Benchmarks: FreeBSD 13 vs. NetBSD 9.2 vs. OpenBSD 7 vs. DragonFlyBSD 6 vs. Linux

[JackLilhammers]

A lot of Michael's benchmarks are eg Python related. This is a programming language that lacks several basic functions (https://medium.com/nerd-for-tech/pyt...ge-2ab73b0bda5) Python is less productive than Lisp, and has much lower performance than Lisp. If we look at how fast AI is growing, we can see that Python can no longer be used in this domain, since it is much slower than Lisp and Julia eg. You're just burning the earth if you continue to use Python for extremely energy-intensive things like AI and data science. A data center will draw significantly more power, and will therefore also need much more energy to stay cool. Python has to stop, or it will stop with humans..

A lot of stuff is done with Python, so it makes sense to have a lot of benchmarks.
I won't argue with the post you linked, because I think it's a matter of tastes and opinions, and so is productivity to some extent.

That said, you're missing the point about Python's speed.
CPython is slow. Mainly because it doesn't have any integrated JIT compilation capabilities.
This should change in the next years, but at the moment nobody disputes that.
However, when you're doing computationally intensive stuff you're not using it.

All scientific tools in Python are essentially wrappers around C/C++ code.
This way users can have strong performance in a beginner friendly ecosystem.
So, no, Python is not burning Earth...
(and in a way it is already not being used for the tasks you don't want it used for)

[looserouting]

IIRC Zstd on DFBSD is not multithreaded. but you can turn it on.

[trasz]

Sadly, this is yet another Phoronix benchmark which is obviously miscompiled (see zstd runs) and thus useless.

[Michael]

Sadly, this is yet another Phoronix benchmark which is obviously miscompiled (see zstd runs) and thus useless.

As has been discussed already, the Zstd tests for this comparison were the system/zstd test profile making use of the zstd binary shipped by the OS itself, not compiled by PTS.

[trasz]

>As has been discussed already, the Zstd tests for this comparison were the system/zstd test profile making use of the zstd binary shipped by the OS itself, not compiled by PTS.

Honestly, this doesn't matter. It's not important if you built it incorrectly (this problem applies to all Phoronix tests), or if you benchmarked it incorrectly (again, you fail to provide even just error bars for a start), or even if you measured something else than you think you did. It's a systemic problem with LTS: you just don't understand what you are measuring.

[looserouting]

I appreciate Micheals work. I wasn't complaining at all. I just thought that it should be noted that zstd can run much faster on DFBSD so one can fully understand the results. By the way, it's not the way it was build. You have to use the -T switch. see https://lists.dragonflybsd.org/piper...il/770385.html. I've never tested it. I juste remembered that I read something about that.

[Michael]

I appreciate Micheals work. I wasn't complaining at all. I just thought that it should be noted that zstd can run much faster on DFBSD so one can fully understand the results. By the way, it's not the way it was build. You have to use the -T switch. see https://lists.dragonflybsd.org/piper...il/770385.html. I've never tested it. I juste remembered that I read something about that.

(Yes the -T switch is used across all platforms https://openbenchmarking.org/innhold...421cfce311432e )

[Classical]

So, you changed rhetoric, because your original point got burned? KVM is superior in performance so your 'start up' benchmark (valid or not) is POS like the rest of your comment. KVM is superior in memory footprint, but I bet you have no idea why it's like that.

I didn't change rhetoric. I nuanced which performance I was talking about. Start and stop times are definitely in the 'performance' segment. And KVM is not competitive in this area. In terms of memory usage, the reddit link you provided shows that KVM uses more than twice as much memory as bhyve for the same task. So you can always explain your statement further, but it seems that KVM uses more than 2x as much memory than bhyve. Ultimately, most computer engineers use VMWare. Well here's a performance comparison of VMWare with bhyve: https://b3n.org/vmware-vs-bhyve-performance-comparison/

Even netflix storage runs on Linux and Linux PostgreSQL performance is superior to FreeBSD, so everything is as it should. Linux for serious business while BSD for few fanboys. Linux is more than 2x faster than FreeBSD in PostgreSQL:
https://www.phoronix.com/scan.php?pa...ebsd-bsd&num=5

If you had read my link from redbyte, you would have understood that read+write is the most realistic scenario for most databases. Linux is indeed faster in a 'read only' scenario, but this is less important because most databases are configured in read+write mode. The phoronix link you provide is indeed about this 'read only' mode. But FreeBSD is 2x faster than Linux in PostgreSQL in read+write mode.

Linux performance is superior to FreeBSD: https://matteocroce.medium.com/linux...g-cbadcdb15ddd

FreeBSD's network stack has always been faster than Linux's. Facebook has tried to make Linux more competitive in this area, but their efforts have had little effect. Even phoronix has admitted that FreeBSD has the fastest network stack: https://www.phoronix.com/scan.php?pa...sd-linux&num=3

And most of NginX is running on Linux just like Apache, so what's your point?

FreeBSD is faster than Linux in the two most popular web servers and thus is the best operating system for the cloud (FreeBSD also has a more stable implementation of ZFS). In the case of NginX (the best web server out there), FreeBSD is 4 to 5 times faster. Here is the explanation why this is so:

https://long-zhou.github.io/2012/12/21/epoll-vs-kqueue.html

Since FreeBSD 1) is usually slightly more secure than Linux 2) has the highest performance in NginX and 3) has the most stable ZFS implementation, we can conclude that FreeBSD is unquestionably a higher-up operating system for the cloud compared to Linux.

FreeBSD also has important advantages over Linux for regular desktop users. Most people I know like to listen to music and also in this area is Linux inferior:

https://linuxreviews.org/OSS

http://ossnext.trueinstruments.com/forum/viewtopic.php?t=5811

https://www.audiocircle.com/index.php?topic=95824.0

If you have any further questions about these topics, you can always ask me for a more in-depth explanation. I think it's also helpful to explain why Linux has been more popular than FreeBSD for years. Popularity of a topic usually has little or nothing to do with how good something is. For example, think of McDonald. It is the most popular fast food chain, but a survey found that their products have the lowest quality of literally all fast food chains tested.

[trasz]

As just reported by [email protected] on feeebsd-current@, the numbers for zstd, the results for zstd are plainly bogus: there is no performance problem there, just broken reporting.

This kind of proves my point.

[abu_shawarib]

Python is less productive than Lisp, and has much lower performance than Lisp. If we look at how fast AI is growing, we can see that Python can no longer be used in this domain, since it is much slower than Lisp and Julia eg. You're just burning the earth if you continue to use Python for extremely energy-intensive things like AI and data science. A data center will draw significantly more power, and will therefore also need much more energy to stay cool. Python has to stop, or it will stop with humans..

I've seen many ignorants spreading this misinformation and no matter how many times its corrected they just keep coming back.
First of all, most of AI is run on domain specific hardware that don't accept general purpose programming languages, which means that there's a library that compiles special code optimized for that hardware. Second, Python is only a front end to lower level libraries and doesn't actually do any heavy lifting by itself, it's all done in complied c/c++ library. Python being slow or fast have marginal effect on performance.