special tool. It takes a lot of samples and takes care about the accuracy, so there’s no need to run it five times. Also, as always, before letting the computer idle, I turn ON the Wi-Fi adapter and keep it connected to the wireless router, turn OFF the Bluetooth adapter and set the screen to its maximum brightness (and disable anything that could interfere such as screen-dimming, turning off, screen lock, screensaver etc). Once I’m satisfied with the setup, I let the OS idle and start measuring the power usage. Below is the actual screenshot that includes the summary of the power usage measuring utility.
In my experience, anything around 12 Watts is a pretty well optimized power usage consumption for my Dell notebook computer. As it turned out, Manjaro 15.12 comes with TLP (automated power usage optimizer) so I’m not surprised at this result. That said, in the below graph, Kubuntu 15.10’s readings are also from after manually installing TLP (as Kubuntu doesn’t ship it by default), but Manjaro 15.12 still does it better.
The most impressive one is still Manjaro 0.8.9 (I don’t know if TLP was included in that release though), that 11 Watts at idle is an all time impressive one in my experience!.
CPU Usage at Idle…
At idle, only the system monitor process kept using 1% of the CPU time, most other applications leave it alone for long periods, though Xorg and plasmashell processes used 1% of the CPU (each) here and there only, as well. No complains here.
Hardware Recognition and ACPI…
In Manjaro 0.8.9, I had a small issue while trying to make the Bluetooth adapter work and I was able to fix it in the end. But I came no such issue in Manjaro 15.12 and the Bluetooth adapter worked flawlessly.
Except for the proprietary fingerprint reader, the rest of the hardware was recognized and configured correctly by the kernel and the KDE desktop was able to successfully restore the previous values of the screen brightness, Bluetooth adapter and the Wi-Fi adapter as well.
Other features such as Suspending also worked where both Suspending and waking up happened quite fast. KDE Plasma 5.5 is seemed very stable, and except for once where it threw me an error while trying to open the system monitor through the task-bar icon, so far zero crashes which is an impressive fact as far as my KDE experience is concerned.
I try to get a sense of an operating system’s responsiveness by making the hard disk drive busy (usually achieved by copying a somewhat large file within the same partition) and then trying to open a couple of programs and by trying to play a multimedia file.
If the OS opens up most of the programs, carries out the multimedia playback without major interruptions and doesn’t lose the cursor sensitivity that much when all this is happening, then I consider the operating system to be a responsive one.
So I did the same thing in Manjaro 15.12 and found out the responsiveness to be a good one.
Not exceptional, not pretty good, but good enough. Manjaro 15.12 was able to open most of the programs before the file was finished copying, the multimedia playback was interrupted about two times and each time this happened the whole system got stuck (cursor included), each instance lasted about 2-3 seconds, and then came back alive. And as mentioned early in the review, this is not the first time KDE has done this and when it does it, this is pretty much the order of how things take place. That said, I’m happy with the results. Older versions of KDE used to suck at this (such as the Manjaro 0.8.9 KDE release) but I’m seeing positive outcomes in newer KDE versions such as Manjaro 15.12 and Kubuntu 15.10.
P.S: Manjaro 15.12 also uses the ‘BFQ’ I/O scheduler (known for it ability to make an OS responsive under heavy disk load, but there’s little it can do if the problem lies elsewhere) by default. I also noticed that, the less I involve the start-menu for launching applications, the more responsive the OS becomes.
So perhaps after all it has something to do with KWin’s compositing method, because KWin compositing will be more involved in action the more we use the start-menu as all the transparency, or updating the screen for displaying the searched results (which is mostly when the system gets stuck!) etc are rendered by it. I can’t say anything for sure though, because I tried the same test with software rendering (XRender) opposed to the hardware accelerated rendering, and things actually were slightly improved, but it didn’t really make a big difference. So again, I just don’t have an exact answer, who knows how the KDE mind works 😛 .
Manjaro 15.12 was about 59.5% slower while shutting down compared to the older 0.8.9 KDE version and here too compared to the newer Kubuntu 15.10, the margin is a close one.
I love how Manjaro developers have presented the KDE Plasma 5.5 desktop. It’s a beautiful looking, responsive, power efficient, and a stable desktop. I’m also okay with it using a bit of memory as well. But you know, I can’t wait for 50+ seconds for an operating system to boot (again, part of that has to be blamed upon systemd developers) and 12.6 seconds of shutdown times is also a bit high for my taste, it just ain’t my cup of tea. I like lean & fast operating systems. But hey, that’s just me. And these days, one doesn’t get to see blisteringly fast booting KDE distributions either (in my short experience).
And, in my opinion, I’m still of the belief that GNOME developers are more insightful at seeing solutions from a technological point of view compared to the KDE developers, and I think this is the precise origin of this lag in performance of KDE, when the two desktop environments are compared.
Take MySQL as an example. Some KDE programs use it as their database handler. But the problem is, in its very essence, MySQL is designed to handle large number of data and thus is not optimized to have a small footprint. The now outdated search index service of KDE 4 known as Nepomuk required MySQL as a dependency. Thankfully, the new one in Plasma 5 called Baloo, according to its official page has a database engine of its own which has a small footprint. So it’s apparent that after a while KDE saw the mistake and corrected it.
But not every aspect of KDE is freed from this fundamental mistake as KDE software such as Akonadi still requires MySQL and this MySQL service by default uses about 95 MB (roughly), that’s a heavy burden as far as system optimization is concerned. And yes, Manjaro 15.12 comes with MySQL service enabled. The responsiveness related issue is another example, although with recent changes, things seem to be getting better which if it’s true (in the sense that they’ve at least identified it as a bug and trying to fix it), KDE developers deserve some praise for it.
GNOME on the other hand, has not made such costly technical mistakes. For instance, their (GNOME3’s) desktop search service is called Tracker and according to them it has a memory footprint of 4-6 MB!, that’s a clever design. But that’s not to say GNOME is better, because ‘better’ is a complex word. The term ‘better’ I think is measured (in this context) by the overall success of the whole ecosystem. In its very essence, it’s entirely a mere perspective, not an abstract concept, well, at least in my opinion anyway. Because after all, beauty is in the eye of the beholder.
Now I could go all day long, but I’ll stop here and would say the following thing about Manjaro 15.12. If you’re a KDE fan, take Manjaro 15.12 for a spin, it’s good. You know what, it’s more than good because they’ve worked hard and improved things as far as they could as outside developers, I can see that, and I appreciate it. If interested, get it from here. Thank you for reading.