DragonFlyBSD vs. FreeBSD vs. Ubuntu 20.04 On Intel's Core i9 10900K Comet Lake

[phoronix]

Phoronix: DragonFlyBSD vs. FreeBSD vs. Ubuntu 20.04 On Intel's Core i9 10900K Comet Lake

One of the areas where Intel has the leg up over AMD when it comes to open-source software support is on the BSD side where generally the likes of FreeBSD and DragonFlyBSD often exhibit better out-of-the-box support at launch. Here is a look at how DragonFlyBSD and FreeBSD are running on the Core i9 10900K "Comet Lake" processor with Z490 motherboard.

http://www.phoronix.com/vr.php?view=29301

[andyprough]

Nice, found some systems that Ubuntu can actually beat. Folks at Canonical will be a bit happier today.

[Tuxee]

Nice, found some systems that Ubuntu can actually beat. Folks at Canonical will be a bit happier today.

Is this Canonical bashing habitual or a pathologic? It can't be "facts" since Ubuntu is pretty much in line with other distros: https://www.phoronix.com/scan.php?pa...g-2020-distros

[Steffo]

That's sad. DragonFlyBSD has been forked from FreeBSD, because it wanted to be faster.

[barti_ddu]

That's sad. DragonFlyBSD has been forked from FreeBSD, because it wanted to be faster.

IMHO, its killer feature is HAMMER (no pun intended) and it is highly interconnected with the kernel (i.e. would be impossible as FreeBSD component).

[Steffo]

IMHO, its killer feature is HAMMER (no pun intended) and it is highly interconnected with the kernel (i.e. would be impossible as FreeBSD component).

A killer feature that nobody uses, which is even more sad.
And what means impossible? DragonFly BSD is a fork of FreeBSD. It is surely possible with a certain amount of work.

[barti_ddu]

A killer feature that nobody uses, which is even more sad.

But nevertheless it is much better than it wouldn't exist at all. Even as an experiment.

And what means impossible? DragonFly BSD is a fork of FreeBSD. It is surely possible with a certain amount of work

Sure, everything is theoretically possible; however, I doubt that it will ever happen in practice: even re-writing the VFS layer would be not an easy undertaking (both technically and politically), but there are more core differences (e. g. kernel thread locking mechanism).

[JustinTurdeau]

The reason FreeBSD/Clang beats everything else in some of the GraphicksMagick tests is probably due to the fact that Clang enables auto-vectorization at -O2 whereas GCC doesn't. Comparing Clang and GCC at the same -O level is never an apples-to-apples comparison, as the Clang developers include extra -f* options at -O2, in order to make their compiler seem faster than it really is. As usual, Michael just throws some random shit at the wall and thinks it's demonstrating some real performance difference.

[zexelon]

The reason FreeBSD/Clang beats everything else in some of the GraphicksMagick tests is probably due to the fact that Clang enables auto-vectorization at -O2 whereas GCC doesn't. Comparing Clang and GCC at the same -O level is never an apples-to-apples comparison, as the Clang developers include extra -f* options at -O2, in order to make their compiler seem faster than it really is. As usual, Michael just throws some random shit at the wall and thinks it's demonstrating some real performance difference.

That is an unfortunate lack of observation. The goal from what I understand is to test out of the box performance. Now if you can send me your site where you do a better job of detailed analysis I would be very interested in comparing if you do a better job.

[willbprog177]

Wow, the Ubuntu VS FreeBSD 12.1 + GCC9 difference for LAME encoding is crazy! In my mind, FreeBSD was always the faster OS compared to Linux, but I guess times have changed and my personal experience is outdated now.