Arch Linux

secsaba

Member

From: Espoo, Finland

Registered: 2003-11-06

Posts: 19

Beaver Challenge

Peaple at http://osuosl.org/ are going to benchmark some Oses.  Here http://osuosl.org/benchmarks/bc/ you can read about the Beaver Challenge 2004. The OSes until now are:
    *  Debian GNU/Linux
    * Dragonfly BSD
    * Fedora Linux
    * FreeBSD
    * Gentoo Linux
    * Mandrake Linux
    * NetBSD
    * OpenBSD
    * Red Hat Linux
    * Slackware Linux
    * SuSE GNU/Linux
They accept suggestions for other OSes. Can I suggest ArchLinux? There is any Arch guru who can setup the machine?

dp

Member

From: Zürich, Switzerland

Registered: 2003-05-27

Posts: 3,378

Website

Re: Beaver Challenge

great idea proposing arch for this benchmark

... unfortunately i'm not near this place and dont have much time to spare, sorry

---

The impossible missions are the only ones which succeed.

Xentac

Forum Fellow

From: Victoria, BC

Registered: 2003-01-17

Posts: 1,797

Website

Re: Beaver Challenge

Do you have to be near the place?  I didn't think so...

Are we going to have a team or should one person just step forward?  It's looking like we'd need someone good at tweaking mysql, the kernel (2.4 and 2.6), filesystems, and probably others.  We'd also need people with lots of time (I have some... not lots).

Though, as strange as it may sound, I could probably be the team lead.  Anyone with me on this?  If I get two other capable people, I'll talk to them about making a team.

---

I have discovered that all of mans unhappiness derives from only one source, not being able to sit quietly in a room
- Blaise Pascal

Dusty

Schwag Merchant

From: Medicine Hat, Alberta, Canada

Registered: 2004-01-18

Posts: 5,986

Website

Re: Beaver Challenge

Er... if nobody else wants to help you, I might. But I'm probably more of a hindrance than a help.

Plus I haven't got a lot of time... but I did make the offer.

---

** Arch Linux Apparel and more ** Arch Linux stickers, laptop bags, jewelery ** I wrote a book on Python 3 ** and one on mental health **

Guest

Guest

Re: Beaver Challenge

The goal of these tests is to finally determine, based on skilled configuration of each distro, which is the fastest and most scalable using todays fastest hardware.

Alright... So they use 2.8Ghz Intel Xeon processors instead of AMD Opterons.

Imho, Linux is Linux, so they probably won't be really testing the distros, but which distro has the best configuration (kernel options, filesystem, cflags etc) for that particular system and test. They let people tweak it, so then it's even less a comparison of distros but even more of the configuration skills of some people. It would be more fair to do a more or less default install of each distro and then test it, without giving all the gurus the time to tweak it to death.

sarah31

Member

From: Middle of Canada

Registered: 2002-08-20

Posts: 2,975

Website

Re: Beaver Challenge

The goal of these tests is to finally determine, based on skilled configuration of each distro, which is the fastest and most scalable using todays fastest hardware.

Alright... So they use 2.8Ghz Intel Xeon processors instead of AMD Opterons.

Imho, Linux is Linux, so they probably won't be really testing the distros, but which distro has the best configuration (kernel options, filesystem, cflags etc) for that particular system and test. They let people tweak it, so then it's even less a comparison of distros but even more of the configuration skills of some people. It would be more fair to do a more or less default install of each distro and then test it, without giving all the gurus the time to tweak it to death.

well put.

---

AKA uknowme

I am not your friend

secsaba

Member

From: Espoo, Finland

Registered: 2003-11-06

Posts: 19

Re: Beaver Challenge

I think it's a good idea to subscribe to the mailing list at http://lists.oregonstate.edu/mailman/listinfo/bc2004 Here somebody with a better english can suggest the Arch Linux distro also.

andy

Member

From: Germany

Registered: 2002-10-11

Posts: 374

Re: Beaver Challenge

Alright... So they use 2.8Ghz Intel Xeon processors instead of AMD Opterons.

and with respect to distros that means .... ?

Imho, Linux is Linux, so they probably won't be really testing the distros, but which distro has the best configuration (kernel options, filesystem, cflags etc) for that particular system and test. They let people tweak it, so then it's even less a comparison of distros but even more of the configuration skills of some people. It would be more fair to do a more or less default install of each distro and then test it, without giving all the gurus the time to tweak it to death.

These standard tests have been done over and over. One of the criticism they get is "the tester has no clue about this OS". If you carefully read the Beaver challenge web site, you would have noticed that this is exactly the point these guys want to address. Especially the benchmark they refer to was one such example where the tester obviously had no clue about the various BSD's (He used OpenBSD CURRENT - that's like running Linux kernel 2.5.8 or 2.7.3  ....).
And secondly, there WILL be benchmarks with a default install. And only after that the teams will tweak.

I think this is a very useful benchmark. And if you know how to read benchmarks, you probably can learn something from it. Unfortunatly most people can't read benchmarks.

Enough ranting. I would volunteer but I don't have time - especially this weekend. Also, I have no experience with tweaking for database and/or disk IO.

i3839

Member

Registered: 2004-02-04

Posts: 1,185

Re: Beaver Challenge

Ok, I'll try to keep it short.

and with respect to distros that means .... ?

Nothing, but then they shouldn't claim to use today's fastest harware either (which in itself is already debatable, it's far more useful to use the average system because that's what most people use).

These standard tests have been done over and over. One of the criticism they get is "the tester has no clue about this OS". If you carefully read the Beaver challenge web site, you would have noticed that this is exactly the point these guys want to address. Especially the benchmark they refer to was one such example where the tester obviously had no clue about the various BSD's (He used OpenBSD CURRENT - that's like running Linux kernel 2.5.8 or 2.7.3  ....).

I do know that, but people will always complain about something, so giving all whiners what they want will give you a worthless benchmark.

And secondly, there WILL be benchmarks with a default install. And only after that the teams will tweak.

That is good, I didn't read the Methodology page, so missed that. Reading that page it seems that the benchmark isn't as bad as it looked at first sight.

I think this is a very useful benchmark. And if you know how to read benchmarks, you probably can learn something from it. Unfortunatly most people can't read benchmarks.

Unfortunately most people can't benchmark well. First rule of every good benchmark is to keep only one thing variable (if you consider something as vague as a distro one thing, well..). This will be more a contest for the best tweakers than a real test, doesn't mean it's bad or shouldn't be done, but I'm afraid that people will take it way too seriously. It should be seen as a fun tweaker contest, not as a "which OS is best" benchmark. So I don't have problems with the challenge itself, but how people will treat the results.

It would be also fun if there would be a third round (first round is basic install, second is 3 days time to tweak the system), where they can see eachother's tweaks and use those to do it better. I don't expect any significant differences, and if there are any then the heavy kernel patchers probably won (a reason to do a very good stability test, like running some benchmarks for a week).

I think that it would be a very good adverticement for any distro that participates (except if they screw something up of course), so if I were Arch I would sign up. Unfortunately the deadline is today.

sarah31

Member

From: Middle of Canada

Registered: 2002-08-20

Posts: 2,975

Website

Re: Beaver Challenge

the problem i have with benchmarking is that it proves nothing and serves absolutely no purpose. the idea of any OS is to make people more productive. any fricken distro could optimize to death for each type of architecture out there but who really gives a blow? i sure don't .

benchmarks are like talking about resource use. i got news for those people that talk about resource use.... linux is just as much of a resource hog as any other  OS i use.

in the end i used arch because it was a good distro.. i wanted an i686 optimized distro with good package management tools and a nice slim package set that i had control over. but the biggest thing for me was ABS that is definitely what made me move to arch from debian or debian based distros. it was nice getting the speed bump but in the end it meant nothing.

speed bench marking like this does nothing but create bad blood and switch user/developer focus to pointless optimizations and not the quality of the overall system. this is like developing your system based on statistical out put. if you turn your focus to one area the rest gets neglected.

stats can be useful in some ways but speed benchmarking stats are really not something arch linux should be involved in when there are other more pressing issues.

---

AKA uknowme

I am not your friend

Dusty

Schwag Merchant

From: Medicine Hat, Alberta, Canada

Registered: 2004-01-18

Posts: 5,986

Website

Re: Beaver Challenge

speed bench marking like this does nothing but create bad blood and switch user/developer focus to pointless optimizations and not the quality of the overall system. this is like developing your system based on statistical out put. if you turn your focus to one area the rest gets neglected.

Hmmm. Well put!

My experience with distros is that if you mess with them enough, they look absolutely nothing like the original distro. If I had the time, any distro I used would be "Dusty's Linux".

The reason that I like Arch so much is that it's a lot closer to "Dusty's Linux" to start with.  It does what I want' what more can I ask of it?

Dusty

---

** Arch Linux Apparel and more ** Arch Linux stickers, laptop bags, jewelery ** I wrote a book on Python 3 ** and one on mental health **

sarah31

Member

From: Middle of Canada

Registered: 2002-08-20

Posts: 2,975

Website

Re: Beaver Challenge

it could make you richer so you could move out of that hellish province

just kidding.

---

AKA uknowme

I am not your friend