Phoronix FreeBSD 8 Performance Test

Phoronix posted a performance tests of FreeBSD 8.0 today. FreeBSD doesn't do very well there but the reason could be that it's first stable release of 8 branch and that it's currently stabilized and not that optimized for performance. Of course that is something else. As far as I know Linux distros use a lot newer and better GCC and FreeBSD uses a long outdated gcc 4.2.1 (20070719).
 
As far as I know Linux distros use a lot newer and better GCC and FreeBSD uses a long outdated gcc 4.2.1 (20070719)

And as far as i know the gcc version will stay at that version.
If i recall correct it had something to do with the new v3 GPL license.

And maybe FreeBSD is slower at some points, it remains my main system. Just because it is clean, manageable (for me at least;) )
and i feel very comfortable with it.

tried some linux distro's even did RHCE when RedHat was at 7.0, but when i came acros FreeBSD it never let me go.
regards
Johan
 
Sylhouette said:
...
And maybe FreeBSD is slower at some points, it remains my main system. Just because it is clean, manageable (for me at least;) )
and i feel very comfortable with it.
...

Nobody talks about switching OSes. I'll probably never switch even if FreeBSD 10 is 3x slower than windose ;) (all my ~100 servers and desktops are FreeBSD).
But it's always better to be faster than not, right ?
 
i would say fast and stable. check questions@ and performance@ there is discussion exactly about that bench. iirc one topic was named "specific example of io(was zfs bla bla..)"
 
mk said:
... check questions@ and performance@ there is discussion exactly about that bench. iirc one topic was named "specific example of io(was zfs bla bla..)"

The only thread that looks like that is 'ZFS Re: A specific example of a disk i/o problem' which has nothing to do with that and is a month old. That article is from today :)
 
tbyte said:
The only thread that looks like that is 'ZFS Re: A specific example of a disk i/o problem' which has nothing to do with that and is a month old. That article is from today :)
this phoronix bench fbsd vs ubuntu - this is from september 28. last time that make performance@ noisy and in the same time that specific zfs io example appeared so i remembered as they were the same thing. i think that there was even redirect to questions@. i can't remember and i'm lazy to search now.
 
I think this phoronix benchmarks linked here are completely meaningless. It's like trying to impress a little girl with statements like "...my penis gets hard in 5.123 seconds instead of 5.789 second but only when the sun shines in Paris and I have Angelina Jolie as desktop background..."
 
I read the article on Phoronix just today and it was a pretty interesting read.

However, just recently making the switch from Linux to FreeBSD, I am pretty satisfied with the way things are in the FreeBSD operating system. Everything seems so much more "simplified & organized".

Linux too me now just seems so unorganized and chaotic and lots of unnecessary junk :)
 
Penel said:
However, just recently making the switch from Linux to FreeBSD, I am pretty satisfied with the way things are in the FreeBSD operating system. Everything seems so much more "simplified & organized".

Linux too me now just seems so unorganized and chaotic and lots of unnecessary junk :)

That's what I'm trying to explain to linux users every time I end up in a linux vs FreeBSD argument. But as they have never tried FreeBSD ... good luck trying to explain that to the linux zealots ;)
 
tbyte said:
As far as I know Linux distros use a lot newer and better GCC and FreeBSD uses a long outdated gcc 4.2.1 (20070719).
The FreeBSD team is working on that problem with Clang/LLVM being the solution. Check out the FreeBSD wiki page at http://wiki.freebsd.org/BuildingFreeBSDWithClang. Once FreeBSD can be built with Clang/LLVM, then the port maintainers can work on getting their ports to build with Clang/LLVM, or if they still need to be build with GCC, then they can work on getting them to build with the latest stable version of GCC.
 
troberts said:
The FreeBSD team is working on that problem with Clang/LLVM being the solution.

That doesn't mean that Clang/LLVM will be faster than the current version of GCC (4.5). As far as I know the current focus for Clang/LLVM is making it stable and usable at all. So I'm guessing there will be a few years before the focus shifts to performance.
So I'm not exactly sure it will be "the solution" :)
 
tbyte said:
Phoronix posted a performance tests of FreeBSD 8.0 today. FreeBSD doesn't do very well there but the reason could be that it's first stable release of 8 branch and that it's currently stabilized and not that optimized for performance. Of course that is something else. As far as I know Linux distros use a lot newer and better GCC and FreeBSD uses a long outdated gcc 4.2.1 (20070719).

This is a nonsense benchmark. Nobody e.g uses Povray anymore, apart from some hobbyists, most people are using yafray as render engine for prof. work in FOSS or Blender and so on. Furthermore who uses gzip? For really big data people are using different applications, so again a test of no relevance.

A benchmark is usable only at your very own hardware and among the same line of operating system. Last not least we see heavily tuned operating systems like Fedora and Ubuntu using alpha/beta software compared to a stable and reliable operating system like FreeBSD.Even OpenSolaris is tuned by default. But FreeBSD is by default maybe ready for a small headless server. So guess what, this benchmarks stinks.
 
FreeBSD 8.0 Benchmarked Against Linux, OpenSolaris

Any thoughts?

http://www.phoronix.com/scan.php?page=article&item=freebsd8_benchmarks&num=1

Out of the 26 graphs shown in this article, how many times did FreeBSD 8.0 pull ahead of Linux / OpenSolaris? Just one. This win was with the C-Ray ray-tracing engine. However, in this test, it actually illustrates a performance regression where FreeBSD 7.2 was even faster than the newest 8.0 release. Between FreeBSD 7.2 and 8.0, the week-old operating system did offer nice performance improvements in a few areas like MAFFT. The real performance race though came down between the OpenSolaris 2010.02 development build (derived from b127) and Ubuntu 9.10 / Fedora 12
 
Benchmarks aren't per se worthless, especially not in certain environments. But on a broader basis they're mainly entertainment, like Phoronix itself.
 
Back
Top