Slashdot Mirror


Vector Linux 4 Reviewed

SilentBob4 writes "On October 7th, the developers at Vector Linux released the latest version of their lightweight Linux distro, version 4. Vector has always been built upon the Slackware Linux framework and this time around it is based on Slack 9.0. The interesting thing here is that there was quite a delay between releases from the Vector camp, so as they were readying version 4, Pat Volkerding was releasing version 9.1 of his Slackware distro. This past Friday, the first review of Vector Linux was released (Distrowatch.com posted a link to it today). It was a pretty good review for the most part, but the interesting thing about it was that they actually benchmarked it against Slackware 9.1 and posted the results. I'll spoil the ending right now and tell you that Vector Linux won, but you should check out the findings. There are some pretty interesting numbers obtained from the two distros. The reviewer has published three PDF documents detailing the results. Everything was tested from the kernel to filesystem performance. It is interesting to say the least. Even if you don't have to time to read the whole article (it's two pages long), do check out the benchmark results. "

30 of 175 comments (clear)

  1. Okay, "stupid question" time by Illbay · · Score: 5, Interesting
    Being software-illiterate, will someone please explain to me how this really matters?

    A Linux "distribution" is (or so I have always thought) the kernel, plus system support files, plus all the tools (typical "GNU") that go into making a working OS. So how is it that you can have significant differences between distros in "performance"? And how does that matter if you build the system on your own hardware (a la Gentoo)?

    Is this just another example of irrelevant Geek pissing contests, or is there some actual significance here?

    --
    Any technology distinguishable from magic is insufficiently advanced.
    1. Re:Okay, "stupid question" time by daBass · · Score: 3, Informative

      Performace won't matter much, except for self-compile distros, where you can get about 10% better performance. But it is a bit silly, on any distro the kernel is easy to compile and so are any applications you really use (database, webserver). But I don't see the point in making vi run 10% faster...

    2. Re:Okay, "stupid question" time by Anonymous Coward · · Score: 4, Interesting

      The significance is this: I just installed Vector 4 on a 233MHz machine with 32 meg of RAM that I bought for 1 (One pound sterling) from my employer who was chucking it out. I will use it for word processing and other basic tasks and it will be just fine. It has the speed of Windows 95 (which was previously on board) but it's a modern operating system with all the capabilities that implies. I am well pleased.

    3. Re:Okay, "stupid question" time by Valar · · Score: 2, Informative

      Well, the issue is that different distrobutions include different sets of tools. Some, for example, include many more daemons (think apache, mysql, etc) and turn them on by default. Also, there will be a noticible difference between Gentoo built for Athlon versus Gentoo built for i386, for example. Similarly, the default build instructions are different for different distros. You might download a Redhat ISO targetted for the i586 (pentium) or perhaps Mandrake targetted at i686. So in those cases, often, it would matter.

    4. Re:Okay, "stupid question" time by BrokenHalo · · Score: 5, Informative
      So how is it that you can have significant differences between distros in "performance"?

      Quite easily if you use them as they come out of the box.

      I'll probably get flamed for saying this, but in the past when I played around with RedHat and Mandrake before going back to Slackware, I noticed a general sluggishness about those distros.

      Most of that got fixed when I rebuilt the kernel my own way, but other aspects such as slow init loading never got fixed until I threw those distros out.

      I haven't been able to read the article yet, since it is presumably slashdotted, but I would have thought most benchmarks against Slackware 9.1 would be irrelevant, since the majority of Slackware heads compile their own apps and kernels. In other words, it makes more sense to compare it against Slack in a way that the latter might be assumed to be implemented. Yes, I know that's hard, since Slackware fans tend to be an individualistic lot, but that's too bad.

      And yes, I know Gentoo does that, but Slackware gives you a system that works so that you can compile stuff at your leisure, rather than having to leave a machine out of action for hours/days while everything gets built.

    5. Re:Okay, "stupid question" time by watzinaneihm · · Score: 2, Informative

      If you take a typical redhat kernel, it has support for AMD and Intel both built in,SSE2 and MMX are both on etc., meaning it is not optimised for either.It also enables a lot of stuff like i2c bus support on, which is useful for servers etc. Then there are stuff like patches for faster GUI response etc. which can be added to improve apparent speed. Modules being built in versus loaded matters a lot too.This is for the kernel.
      When it comes to tools, there are so many tools available in the typical GNU stuff that choosing a different one can change performance for different stuff. A change in init can increase bootp speed, since this distro is based on another distro here this would not matter. What they probably did is to change a few package options to tweak the speed.For webserver performance you can tweak Apache (this ofcourse depends on knowing your typical load etc.).
      Certainly there cant be a significant speed increase between distros, but if you know what you want, then you can compile your packages and set options well enough on YOUR system and get a significant boost.

      --
      .ACMD setaloiv siht gnidaeR
    6. Re:Okay, "stupid question" time by yarbo · · Score: 2, Insightful

      You can pick your own compile settings. Normally people don't pick these intelligently, but that's beside the point. There can be performance differences, but there are people out there who spend a ton of money on cooling for a small overclock, so what's so bad about compiling for a little extra performance? Also, there are USE flags which can set the functionality at compile time. Want to use a program without a front end? set -X, want to use ungif instead of gif? set -gif, etc...

    7. Re:Okay, "stupid question" time by fdawg · · Score: 3, Interesting

      I agree, what really sets this distribution apart from any other distribution besides the obvious bootup and daemon start up tweaks?

      Most of the benchmarks on this review are all kernel dependant. Wouldnt it be safe to say that the same numbers can be found in any modern distribution if they use the same kernel patches/optimizations?

      This leads me to my next question. Its true not all distributions are created equall and all distro maintainers do their own black magic to tweak their kernels. But, rarely do I see these optimizations published; as in a list form or even a howto. I am a longtime Slackware fanatic (circa slack2.1 to present) and have always swore by it if for nothing more than the fact that its kernels are straight vanilla with no patches. Now here comes VL which is said to be "faster". How? If i were to remove all running daemons that I dont use (which I do upon a knee-jerk reflex response after a fresh install anyway), wouldnt I get the same speed ups? What did they do to their kernels? How do I find out?

      I switched from Slackware to Debian last week (oh how painful it was...really) on my desktop and I am keeping slack on my laptop. All my development is done on the laptop and Id rather not have to figure out why code I wrote last week no longer wants to work because of some strange header problem between distros. First thing I did was install a vanilla kernel since I really dont trust any distribution's kernel unless I KNOW its vanilla. Anyway, I notice no speed difference. What does the slackware world think about this? Am I missing out by using the vanilla sources? Are there some optimizations that the rest of the world swears by that may increase my speed somewhat?

    8. Re:Okay, "stupid question" time by GreyWolf3000 · · Score: 2, Informative

      What is cooler than the local building about Gentoo is that I find portage personally easier to use than apt, and Gentoo's init & rc setup is IMO just awesome.

      You don't have to stick to your distros init & rc setup. You could even make "init" a bash script containing functions for every runlevel, and have it call the appropriate one, based on the parameters passed.

      Personally, I use just 3 bootscripts, one to start, one to stop, and one to enter multi user mode. For me, the best distro is the one that has the least amount of bloat to whine about me changing things around.

      I use Crux, a much more user-managed distribution than Gentoo, and have taken out devfs, hacked the init scripts, hacked the ports tree to my liking, etc. Using Gentoo from Crux is honestly like moving from a manual transmission to an automatic (something I'll never, ever do.

      I like ports trees that are easy to hack, not done for you (there are always times when you really want ports to behave differently than the official tree).

      --
      Slashdot: Where people pretend to be twice as smart as they really are by behaving like children.
    9. Re:Okay, "stupid question" time by mav[LAG] · · Score: 2, Informative
      I call bullshit. Unless you're doing something stupid, like using bog i386 binaries on a PIV or PIV optimised binaries on an Athlon, you'll get nowhere near 10% performance increase just by recompiling from source.

      I've easily got 10% performance improvement from compile-it-yourself distros (LFS and gentoo are the two I run here) for a number of reasons:
      • The cumulative effects of optimisations to a number of packages can be significant. When X and Gnome and ImageMagick and the Gimp and glibc and the kernel are all optimised for this exact machine (and yes I know what I'm doing with flags), then its faster - starting with kernel block memory copies and working up from there.
      • Code and shared library sizes are smaller because I didn't include the --with-kde flag or whatever when I knew I wasn't using it.
      • Some obscure flags make a huge difference - like --no-g++-exceptions on Qt and KDE. I've found the speed increase to be 50% in some cases which you just don't get when installing a bog-standard binary without those flags enabled.

      --
      --- Hot Shot City is particularly good.
    10. Re:Okay, "stupid question" time by dollar70 · · Score: 2, Interesting
      Why don't you just shell out on a decent computer?

      Oh, that's an easy one! I love buying new computers. This computer that I'm on now is a "Belchfire" computer with GHz to spare, and the one in the other room is an amazing little devil with a lot of spunk. But down in the basement is where my "server box" is. This is a computer I don't want to have to touch. I just want it to quietly do its job 24/7. The fact that it is an older technology doesn't prevent it from performing exactly the same as when it was a new technology.

      Do I need to spend another $500 for yet another full featured PC when all I really wanted was a box that took care of a rather menial webserving task? No. I want a cheap reliable OS with staight-forward tools that will do their job in the dank darkness of my dungeon while I enjoy the cozy comfort up here in my living room.

      Another point is that modern computers come with bells and whistles to entice consumers. These things can get in the way of straight forward tasks where a person isn't going to be there to respond. (Pop up dialogs, commercial offers, etc...)

      And remember: There's no such thing as a stupid question. Only stupid people asking needless questions. ;-)

    11. Re:Okay, "stupid question" time by Cloud+K · · Score: 2, Interesting

      "Why don't you just shell out on a decent computer?"

      A fair enough question.

      My answer: I work for a charity. They can't afford to shell out on a decent computer. Heck at the moment they even struggle buying inkjet cartridges.

      Currently we're running a PC refurbishment scheme where, older systems (e.g. currently a lot of K6-2/300s for some reason) which would've been on their way to landfill are donated to us. I then install Linux onto them and either sell them on cheap to support the charity's income (for members of the public who simply don't need anything more) or give them away to disadvantaged people and families through referral schemes.

      It's a fairly new project, and I've been through a good few distros (Slackware, Gentoo, Peanut...) looking for one that's both simple to use for the target audience (everyday people) and runs smoothly on this older hardware. Obsolete versions of RedHat didn't really appeal to me. Peanut has been the choice so far, but Vector definitely sounds right on the mark.

  2. Don't like the website by Ianoo · · Score: 4, Insightful

    Yet another PHP-Nuke-alike that seems to be so obfuscated it's impossible to find anything...

    However: looks like a fairly good distribution with a good set of tools for the space. But I find myself asking why the 450MB number? Too little for a compressed CD-ROM (like Knoppix), and I haven't seen a hard disk in a machine (even consumer devices) that's under 1GB in years.

    It could be the ideal candidate for a 512MB CF card or Microdrive, but then again, it only runs on Intel x86, so ARM-based XScale, StrongARM, OMAP etc devices are out of the picture.

    So my question is this: it looks pretty good and seems to have quite a bit of support, but what's it's niche? Older machines, like 386s?

    1. Re:Don't like the website by Rutulian · · Score: 3, Insightful

      and I haven't seen a hard disk in a machine (even consumer devices) that's under 1GB in years.

      I think that's the point. This distribution is meant for older hardware. When a 1GB hard disk was as large as you could get.

  3. we need a new cliche by rootofevil · · Score: 2, Funny

    every time you slashdot a server, god kills a kitten.

    please, think of the kittens.

    --
    turn up the jukebox and tell me a lie
  4. The 4th kind of lie by tangent3 · · Score: 4, Funny

    Slashdotters should know by know that there are now 4 kind of lies. Lies, damned lies, statistics and benchmarks.

  5. Great distro by Beek · · Score: 4, Interesting

    Runs comfortably on my P133 with 48 MB of RAM.

    Although I am curious what other modern distros will run on such a machine?

  6. 233Mhz is slow? by Anonymous Coward · · Score: 4, Insightful

    My God, kids these days are spoiled. How about starting the benchmarks on a 386 with 4 megs of RAM and working up to that mighty 233 monster. I'd be most interested in the differences between say a 486DX66 with 128Megs and a P90with 32Megs. Perhaps these specs will make some snicker, but when you're working on making old hardware do something cool these are the specs that are interesting. If you've got a 1GHz plus machine with tons of RAM you don't need stripped down distro.

    1. Re:233Mhz is slow? by chasec · · Score: 2, Funny

      You have a 386? In my day, we didn't have no fancy-schmancy 386s. We had an abacus, and we liked it!

      I overclocked my abacus, and that thing was fast. It runs linux as fast as you can move the beads.

      Kids these days...

  7. Good by d-Orb · · Score: 4, Interesting

    Vector Linux is good because there are still loads of P2-2xx/3xx around. While you can always run XFCE on them, dillo and so on, they do have hardware limitations which might be difficult to overcome in order to have something normal people can use. If you get a distro that runs a modern desktop (KDE, GNOME) with some response (this is what the reviewer says; what he means is anyone's guess), then you can effectively extend the life of these older machines, save money, and so on. While many gamers may not be aware of this, people in small/family-run companies could definitely use something like vector for these purposes.

    Unfortunately, I think that most of the people who would benefit most from things like Vector will never hear of it, and if they do, they will probably be overwhelmed by difficult installers and so on. If the VL people could come up with their distro packaged so that it effectively is a domestic distro (put CD in, wait, enjoy), then they would have a great product that many SMEs would use. They might also want to get some other software (accounting and that sort of stuff), but there is definitely a niche there for them to occupy. I wish them well :-)

  8. I am rolling it out in a school by codepunk · · Score: 5, Interesting

    I was approached by a private grade school that had a bunch of old gear and no money to upgrade software much less the hardware to run the latest windoze. I had them buy a single poweredge server and loaded up mandrake. Turned on XDMCP and are currently loading the clients with Vector Linux. It works really nice on old hardware and boots quickly. So for the cost of a single server they get to use the old gear plus all of the open source software they could ever want. Kudo's to the vector linux developers they saved this schools 50 workstations.

    --


    Got Code?
  9. Every time you make that joke... by yerricde · · Score: 2, Funny

    Every time you make that f*@!(%# joke, God kills a Domo-kun.

    Please, think of the Domo-kuns.

    --
    Will I retire or break 10K?
  10. Lightweight Distros by MuParadigm · · Score: 3, Informative


    Well, we all know that most of the distros *can* be configured to run on older machines -- if you change all the defaults to load programs and interfaces running with lower memory requirements. Slack and Debian are probably the best for tis task.

    But if you're looking for something designed for older PC's "out of the box" then Deli, Damn Small Linux, or most of the other live CD distributions designed for business card CD's will also work, since they're all designed for small memory footprints, which is probably the main constraint on older systems.

    Deli, in particular, makes the interesting choice of using the 2.2.25 kernel, which should be good for older machines.

  11. For those who do not want to build their own by the_crowbar · · Score: 4, Interesting

    Reviews highlighting a distro built for speed are useful for those who are unable/unwilling to build their own Linux system.

    Recently I bought a mini-itx system to put in my car. At that point, I wanted something lighweight (runs on VIA C3 Proc) and fast. I ended up building a Linux From Scratch system, but if I did not have the skill or inclination for something like that a distro optimized for slower hardware would be a great choice.

    Even though I built a LFS system, I could have saved a bit of time by installing a binary distro. LFS took me several days of steady compiling to complete. (I did most of the build work on my desktop and simply copied the files over, but still.)

    the_crowbar
    --
    Have you read the Moderator Guidelines
  12. Re:Yeah? So? by fdawg · · Score: 2, Funny

    Itd be a joke if it wasnt true.

    "#apt-get install flame-armor increased-karma"

  13. How very... (what's the word I'm looking for) by joeslugg · · Score: 4, Funny
    The interesting thing here is...
    ...but the interesting thing about it was...
    There are some pretty interesting numbers...
    It is interesting to say the least.

    I'm searching for an adjective to describe this article. Somebody help me here...

  14. packages? by Danathar · · Score: 2, Interesting

    Is vector compatibile with packages designed for slackware 9.0?

  15. Re:Uggh by stm2 · · Score: 2, Informative

    Maybe Morphix Linux is what you are looking for. It has a "base" system and then you could add "modules". there are some modules ready for download or you could download an ISO with some modules installed.

    --
    DNA in your Linux: DNALinux
  16. OT -- Back to Basics by digitalhermit · · Score: 2, Informative

    Somewhat OT, but the idea of a toned down but still functional Linux appeals to me. I had a similar idea of doing this to a distro not well known for being slim -- RedHat 9. For partially philosophical reasons, and partly because I was not getting any work done with all the gizmos cluttering my desktop, I decided to remove the clutter but not lose any functionality. (In my defense, I'm comfortable with Debian and some embedded, minimal distros and have built a Linux from the kernel sources to X).

    The hardware for this ongoing project is a 333MhZ PII laptop, 192M. I started by getting rid of the Gnome and KDE environments (well, most of it -- I kept the libraries and some select applications around). In their place I put in Fluxbox, choosing ./configure options carefully to minimize memory usage.

    Next, because I spend the majority of my time in the shell, I looked at some of the different xterms around. I was surprised that the native xterm, though much smaller than konsole or gnome-terminal, was still somewhat bloated in comparison with others such as aterm or rxvt. They didn't support transparent terminals but that's no biggie. The important thing was that they could do green-on-black terminals; also no biggie, but I was thinking about this because I made an assumption that a black background would use less battery than a white one. Of course, you could also ditch X entirely and run from a console but browsing the web in elinks or links, though great for documentation, kinda sucks for looking at Dolphin cheerleaders.

    Next, I exchanged the stock RedHat kernel with a 2.6.0-test kernel (test9 at this writing). It does seem a lot faster, but I am still working out some module loading issues so there is some functional loss until I get these working. This is important because the goal is not to lose functionality for performance.

    I've been testing different journaling modes for the ext3 filesystem. No benchmarks yet, but I understand that there's a decent performance boost to be had from using a different writeback mode.

    These are all in addition to the standard tweaks such as using a lower bit depth on the X session, replacing apps with slimmer alternatives (Firebird for Mozilla, etc.). There are also dubious claims of speedups by just recompiling but in my case these gains weren't perceptible.

    I'm about to replace syslog with one that batches writes. This will allow the drives to spin down. Since power management is otherwise functional it might gain some performance.

  17. Real Performance Tests Are Subjective by reallocate · · Score: 2, Insightful

    Setting the right compiler options for your hardware can make a small difference. Most people, though, don;t have a clue about setting compiler options.

    Hardware itself makes a much greater difference. Any OS running on a Pentium 4 3 gHz with a large UDMA133 drive and a $500 video card will be faster than the same OS running on a 486SX-25 an ancient drive and a $35 video card. One would think that's obvious.

    Most of these so-called performamce tests are silly. One guy finds Thing A is faster on his hardware than Thing B. Doesn't mean that will happen on my hardware.

    The real test is you, and that's subjective. If its fast enough for you, it's fast enough.

    --
    -- Slashdot: When Public Access TV Says "No"