fun in numbers, or what benchmarks do show

[Onno Kreuzinger]

see this comparison between
linux 2.4 / 2.6
and
Free/Net/OpenBSD

http://bulk.fefe.de/scalability/

[Pierre-Marie Baty]

nitpicker

OK, if you want to start a ranged battle then here we go (I suppose posting such a benchmark on Bots United has something to do with my new avatar, what, you don't like it ? )

I don't see what OpenBSD has to do in a scalability benchmark... everybody knows the main concern of this OS is stability and security, not speed of code nor scalability... and furthermore this idiot took OpenBSD-current which is the most unstable WIP, being not recommended for anyone to use BUT the OpenBSD developers, and if it was not sufficient he enables the IPv6 stack, which is VOLUNTARILY broken in OpenBSD because of its security issues !

And the guy does not seem to know that NetBSD and OpenBSD are still using a static library model, unlike FreeBSD and Linux which use dynamic libraries... this falsifies completely the benchmark

He sems to be completely clueless of what's up in the BSD world all along the benchmark, anyway. I seriously doubt even his unpartiality

However these numbers tend to show that Linux seem to have improved a lot since the 2.4 series, which is a good thing (it would be great if it also improved in terms of stability, but one obviously cannot have everything at once )

[Onno Kreuzinger]

*rofl*

no that would be politics!

it just shows that numbers are irrelevant, as long as you don't unterstand what they mean and how to cope with that.
the only point which is clear, is that linux 2.6 scales linear in all but one test.
and well i knew that openbsd does not scale well, but did you how bad / no-so-good it scaled. simply for the reason that a firewall should be able to log it actions i wanted to know the comparison. so it must be in there. also the *BSD family is not lead by freebsd. netbsd looks better. thus freebsd has more features and special stuff, but untill 5.0 is ready netbsd is the bsd to go. and they react much faster and take this critzism serious, while openbsd (like your posting *g*) screams leave us out of here ...
and they have some serious problems, which are not acceptable, they lead to dos weaknesses, not good for security at all :-(.

but anyways, it's just to show the 2.6 can cope with freebsd and that netbsd is more rounded up than freebsd 5.0. the fact that openbsd is not the king of the hill is not of interesst for me, despite your avatar

and i know it's much to read, but you should start with the new results at the bottom of the page, there you wil see he uses static for net/open bsd and compares 4.9 and 5.0 Freebsd. so he reacts quite fast to critisism and redoes the tests as proposed by the maintainers, there you see thet netbsd fixed allmost all "bugs" he found, openbsd only fixed one...

cheers

p.s.
i would like to see AIX and Solaris in that tests too

Originally Posted by Pierre-Marie Baty

nitpicker

OK, if you want to start a ranged battle then here we go (I suppose posting such a benchmark on Bots United has something to do with my new avatar, what, you don't like it ? )

I don't see what OpenBSD has to do in a scalability benchmark... everybody knows the main concern of this OS is stability and security, not speed of code nor scalability... and furthermore this idiot took OpenBSD-current which is the most unstable WIP, being not recommended for anyone to use BUT the OpenBSD developers, and if it was not sufficient he enables the IPv6 stack, which is VOLUNTARILY broken in OpenBSD because of its security issues !

And the guy does not seem to know that NetBSD and OpenBSD are still using a static library model, unlike FreeBSD and Linux which use dynamic libraries... this falsifies completely the benchmark

He sems to be completely clueless of what's up in the BSD world all along the benchmark, anyway. I seriously doubt even his unpartiality

However these numbers tend to show that Linux seem to have improved a lot since the 2.4 series, which is a good thing (it would be great if it also improved in terms of stability, but one obviously cannot have everything at once )