AMD Ryzen 7 5800X3D - Windows 11 vs. Ubuntu 22.04 LTS Linux Benchmarks

[phoronix]

Phoronix: AMD Ryzen 7 5800X3D - Windows 11 vs. Ubuntu 22.04 LTS Linux Benchmarks

With the AMD Ryzen 7 5800X3D as the first consumer processor with AMD 3D V-Cache technology that launched earlier this year, the Linux performance has been fantastic for a variety of workloads especially in areas of technical computing and other non-gaming workloads -- similar to the great results we've enjoyed with AMD EPYC Milan-X processors too. One of the areas I hadn't had a chance to look at until recently was how the Windows 11 vs. Linux performance is looking for this Ryzen CPU with its 96MB L3 cache. Here are those quick benchmarks.

https://www.phoronix.com/review/5800x3d-windows11-linux

[Volta]

Nothing surprising here. Slaughter.

[gnwiii]

It would be nice to have power consumption stats to go with benchmarks. My subjective impression is that Windows runs hotter for the same computation (Fortran, same gcc version on both platforms), but lower performance could come from different power/performance tradeoffs.

[Linuxxx]

Nothing surprising here. Slaughter.

All those victories while crippling the CPU with the broken-by-design schedutil governor on Linux!

Michael probably thought the bloodbath with the performance governor would be too much to endure for Windows users...

[birdie]

I wonder if Michael has enabled any of the optimizations (a list for W10) required to carry out benchmarks on Windows.

Windows runs a ton of random crap in background which may roally skew the results.

[Volta]

I wonder if Michael has enabled any of the required to carry out benchmarks on Windows.

Windows runs a ton of random crap in background which may roally skew the results.

It's obvious he didn't, because defaults are tested. The same rule can be applied to Ubuntu which configuration is far from optimal.

[Volta]

All those victories while crippling the CPU with the broken-by-design schedutil governor on Linux!

Michael probably thought the bloodbath with the performance governor would be too much to endure for Windows users...

It's amazing this POS governor is default one. It was broken for years and it still sucks. It even makes games like Terraria unplayable. What a mess.

[Michael]

It would be nice to have power consumption stats to go with benchmarks. My subjective impression is that Windows runs hotter for the same computation (Fortran, same gcc version on both platforms), but lower performance could come from different power/performance tradeoffs.

Unfortunately not possible for lack of PHP PCNTL extension on Windows for being able to concurrently and accurately poll the power data while running benchmarks.

[CochainComplex]

I wonder if Michael has enabled any of the optimizations (a list for W10) required to carry out benchmarks on Windows.

Windows runs a ton of random crap in background which may roally skew the results.

Then it is not the out-of-the-box benchmark. Setting the cpu gov to performance or windows to "game mode" would be considered as fair.
But looping through all regedit win 11 tweaks is a bit beyond the default. (Although those benchmark results would be very interessting)

If windows has a lot of crap running by default than it shall slow down its performance. Thats how they have delivered and that is how it is supposed to be used.

[Linuxxx]

It's amazing this POS governor is default one. It was broken for years and it still sucks. It even makes games like Terraria unplayable. What a mess.

That's exactly the reason I ask myself why Valve made it the default on SteamOS, with no easy option to switch over to the performance governor via GUI.

If it stays that way once the official SteamOS ISO is released to the public, expect alot of YouTube videos where people will claim that SteamOS performs worse & has more stuttering than Windows...

Hopefully Valve atleast installs Feral's GameMode by default!