Performance compared to other Linux distributions

Yet another thing I love about this distribution is that it appears to be more lightweight than other Linux distros I toyed with. While I’m still hyped about having switched to Manjaro, I figured I’d take this opportunity and ask: How does performance truly compare to other Linux distros? I’m honestly intrigued to see a proper benchmark for this OS! Not just in terms of used idle resources, but practical performance including FPS in games.

I guess I can start with mine; Having moved from openSUSE to Manjaro on my main system this week, I can do a partial comparison, especially as I’m using the same software and even ported my home directory with application settings. Of course this isn’t a proper benchmark, I’m citing what I remember from memory. Here’s a rough description of what I’m seeing so far:

  • General system performance is roughly the same, possibly a bit faster by a barely noticeable amount. It’s hard to sense the difference between boot and login speeds as well as how quickly applications open, but I can say for sure it’s definitely not any slower.
  • Memory usage seems to be a tad better on Manjaro. I’d estimate 300 MB to potentially 1 GB lower on average after many hours of uptime (5.0 GB versus +5.6 GB).
  • CPU usage when idle seems to be a fraction better on Manjaro. The plasma widget shows roughly 0.3% to 0.6% CPU while nothing important is working… this may be a decimal or two lower than what I remember observing on openSUSE.
  • Same amount of background processes kept open by the system: Roughly 400 processes (user with root) listed by KSysGuard, with about 440 in the first minute(s) after login.

On YT you have many compares, but performance depends on many things:

  • hardware
  • hardware modifications (overclocking, underclocking etc.)
  • kernel version
  • DE (KDE will have other performance than XFCE)
  • configuration (like programs in autostart, enabled addons)
  • processes in background
  • programs and games you run
  • software version (for example today you have 100fps in game, but week later after update you may have 110fps or 90fps)
  • many, many other things, even temperature in your room may be important :stuck_out_tongue:

so it’s hard to say which distro is faster.

Thanks. I also found this benchmark but it’s from 2014 so very old.

To add to the confusion, distros “roll” at different steps and different paces. It’s hard to do a real side-by-side when distros don’t line up in their versions of the kernel, libraries, and other packages.

The real question is, how well does XYZ distro work for you?

  • The update / upgrade cycles and iterations
  • The default settings
  • The default configuration for your preferred desktop environment
  • The extra unique tools supplied by the distro
  • The availability of software and packages
  • The ease of managing software and packages on your system
  • The community
  • …and most importantly, THE DEFAULT WALLPAPER!!! :love_you_gesture:

The last point is a deal-breaker for me. I’ve thrown many computers from my rooftop and screamed to the heavens because of the last point. :face_with_symbols_over_mouth: :triumph: :tired_face:

5 Likes

In my opinion, performances of distros might have very few differences if they were configured with nearly the same environment at the same time. I tried to run similar molecular dynamics simulations on several nodes with the same hardware configurations but different distributions several years ago, and all the simulations took almost the same time.

Hehe. I am assuming the main difference a distribution makes on performance is what services it installs and enables by default (has running in the background) as well as how they’re configured. Not counting drivers and components as those are usually the same, especially on rolling release distros where they’ll typically be the latest versions… I also choose to install root on ext4 so that sticks too.

One thing I learned is that Manjaro does have a reduced memory footprint, including in comparison to Ubuntu: Read other articles confirming my observation, definitely something good to hear :slight_smile:

About the benchmarks, something people seem to often forget when they do comparison is that to have relevant results, you need to have proper methodology, and also do many many runs to average them all, as many games can have run to run performance variations, it is not as easy as to run a one minute test on each distro and compare the results.

Benchmarks can be good but if they are not made ‘properly’ they are not really relevant because of the random things happening on the computer in the background, the run to run variance, the subtle margin of error, and so on… Here is an example of rigorous benchmark methodology, for an example (it is about benchmarking CPUs but the principle can be applied generally).

A good example of bad methodology is that recent Reddit thread where guys “fixed” the game Cyberpunk 2077 by modifying a config parameter in a file. It turned out that there proofs and benchmarks where completely messed up and this parameter wasn’t even in use by the game so they changed nothing at all, they just experienced run to run variance, which Cyberpunk game had A LOT, it was very inconsistent.

As said in previous post I’m not sure benchmarking distros is really relevant if not done in proper condition that would still need to be thought of and defined.

Benchmarks provided online never been relevant for me and my installations, not for my work or work i do. The empirical experience and workflow made me stick to Manjaro for almost 3 years, with just on breakage in all this time, and all that because of something i did wrongly. For me that is more important than any other numbers displayed in a sheet :slight_smile:

After having read all the answers, I have come to the conclusion that there’s no correct answer and, ultimately, it’ll depend on you, what you, as an individual, prefer.

I have a ‘friend’ who uses Windoze :face_vomiting: because that’s what works best for him. I have another friend that used to use Mac, for much the same reason and now uses Kubuntu, because that’s what works for him.

1 Like

you can install and use phoronix benchmarks tests
see in AUR phoronix-test-suite

it works on linux , BSD , macos X , windows

last benchmark by phoronix
https://www.phoronix.com/scan.php?page=article&item=endeavour-salient-ryzen&num=1

What is your opinion of Clear Linux, benchmark-wise ?

I know YouTube links are frowned upon but this video sums it up pretty well. The differences between the distros is negligible and depending on what’s important to you what differences there are

1 Like

Just installed manjaro over the weekend and noticed it using more RAM than endeavour OS. Without starting any application I get>1Gb, even with i3wm version.

If that could also be classified in the “performance” category.

Don’t get me wrong, manjaro feels snappy and I like it. Will have to check what uses RAM in the background. Not really worried since I got 32gb better use them.

Personally the only time I would be interested in a benchmark is if it almost exactly matched my use case - perhaps a game that won’t run too well on my ancient i3 graphics needs every bit it can get. I guess a few more years might show my 8GB Ram needs bumping up to 16GB - but by then I’m expecting a complete Mobo/processor swapout might be a better option (that’ll be the second since 2007).

I’m always impressed with Chrome vs Firefox because however ‘slow’ they show it to be, when I use it - it’s just fine.

Stability is probably fine, though I heavily corrupt my native KDE install - so I do hit issues sometimes and find they’re easier to fix now that I’m more accustomed to the Arch-itechture (like not being able to do dpkg-reconfigure etc) and so now it’s comfortable.

Especially good is the option to install Plex - the server and media player - as binary packages and have them work the way they’re intended (unlike Mint - that needed the appimage and function was different viz sound output).

The short answer - it doesn’t.

The longer answer is - depending on your setup - your hardware and the services loaded.

You can create a true minimalist thing which consumes very little memory and outruns everything. But as soon as you add dock(s), task bar(s), printing services, file services, indexers, game loaders, convenience tools of various kinds with background services - it adds up - so there is no one size fits all - the force of Manjaro is inherited from Arch - the ability to rip out components to create THE system which fits your exact workflow - but that treat only comes with experience …

Another thing to note is - while some think - the less ram the system uses the better - this is only true for low ram systems - in fact if you have a high-end system you will benefit if you let the system get more memory to juggle with - memory is way-way faster than even the fastest NVME-SSD.

Why is Manjaro using all my RAM?

–Essentially, unused RAM is wasted RAM.

Many new users notice how the Linux kernel handles memory differently than they are used to. Since accessing data from RAM is much faster than from a storage drive, the kernel caches recently accessed data in memory. The cached data is only cleared when the system begins to run out of available memory and new data needs to be loaded.(from Arch wiki)

2 Likes

That’s an useful video which says a lot. All in all Manjaro seems to be one of the more lightweight distros which is awesome! It’s not Clear Linux obviously but does put itself on the more optimal side.

Interesting to see this guy playing - synthetic benchmarks are a bit of a waste of life IMO.
Nice boot-reboot test, I was getting rather over-excited at how Manjaro held up it’s end, beating Endeavour and Garuda in many tests, including the file transfer - Garuda slow, but Endeavour was way behind, but then in came a very disappointing result for heavy video work; why would Manjaro end up so far behind?

I wondered about that as well. Manjaro seems more lightweight, using somewhat fewer processes and lower memory, so I couldn’t blame it on system bloat. It could be due to how some applications are compiled and configured, potentially newer / older libraries slowing down the process.

Complicated issue - BLOAT generally means how many things are installed and how much is running from a fresh boot…
Garuda has the feature of trying to use all the RAM, so it always shows HIGH use, however this usually has the effect of causing things to launch faster… RAM is only really an issue when it runs out - so seeing 450, 0r 600, or 1400 MB in use after a boot isn’t really interesting at all.

There could be some issue with the filesystem - but that’s huge!

1 Like