- Site menu:
- Latest Discussions
- Latest Newswire
- Archives
- Submit Story
- Story HOWTO
- Syndicate
- E-mail Alerts
Taking a break from Ubuntu
|
|
|
Comments (31)
11 earlier comments in the full thread.
Regarding my Karmic issues, I'm going to try turning off kernel mode setting again.
@Steven: Take a look at an HP netbook in the store. They have the largest keyboards I have found on netbooks and touch typing really isn't a problem, at least for me. The resolution is only 1024x576, though.
I do like the HP netbooks. My shoulder is definitely tired of what I'm carrying around now, and for the moment I have two laptops with me (and a third on the floor, two Sun Sparcstations on the desk next to me ... and yes, I probably do need help).
OK, so here's a question: Why is Slackware faster?
Given Slackware, and after you installed the exact same software as with Ubuntu, is it still faster?
Or, in other words, is Slackware faster because of less bloat, different software set (KDE is faster than Gnome but not less bloated), more optimized software (patches / newer versions) or other compile options? Or something else?
Given Slackware, and after you installed the exact same software as with Ubuntu, is it still faster?
Or, in other words, is Slackware faster because of less bloat, different software set (KDE is faster than Gnome but not less bloated), more optimized software (patches / newer versions) or other compile options? Or something else?
Slackware still uses BSD init. Patrtick V. has done some to put intense disk I/O into the background, but hardware configuration and server daemon startups are still serialized, for the most part.
However, the stock Slack kernels support SMP by default, so that's a big help.
However, the stock Slack kernels support SMP by default, so that's a big help.
Hmmmm, maybe I should actually consider the question on the table. :)
I've tried both on the same hardware.
Slackware's boot is slower, even taking into account the Ubuntu trick of putting up a login screen before everything's up and running. However, after logging in, and starting the services needed by the desktop, I see no difference in desktop responsiveness.
One thing that may cause an observable slowdown is Ubuntu's use of AppArmor. Slackware by default uses no non-traditional security mechanisms; it's all user:group and ugo+-rwx. No SELinux, no grsecurity, no AppArmor. Libsafe was available for a while in -extra, but improvements in compilers and kernel design made it redundant.
(For reasons unrelated to these, I am no longer using Ubuntu.)
I've tried both on the same hardware.
Slackware's boot is slower, even taking into account the Ubuntu trick of putting up a login screen before everything's up and running. However, after logging in, and starting the services needed by the desktop, I see no difference in desktop responsiveness.
One thing that may cause an observable slowdown is Ubuntu's use of AppArmor. Slackware by default uses no non-traditional security mechanisms; it's all user:group and ugo+-rwx. No SELinux, no grsecurity, no AppArmor. Libsafe was available for a while in -extra, but improvements in compilers and kernel design made it redundant.
(For reasons unrelated to these, I am no longer using Ubuntu.)
> Given Slackware, and after you installed the exact same software as with Ubuntu, is it still faster?
If it's a reasonably fast machine with 512 MB or more of memory, you probably won't notice any significant difference (except for boot times, as noted above).
If it's a slower machine less than 512MB of memory, I'm almost positive you'll find Slackware significantly faster, even with the same lower overhead windows manager (comparing Slackware with XFCE to Xubuntu, for example).
Others have reported that a minimal Debian install is about the same speed as Slackware, FWIW.
If it's a reasonably fast machine with 512 MB or more of memory, you probably won't notice any significant difference (except for boot times, as noted above).
If it's a slower machine less than 512MB of memory, I'm almost positive you'll find Slackware significantly faster, even with the same lower overhead windows manager (comparing Slackware with XFCE to Xubuntu, for example).
Others have reported that a minimal Debian install is about the same speed as Slackware, FWIW.
@Hans: First, Ubuntu starts more daemons and services by default. You need to strip those out to get closer in performance. Second is the security overhead: not just SELinux and AppArmor but even simpler things like PAM are missing in Slackware. Now, if you do a minimal Ubuntu install from the alternate (text based) installer and don't install ANY of the desktop metapackages it is entirely possible to get Ubuntu to run as fast as Slackware. I've had Ubuntu running in as little as 32MB of RAM successfully. Doing so takes work but it can be done.
At 512MB RAM the difference between the two out of the box is significant. At 1GB RAM much less so.
At 512MB RAM the difference between the two out of the box is significant. At 1GB RAM much less so.
I'm one of those who found Slackware with Xfce to be fairly equal to Debian with Xfce. But that doesn't mean equal to Xubuntu, which is the Ubuntu desktop with Xfce instead of GNOME.
In my same tests I didn't seem much if any speed bump between Ubuntu (GNOME) and Xubuntu (Xfce).
If you run top in a terminal, you'll see many fewer services running in a stock Debian or Slackware Xfce system.
I agree with Caitlyn that these things are way more important when you have less memory. I also agree with her about Slackware-derived Zenwalk and Vector, both of which seem super-fast in comparison to just about anything else I've tried. ... I should probably burn a Zenwalk live CD and give that a test.
I run a 10-year-old laptop with Pentium II at 233 MHz with 144 MB of RAM. It runs better with Puppy Linux on live CD. But performance with Debian and Xfce is more than acceptable. I can't even boot Xubuntu on this machine let alone install and run it.
On Friday I was testing my Debian Lenny laptop with the standard GNOME desktop. It compared very favorably to my identical laptop running Ubuntu Karmic (the one that's giving me screensaver trouble at present; and no, turning off KMS didn't help. Instead I'm messing with GNOME settings tonight and will see how that went in the morning).
Then I forgot: The Ubuntu laptop has 1 GB of RAM; the Debian one has only 512 MB.
And while OpenBSD is no speed demon, especially on multicore processors, for single-core CPUs with low memory, it installs with a very small footprint and uses relatively few services. While I don't begin to understand exactly how Linux uses memory, I can tell you that OpenBSD is quite sparing of it. Running a 768 MB laptop, I don't think I needed swap once in the entire six months I ran OpenBSD 4.4. ... Then as now, it was Xorg that caused everything to go to (insert curse word).
In a related manner, I actually had a commenter on the blog tell me that Toshiba laptops are just hunks of (curse word) and that Linux needs to leave them behind in order to make the experience better for users of newer hardware (presumably including himself). Nice.
That said, I left all the laptops at the office, and it looks like I might need to work at home (but won't be able to). I didn't want to carry one of those heavy hunks of (you know what) home in the rain today (yes, it rains about once a year in L.A., and this is that once). ... wish I had a netbook now.
In my same tests I didn't seem much if any speed bump between Ubuntu (GNOME) and Xubuntu (Xfce).
If you run top in a terminal, you'll see many fewer services running in a stock Debian or Slackware Xfce system.
I agree with Caitlyn that these things are way more important when you have less memory. I also agree with her about Slackware-derived Zenwalk and Vector, both of which seem super-fast in comparison to just about anything else I've tried. ... I should probably burn a Zenwalk live CD and give that a test.
I run a 10-year-old laptop with Pentium II at 233 MHz with 144 MB of RAM. It runs better with Puppy Linux on live CD. But performance with Debian and Xfce is more than acceptable. I can't even boot Xubuntu on this machine let alone install and run it.
On Friday I was testing my Debian Lenny laptop with the standard GNOME desktop. It compared very favorably to my identical laptop running Ubuntu Karmic (the one that's giving me screensaver trouble at present; and no, turning off KMS didn't help. Instead I'm messing with GNOME settings tonight and will see how that went in the morning).
Then I forgot: The Ubuntu laptop has 1 GB of RAM; the Debian one has only 512 MB.
And while OpenBSD is no speed demon, especially on multicore processors, for single-core CPUs with low memory, it installs with a very small footprint and uses relatively few services. While I don't begin to understand exactly how Linux uses memory, I can tell you that OpenBSD is quite sparing of it. Running a 768 MB laptop, I don't think I needed swap once in the entire six months I ran OpenBSD 4.4. ... Then as now, it was Xorg that caused everything to go to (insert curse word).
In a related manner, I actually had a commenter on the blog tell me that Toshiba laptops are just hunks of (curse word) and that Linux needs to leave them behind in order to make the experience better for users of newer hardware (presumably including himself). Nice.
That said, I left all the laptops at the office, and it looks like I might need to work at home (but won't be able to). I didn't want to carry one of those heavy hunks of (you know what) home in the rain today (yes, it rains about once a year in L.A., and this is that once). ... wish I had a netbook now.
@Steven: I have two 11 year old Toshiba Libretto SS1010s (one has issues), which have Pentium 233MHz MMX processors and 64MB and 96MB RAM respectively. The best/fastest distro I've found for these machines is Vector Linux Light. It's absolutely amazing how usable those systems still are with the right distro.
Vector Linux Standard seems to have some performance tweaks that Zenwalk lacks, everything from VL's custom init scripts to some differences in compiler flags when the packages are built does seem to give Vector an edge over anything else when it comes to speed. When comparing Zenwalk to Vector I think you'll find it only becomes noticeable on very, very low spec machines like my old Librettos. (Yes, I know the correct Italian plural is libretti but I'm using a proper name in English.)
Also, PII 233? Are you sure? I thought all 233MHz machines would be classic Pentium or Pentium MMX, meaning i586 rather than i686. A lot of distros compile with march set at i686 which means they absolutely will not work in a classic Pentium. Slackware uses march at i486 but mtune at i686, meaning it will even work on AMD K5 and Intel 486 processors provided you have at least 32MB of RAM.
Oh, and yeah, netbooks are really nice. Check out the keyboard on the HPs. I know they aren't available preloaded with Linux any longer but the larger keys may be worthwhile enough for you to grit your teeth, hold your nose, and buy a Windows box.
Vector Linux Standard seems to have some performance tweaks that Zenwalk lacks, everything from VL's custom init scripts to some differences in compiler flags when the packages are built does seem to give Vector an edge over anything else when it comes to speed. When comparing Zenwalk to Vector I think you'll find it only becomes noticeable on very, very low spec machines like my old Librettos. (Yes, I know the correct Italian plural is libretti but I'm using a proper name in English.)
Also, PII 233? Are you sure? I thought all 233MHz machines would be classic Pentium or Pentium MMX, meaning i586 rather than i686. A lot of distros compile with march set at i686 which means they absolutely will not work in a classic Pentium. Slackware uses march at i486 but mtune at i686, meaning it will even work on AMD K5 and Intel 486 processors provided you have at least 32MB of RAM.
Oh, and yeah, netbooks are really nice. Check out the keyboard on the HPs. I know they aren't available preloaded with Linux any longer but the larger keys may be worthwhile enough for you to grit your teeth, hold your nose, and buy a Windows box.
Thanks Caitlyn, I'd qualify the things you are referring to why Ubuntu is slower as 'bloat', though that's personal.
There has been an interesting article somewhere on LXer as to which program is the best for showing the amount of memory Linux 'really' uses. The answer was, there is none such program - because memory in Linux is rather complex. Only if you patch existing programs with 'hacks' you can get close to the 'real answer'.
Which Xorg version? Same happened here with Linux and Xorg 1.6, but I've been able to repair it (don't ask me how, it was really... insert curseword again)
Quoting:While I don't begin to understand exactly how Linux uses memory, I can tell you that OpenBSD is quite sparing of it.
There has been an interesting article somewhere on LXer as to which program is the best for showing the amount of memory Linux 'really' uses. The answer was, there is none such program - because memory in Linux is rather complex. Only if you patch existing programs with 'hacks' you can get close to the 'real answer'.
Quoting:Then as now, it was Xorg that caused everything to go to
Which Xorg version? Same happened here with Linux and Xorg 1.6, but I've been able to repair it (don't ask me how, it was really... insert curseword again)
Quoting:Also, PII 233? Are you sure? I thought all 233MHz machines would be classic Pentium or Pentium MMX, meaning i586 rather than i686. A lot of distros compile with march set at i686 which means they absolutely will not work in a classic Pentium. Slackware uses march at i486 but mtune at i686, meaning it will even work on AMD K5 and Intel 486 processors provided you have at least 32MB of RAM.
Thanks for catching that. I think it's a Pentium Mobile MMX 233MHz. I believe this is the Compaq part: http://www.impactcomputers.com/291758-001.html
Quoting:Which Xorg version? Same happened here with Linux and Xorg 1.6, but I've been able to repair it (don't ask me how, it was really... insert curseword again)
It's the current Xorg in Ubuntu Karmic, according to the Web pages http://packages.ubuntu.com/karmic/xorg and http://packages.ubuntu.com/karmic/xserver-xorg:
Package: xorg (1:7.4+3ubuntu7)
I meant in OpenBSD 4.4 believe, because that's the one who went *poof* IIRC.
OpenBSD 4.4 was great for me. I ran it the full six months with no hardware issues.
It was 4.5 that totally hosed X.
OpenBSD 4.4 http://openbsd.org/44.html had: Xenocara (based on X.Org 7.3 + patches, freetype 2.3.5, fontconfig 2.4.2, Mesa 7.0.3, xterm 234 and more)
OpenBSD 4.5 http://openbsd.org/45.html had:
Xenocara (based on X.Org 7.4 + patches, freetype 2.3.7, fontconfig 2.4.2, Mesa 7.2, xterm 239 and more)
And hey, it's XOrg 7.4 in Ubuntu Karmic ... so that's where the trouble started for this particular Intel-video-equipped (82830 CGC, aka 830m) Toshiba laptop.
Now all supported versions of Ubuntu except for Hardy use Xorg 7.4: http://packages.ubuntu.com/search?keywords=xorg
Lucid has Xorg 7.5, but will that only make things worse?
In Debian, http://packages.debian.org/search?keywords=xserver-xorg
Lenny = Xorg 7.3 Squeeze + Sid = Xorg 7.4 Experimental = Xorg 7.5
It was 4.5 that totally hosed X.
OpenBSD 4.4 http://openbsd.org/44.html had: Xenocara (based on X.Org 7.3 + patches, freetype 2.3.5, fontconfig 2.4.2, Mesa 7.0.3, xterm 234 and more)
OpenBSD 4.5 http://openbsd.org/45.html had:
Xenocara (based on X.Org 7.4 + patches, freetype 2.3.7, fontconfig 2.4.2, Mesa 7.2, xterm 239 and more)
And hey, it's XOrg 7.4 in Ubuntu Karmic ... so that's where the trouble started for this particular Intel-video-equipped (82830 CGC, aka 830m) Toshiba laptop.
Now all supported versions of Ubuntu except for Hardy use Xorg 7.4: http://packages.ubuntu.com/search?keywords=xorg
Lucid has Xorg 7.5, but will that only make things worse?
In Debian, http://packages.debian.org/search?keywords=xserver-xorg
Lenny = Xorg 7.3 Squeeze + Sid = Xorg 7.4 Experimental = Xorg 7.5
Quoting:Lucid has Xorg 7.5, but will that only make things worse?
You'll never know until you try.
Speaking of trying... when will we get you to try a Slackware derivative like Zenwalk or Vector Linux of SalixOS?
I really need to give Zenwalk and/or Vector a try. It has been some time since I have.
In other news, I burned a live DVD of OpenBSD 4.6 from jggimi http://jggimi.homeip.net/livecd/downloads.html
I'm running the GNOME version, and thus far things are looking pretty good, and it's Xorg 7.4. I'll wait for the screensaver to pop in and think again.
Yeah, if I could live without Flash 9/10, I could return to OpenBSD (and if I could figure out how to reinstall w/o wiping /home; I have instructions, I just couldn't follow them).
(Later)
Yep, X runs pretty darn well with Xorg 7.4 in OpenBSD 4.6 with no xorg.conf. I have a little mouse-pointer glitch on the CRT monitor, but on the laptop's own LCD screen, it's perfect. And it comes back from screensaver fine ... and this is GNOME. (And after my OpenBSD 4.5 debacle, it's nice to see the OS working well on this hardware)
It means that I do have a snowball's chance with Debian Squeeze (and other things that are not Ubuntu).
In other news, I burned a live DVD of OpenBSD 4.6 from jggimi http://jggimi.homeip.net/livecd/downloads.html
I'm running the GNOME version, and thus far things are looking pretty good, and it's Xorg 7.4. I'll wait for the screensaver to pop in and think again.
Yeah, if I could live without Flash 9/10, I could return to OpenBSD (and if I could figure out how to reinstall w/o wiping /home; I have instructions, I just couldn't follow them).
(Later)
Yep, X runs pretty darn well with Xorg 7.4 in OpenBSD 4.6 with no xorg.conf. I have a little mouse-pointer glitch on the CRT monitor, but on the laptop's own LCD screen, it's perfect. And it comes back from screensaver fine ... and this is GNOME. (And after my OpenBSD 4.5 debacle, it's nice to see the OS working well on this hardware)
It means that I do have a snowball's chance with Debian Squeeze (and other things that are not Ubuntu).
OK, WRT to Xorg: Now I'm lost.
Xorg 7.4 came with Xorg-server 1.5, and it worked fine for me.
That's when I migrated to Xorg-server 1.6 and problems started. More people had those problems.
Now I read Xorg 7.5 comes with Xorg-server 1.7.
So there's no Xorg version that comes with xorg-server 1.6? So when you use Ubuntu or OpenBSD and upgrade from xorg 7.4 to 7.5 you go directly from server-1.5 to server 1.7 without using 1.6? That sounds odd to me. Nonetheless, I hope 1.7 doesn't bring the problems that 1.6 did.
Xorg 7.4 came with Xorg-server 1.5, and it worked fine for me.
That's when I migrated to Xorg-server 1.6 and problems started. More people had those problems.
Now I read Xorg 7.5 comes with Xorg-server 1.7.
So there's no Xorg version that comes with xorg-server 1.6? So when you use Ubuntu or OpenBSD and upgrade from xorg 7.4 to 7.5 you go directly from server-1.5 to server 1.7 without using 1.6? That sounds odd to me. Nonetheless, I hope 1.7 doesn't bring the problems that 1.6 did.
I've had nothing but headaches and consistent instability with 64-bit Karmic. After significant research I went back to try my old friend Mepis (8...Lenny-based). I am now in 64-bit heaven...with KDE3.5, to boot!!!!!!
Post a comment
A free account is needed to post. Write your comment here first: your text is kept in this browser while you sign up or log in, and is waiting for you when you come back.
Your message:Markdown: **bold**, *italic*, [text](https://link), - lists, > quotes, $x^2$ for math. Formatting help
Sign up and post (it is free) or Log in and post
Just in case anyone is wondering, due to personal bias, I really shouldn't say anything concerning the above comment. I'm sure pretty much everyone here knows my opinion anyway. :)
I've installed Ubuntu 9.10 on a machine at work for testing, and it's working well. I agree that for a production machine, I'd use ext3 rather than ext4 though.