GNU/Linux Versions Performance Comparison (older hardware)

Questions and thoughts about present and future editions
Forum rules
Before you post please read this
exploder
Level 15
Level 15
Posts: 5500
Joined: Tue Feb 13, 2007 10:50 am
Location: HartfordCity, Indiana USA
Contact:

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby exploder » Sun May 31, 2015 3:24 pm

Take your time! I just find your testing interesting. There are very few that do performance testing and you seem to have a real knack for it. I see a couple of reviewers that do this but not quite the same way as you. You kind of dispelled the myth that Cinnamon was a more demanding desktop environment. :D

Dave B
Level 4
Level 4
Posts: 339
Joined: Thu Jan 08, 2015 10:49 pm
Location: UK

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby Dave B » Thu Jun 04, 2015 7:44 am

Thanks exploder, your kind comments are appreciated. :)

Just finished updating the second set of test results (if helpful, quick link provided below), thought may as well add MATE and KDE before upgrading to an SSD, and memory back to 2GB.

Linux Mint Performance Comparison - Part 2 (updated)

Please will everyone let me know if they prefer the old style text based (I've saved a backup just in case), or the new style with performance comparison charts, also (due to dyslexia) if you spot any errors.

Thank you kindly,
David

exploder
Level 15
Level 15
Posts: 5500
Joined: Tue Feb 13, 2007 10:50 am
Location: HartfordCity, Indiana USA
Contact:

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby exploder » Thu Jun 04, 2015 9:12 am

All I can say is, I'm impressed! The charts are impressive! You really know how to present data in a very clear and easy to grasp way. I do not know what you do for a living but if I was looking for someone to analyze data you would be the guy I would hire!

Your results are interesting too, there is not as big a difference in these desktop environments as many would believe. Also, LMDE 2 is not really lighter or faster than the main edition. There are certain variables that can change the results, different kernels, turning off un-needed services, etc but no one can take those kinds of things into account when doing these types of comparisons. In my mind I have felt that Cinnamon was not as heavy a desktop environment as many believed it to be and your results clearly demonstrate this.

You do nice work my friend and I appreciate your time and effort and sharing your results with the community. :D

Edit: I hope Clem comes across this thread. I think he would find this data very interesting. :D

KBD47
Level 6
Level 6
Posts: 1252
Joined: Fri Jul 29, 2011 12:03 am

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby KBD47 » Thu Jun 04, 2015 10:18 am

Yes, very interesting.
I suspect when systemd hits Mint Debian the bootup time will be quicker. I can't help but wonder if something else is going on with LMDE2. I can't imagine why it would lag Ubuntu/Main Mint. Yet I noticed recently on a netbook that LMDE2 was running at 55 C temp while Xubuntu 14.04 was at 49 C temp. Very curious.

mr_raider
Level 6
Level 6
Posts: 1271
Joined: Sun Jun 20, 2010 9:50 am
Location: Montreal, QC

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby mr_raider » Thu Jun 04, 2015 11:53 am

As I suspected, KDE is not as FAT as people say. I've always used KDE, and by tweaking the right settings, I can get it to run on a vast array of hardware. Once you become familiar with it, you can really customize KDE down to the bone.
Image

Dave B
Level 4
Level 4
Posts: 339
Joined: Thu Jan 08, 2015 10:49 pm
Location: UK

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby Dave B » Thu Jun 04, 2015 11:54 am

@ exploder: Thank you kindly, all in a days work, just pushed a few buttons and out popped all the charts (or at least that's some people's impression of how computers work). :D As for work, presently looking for employment, thanks, when do I start?. :)

Agree, results are very different than I expected. Also need to test LMDE 2 MATE but might wait until 17.2 equivalents are available then re-test with all versions. In the meantime, for comparison, I will still test with SSD and 2GB memory. Thank you again.

Yes, it will be interesting to hear if Clem has any comments (except for... What are you doing posting so many images on my forum!). :D

@ KBD47: Interesting, (apart from 686 kernel upgrade) I tried to test versions vanilla as possible, did consider testing after applying updates, but since updates can change daily, it would be harder to apply a level playing field. So saying, I do appreciate LMDE 2 already contains newer software compared with 17.1 versions. Let me know if you find what's causing the higher temps, there are so many contributing factors involved.

@ mr_raider: Very true, I've read so many comments where people believe KDE is the most demanding that I approached the tests expecting it to perform poorly on this level of hardware, and results are of course without major modifications (minor changes mentioned in with the test results, disable screen saver and so on...). I suspect KDE is constantly being optimised.

Dave B
Level 4
Level 4
Posts: 339
Joined: Thu Jan 08, 2015 10:49 pm
Location: UK

Linux Mint Performance Comparison - Part 2 (updated)

Postby Dave B » Fri Jun 05, 2015 7:05 am

Actually have one further question, now other Mint versions have been compared, what are your thoughts on renaming the thread, such as...
Linux Mint Versions Performance Comparison (older hardware)?

Thank you all for your time and replies,
David

User avatar
phd21
Level 14
Level 14
Posts: 5020
Joined: Thu Jan 09, 2014 9:42 pm
Location: Florida

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby phd21 » Tue Jun 09, 2015 6:41 pm

Hi David Black,

Great work David. I thought this was a good and useful thing to do for everyone. Your graphical presentations are very well done too. :)
Phd21: Mint KDE 17.3 & 18.2, 64-bit Awesome OS, Ancient Dell OptiPlex 780 Core2Duo E8400 3GHz,3gb Ram,160gb hdd, Video: Intel 4 Graphics, DVD Lightscribe. Why I use KDE?:https://opensource.com/life/15/4/9-reasons-to-use-kde

User avatar
MtnDewManiac
Level 6
Level 6
Posts: 1249
Joined: Fri Feb 22, 2013 5:18 pm
Location: United States

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby MtnDewManiac » Tue Jun 09, 2015 10:01 pm

First, thanks for your efforts.

David Black wrote:@ mr_raider - Idle CPU usage is higher in Cinnamon even with fluctuations, but as seen from the performance tests, the impact upon overall general performance is not so obvious.


I could be wrong (and often am :roll: ), but I suspect that if you take your netbook somewhere but forget its power cord, the increased CPU usage might, soon enough, ultimately show you decreased performance (when the battery dies earlier :lol: ) .

I find myself wondering... You used as one of your tests, time to load the Firefox web browser. I would generally expect times to be similar because I wouldn't expect drive-to-memory throughput to be greatly affected by DE choice. To be honest with you, unless I find myself in the (theoretical) situation of needing to load that "Heart defibrillator app" quick enough that I can click on its "Attempt to Save My Life" button before I hit the floor, I don't consider application load times to be greatly relevant. Perhaps if something takes ten seconds to load in one DE and long enough to go put together and eat a sandwich in the other, I might find myself feeling differently (or I might just go get that sandwich :wink: ). Still, I realize that others might(???) consider a difference of seconds to be of high importance (<SCRATCHES HEAD> which makes the fact that so many people spend time using Facebook, MySpace, Twit, et cetera even more confusing to me...), since such a difference over an entire lifetime might actually amount to, IDK, enough time to eat a second sandwich (lol). But, either way, I wouldn't consider this a "performance" thing. How about... Hmm... Okay, how about opening your bookmarks, highlighting a few (or 50, IDK) entries and choosing to open them all at once in tabs? See how long it takes until every page has completed its load. If you aren't using some kind of basic common sense script blocker, you might find that your CPU usage is causing your netbook to do double-duty as a hand warmer... Of course, with all the scripts out there - which are routinely found on websites (present, err, company excluded, thankfully!) - that track users and what they do from one website to another (regardless of whether or not there is a connection in subject matter, website ownership, et cetera), everyone (IMHO) should be running a script-blocker. But this might still be a "torture-test" of sorts. At one point I had a slow (and old) computer. Being in the habit of having lots (IDK, a few hundred) of tabs open, I noticed that my computer would get sluggish. Drive caching would have had something to do with this of course....

Just rambling,
MDM
Mint 18 Xfce 4.12.

If guns kill people, then pencils misspell words, cars make people drive drunk, and spoons made Rosie O'Donnell fat.

KBD47
Level 6
Level 6
Posts: 1252
Joined: Fri Jul 29, 2011 12:03 am

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby KBD47 » Wed Jun 10, 2015 9:54 am

I have 3 netbooks and I have noticed even small increases in ram and cpu use increases the temperature, and in my mind when I think of something running 'heavier' on those I'm usually thinking hotter temp. Those little beasts don't handle temp rises all that well, so even incremental increases in ram and cpu use can be a big deal compared to a desktop or larger laptop.

exploder
Level 15
Level 15
Posts: 5500
Joined: Tue Feb 13, 2007 10:50 am
Location: HartfordCity, Indiana USA
Contact:

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby exploder » Wed Jun 10, 2015 11:24 am

I have a very low end HP laptop with an AMD E1 processor, yeah the graphics are built into the processor. The laptop runs the same temperature regardless of the DE I am using. Battery life is also the same regardless of the DE used. The kernel and power management are are what influences these and are going to be the same no matter what DE you use. I get the same 3 hrs 45 min of battery life and 53c temps no matter what DE I use on this hardware.

The laptop only runs @ 1.2 GHz, you can't get much lower end than that!

KBD47
Level 6
Level 6
Posts: 1252
Joined: Fri Jul 29, 2011 12:03 am

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby KBD47 » Wed Jun 10, 2015 5:44 pm

I definitely notice temp differences on netbooks with different distros and desktops. There is not much space and everything is close together. KDE almost always runs warmer. Cinnamon runs a bit warmer than Xfce.

Dave B
Level 4
Level 4
Posts: 339
Joined: Thu Jan 08, 2015 10:49 pm
Location: UK

Linux Mint Performance Comparison - Part 2 (updated)

Postby Dave B » Sat Jun 13, 2015 8:24 pm

@ phd21: Thank you for your kind feedback. It's nice to be able to help (or at least try and help). First time produced a more advanced spreadsheet with LibreOffice Calc, in fact, now I think about it, last time was several years ago and something rather simplistic using OpenOffice Calc, before LibreOffice came into being.

@ MtnDewManic: You're welcome.
MtnDewManiac wrote:I could be wrong (and often am :roll: ), but...
You are wrong, just kidding! :)
If I took my netbook somewhere and forgot its power cord, it simply wouldn't work, the battery stopped working around four years ago. Due to bereavement, the netbook remained unused for a few years. Probably not worth buying a new battery. Old battery has been revived three times, not sure it’s worth doing again.

If there had been no variation in overall application launch times, I would agree with you, but even a few extra seconds repeatedly observed is an indication there are discernible performance differences.
While it would be interesting to compare how long fifty bookmarks take to open, I am not so sure it would help reflect a real world test, not everyone has many tabs open at once. Plus, a fifty bookmark test introduces several uncontrollable variables, server/s performance due to constantly fluctuating network traffic, (some) ISPs using traffic shaping at certain times of day, radio interference and signal strength if using a wireless connection. Probably best to keep it simple. Do agree with you about script blocking, its interesting to observe the kind of connections being made and data harvesting, but that's a topic probably best saved for another thread.

@ KDE47 and exploder: If netbooks CPU throttle sooner due to load with their less adequate cooling, it will be interesting to see how Linux Mint 17.2 and Cinnamon 2.6 closes the gap. Even while testing Cinnamon 2.6.7 in a virtual machine, there is a definite idle CPU load reduction.

Updated first post thread title, so others will realise further versions of Linux Mint have been compared.

Dave B
Level 4
Level 4
Posts: 339
Joined: Thu Jan 08, 2015 10:49 pm
Location: UK

Linux Mint Performance Comparison - Part 2 (futher updates)

Postby Dave B » Sat Jun 20, 2015 9:40 pm

Hi all,

New updates, as follows...

Linux Mint Performance Comparison - Part 2 (further updates). Now includes 17.2 Cinnamon RC and 17.2 MATE RC.

Idle CPU resource tests, added MATE, KDE, 17.2 Cinnamon RC and 17.2 MATE RC.

Thank you,
David

User avatar
BigEasy
Level 5
Level 5
Posts: 994
Joined: Mon Nov 24, 2014 9:17 am
Location: Chrząszczyżewoszyce, powiat Łękołody

Re: Linux Mint Versions Performance Comparison (older hardwa

Postby BigEasy » Sun Jun 21, 2015 12:28 am

Perfomance comparation of different desctops should be comparation of cold boot times, RAM required for DEs itself, response time for open default DEs menus and utilites. That's all. Any other comparation has no technical sence.
I like your Memory: 2GB in sence of older hardware :) Try 1Gb or even 512Kb then I promise you that range of tests results will from some programs opens and works to not works at all.
Windows assumes I'm stupid but Linux demands proof of it

Dave B
Level 4
Level 4
Posts: 339
Joined: Thu Jan 08, 2015 10:49 pm
Location: UK

Re: Linux Mint Versions Performance Comparison (older hardwa

Postby Dave B » Sun Jun 21, 2015 3:14 am

Hi BigEasy,

Thanks for stopping by and providing feedback. I could integrate the RAM required for each DE from Idle CPU resource test images. :)
Still feel Blender tests are worthwhile, as they clearly show how each DE performs while under constant full CPU load, and performance differences between the various Mint versions are not as great many believe, or once was.

Only tests in the very first post used 2GB of memory. For all subsequent tests (Linux Mint Performance Comparison - Part 2 (further updates)) the netbook was downgraded to 1GB of memory.

What are others thoughts, I value everyone's constructive opinions, should the Firefox, LibreOffice Writer and GIMP launch times be removed?

User avatar
RacerBG
Level 4
Level 4
Posts: 397
Joined: Sun Mar 03, 2013 4:31 am
Location: Bulgaria

Re: Xfce vs Cinnamon, performance comparison (on older hardw

Postby RacerBG » Sun Jun 21, 2015 10:20 am

exploder wrote:All I can say is, I'm impressed! The charts are impressive! You really know how to present data in a very clear and easy to grasp way. I do not know what you do for a living but if I was looking for someone to analyze data you would be the guy I would hire!

Your results are interesting too, there is not as big a difference in these desktop environments as many would believe. Also, LMDE 2 is not really lighter or faster than the main edition. There are certain variables that can change the results, different kernels, turning off un-needed services, etc but no one can take those kinds of things into account when doing these types of comparisons. In my mind I have felt that Cinnamon was not as heavy a desktop environment as many believed it to be and your results clearly demonstrate this.

You do nice work my friend and I appreciate your time and effort and sharing your results with the community. :D

Edit: I hope Clem comes across this thread. I think he would find this data very interesting. :D


Good to see you here, Exploder. :)

For me usually the speed matters a lot and I'm not surprised with the results because Cinnamon is already very stable and fast but XFCE is still at version 4.10 on LM. I suggest that XFCE 4.12 will be a little faster but not blazing fast. Also when XFCE 4.14 goes to GTK 3 we could see perfomance drop of about 1-2 seconds I believe.

As you know nowadays I'm playing a lot with Debian and I have a few things to mention because of that:
1. Mint - XFCE 4.10 (default config, no extra programs running, cold start) takes around 150 MB RAM. The CPU usage goes about 1-4 %.
2. Debian - XFCE 4.10 default config and cold start, without panels + Docky takes around 115-120 MB RAM. The CPU usage goes about 1-4 %. This is my current DE on that machine (Docky is not very memory intensive, only the CPU usage can go up for a few seconds).
3. Debian - KDE 4.14.2 (default config, no extra programs running, cold start) takes around 180-250 MB RAM. The CPU usage goes about 7-12 %.

The PC:
3 GB RAM
Pentium 4 2 GHz single core CPU
ATI Radeon X1200
80 GB HDD
Proud of Linux Mint!

User avatar
BigEasy
Level 5
Level 5
Posts: 994
Joined: Mon Nov 24, 2014 9:17 am
Location: Chrząszczyżewoszyce, powiat Łękołody

Re: Linux Mint Versions Performance Comparison (older hardwa

Postby BigEasy » Mon Jun 22, 2015 1:57 am

David Black wrote:Hi BigEasy,
Still feel Blender tests are worthwhile, as they clearly show how each DE performs while under constant full CPU load, and performance differences between the various Mint versions are not as great many believe, or once was.

DEs nothing do with perfomances of Linux applications. DE just providing GUI for system.

What are others thoughts, I value everyone's constructive opinions, should the Firefox, LibreOffice Writer and GIMP launch times be removed?

Yes, DEs nothing do with launch times of applications.
Windows assumes I'm stupid but Linux demands proof of it

Dave B
Level 4
Level 4
Posts: 339
Joined: Thu Jan 08, 2015 10:49 pm
Location: UK

GNU/Linux Versions Performance Comparison (older hardware)

Postby Dave B » Thu Jul 16, 2015 3:48 pm

Based upon replies, completed a new set of performance comparisons, removed all application launch tests except for GIMP.
17.2 KDE RC and 17.2 Xfce RC tests included

GNU/Linux Versions Performance Comparison - Part 3 (tests re-run, couple of other Linux versions added)
All tests performed using 1GB of memory.


Test system:
NC10 Netbook (from 2008)
CPU: Intel Atom single core Hyperthreaded 1.2GHz
GPU: Intel 945GSE
Memory: 1GB
HDD: 160GB

Extra Information:
HDD set-up as GPT, 1MB boot, 22GB ext4 and 2GB Swap.
Tests performed after install (except for LMDE 2 Cinnamon / MATE 686 kernel upgrade), 32-bit versions used.

Testing Methodology:
After installation, once the desktop has loaded, GIMP is started, then closed, allowing the application to set-up and provide more realistic real world test results. Blender archive added to desktop. OS is shut-down and performance tests commence. All tests performed two or more times.

Test Results:

Cold boot
Image

Launch GIMP
Image

Extract Blender
Image

Due to potential for human error (stopping a digital stopwatch), Blender tests have been performed comparing CPU capabilities while under full load and effect of resources within each desktop environment.

Blender Testing Methodology:
Mike Pan's (updated BMW scene) BMW27.blend

Blender settings default except for:
  • Greg Zaal's 'Auto Tile Size' add-on (default 32x32 tile size setting used)
  • 'Bottom to Top' tile setting
  • To save time, sample 'Render:' setting reduced to 10 (100 AA samples)
GNU/Linux:
  • Screen saver disabled
  • Power Management settings changed to prevent screen switching off
  • Computer rebooted just before each render

Image

Render Results
Image

Conclusion
Image

Surprisingly, MATE is the fastest overall (by a tiny, tiny margin). Also of surprise, how close all Mint results are on this level of hardware!
While interesting testing other Linux versions, Mint feels far more polished and user friendly. :)

Tip:
CTRL+click, or middle mouse click thumbnail images, full size image opens in new tab (forum maximum 320x240 for images).

David

Edit: Fix typos
Last edited by Dave B on Sun Jul 19, 2015 2:55 pm, edited 1 time in total.

User avatar
xenopeek
Level 24
Level 24
Posts: 21458
Joined: Wed Jul 06, 2011 3:58 am
Location: The Netherlands

Re: GNU/Linux Versions Performance Comparison (older hardwar

Postby xenopeek » Thu Jul 16, 2015 4:32 pm

And there goes the theory often heard here that LMDE is faster, or that Xfce is faster :wink: Thanks for providing some facts, even if this gives just a glimpse of performance on one system and one set of tests.

I don't know how you're doing these tests, but if you weren't aware of them yet https://openbenchmarking.org/ and http://www.phoronix-test-suite.com/ would be interesting for you I think.

Interesting statistics and giving some food for thought (if not ammo to revive the desktop wars -- I hope not :wink:).
Image


Return to “About Mint Editions”