It doesn’t work natively on Windows without stressing even the highest end machines. I guarantee if you try to emulate it, the result will be even worse than the already terrible native experience.
API call translation is often very inexpensive and, particularly in the case of DXVK for graphics calls, sometimes actually results in faster code if the underlying API implementation is more performant than the original Win32 equivalent – see Elden Ring launch day performance on Linux vs. Windows for an example of this.
Proton is good enough that I’m extremely confident that games will “just work” even if it’s a fresh release. If it for some reason doesn’t, refunding is free and easy. Though it sounds like it’s an unoptimized mess regardless lol.
Typically it’s only “good enough” because the community has rallied and written the scripts to tailor the wine prefix for that specific game’s peculiarities.
A demo would probably do it, or an advance copy, or a leak. But it might not “just work” from 12:01 after the midnight release.
Uh when is the last time you used proton? That’s not even slightly true. I just beat Baldur’s Gate 3, it “just work”, zero issues from proton’s side of things. I play Genshin Impact too, which works with zero configuration or anything. Even the anti cheat they made compatible with it.
I can attest to this. I don’t play the absolute newest games, but I can’t think of a single title that did not work with Proton-GE, without any tinkering past like… early 2022. It has been very smooth, no matter the kind of game. Of course, minus the known offenders like Destiny 2 or R6 Siege
Every other game I play needs some special command line argument.
Baldur’s Gate, for example, but that was the other comment that used that as their example.
Yes, but generally it’s not “the community” like how mods work, it’s a project run by Valve who controls the review and release process. Sometimes patches come from “the community,” and sometimes they come from paid developers, but they all go through Valve.
You can choose to use a community Proton build like GloriousEggroll’s builds, but them that’s something more like a mod than a patch to an official project.
You’re confusing Proton with community efforts like Lutris. Proton is a package of technologies (Wine, DXVK, Vessel), not a configuration manager. Each individual game gets an identical, isolated runtime environment without any bespoke modifications except for downloading precompiled shaders (if available).
It’s certainly true that Proton has hardcoded quirk flags for specific applications, but these are exceptions which prove the rule – there are <200 of these compared with thousands of Verified status games. Almost always, Valve prefers to fix the upstream Wine/DXVK bug rather than hacking around it. Any hacks which Valve does ship are in the Proton source code, not per-game environment scripts.
I just installed it and it seems to work out-of-the-box for me on Ubuntu 23.10 with a AMD 7900 XTX and AMD 5800X. I’m getting around 55 fps at 3440x1440 with almost everything high beside having disabled V-Sync, DoF and Motion blur. AA set to TAA 2x.
Runs like hot garbage on medium 4K on a 6900XT. 25fps with an empty city. On launch, it defaulted to ultimate settings, and had the menus chugging at 3fps.
Feels like they accidentally shipped a debug/non-optimized build to customers.
They flatout said the performance is not where they wanted it to be. This state was to be expected. Nonetheless glad it’s on gamepads and hope that I’ll be able to run it some day.
I’d say Tumbleweed is what you’re looking for. They have some sort of automated testing process (OpenQA, I think) and are far more stable than Arch, while oftentimes having newer versions of packages before Arch.
Haven’t watched the video yet but I’d like to add from my very limited experience. I recently switched to Kubuntu (still have my windows boot) and the one game I play (Red Dead Redemption 2) seems to be running worse. I haven’t done much testing at all so it could be something I can adjust and get running better.
Having said that, general day-to-day performance is miles ahead of my Windows install.
If I could get RDR2 to run better on Linux and DaVinci Resolve to run I’d have no need to keep my Windows install.
Have you tried something like Nobara? I’m pretty sure DaVinci Resolve works on Fedora (which Nobara is based on) and you will get the latest optimizations as well. I am on OpenSUSE Tumbleweed just cause best performance on my system.
Nobara comes with DaVinci Resolve out of the box (or in the post install configuration screen at least).
That said I saw problems on Nobara I don’t have in arch that made me almost switch back to windows.
Decided to try arch before I switch back to windows, long story short have been on linux for two months without any plans of going back, the idea of windows now makes me wince.
Archinstall is perfectly fine, or EndeavourOS, they will make installing easy, i only use pacman and AUR for packages, anything not there Ive managed to build myself. This is the main reason I love arch, pacman + AUR are amazing.
You will probably want an AUR helper like yay or paru (doesn’t really matter which one for you, i prefer yay for sounding fun).
How’s the state of Nvidia’s drivers? Do the shiny new features work? Things like RT, frame gen, ray reconstruction, and randomly crashing the game because the driver has tripped TDR yet again?
The 545 Beta drivers apparently support a ton of new features on Linux like VRR. Frame Gen is the only thing not supported by Proton yet I think.
Tbh though, if you’re on Nvidia I would stay on Windows. It’s definitey doable to use Linux and get equal performance. Until Nvidia is ready for Wayland though I wouldn’t switch.
Good news is NVK is a new Open Source Nvidia Vulkan driver so in like a year or two things should be as good as on AMD. The driver already loads most games with DXVK 1.5.1 but runs them at like 1 FPS if they have anything like Valheim Graphics and above. Reclocking should be coming soon though so the situation will improve. After that it’s just working to get all features in and optimize.
As someone who’s really into gaming and gives NVIDIA on Linux a try now and then, the one thing that really bugs me is DLDSR isn’t available at all, nor is plain, non AI DSR. The latter isn’t hard to replicate, but I miss the extra bang for your buck of the DL variant.
Granted, mine is a very niche use case but I rely on it a lot since it works great on older titles or ones with bad or no native AA and such.
I’m as much a believer that Linux can get better performance than Windows because the less bloat, the best example is Blender which works almost twice as fast on Linux. That being said 25% increase on a game running on wine seems fishy.
Your video did not play correctly, also you didn’t synced properly between the two at the start so it’s hard to compare that both have the same settings, and on the screen at the end it shows windows is running in full screen and Linux in Borderless, not sure if this should make a difference but showcases that not every setting is the same. After the video crashed for the first time I skipped a bit ahead and saw that at one point you put the screen half-half, that’s a good approach, but I also noticed that the right side had a character the left side didn’t right at the start, that means that Wine is failing to render some stuff, or disabling some features which is usually what’s happening when you get this massive performance differences, so the comparison might not be valid. An example would be if DXVK ray tracing implementation bounces the light less times than DX12 does, it would be almost indistinguishable but would have a performance boost (at which point my question would be to show me the benefits of bouncing the light more, but that’s my opinion and not a technical analysis).
In any case, great video, even if something is different I couldn’t see any significant difference in the screen when doing the side-by-side, and I don’t think people who claim Linux is always worse would even know of the possibility of wine lacking some implementation therefore not rendering that.
The video is still being transcoded, check again later.
You can pause the video to check the settings, timing things like this properly is almost impossible but for next videos I will edit properly. Thank you for the feedback.
The character is also at the beginning on the Linux side but he just walks on Windows. This is a dynamic scene so details like that are expected to differ.
AC Odyssey doesn’t have Ray Tracing and DXVK is mature enough to render everything properly (and at better frames).
You are most welcome. I really think disbelief in how much better Linux is derives from a really cumbersome past. I’ve been benchmarking games on Windows and Linux for 3 years now.
At first performance was a little better/same on Linux, then it improved and then it improved vastly.
Don’t fear that Proton is a compatibility layer. Linux overall (with its lightness, better Filesystems and optimizations) can achieve great results like this in most DX11 games. I will do a MIrage Benchmark as well on Tumbleweed vs Windows 11 to see how things are on he DX12 side. Ray Tracing is not ready on Linux on AMD yet so that will have to wait.
Exactly this. I am willing to believe these people either don’t game on Linux or just use Nvidia which has less performance on Linux. In any case Apex is a good idea. After Mirage which is on the schedule perhaps Apex will be the next one.
Because just as many people have had a complete opposite experience
When I tried Apex on Manjaro a few weeks back I saw a ~15% decrease in frames and major stutters.
A single system, running untested benchmarks and without any external validation from a trusted source doesn’t mean anything. Just like my experience with it isn’t universal, neither is either of yours.
Yeah I got worse performance on some Linux OSs. This was on PopOS. Every Linux distro will be different, I suggest sticking with the current best gaming distro… not sure what that is currently, but I think Arch was previously one of the best distro for gaming. Once you have that and proton going, you will should see a difference.
Unless your on Nvidia: in which case I hear it’s hit and miss for performance gains, I see no reason why it shouldn’t perform better in the majority of cases.
There is absolutely no way 25% is realistic in this scenario, it’s most likely, as you said, a certain characteristic/feature is interpreted/skipped/handled differently by WINE.
the best example is Blender which works almost twice as fast on Linux
People say this, but what exactly do you mean? I mostly model on windows because it's my primary system (I use applications that simply don't work well enough with wine), but mostly finish and render stuff on linux because of windows' retarded automatic updates etc. that can just cancel rendering without asking. And the only difference I've seen is how fast Blender starts - I'd say that's more than 2x as fast on linux, it's a huge difference. But rendering is the same (NVidia GTX GPU) and other work inside blender also seems to be about the same.
Nvidia drivers don't tend to be as performant under linux.
With AMD instead of using the AMD VLK driver, you would use the RADV (developed largely by valve). Which petforms better.
Every AMD card under linux supports OpenCL (the driver is more based on graphics card architecture) and you install it very easily. Googling it with windows found pages of errors and missing support.
Blender supports OpenCL. I bet the 2x improvement is Blender being able to ofload rendering to the AMD graphics card.
Also this represents the biggest headache in Linux, lots of gamers insist they can only use Nvidia cards. Nvidia treats linux as an afterthought as best or deliberately sabotages things at worse.
AMD embraced open source and so Linux land is much nicer on AMD (and to a less extent Intel).
The results here will probably be a DxVK quirk, lots of "Nvidia optimised" games have game engines doing weird things and the Nvidia driver compensates. DxVK has been identifying that to produce "good" vulkan calls.
Afaik Blender since 3.0 does not support OpenCL anymore and AMD rendering uses HIP instead. I have not found any information about dramatic performance differences, though CPU rendering does seem to be somewhat faster on Linux - but more like 10% faster and the amount of computation practically done on the CPU is not that big.
Also this represents the biggest headache in Linux, lots of gamers insist they can only use Nvidia cards. Nvidia treats linux as an afterthought as best or deliberately sabotages things at worse.
Personally I use NVidia because of CUDA, gaming is an afterthought. I wish CUDA just fucked off and we got some universal compute API instead, because that's what would reduce the NVidia stranglehold on the market, perhaps OneAPI is going to catch on at some point, but at this moment those options are not practical.
We’re referencing a somewhat old video of a benchmark ran in both systems www.youtube.com/watch?v=cpE2B2QSsa0 that’s likely not still true, possibly devs figured out what was the issue on windows and circumvented it somehow.
Yeah, don't do that anymore then. Firstly the video doesn't really find overall 2x speedup, but mainly Cycles X came out since then, where most of the codebase has been rewritten from scratch, and after that numerous significant optimizations happened as well. That video is pretty much irrelevant now.
I want to switch to Linux and I would love to game on it daily, but just like so many people, software incompatibility is holding me onto Windows.
In my case, it’s Parsec that I need, because I game a lot with friends who live in other countries. And unfortunately, Steam’s remote play together feature is very broken on Linux (I remember even filing bug reports about it when I was daily driving Linux two-ish years ago.
Longtime windows and Linux user, my last several machines have all been dual-boot.
I’ve tried multiple times to get gaming to work right with Linux, whether it be Unbuntu or just plain Debian, but something always gets in the way. Graphical issues, sound issues, controller incompatibility, platform incompatibility, unable to launch game, whatever. I give up and just stick with windows and use Linux for other things.
I’m sorry, but if you see a 25% difference in a benchmark, that means your methodology is somehow flawed. A few percentage in either direction would be believable, but this difference would be so comical if true, that extra wariness is needed.
There’s a few thing that look a bit off to me, but most importantly it seems like your OBS settings are wildly different between systems. It’s a bit hard to make out, but it seems like you’re doing CPU-based encoding on Linux and GPU-based encoding on Windows.
I am not doing CPU Encoding on any system but there is a difference indeed.
Linux is Gstreamer VAAPI H265 and Windows GPU Encoding H264. In fact, Windows should have had an easier time encoding, I didn’t realize that until now. Also asI have commented on the video the game is on a 980 Pro on Windows and on an HDD on Linux so Linux can be much faster. I will rectify that by getting an SSD to put all my games on in the future.
Beyond that, the methodology is not flawed, if you can even believe that. Everything is on the video for comments exactly like this one.
I see. As I said, it was a bit hard to make out in the video.
In fact, Windows should have had an easier time encoding
Granted, I don’t know too much about AMD’s video encoding solutions, but from a cursory glance on the internet, it seems like their H.264 solution is quite bad compared to H.265. Given that the game is GPU-bottlenecked and your CPU isn’t stressed at all anyway, I’d recommend recording these tests using the CPU to eliminate more variables.
Beyond that, the methodology is not flawed, if you can even believe that.
Well, yeah. As much as I’d like to believe, these differences are way too big for me to do that, even with everything you’ve shown in the video. Occam’s Razor would suggest that it’s much more likely that the benchmark/setup is simply flawed in some way, rather than multiple teams of OS-, hardware-, and game developers not realizing a gigantic 25% performance improvement on the table that’s somehow more or less “accidentally” fixed just by using Linux/Proton/DXVK.
Not saying you’re wrong, but it’d need a good chunk more evidence for me to believe that.
I just edited my comment right as you posted, so I’ll put it as a separate comment now:
It would also be interesting to see this game running through DXVK on Windows. That way the calls made to the GPU should be virtually identical, eliminating possible problems with DX11 in the AMD driver.
yeah same, I used to dual boot win10 for certain games and slightly better performance but since kernel 6 something and the 7900xtx I’m gaming on Linux full-time including Starfield
I do all my gaming on Linux EXCEPT there is this one niche game that, at a certain point, needs so much repetitive grinding for resources that pretty much everyone uses macros for it. Guess what? I have gotten to that point, and unfortunately, the only macro that’s efficient enough (because it uses AutoHotKey for detecting elements in a window) is Windows only, because AutoHotKey is Windows only. And no, it cannot be rewritten in AutoKey, it cannot be rewritten in Python (it probably can, but the project is so massive that it would be a near-impossible task, and there is neither enough supply of people willing to do it, nor enough demand from users), it cannot run under AutoHotKey in the same WINE prefix OR in a different WINE prefix as the game (I tried both, Window detection doesn’t work), I have tried everything and nothing seems to work. In terms of less efficient macros, there are 3 projects listed on the official “[Insert game here] Macro Community” Discord server under the Python-macros channel: 1 of them is supposed to be a macro working for both Windows and Linux, but it has been abandoned (I even contacted the developer), the 2nd one is MacOS only, with the Dev stating “retina display” as the reason behind it. Still , I tried it and couldn’t get it to work. The third one was a project that started some time ago, but then there is now a message by the dev stating “I’m now pausing the development of this macro” so I contacted them a few weeks ago to see if I can get their incomplete source code to use as inspiration when writing my own macro, no response. And yes, I tried writing my own macro, and failed miserably. It is far more difficult than I thought.
So yeah. I’m dualbooting a debloated Windows 10 (thank you CTT and winutil) alongside Linux. And Windows is, in fact, the secondary OS i.e I installed Windows after Linux.
linux_gaming
Oldest
This magazine is from a federated server and may be incomplete. Browse more on the original instance.