Windows 10 vs. Ubuntu 19.10 vs. Clear Linux vs. Debian 10.1 Benchmarks On An Intel Core i9

[rabcor]

Man, computer performance is such a mess,, one different release of an OS to the next can completely alter the landscape and one program may work better on one os while another works better on different os... Why can't this shit be more simple.

[stormcrow]

Why is Ubuntu ranking so much lower on these and previous benchmarks? Would Ubuntu win, on the other hand, in things like interactive responsiveness?

Interactive responsiveness is notoriously difficult to quantify, and I wouldn't bet on Ubuntu wining on them either. One program can be fine and dandy and quickly responsive while another can be dog slow on the same hardware and platform. Gonna enter into flame bait territory here, but I've yet to see any Gnome 3 based desktop that I could consider "snappy". And yes that's just my own opinion, but dissent is almost guaranteed to prove just how subjective what acceptable input lag can be.

[Linuxxx]

Interactive responsiveness is notoriously difficult to quantify, and I wouldn't bet on Ubuntu wining on them either. One program can be fine and dandy and quickly responsive while another can be dog slow on the same hardware and platform. Gonna enter into flame bait territory here, but I've yet to see any Gnome 3 based desktop that I could consider "snappy". And yes that's just my own opinion, but dissent is almost guaranteed to prove just how subjective what acceptable input lag can be.

Ubuntu is actually your best bet when it comes to interactive responsiveness:

No other distro gives one the option of easily installing a "lowlatency" kernel!

Give it a try; it really does make a difference.

[andyprough]

Ubuntu is actually your best bet when it comes to interactive responsiveness:

No other distro gives one the option of easily installing a "lowlatency" kernel!

Give it a try; it really does make a difference.

Depends on your definition of easy. I've got multiple repositories I can add to any Debian-based distro for a low-latency kernel in about 4 or 5 commands. Seems pretty easy to me.

[mgtelve]

Not surprised by these results. We ran about a month of internal tests on home-spun benchmarks for matrix work - mainly performance with a little usability - and found Debian 10 to be the winner by quite a bit. Debian 10 even beats headless Arch installs on pure math runs. This was a shock to me, as I had spent nearly two years in the Clear Linux ecosystem and knew the internals pretty well.

Clear Linux SOUNDS good on paper, but the optimizations it often advertises are only for the packages in its "bundle" system. And the system has a hard cap on the number of bundles offered. This is the reason that Clear Linux devs will reject requests for math libraries that are clear candidates for optimization (pun intended!), instead telling you to "go get the Flatpak".

Anyone who has used Fedora SilverBlue probably saw the same pattern. The reason isn't because the devs are arrogant jerks. Its because "performance" and "optimization" were never goals of the OS. The goal was containerization and infrastructure. Meaning, the smaller their OS install image is, both real and potential, the easier the image is replicated across instantiations. Same goes for the "stateless" philosophy. The OS image is everything you need.

TLDR: Seekers of the modern Gentoo should stay away from Clear Linux, and stick to Debian. Consider that Michael's results were obtained with the generic kernel and glibc, both of which can be re-compiled and optimized for decimal dust levels of gain.