Randy Appleton wrote:
Hello.
I'm a computer science professor at Northern Michigan University. I found your email address at http://en.opensuse.org/OpenSUSE_Testing_Core_Team. I'm considering offering a student research project and thought you might be able to provide some input. Or, could you suggest someone better to contact?
Has anyone timed various operations in Suse vs Ubuntu vs Fedora? I'm curious if it's quicker to log in, find a file, open a document, etc. in Suse, Ubuntu or Fedora?
Does anyone already have data on these questions? Or, if we do the benchmarking myself, is anyone interested in the results? Finally, is there some particular operation you'd like to see benchmarked?
I'm copying this reply to the opensuse-testing mailing list.
I have heard anecdotal reports that distro XX is faster than YY while doing ZZ, but I am not aware of any properly designed studies. I am confident that the other members of the Testing Team will post such studies if they exist.
Certainly, the distro packagers would be interested in the results. My suspicion is that the most important factor will be the kernel itself, and that there will be only small differences when all are using the same kernel version; however, that is certainly a testable hypothesis.
Larry
2009/9/29 Larry Finger Larry.Finger@lwfinger.net:
Randy Appleton wrote:
Hello.
I'm a computer science professor at Northern Michigan University. I found your email address at http://en.opensuse.org/OpenSUSE_Testing_Core_Team. I'm considering offering a student research project and thought you might be able to provide some input. Or, could you suggest someone better to contact?
Has anyone timed various operations in Suse vs Ubuntu vs Fedora? I'm curious if it's quicker to log in, find a file, open a document, etc. in Suse, Ubuntu or Fedora?
Does anyone already have data on these questions? Or, if we do the benchmarking myself, is anyone interested in the results? Finally, is there some particular operation you'd like to see benchmarked?
I'm copying this reply to the opensuse-testing mailing list.
I have heard anecdotal reports that distro XX is faster than YY while doing ZZ, but I am not aware of any properly designed studies. I am confident that the other members of the Testing Team will post such studies if they exist.
The Phoronix test suite ( http://www.phoronix-test-suite.com ) allows to benchmark many things on UNIX systems (disk, compilation, SQL, 3D, ...) for free. They frequently post the results of some benchmarks in their articles site ( http://www.phoronix.com ). They say they are going to post a big test with many OSes soon. These tests does *not* include boot times, and no specific GUI application is tested (like, say, firefox); GUI tests measure things like 3D performance and GTK+ performance.
You can use bootchart to test boot times, so you can measure the various components of boot time. I think testing specific GUI applications like, as you say, «open a document» would require support by the application itself (if you don't manually measure the time with a chronometer), so this could be difficult to do.
Certainly, the distro packagers would be interested in the results. My suspicion is that the most important factor will be the kernel itself, and that there will be only small differences when all are using the same kernel version; however, that is certainly a testable hypothesis.
Larry
-- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
2009/9/29 Larry Finger Larry.Finger@lwfinger.net:
Randy Appleton wrote:
Hello.
I'm a computer science professor at Northern Michigan University. I found your email address at http://en.opensuse.org/OpenSUSE_Testing_Core_Team. I'm considering offering a student research project and thought you might be able to provide some input. Or, could you suggest someone better to contact?
Has anyone timed various operations in Suse vs Ubuntu vs Fedora? I'm curious if it's quicker to log in, find a file, open a document, etc. in Suse, Ubuntu or Fedora?
In my experience, the only real variances in speed are for the startup, as distributions use different methods and start different processes. With regard to things such as launching basic applications, I've never seen a serious statement about this backed up.
The general claims are "I just installed distribution X" and things start up so much more quickly!!" Well, yes of course when you install a new distribution it will be faster. You haven't done any updates so the disk is not as fragmented, you haven't personalised your settings so less applications start up, you have less fonts installed[0], etc. So how is it on a clean install of distro X and distro Y?
There will be variances if you are using different desktop environments, and those kind of results are interesting to see -- particular with regards to memory. Lubos Lunak (an openSUSE/KDE developer) did some testing into this a _long_ time ago[1].
However, if you are launching the same application in the same desktop environment on a different distribution (making sure it is the same app -- which wouldn't work for i.e. searching), I very much doubt there will be a significant or material difference. But there are some possibilities, and it might be interesting to just verify or disprove that, but only if you would investigate why ;-).
Anyhow, with regard to distribution boot times, I ran a test some time ago on the differences between the boot times from some of the popular distributions[2]. An update on something like that probably would not take too long and it would be interesting.
There is some work being done on the boot time in openSUSE at least, particularly within the Moblin project. I also think they're addressing some of the inefficiencies / bugs with bootchart (I recall Michael Meeks mentioning some in his talk at the openSUSE conference...slides should be available).
[0] http://www.kdedevelopers.org/node/1654 [1] http://ktown.kde.org/~seli/memory/desktop_benchmark.html [2] http://francis.giannaros.org/blog/2007/08/23/comparing-distribution-boot-tim...
-
Francis Giannaros -
Larry Finger -
Marco Poletti