I’m just going to go ahead and say this now, do not expect most windows games to run better on Linux than windows. Typically the case is when you find a well optimized game that is CPU bound, or is natively vulkan. Anything else, expect comparable framerates.
It’s comparable more often than not, but honestly even if it was 17% worse on average I would still stick to Linux and just build a better computer. Which is what I did before proton.
No doubt and I’m the same way, I’m just trying to say that one shouldn’t try to sell Linux solely based on “gaming performance” when it is definitely not the case most of the time.
Linux is not used like windows or macos at all, and new users will definitely be frustrated enough just learning to use the operating system. Believe me, I think it is awesome that we are finally getting another gaming revolution in the community (Linux gaming actually used to be pretty good before around 2010), but keep in mind that these efforts are for the community and steam deck users. Anyone who wants to have it too will ultimately have to join the community and learn the ropes.
Honestly, if you use Proton-GE's FSR feature for games that don't offer built-in FSR/DLSS + GameMode, you can def beat Windows performance in some Windows-only games. I know it's kinda cheating, but it does net you higher FPS on the same graphics settings.
I will disagree and that’s why I made this video. Been benchmarking games for 3 years now, mostly on AMD systems. It went from about same performance, to slightly better, to this. 17% average improvement is nothing to laugh at. It’s the difference between a 4090 and a 7900XTX on Windows. So people can literally save $1000 just by using Linux.
What you say, does mostly apply to Nvidia users though.
Man look, I’ve been using Linux as a daily driver for 18 years, people have been saying exactly what you’re saying since before performance was even comparable.
You’re not going to get 17% better performance on the GPU just because you’re using another operating system, it’s not going to happen unless you’re running a Linux native version of the game. Often times, that is not even the case.
Performance can be a little bit better if the game is natively opengl or vulkan, but if it is directx (the vast majority of windows games) then it is going to be comparable at best in GPU-bound scenarios, I.E. most of the games people are playing on PC.
You can’t just magically put more transistors in a GPU just because you are running a different OS. CPU bound games run better on Linux because of the god-tier scheduler, but a GPU is essentially a computer in itself, all drivers do is tell the GPU to take this information and translate it into something you see on a screen.
By the way, the Nvidia thing has been false for quite some time now. I primarily use AMD on Linux, but the only place you will run into issues with Nvidia is wayland, otherwise it works perfectly fine everywhere else.
I don’t see an argument which disproves my results apart from you disbelief. But I like the Nvidia comment. I’ll do a video of Linux vs Windows on my 3080M laptop. We’ll see how true is that Nvidia works as well as AMD on Linux. :)
Go right on ahead, I’ve done the tests myself already.
Keep in mind though that if you are using a laptop, nvidia tends to work better when paired with Intel vs amd for the sake of graphics offloading.
I don’t think you understand how this works, I’m not trying to disprove anything, you are the one trying to prove something. You chose 10 very specific games to run these tests, some of them being heavily CPU bound, and state that you are receiving an increase in GPU performance when it is simply not the case. All of these games are also optimized for proton, which does not help your case.
Tell you what, why don’t you give something like “Spec Ops: The Line” a test? Halo Infinite? 40k Darktide? Vermintide 2? Dying Light? Hell, infinite and darktide are very popular in the Linux gaming community, I was even one of the beta testers for darktide.
You say that like I’m afraid to do it. You’re missing the point that these games don’t have benchmarks lol. If you want I can do a gameplay comparison but don’t tell me, the areas or movements are not the same. :)
Also these games couldn’t be more diverse. I tested DXVK, VD3D and Vulkan (both on Linux and Windows) with these games. If you can find a more diverse benchmark please let me know, cause I haven’t found one.
Also, I’m already doing benchmarks on my i7-10870H and 3080 laptop. Linux won’t go above 80W, cause of the Nvidia Drivers (545 Beta btw) so the difference will be IMMENSE for Windows there.
You don’t need a specialized benchmark to do a benchmark, you can use a realtime rendered cutscene, you can do an average over several games. That’s how they have been done for like a decade and a half at this point.
Also, I’m not referring specifically to mobile graphics nvidia, but nvidia altogether. Linux laptop gamers make up a very very small amount of total Linux gamers, it is an incredibly small niche of two already small niches, both being Linux and laptop gamers. Yes, of course if you have a limit to the total amount of power, it will lag behind.
I gave you a list of games, start there, my list is also diverse and includes all of those except for vulkan, which if you want, throw doom eternal in there, though as I have already stated vulkan will get a small increase on linux over windows in terms of GPU performance, so that’s not really proving anything anyone doesn’t already know.
If you want a fair comparison, limit it to 80 watts in windows as well. Remember though that power is NOT EVERYTHING when it comes to GPU performance. All of the games I detailed above are GPU bound games and will be a fair comparison. Just a heads up darktide may or may not have graphical glitches on your system if you are running amd (both operating systems, it is hardware related), I’ve worked with the devs to fix it in the past but it seems like recently people have been having issues with it again.
It sounds like some time in that 18 years, you solidified this impression, and are choosing to not recognize the advancements in Proton and drivers that have occurred post-Steam Deck.
I’ve been using Linux since before Xwindows existed, and I am open to OPs research. Just because we’ve used it longer, doesn’t make either of us right without proof. OP supplied evidence. Prove them wrong.
I’ve been using Linux since before Xwindows existed
Why are you blatantly lying like this? X came out seven years before the Linux kernel was even released. And even then, there wasn’t a working system for the Linux kernel when it was released. Keep in mind I said DAILIED Linux for 18 years, I didn’t say USED, I’ve been using Linux for 27 years now. I actually remember a time when Linux was not an operating system that people would use to play games on.
I’m using my time specifically in the community as an example to show that this is not the first time I have heard this. OP supplied evidence in ten very specific games here, there are over 12000 games on protondb that are “playable”, not even verified. I have run across myself quite many games that run at half to three quarters the performance that it does on windows, and that is absolutely fine.
Telling people that using Linux will get you a “free performance boost as much as 17%” when it very likely will NOT, will create a lot more angst towards the Linux community than it already is. The elitists are already doing that for us, we don’t need more of it.
We should be pushing people towards Linux for digital privacy+security and free software, not cherry picked performance boosts.
Yes, I very well recognize the black magic sourcery of proton and wine, but you are sitting here and trying to tell me that proton is somehow going to make your GPU somehow physically push more calculations per cycles just because it is running Linux. Not even giving me the “mesa drivers” spiel which is also BS, as performance is not the main area that the Foss drivers are better in.
Linux is not going to break the laws of physics buddy, I’ve already said what I said, boost in CPU bound games, little to no boost in GPU bound games. If you’re seeing a boost, it’s because you have a CPU bottleneck and you are getting it because of the scheduler.
“works fine” is very different than “is equivalently optimized.”
Valve has done a lot of work to get games to work well on the Steam Deck, and that likely translates to other AMD GPUs. So it makes total sense that Valve would optimize the Proton translation layer for DirectX calls to the AMD driver differently than the NVIDIA driver (or rather, in a way that AMD handles better). A big issue in GPU optimization is keeping it busy, so perhaps the AMD driver working with Valve’s patches on the DirectX to Vulkan layer improve utilize m utilization. That could translate to a modest performance improvement even on well optimized games (perhaps 5-10%, probably not more than 20%).
I don’t know if that’s what’s going on here, but it’s a plausible explanation.
I can see why you’d think that, but what you fail to understand is that valve is not the only one working on proton, and valve themselves did not even make DXVK. Those are free and open source efforts and valve even pays external devs to commit to that software. I’m telling you that DXVK itself is not going to give a boost to graphical performance because it literally cannot, those are extra instructions that your GPU has to perform in order to send out frames.
Directx to vulkan translation is exactly that, translation. It receives directx calls and translates them to vulkan. For one, it has overhead, two, if the game is optimized, it is already going to be running at max performance on windows, using DXVK is going to slow the GPU time down because it will have to perform more calculations. No scheduler will save you from that, not even the Linux one, because it isn’t something that is handled by the scheduler.
I never said Valve built DXVK or even WINE, just that they have a vested interest to ensure it works well on their AMD-based hardware
I never mentioned anything about the scheduler or any Linux imtrinsics other than the AMD GPU kernel module
DirectX -> Vulkan isn’t a direct translation since the APIs aren’t 1:1, so there’s going to be some tuning in how APIs are mapped, and the tuning can differ depending on the GPU driver you’re using.
It’s the same with processors, you can optimize a compiler to work better on AMD vs Intel or vice versa (look at Intel C++ compiler benchmarks for an example of that), even if they use the exact same set of instructions because the microarchitectures are optimized differently. This is because the way the instruction set gets mapped to the microarchitecture can impact performance significantly (something like 10% is possible, depending on the benchmark).
GPU drivers are complicated, and there are a lot of areas where the interaction between the driver, software, and system services can be optimized. AMD’s drivers are open source, which helps with those optimization efforts. Then you throw in a big, well-funded, and motivated company like Valve funding development (both through salaries and donations) and you end up with AMD GPUs getting extra attention for things like DXVK.
So I would expect AMD on Linux to perform better vs NVIDIA on Linux when compared to AMD vs NVIDIA on Windows. As in, the performance difference on Linux vs Windows would be more favorable for AMD cards than NVIDIA ones because AMD on Linux gets more attention than NVIDIA on Linux. I don’t expect the same for compute, since NVIDIA invests heavily in that space on Linux, so it’s not an inherent advantage of the platform (e.g. the scheduler discussion), but a question of where optimization efforts are focused.
Alright look, I’m not going to argue about who said what because we both know what we said and it is unrelated to the topic at hand.
The reason the windows amd driver is bad is not due to performance, it is the very same reason why the proprietary driver is bad on Linux, it is horrible reliability.
There are circumstances where they trade blows and circumstances where they perform similarly. If you really want to compare the two based on OS alone, you need to compare the equivalent drivers which is the proprietary one.
We’re already not doing an apples to apples comparison here because we’re comparing WINE+DXVK vs DirectX. Comparing the OS itself isn’t that interesting, at least from an end-user perspective, what is interesting is comparing the typical user experience on both platforms. As in, no tinkering with stuff, just installing in the most obvious way.
Valve is optimizing for that typical user experience on their Steam Deck, and that translates to the desktop fairly well. They’re not really doing the same on Windows, so it’s interesting to compare devs+manufacturers optimizing stuff on Windows vs the community+Valve optimizing stuff on Linux.
Why would not comparing the OS itself be interesting? That is literally the foundation of everything you are seeing on the screen.
You also can’t just compare WINE+DXVK to DirectX, because you can actually use DXVK on windows. If the video title was “directx vs dxvk” then that would be totally fair, but it is not, it is called “windows vs linux”. I’m simply trying to say that the vast majority of games are not going to see a 17% increase in GPU performance, your biggest boost is going to lie with CPU bound games because it is the truth.
The only time you’ll see a game perform better on a GPU on Linux is when the game has a native version, and even then that only applies if they actively develop that version, many games are not actively developed and are even a few versions behind.
Because regular users aren’t going to be changing drivers based on the game, or doing a ton of system-level configuration to get a bit better performance.
So it should be defaults vs defaults.
If we want to compare OSes, we should do targeted benchmarks (Phoronix does a ton of those). There are far more interesting ways to compare schedulers than running games, and the same is true for disk performance, GPU overhead, etc.
you can actually use DXVK on Windows
How many people actually do that though? I’m guessing not many.
“Windows vs Linux” is comparing the default experiences on both systems, and that’s interesting for people who are unlikely to change the defaults (i.e. most people).
The only time you’ll see a game perform better on a GPU on Linux is when the game has a native version
That’s just not true, as evidenced by this video. If you take the typical setup on Windows vs the typical setup on Linux, it seems you get a 17% average performance uplift on Linux on these games.
That doesn’t mean Linux is 17% faster than Windows, nor does it mean you should expect games to run 17% better on Linux, it just means Linux is competitive and sometimes faster with the default configuration. And that’s interesting.
Linux does not have a default configuration, that’s why we have over 600 distros. If you want to have a baseline “default configuration” then fedora would be the way to go, which he has not used.
Yes, he got a performance uplit by 17% on average in these games, the point he is trying to make is that you can get this in every game on Linux which is what is not true.
Most of those games are also CPU bound, an area that Linux is going to destroy windows. Once again, I am referring to GPU performance specifically, as that is the general point that OP makes with these posts.
That may be true, but de facto defaults today is Proton experimental on Steam with the a recent Linux kernel. That’s pretty much the same across all distros.
Yup, the difference between Ubuntu, Fedora, and Arch or whatever isn’t going to be all that big, assuming you’re working with each distribution’s default kernel and running with a Steam’s provider runtime. You might get 1-2% here and there, but that’s pretty much within run to run variance anyway.
That’s not all the factors that play a role in performance in games.
For instance, what fork of the kernel are they using? Are they using zram? What graphics driver are they using? Gamescope? Gamemode? All of those things affect performance of a game to varying degrees.
Also, Proton experimental is definitely not the default on any system, that would be Proton 8.
Sure, but each distro has a default configuration, and distros don’t vary that much in terms of performance with those default configurations for playing games. If there is a consistent performance difference, it’ll likely be something like 1-2%, which should be within run-to-run variance and not really impact the results.
And if anyone assumes that an average between 10 games represents the difference you’ll see on average for your own games doesn’t understand statistics because 10 games is not enough to be a representative sample, especially since they weren’t even randomly selected to begin with. It’s still an interesting result.
CPU bound… Linux is going to destroy Windows
You’re being hyperbolic here.
The differences, all else being equal, should be pretty small most of the time unless there’s a hardware driver issue (e.g. when Intel’s new p-core vs e-core split came out, Windows had much better support).
If we’re seeing a huge difference, more is going on than just a “better” scheduler or more efficient kernel or whatever. It’s much more likely Windows is using DirectX and Linux is using DXVK or something. The bigger the gap, the less likely it’s the kernel that’s doing it.
As someone who has used Linux exclusively for ~15 years, these kinds of benchmarks are certainly exciting. However, we need to be careful to not read too much into them.
Not enough people running nvidia realize just how much nvidia does to make sure you stick to their proprietary software. That you can close most of the performance gap with FOSS on AMD is an amazing finding.
Unfortunately it won’t convince many who haven’t already seen the benefits of a more open system.
yeah i have a work college that i got to use vim keybinds in most software but when i tell him that he could control his whole system like that if he switched to linux he dosen’t like the idea because he isn’t used to using linux
Not enough people realize how much AMD does to make sure people stick to their proprietary software. Nvidia software that is.
A lack of ROCm support on consumer hardware is simply inexcusable. Nvidia makes a shit ton of money with the AI boom, because people like to work with stuff they already know. And it’s infuriating, because Nvidia cards have way less VRAM.
how much nvidia does to make sure you stick to their proprietary software
Holy fucking shit you are extremely misguided. Are you not aware that the Linux nvidia drivers are proprietary? The only reason that the mesa drivers are awful is because they barely support the 10 series and they don’t support the new instruction set of the 20’s and above.
If you are running Nvidia on your system and it is above a 10 series, you are running proprietary software. Big whoop, steam is proprietary too, so are the vast majority of the games you play on steam.
Hell, nvidia used to be the ones supplying an open source driver on Linux like 14-15 years ago, AMD didn’t have that, only the proprietary driver. DO NOT OWE ALLEGIENCE TO ANY PUBLICLY TRADED COMPANY, that’s exactly why we don’t have good FOSS drivers for nvidia now.
Those are the mesa drivers, not the “amd” drivers.
Those very same drivers work on Intel cards and pre-20 series nvidia cards. Mesa is not an AMD project or an Intel project either, that is an independent team.
Even then, those drivers are for allowing the GPU to display to a screen and interact with the system. They are pretty much the same idea as the Microsoft basic display adapter. You still need the xf86 drivers to display X, the opengl drivers for opengl, cuda for cuda, vulkan for vulkan, etc. Those are all separate components because they have libraries included with them.
If all of those extras were built into the kernel, the kernel would be like 2 gigabytes, not 150ish megabytes. It is literally enough to get you going with a getty and that’s about it.
Mesa isn’t a kernel driver. AMDGPU is the name of the kernel module and it’s primarily developed by AMD. Mesa provides OpenGL, Vulkan, etc. implementations and is funded by AMD, Intel and Valve (among others). There’s also AMDGPU-PRO which is a proprietary alternative to Mesa from AMD.
That does not change my point in any way, mesa is not built into the kernel, which you need as a dependency to use X, which is required to run a window manager and/or WINE. I never ever said mesa was a kernel driver.
The drivers in the kernel (MESA) work with AMD out of the box. If you have AMD hardware, you don’t need anything else. What I said was “AMD drivers are currently in the kernel”. I did not specify that these drivers are developed by AMD – you seem to care a lot about that, but it’s not part of the argument I was making.
Again, you seem to have misread my first comment, which on the Linux side means: you still need proprietary nvidia drivers on Linux. This is also true for Windows, where many folks are perfectly happy to continually update GeForce Now and stay in that ecosystem. That was the point of the comment.
I’m not coming at you with hostility, I am informing you that what you are saying is incorrect. If you keep on skimming over everything I say, then perhaps I may get hostile because that is extremely annoying.
If you are so sure of yourself on the kernel driver front, then do me a favor and fire up gentoo or arch and try to run a desktop environment or window manager without the mesa packages installed. You’ll find that xorg has mesa as a dependency, and there’s a very good reason for that: it’s because that’s not what the kernel driver is for, mesa itself is larger than the kernel itself. The kernel driver is exactly what I said it is, it allows the operating system to see and interact with the device, it doesn’t tell the device how to do its job, it tells it “here are some pipes, you will receive information from certain ones, and send it through others”. That’s exactly what a kernel driver does, there are no libraries or anything of that nature which is the overwhelming bulk of what makes a graphics driver.
Also, geforce now is optional, you can as always install the drivers without the useless spyware application that nvidia provides.
If you keep on skimming over everything I say, then perhaps I may get hostile because that is extremely annoying.
What is annoying is getting novel after novel of irrelevant information. Can you actually tell me what part of my first comment you are referring to?
The closest I’ve seen is that you took “AMD drivers” meaning explicitly developed by AMD, but that’s not how adjectives work. Now you are all about needing more than MESA, which is also fine and correct, but irrelevant to my comment about nvidia drivers.
Go read this and do not reply back until you understand what it is we are talking about.
RADV is not built into the kernel.
Stop trying to sidestep and make it seem like I’m misunderstanding, you know full damn well that when I say AMD drivers that I am referring to drivers for AMD hardware. You have the Foss drivers (mesa), the open source drivers, and the proprietary drivers. All of these are AMD drivers.
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.
Maybe I’m missing something, but it didn’t look like the GPU was engaging at all. 15fps? That doesn’t seem right. Given that the overlay had a battery percentage reading, and the GPU readings were all zero, I’m guessing this was on a laptop, and it didn’t switch from using the iGPU to using the GPU.
Either that, or NVK has a lot of work to do, and that doesn’t track with recent news about its progress.
OP, is this your test? Can you shed light on the results?
It is with a heavy heart that I have to tell you that NVK’s latest progress has little to do with performance, and that it will take years to achieve performance parity or even remotely near the proprietary drivers.
One thing that should raise morale for us all, is that Nvidia is now actively contributing to the respective FOSS drivers.
Oh really? I thought I saw elsewhere that it had achieved and even exceeded performance of the proprietary driver in another test. Maybe it was just that specific case.
It can be fast in one test and slow in 49 on one generation and on another generation it could be faster on 10 and slower 40. Just the nature of complex software supporting various generation of hardware.
I believe you’re correct that the GPU isn’t being engaged at all. For one I highly doubt it’s NVK because it’s certified as “Vulkan 1.3 conformant” meaning it’s a complete to spec implementation. The overly also say it’s using a “NVIDIA GeForce GTX 4090 Laptop GPU”… Then it also says that there’s 0 watts being drawn from both the GPU & CPU which is extremely odd.
Maybe it’s a driver issue?
Nouveau doesn’t support power/load reporting yet. Some games will work better than others. Just because it is comformant does not mean it will be fast either. Comformance comes first then optimizations.
That said, it could still be running on the wrong gpu, but i don’t know. My experience when I’ve tried it on my 4080 hasn’t been mind blowing performance, but much better than before all this work started landing.
What about the proprietary driver? IDK if they’re even using Nouveau or not. NVK is apart of Mesa… Also just noticed that they’re running it under Wayland directly… And with HDR. Kinda in a very experimental area. Point is we need more specs. Would love to see a hw-prope.
It is definitely Nouveau. NVK is just the Vulkan side of it. NVK is in a experimental level. I personally would not run it daily yet, ignoring the lack of some video features and dlss in games. They have made huge strides in the last couple years with it though,
Edit: He has been posting videos for a while now running NVK, btw.
I’m aware. I’m just pointing out that he needs to publicize his specs hw-prope and that he’s stacked on 3 layers of experimental software. Nvidia on Wayland is WIP, HDR on Wayland is WIP and NVK is work in progress, even VKD3D with NVK is experimental.
I tried to play Vermintide 2 on Linux after Windows kept giving me issues, but I couldn’t get the Easy Anti-Cheat to stop freaking out about me playing on Linux.
I never found a fix for it, has that changed or is it just luck?
Yeah, if you go to your instance’s Trending, Recently Added or Local Videos page, it will allow you to set such filters (all three are the same page with different filter presets). So, on the instance where this video is hosted, that’s for example here: video.hardlimit.com/videos/trending
It should be said that PeerTube’s federation isn’t as tightly knit as Lemmy’s, though, so you’ll only see a small portion of the network from there.
If you really want to see pretty much everything, you can use SepiaSearch, although well, it is moreso a search engine than a feed…
Impressive to see it correctly render such a modern game and even at something approaching playable frame rates at high resolution? What is this magic?
IIRC, nvidia linux drivers benefit greatly from rebar (10 ish %) while the windows driver doesnt, maybe this could bring linux closer to windows on nvidia.
Also, just a stable accelerated desktop on install. Basic home/office and web/media consumption use.
Regarding cuda, yeah never know. Probably not for a long while as one of the benefits of their current driver is that it uses the same codebase between linux, freebsd and windows, so they should have feature parity in that regard. There are definitely pro/cons for their driver though.
video.hardlimit.com
Hot