DragonFlyBSD 6.0 Is Performing Very Well Against Ubuntu Linux, FreeBSD 13.0

[phoronix]

Phoronix: DragonFlyBSD 6.0 Is Performing Very Well Against Ubuntu Linux, FreeBSD 13.0

Earlier this month in our initial benchmarking of DragonFlyBSD 6.0 we found DragonFlyBSD 6.0 performing much better than DragonFlyBSD 5.8, but how does that put its performance up against FreeBSD 13.0 and Ubuntu Linux for reference? Here are such benchmarks in our latest benchmarking of DragonFlyBSD 6.0, FreeBSD 13.0 (with both GCC and Clang), and Ubuntu Linux.

https://www.phoronix.com/vr.php?view=30208

[schmidtbag]

Kind of weird - where Ubuntu does better, BSD typically isn't too far behind. When BSD is ahead, it's pretty far ahead. In many of the tests were BSD won, I wouldn't consider the performance difference ignorable.

[Volta]

It seems schedutil governor still isn't the best choice for every workload, but for the real world ones it's quite decent.

https://www.phoronix.com/scan.php?page=article&item=amd-linux511-perfgov&num=1

[C8292]

Does dragonfly have all the cpu vulnerabilities mitigations?

[MadCatX]

Intel on schedutil? What am I missing here, aren't Intel chips supposed to use intel_pstate? Considering how all over the place the results are, there must be something fishy going on with either frequency scaling on task scheduling. The x265 is quite telling, 1080p is lead by Ubuntu but in 4K Ubuntu trails behind by a lot? EDIT: It's actually the other way around with DfBSD slightly edging out Ubuntu in 4K.I wonder what would the results look like on a Ryzen.

[ayumu]

I wonder what would the results look like on a Ryzen.

Considering Matt and the team personally use and are fond of Ryzen CPUs, I'd say probably better.

[Steffo]

I have never expected this! Especially when I look at the man power and the benchmarks in the last years! Maybe I give DragonFlyBSD a shot!

[ayumu]

I have never expected this! Especially when I look at the man power and the benchmarks in the last years! Maybe I give DragonFlyBSD a shot!

The superior performance despite difference in manpower is easily explained: Superior architecture.

Linux has hit a wall, and won't magically get any faster overnight, no matter how much manpower is thrown into it.

Dragonfly absolutely has plenty of headroom. Imagine if the development team grew at all.

[Steffo]

The superior performance despite difference in manpower is easily explained: Superior architecture.

Linux has hit a wall, and won't magically get any faster overnight, no matter how much manpower is thrown into it.

Dragonfly absolutely has plenty of headroom. Imagine if the development team grew at all.

Why has DragonFly a superior architektur?

[ayumu]

Why has DragonFly a superior architektur?

Because FreeBSD chose to copy Linux in their approach to SMP (fine-grained locks).

Matt used to be FreeBSD's technical lead, when FreeBSD was actually outperforming Linux, around 20 years ago. There was quite the argument as he'd reverted/rejected patches that simply went ahead with Linux's approach before a technical discussion on the matter could be held.

It ended with Matt being ejected from FreeBSD. Some devs who also disagreed with the approach followed him to Dragonfly, which forked from FreeBSD.

Slowly, with a small team, Matt's alternative approach, with a focus on low IPC cost and implementing lockfree/lockless system servers instead of locks, got implemented, furthered and to the point it is today, where Linux (and FreeBSD) are actually behind.

Thus Matt's point can now be considered proven.