Windows 10 Build 21370 vs. Ubuntu 21.04 Linux On AMD Ryzen 5900X

[phoronix]

Phoronix: Windows 10 Build 21370 vs. Ubuntu 21.04 Linux On AMD Ryzen 5900X

Last month when carrying out tests of Windows 10 vs. Linux on the Intel Core i9 11900K "Rocket Lake" processor we were very surprised to see Windows 10 frankly performing so well compared to Ubuntu and picking up more wins than usual. That unexpectedly strong showing for Windows 10 might be due to Intel's P-State behavior with Rocket Lake or other power management tuning or there the lack of on Linux at this time. But it led me to wondering if the latest Windows 10 updates spelled out anything different on the AMD Ryzen side... So here are some benchmarks of the latest Microsoft Windows 10 against Ubuntu 21.04 on the same AMD Ryzen 9 5900X system.

https://www.phoronix.com/vr.php?view=30172

[ddriver]

Is there any merit to including the firefox tests? Since the implementations themselves vary to such an extent, it defeats the purpose of cpu testing - as it doesn't reflect it whatsoever. Besides, are people still that interested in browser performance in this day and age, and if so, why is it such a critical matter in a workflow that is ultimately bottlenecked by human interaction.

[birdie]

1. I'd like to see power consumption and CPU frequency metrics for each test.
2. I'd like to see applications being tested compiled the same way (the same compiler and compilation flags)
3. I'd like to be sure that Linux and Windows use the same or similar power profiles
4. I'd like to be sure Windows doesn't perform upgrades/defragmentation/virus scanning or any other background tasks. Linux is easier to control.

Otherwise this comparison is ... moot and drawing any conclusions from it feels wrong.

[schmidtbag]

C'mon Firefox, you're supposed to be the better choice for open source!

[ustolemyname]

1. I'd like to see power consumption and CPU frequency metrics for each test.
2. I'd like to see applications being tested compiled the same way (the same compiler and compilation flags)
3. I'd like to be sure that Linux and Windows use the same or similar power profiles
4. I'd like to be sure Windows doesn't perform upgrades/defragmentation/virus scanning or any other background tasks. Linux is easier to control.

Otherwise this comparison is ... moot and drawing any conclusions from it feels wrong.

It's not a measure of what's possible by tuning each system, it's a measure of what users will experience out of the box.

Which is how most users will experience their systems. Few users are turning off their virus scanner to make their web browser faster...

[andre30correia]

firefox is desilution in linux, not a new storybut with google stopping sync services in chromium and others... well I use chrome and firefox with ubuntu and now i use edge to develop reason but I can say some things work much better with chrome

[HEL88]

Linux is a bit faster but:

Linus Torvalds on why desktop Linux sucks - YouTube

Even Linus not delivery your app for Linux but only for Windows and Mac.

[pmorph]

It's not a measure of what's possible by tuning each system, it's a measure of what users will experience out of the box.

Which is how most users will experience their systems. Few users are turning off their virus scanner to make their web browser faster...

I'd think they are not very useful in even that. If this kind of tests have value, it's mostly in noticing unexpected problems that need fixing.

[skeevy420]

Linux is a bit faster but:

Linus Torvalds on why desktop Linux sucks - YouTube

Even Linus not delivery your app for Linux but only for Windows and Mac.

The part about building packages for Debian Stable had me ROFL

[skeevy420]

firefox is desilution in linux, not a new storybut with google stopping sync services in chromium and others... well I use chrome and firefox with ubuntu and now i use edge to develop reason but I can say some things work much better with chrome

I've been a long time Firefox user but I use Edge on Windows and Linux for the sites that don't work on Firefox. I was expecting ClippIE, but, surprisingly, Edge doesn't suck.