Linux Benchmarking

[Sy]

On 8/19/05, Walter Dnes <waltdnes-SLHPyeZ9y/tg9hUCZPvPmw at public.gmane.org> wrote:
> On Thu, Aug 18, 2005 at 05:26:55PM -0400, Sy wrote
> 
> > humour:  BUT, if you manage to get a "linux distribution speed
> > comparison" website put together, I'd love to see "proof" that
> > stock-Slackware is better than the best-tweaked Gentoo setup.  That'll
> > shut 'em up!  ;)
> 
>   Speaking from the Gentoo side of things, I'd be interested in the
> results too.  Maybe we could get together and produce a Slackware vs
> Gentoo showdown... for the April meeting<g>.

Except I dropped Slackware for the desktop variation of its "it just
works" philosophy.. pclinuxos.  I'm now liking the idea that crap just
works out of the box without my having to learn a new skill.

For me, speed isn't such a big deal.  It's responsiveness.  The
experience as a user is the frustration in waiting during moments when
waiting is not expected.

I always expect to be able to be able to wiggle my mouse.  I don't
expect to be able to interact with a program which I just told to do
something complecated.

I expect that when hovering over a button which brightens it for me to
interact with, I can click on it to do something.  i.e. an application
shouldn't paint an interactive thing unless it's prepared to have it
interacted with.

Stuff like that.  I was happily running a p266 (96MB RAM) with an
X+blackbox setup.. and I could run a low-traffic apache+mysql server
underneath, while surfing and downloading, while ripping audio tracks
from cd, while encoding multiple tracks simultaneously.

While playing mp3s.. without skipping.  Linux won a lot of contests at
that point.

I still remember when colour and graphics was a wet dream, so I
appreciate what I can get.  Right now I just want a setup to do what I
expect when I expect it.. the rest doesn't matter to me.

So even though there's tweaking that's possible.. and even though I
still have some leftover choices from when speed mattered a lot more
(i.e. blackbox, a tcl/tk file manager, console applications vs
gui..).. even though I'm still stuck in the dark ages as a desktop
user, I no longer care about raw data about speed.

I'd care more about "blackbox is faster than x, and this is why.." or
"tweaking mplayer".

So if someone told me that stock gentoo was faster for an average user
than stock slackware, then I'd listen.  Such a concept doesn't really
exist though.  Ok, there's sortof a stock gentoo.. sortof.

But if someone told me that gentoo has a slick newbie-friendly
installer, I'd perk up.

Sure.. I'd have it compiling stuff in a virtual machine for a few days
to check it out..


Blah, I'm going off onto a tangent.

"distro vs distro" just isn't healthy from the perspective of speeds,
because it's the user experience that matters.  Users are individual,
so there's no real way to satisfy everyone.

HOWEVER, application versus application is more interesting.. tweaking
guides, howtos, etc etc..

or, as I said.. I'd love to see a concise listing of tools and methods
for benchmarking my own stuff.  I wouldn't mind learning to test the
capabilities of my own server in new ways, for example.
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml