I ran the in game benchmark for Cyberpunk 2077 giving Linux Mint a try instead of Artix.

Game From: GOG
Game Version: 2.13
Wine Version: 10.0-rc1
HGL Version: 2.15.2
Game Settings: Highest, AMD Fidelity 3.0 set to quality, 16x AF, motion blur/DOF/Film Grain off
Resolution: 2560x1440


Min FPS: 70.74
AVG FPS: 101.28
Max FPS: 130.24


My system is:

Intel i5-12600K | 32GB DDR4-3200 CL16 | Gigabyte RX 7800 XT 16GB | Western Digital Black SN850X 2TB | Linux Mint 22 | Dasharo 1.1.1 | Mate 1.26.1 | Kernel 6.8.0-49-generic | Mesa 24.0.9-0ubuntu0.2


I'm at a loss to explain the difference between Artix and Mint. I don't think thew newer Wine would make thta big of a difference.

Your both benchmarks seem kinda low for the GPU you have. Maybe CPU limited?
or is that with Raytracing?

I have much weaker GPU for example, but get better results.

Steam, proton-experimental

Game Settings: Highest, DLSS quality, RT off, 16x AF, motion blur/DOF/Film Grain off
Resolution: 2560x1440

Min FPS: 67.91
AVG FPS: 117.18
Max FPS: 238.87

edit: first run min and max is a bit strange. second run



Ryzen 7 5800X3D | 32GB DDR4-3200 CL14 | Gigabyte RTX 3080 OC 10GB | Crucial 2TB SSD | Arch Linux | Mate 1.28.2 | Kernel 6.12.2-xanmod | nvidia 565.77

Last edited by Xpander on 8 December 2024 at 7:44 pm UTC

Quoting: XpanderYour both benchmarks seem kinda low for the GPU you have. Maybe CPU limited?
or is that with Raytracing?

I have much weaker GPU for example, but get better results.

Steam, proton-experimental

Game Settings: Highest, DLSS quality, RT off, 16x AF, motion blur/DOF/Film Grain off
Resolution: 2560x1440

Min FPS: 67.91
AVG FPS: 117.18
Max FPS: 238.87

edit: first run min and max is a bit strange. second run

To protect your privacy, external media requires approval to load. Source: i.imgur.com
View cookie preferences.
Accept & Show Accept All & Don't show this again   Direct Link

Ryzen 7 5800X3D | 32GB DDR4-3200 CL14 | Gigabyte RTX 3080 OC 10GB | Crucial 2TB SSD | Arch Linux | Mate 1.28.2 | Kernel 6.12.2-xanmod | nvidia 565.77

Hard to say. It is with Ray Tracing turned off. My results with it turned on were 7.44/8.94/11.6 for the min/avg/max so basically a slide show. I will rerun it using Proton instead of straight Wine to see if that is the cause. I can't say if it is CPU limited, am looking to upgrade to a i7 or i9 14th gen as i think this board will support it but no idea on time frame for that.

Quoting: XpanderYour both benchmarks seem kinda low for the GPU you have. Maybe CPU limited?
or is that with Raytracing?

I have much weaker GPU for example, but get better results.

Steam, proton-experimental

Game Settings: Highest, DLSS quality, RT off, 16x AF, motion blur/DOF/Film Grain off
Resolution: 2560x1440

Min FPS: 67.91
AVG FPS: 117.18
Max FPS: 238.87

edit: first run min and max is a bit strange. second run

To protect your privacy, external media requires approval to load. Source: i.imgur.com
View cookie preferences.
Accept & Show Accept All & Don't show this again   Direct Link

Ryzen 7 5800X3D | 32GB DDR4-3200 CL14 | Gigabyte RTX 3080 OC 10GB | Crucial 2TB SSD | Arch Linux | Mate 1.28.2 | Kernel 6.12.2-xanmod | nvidia 565.77

So I did a couple of tests. The first thing I did was switch from straight Wine to using Proton Experimental. By doing that my result was 84.3/110.83/144.36. Better than what I had gotten but still shy of your score. Next I turned on FSR3 which I had not turned on previously. This upped my result to 139.79/180.15/227 which is a huge improvement. Frame Tracing maxed out is still low at 22.95/27.12/34.20 but even that is leagues better than the slide show it was.

This is why I post stuff like this here, you never know what is normal or not until we all pool our results. Thanks for pointing out my low results so I could fix them.

Last edited by PublicNuisance on 9 December 2024 at 3:36 am UTC

A couple of questions about your wine set up:

* Are you using vkd3d-proton + dxvk?
* Are you using esync?

Both are very necessary for good performance.

My bet - you missed to enable esync. Without it, the game gets a big performance hit.

You can tell if the game is GPU bound or not by observing your GPU utilization. It should be 100% or close to it. If it's a lot lower - something is wrong with your set up.

Last edited by Shmerl on 9 December 2024 at 3:57 am UTC

Quoting: PublicNuisanceSo I did a couple of tests. The first thing I did was switch from straight Wine to using Proton Experimental. By doing that my result was 84.3/110.83/144.36. Better than what I had gotten but still shy of your score. Next I turned on FSR3 which I had not turned on previously. This upped my result to 139.79/180.15/227 which is a huge improvement. Frame Tracing maxed out is still low at 22.95/27.12/34.20 but even that is leagues better than the slide show it was.

This is why I post stuff like this here, you never know what is normal or not until we all pool our results. Thanks for pointing out my low results so I could fix them.

That seems more in line for your GPU yeah.

by FSR3 you mean framegen? 180 sounds a bit too high, unless its balanced or performance FSR upscaler i guess. But i would expect around ~130 from your GPU with FSR quality with this resolution. And yeah raytracing performance on AMD is pretty bad. Its not great on nvidia either though. i get around ~62 fps with Raytracing and 28 fps with Path Tracing :)

Last edited by Xpander on 9 December 2024 at 6:00 am UTC

It's a shame ntsync takes ages to get upstreamed.

Last edited by Shmerl on 9 December 2024 at 5:58 am UTC

Quoting: ShmerlA couple of questions about your wine set up:

* Are you using vkd3d-proton + dxvk?
* Are you using esync?

Both are very necessary for good performance.

My bet - you missed to enable esync. Without it, the game gets a big performance hit.

You can tell if the game is GPU bound or not by observing your GPU utilization. It should be 100% or close to it. If it's a lot lower - something is wrong with your set up.

Esync was enabled the whole time. It's one of the reasons I like using HGL for Windows games is it is enabled by default. I checked and it was. As for the GPU utilization it was at 99 or 100% most of the time.

It's interesting that difference is so big then. I've just seen that ntsync should help even more. I'm going to test it once it's upstreamed.