Why Slackware runs fast?
 

We've heard a lot that Slackware runs faster than most other Linux distros. My personal experience (not benchmarked) also backs this statement. Slackware and Slackware-based distros (Vector, Wolvix, Zenwalk, ...) usually run very fast. I am trying to understand the reason. Slackware doesn't tweak the kernel. It doesn't use aggressive optimization options to build its packages. I know it uses BSD style startup scripts and doesn't turn on too many services in /etc/rc.d by default. This may speed up the system boot process but doesn't make programs run faster in the system. The file directory structure in Slack is not significantly different from other distros'.

If Slack really runs faster than others, what are the reasons?

It has a special hidden directory where all the magic happens

/.placebo

:p

Personally, I believe it to be from minimal source code patching. Go look at Fedora Rawhide and you'll shutter. upwards to 20 patches on some packages. BTW, Debian does it too. ;)

I have nothing to back that assumption up, but am not sure what else to attribute it to. My DIY build runs just as fast. Something to be said for Vanilla Linux I think.

Maybe because it gets in your way less.

"doesn't turn on too many services" is the main reason, I believe.

Slackware does run faster than many distributions on a base install. I have tweaked a Slackware system to boot on old hardware in 30 seconds.

Slackware does not have as many services turned on by default which is one reason. You also do not get a lot of packages (unless you select the full) option that you don't need. I'm not sure the effect of this on boot time but it can definitely slow your computer down at runtime.

Getting benchmarks of different OS's was on my list of stuff todo, just haven't got around to doing it yet.

Pure and utter magic, I'd say. Almost as good as Gentoo ;)

Mons

Not using SELinux, GRSecurity or AppArmor probably doesn't hurt much either. All those additional authorisation checks don't come for free, though for the most part I doubt the lost clock cycles would be noticeable to a human.

My primary system (4 year old, Pentium 4 3.0Ghz Hyperthreading) just died on me. I'm typing this on an old Dell XPS B Series, P3-800 which I've resurrected to tide me over. I get the feeling it will continue to run forever if I let it. <OLD FART MODE>"They don't build 'em like this anymore, sonny!". It has 128mb of ram and a 32Mb GeForce 256 card. I'm running slack 12.1 (X but no KDE) and even on an encrypted LVM setup its surprisingly usable for a machine of its vintage. Memory is a bit on the tight side, but otherwise it makes me wonder if its actually worth replacing my P4 at all. It runs a damn sight cooler than my P4 too.


To be honest though, I can't say I've ever noticed a speed difference between Slackware and any of the other distros I've tried: though I've never been a fan of the full-fat desktop environments such as KDE and GNOME, which may go some way to explaining that.

I have an old HP Vectra workstation: P3 733mhz, 512mb RAMBUS (lol).

Any recent Fedora release is absolutely unbearable to use. However, Slackware seems to run ok, even using KDE. I would even say that it "seems" faster than my brother's P4 with Ubuntu installed.

simple = fast

My P800 is RAMBUS too Granth. Unfortunately only 128MB (though back in those days, that was a lot of RAM... hehe) and getting any extra for it these days is probably not a viable option.

I was just casting a tentative eye over the Dell Precision Workstation range. XPSes have gotten a little too tricked-up for my liking these days. All those colour changing LED light strips and what have you on the front may make the kids go 'Woah!... Cooooooool", but they don't do anything for me.

Slackware doesn't try to coddle you or save you from yourself.

Eye candy (unless using accelerated hardware) wastes cycles better spent on doing work.

Protecting you from yourself means limiting options and wasting cycles on stuff that should not be needed for someone trying to learn, or for serious admins, gets in your way.

By keeping it simple, it's easier to setup and easier to maintain. Since you don't have extra cruft getting in the way, a byproduct is also extra cycles for getting work done as well as an excellent working example that lets you play (even if it means getting yourself in trouble, like 'rm -rf /').

Honestly, I blame Gnome and all the added crap a typical Gnome install adds. Dumping it was the best thing Pat ever did. I've installed GSB and while that still runs faster than Gnome on other distros, it added way too many useless (IMO) daemons and services. Pare it down to an acceptable level and it fails to run correctly. It runs sluggish under X as well.

When you consider that Gnome is the default DE for Ubuntu, Debian, and Suse (I think), and is installed on many more Distros (even if it isn't the default DE) then maybe there's a connection?

Eh, what do I know ...

;)

So your saying that Slackware didn't become fast until gnome-2.6 was dropped?

Slackware doesn't load a lot of extra services on boot-up, less system overhead, so it runs quite fast:-) Slackware is a lean, durable, secure OS.
Slackware 12.1 with XFce 4.4.2 runs smooth as silk on my main work station, an IBM Celeron 850 MHz, with 768 MB RAM. Slackware 12.1 also runs very well indeed on my IBM Plll 667 Mhz, with 256 MB RAM.

I think it's just about what services are enabled, there's no magic to it. Recently I tried Arch and of course it seems very fast at first because it's minimal. Once you add the same services and use the same applications as in another distro there's no appreciable difference. There are distros like the Suse ones which use their own tools, services and structure and might not be capable of performing as well but I'd guess any distro that is based on traditional commonly used tools can be made to perform pretty much the same. It's really a question of building on a minimal base vs removing from a more comprehensive default set-up.

I listened to a podcast recently where the authors, a Debian user and a Gentoo user, tested Ubuntu Hardy. They found no appreciable difference in performance between Ubuntu and Gentoo using geekbench testing tools except Ubuntu has a bigger RAM footprint, which might be explained by the use of prelinking/readahead. I've recently tested for my own interest Debian Lenny, Arch, Ubuntu Hardy, gNewSense and once they're configured with the same kernel, services and desktop environments (as far as is possible) I found the same thing. The only real variables seem to be boot time and memory footprint which I'm inclined to think is the result of different start up scripts and prelinking.