Hi Forum

Has anyone ever noticed that Phoronix *never* benchmarked Slackware against any other Distro...?

... really, I mean... they benchmarked much smaller distros, like Sabayon, OpenSolaris or FreeBSD... but not Slackware AFAIK...

I guess it would be "unfair" benchmarking Slackware against any mainstream Distro...


BRGDS

Alex

Benchmarking linux vs linux is pretty pointless IMO. People create placebos in their mind that their distribution is faster by adding more aggressive optimizations to the kernel and userland. I think that is silly Gentoo behavior, any performance increases are usually negligable and in some cases just make things worse.

Benchmarking OpenSolaris or FreeBSD against Linux on the otherhand acually makes sense as they are not the same thing.

3 members found this post helpful.

Hi amiga32

I see...

I Understand your point, but Phoronix actually detected performance regressions along the buntu series from 7.04 to 8.10...

On other way, I dunno wether Phoronix test suite is actually reperesentative... :|

BRGDS

Alex

Yeah but I think that probably has to do with regressions in the later kernel.

I assumed from your original post you were talking about overall benchmarks of Slackware vs Ubuntu or whatever, using a similar kernel.

1 members found this post helpful.

Yes, that is correct...

Thanks God, they don't try this nonsense with Slack.

3 members found this post helpful.

My impression is that a lot of the big differences the Phoronix benchmarking throws up are due to the filesystems/settings employed in their benchmarking. So benchmarking One_ix with ext3 against Two_ix with ext4 (or one set to writeback, or with atime off...) is going to show which filesystem/setup is more efficient in that particular use case and not which distro is faster more generally (unless there have been kernel/filesystem regressions and then the chances are all distros will get it, until it is cured).

And that would still be useful, if the Phorons would document which filesystems and what settings they used, but all they'll say is 'we used the defaults'. IMHO, it is too much to expect ordinary Joes to tease out what that means for distros that they don't use. And without that, it is 'one filesystem is faster than another, but we won't tell you which is which'.

Of course, there are also small differences, which may be meaningful (but probably not very useful). Or they may just be 'noise' from the big differences.

2 members found this post helpful.

Phoronix mostly follows Ubuntu, Fedora, and Open Solaris. They sometimes mention other distros. For example, they did some recent benchmarking of Sabayon.
Phoronix will post articles on every alpha and beta release of Ubuntu and Fedora. But they never post articles on Slackware.
They ignore "hardcore" distros like Slackware and Debian.
I would like to see them benchmark Slackware and Debian though. I have always observed that Slackware and Debian are faster and use fewer resources on my systems than Ubuntu. This would likely yield better benchmarking results I would think.
However, comparing Slackware or Debian to Ubuntu means comparing different kernels, different versions of Xorg, gcc, etc. So it would be an apples to oranges comparison at best.

1 members found this post helpful.

If these benchmarks make sense depends on what you want to do with them. If you want to choose a distro with a good overall performance without a lot of tweaking, then the Phoronix benchmarks may be useful. Such benchmarks may reveal surprising fact, such as that Xubuntu (with XFCE) is slower (!) than the standard Ubuntu.

On the other hand: What is the 'default' file system of a Linux distribution? All distros I know, with the exception of live systems running from CD, asked me, what file system I during installation. So what's the default?

But now: If Phoronix would benchmark Slackware, and it would excel above anything else or finish last behind everything else, what would it change for you? In the first case, you know, what you already know --- Slackware is just great. And in the second case, would a (probably) small difference in performance make switch your distro?


gargamel

1 members found this post helpful.

I guess yr point is correct...

I just feel sort of curious about this...

Intuitively... most of my apps compiled from source "feel" snappier in Slackware, but on the other way, I compile them with Intel compilers, using all sort of aggressive optimizations...

And I am not talking vulgar apps... I am talking heavy CPU/RAM consuming stuff, like CFD codes, shock-propagation codes, Monte-Carlo transport codes...

Anyway.... used to run them under Lenny, and all things equal, I "think" they run faster now under the same hardware... ( never benchmerked though ...)


A benchmark from Phoronix would be an objective measure of what i intuitively "feel"...


BRGDS

Alex

Hi Alex,

of course, there *are* differences. And with your type of applications they will become more evident than with the typical application an average user knows. But if you stop all services you don't need for your applications, and if you compile everything including your kernel with the same compilers using the same options and optimisations on the same hardware, using the same file system type, I'd be surprised if there would be any noticeable or measureable difference in speed.

Depending on the overall setup my experience from the past is, that the main difference is at application start, anyway. For example, up to Java 1.4.2 it helped me to compile my kernel to make Java apps and OpenOffice.org start a lot quicker. Once they were loaded, I felt no further difference. Current versions of Java and OOo are so fast, that I don't notice a relevant difference anymore between a stock kernel binary and a self-compiled kernel.

EDIT: Of course, multimedia apps sometimes run better ("smoother"), when compiled from source, too.

Finally, there are always people who just want to live on the edge. The Linux users among them choose Gentoo. I've tried it once, too, and yes, applications *can* run fast in Gentoo, as long as the system is not too busy with compiling something else. And the difference to a Slackware stock binary of the same program wasn't simply worth it.

But I understand your point of being curious, too. Just for curiosity I would read a benchmark, if there was one. But it wouldn't make me switch to another distro, just because it's faster in one benchmark. Slackware (and my other favourite distro, OpenSuSE) shine in so many respects, and especially Slackware has never disappointed me in responsiveness. These two just don't give me a reason to leave. Many others did.

gargamel

1 members found this post helpful.

For 14 years I've noticed a difference.

You got two classes of people. Sheep and Wolves. Sheep sleep and Wolves migrate.

It's hard to prove your applications load faster unless you hook up a camera. Then send the video up on Youtube. Post a link on the Kernel mailing list. Post a bug report then sit back and wait.

I'm a cat, actually. Meow !

I'm a Meerkat: I like things "Simples!" and every so often I pop my head in the air and have a good look around.