LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Slackware (http://www.linuxquestions.org/questions/slackware-14/)
-   -   Phoronix' benchmarks (http://www.linuxquestions.org/questions/slackware-14/phoronix-benchmarks-4175419682/)

s1aw0 08-01-2012 01:39 AM

Phoronix' benchmarks
 
Hello everybody,

Phoronix has published results of their wide comparison of performance of a number of distributions, including Slack-14.0 beta: http://www.phoronix.com/scan.php?pag...xdistros&num=1 Sadly, Slack doesn't seem to be among the fastest in their tests. But I believe the tests performed say little to a rank-and-file slacker. Can anyone please comment on the results and possibly suggest how to improve performance of Slack if it is really that modest (though personally I think it's great but I haven't tried anything but Arch from the list of distros tested).

nixblog 08-01-2012 06:32 AM

Saw this article earlier and Slackware is looking sad against the others in that test but then, it's only a beta and not the final release. Also, I would rather have reliability over speed any day but speed is also nice too. Good to see Arch perform well, as it usually does so no surprise there. To be honest, I was quite impressed by the figures that CentOS returned in this test too.

H_TeXMeX_H 08-01-2012 08:29 AM

I don't trust Phoronix benchmarks. I have tried them myself and they are not accurate in some cases (I benchmarked filesystem performance). I am going to test the final Slackware 14 and maybe Arch and see if the benchmarks are true. I just have to remember to do it.

mrascii 08-01-2012 09:34 AM

I didn't see in the article which Slackware kernel was being benchmarked. That could make a big difference in the results and like H_TeXMeX_H mentioned benchmarks don't always relate to real world performance.

DNA
AKA mrascii

willysr 08-01-2012 09:34 AM

please remember that those benchmarks are hardware dependent, so it may have different results based on whose hardware are being used despite the OS being used.

Martinus2u 08-01-2012 02:42 PM

Huh? differences in Dhrystone can only be attributed to different compilers or different hardware. Obviously the 5 operating systems were run on 5 different machines, and obviously those machines had very different performance despite looking similar on paper. shame on phoronix.

nixblog 08-01-2012 04:20 PM

Perhaps a set of real world tests based on a VM install would be better as a comparison?

adriv 08-01-2012 05:56 PM

The benchmarks were performed by SW 14 on XFCE. I know that I am stepping on some toes here, but in my experience XFCE (on SW) is slower than KDE. Sure, the start-up of the Desktop is faster, but after that...

273 08-01-2012 06:01 PM

Hold on, I just realised this, they were using three systems for the test? I had assumed they installed each distro fresh on the same machine and carried out the tests that way. As it stands I know now Phoronix benchmarks aren't worth my time reading.

nixblog 08-01-2012 06:09 PM

Quote:

Originally Posted by adriv (Post 4743551)
The benchmarks were performed by SW 14 on XFCE. I know that I am stepping on some toes here, but in my experience XFCE (on SW) is slower than KDE. Sure, the start-up of the Desktop is faster, but after that...

Not sure about XFCE on Slackware but I would say the it's a good bit faster than KDE based on my experience.

adriv 08-01-2012 06:15 PM

Quote:

Originally Posted by nixblog (Post 4743556)
Not sure about XFCE on Slackware but I would say the it's a good bit faster than KDE based on my experience.

That's certainly true for the distro's that I've tried, except Slackware (YMMV).
In fact, I've always found XFCE's speed disappointing on SW.

the3dfxdude 08-01-2012 06:19 PM

Quote:

Originally Posted by 273 (Post 4743554)
Hold on, I just realised this, they were using three systems for the test? I had assumed they installed each distro fresh on the same machine and carried out the tests that way. As it stands I know now Phoronix benchmarks aren't worth my time reading.

I couldn't figure out if this were the case that they were different machines, but it looked like so to me. But people are echo'ing it. I would call these comparisons bogus. This isn't the first time I've seen bogus comparisons from Phoronix.

Even if something is found, I don't consider what Phoronix does particularly interesting since it most of the benchmarks I could care less about in a real world sense.

273 08-01-2012 06:29 PM

Looking at the image of system specifications it is at least two different systems (e.g. different sound cards?), making the comparison bogus and the site suspect.

nixblog 08-01-2012 06:42 PM

Perhaps someone wants to come up with a series of real world tests here like opening large files, batch processes and say image manipulation in GIMP etc.. Stuff that most ordinary users and admins may perfom on a daily basis.

Also try to narrow down the base software so at least they may have the same DE and core apps involved in testing plus, absolutely stock installs - not tweaking kernels or stuff like that. Then, either test them as virtuals or use the same hardware (as stated).

A good test choice might well be Slackware (KDE), Chakra Linux (Arch based KDE), Kubuntu, openSUSE and PCBSD (FreeBSD with KDE) as a wildcard - preferably with the same release of KDE too.

disturbed1 08-01-2012 06:43 PM

Quote:

Originally Posted by nixblog (Post 4743497)
Perhaps a set of real world tests based on a VM install would be better as a comparison?

I have never been in favor of Phoronix's "Distro" nor OS (Mac/Win) benchmarks. It only proves that the settings he chose give those results with that hardware using that specific application.

Last time I used their benchmarking application, I noticed it did not always use system applications. Like with ogg-vorbis, it compiled and used it's own version of ogg-vorbis. So in essence, this only compares the differences between the kernel and tool chain options. And to be honest, there should be little to no differences between distro's that use the same versions. If there is -- something is drastically wrong. Either with the installation, test matter/execution, PEBKAC, or the distro royally fubar'd something up.

Benchmarks can be useful for some people, for some things.
Like benchmarking your own PC to see if that recently upgraded GPU driver made a difference and gave you an extra 5FPS.
Benchmarking the difference between the vorbis stack compiled at stock settings, and -O3 -march=native -mtune=native.

I can admit from my own personal experience -- it's takes the same amount of time for oggenc to encode 60minutes on Arch Linux and Slackware, using the exact same PC. vpxenc and x264 also encode at the same fps on Arch Linux and Slackware.


All times are GMT -5. The time now is 08:26 PM.