EndeavourOS in Phoronix test

Lounge EndeavourOS pub
1

out-of-the-box performance of different modern Linux distributions when running the new Intel Raptor Lake processors, here are six different distributions running on the current flagship Core i9 13900K processor. Tested this round was CentOS Stream 9, Clear Linux, Debian Bookworm (Testing), EndeavourOS, Fedora Workstation 37, and Ubuntu 22.10https://www.phoronix.com/review/linux-2022-raptorlake

The Core i9 13900K test system used features the ASUS PRIME Z790-P WIFI motherboard, 2 x 16GB DDR5-6000 memory, WD_BLACK SN850 1TB NVMe SSD, and Radeon RX 6800 XT graphics.

2 Likes
2

Idk how they calculated the geometric mean but somehow we are last despite performing well on many tests

3

I don’t think it’ll matter much, we’re like the place to be these days regardless.

2 Likes
4

“Intel’s Clear Linux” That a shock… NOT!! :rofl: :wink:

(Time tell )

5

Usually Arch does absolutely awful on Phoronix tests, this time it does well but is still Ranked last? lol :joy:

6

I think that the LITTLE.big approach from Intel is not yet full working on Linux an in there Clear Linux Distro, there are for sure Kernel Patches, that we will see later in official Kernel and that performance difference will shrink to measurement tolerances.

For the Moment I use an Zen2 CPU but for Upgrade in the mid-range I would switch to Intel (my perception at this moment), because of that I observe the development in that sense.

7

Well…And :enos: is meant to do easy Arch install, not configure it to be fastest Arch ever existed.

In some of the workloads, Arch Linux based EndeavourOS did win by a narrow margin… But overall wasn’t as performant as many like to talk up Arch Linux as a greater power.

That’s very bad…Arch will take a blame because of :enos: now :rofl:

image

4 Likes
8

Seemingly, Clear Linux does rather well on AMD too. Put this here just for the record :wink: :slightly_smiling_face:

TL;DR >> jump to page 6

9

comments more interesting… :wink: :blush:

10

I’M OVERLY ANGRY FOR NO REASON!!! !

Looks like this sums up the whole comment field :sweat_smile:

1 Like
11

It is as always with Phoronix tests: they compare Apples with Pears (as we say in German :wink:)

Putting the winner Clear Linux aside because it is specifically tuned for performance, which obviously turns out to be successful, all the other distros have equal performance when you consider error distribution. From the slowest to fastest it is just 3,5 % difference. And this difference is within the expected error distribution.

And by the way, concerning the apples and pears metapher, phoronix is bascially comparing different kernel versions and different library versions (glibc, etc.). You can easily achieve a 3,5 % performance difference if you compare the current endeavousos versions with the previous one.

Tests like this on phoronix are completely useless.

6 Likes
12

In that case, couldn’t one argue that comparing distributions is completely useless at the first place since they will always be differing in the aspects you mention?

13

Yes, exactly. It can make sense to compare the performance of different kernels or different libraries but within a given distro. But comparing the performance of distros is meaningless because it depends on too many variables and the result can change every day. Especially when you look at rolling release distros.

Look at the different kernels in that phoronix test:

5.14.0-183.el9.x86_64 (x86_64)
6.0.7-1207.native (x86_64)
6.0.0-2-amd64 (x86_64)
6.0.8-arch1-1 (x86_64)
6.0.7-301.fc37.x86_64 (x86_64)

and filesystems were dependding on the distor either: ext4, xfs or btrfs.

and gnome desktop version was different per distro also.

and not only that. Even the CPU was different per distro: some where at 4.00 GHzs others at 5.50 GHz (e.g. Clear Linux)

See yourself:
https://openbenchmarking.org/embed.php?i=2211153-NE-RAPTOR40362&sha=0e17a2c1d45c&p=2

Publishing such a benchmark is rediculous.

2 Likes
14

Well, overall EOS did pretty well here! Besides that I’m surprised by Debian Bookworm (what a name :smile:) and a bit disapointed by Fedora.

15

I’ve got 4 different kernels on my machine at the moment and I get like up to 10fps difference between them without picom running or up to 40fps difference with picom running using a quick vkmark test. So these results aren’t too surprising really.

But I guess they said it is “out of the box performance” they are testing. shrug

16

Mr. Phoronix got bored of setting up Arch so they went for EnOS :grin:

17

Weird part is the EOS did better in this round of testing than everything but Clear linux yet was STILL ranked last?

How is geometric mean even calculated then?

18

Is there any other way?

19

That’s what I’m wondering because the graph in the article confuses me :joy:

Does second best across benchmarks, dead last on Geo mean

Maybe I don’t understand geometric mean, but doing better across the entire data set shouldn’t put you in last place.

20

Well, as far as i understand data you’re not correct in assumption that :enos: is second best for all data sets:
First Place Finishes :enos: 53 / 210
Last Place Finishes :enos: 23 / 210
Rest of tests :enos: 134 / 210 - most of those are either lower than middle, or near last

So it’s very volatile results on the ends of spectrum, and for most tests it leans towards worse results, hence geometric mean (one of the ways to calculate sensible average / overall trend) shows :enos: last…

It’s kinda smooths out extremes, if it makes sense.