วีดีโอ

I actually show in the video that the total power pulled from the wall is the same in both OSs, it just resorts differently l. This is why I suggested to pay more attention to the GPU clocks

@@eXxSe not really. Power consumption is most of the time on par between linux and windows when i measured power at wall. It's just power consumption monitoring usually isn't calibrated that well.

Looking again at the numbers Windows shows TBP while the linux mangohud shows TDP. It's a different metric and you can't compare then directly

You are correct. This is why I shwoed in the video the power pulled from the wall by both OSs, and there was no difference

Amd boosting bug. Fixed in kernel 6.13

Yes, that is the case with Linux, it may have issues like this. I don't use VRR to be honest, but I have the option over DP on the AMD GPU

It is a catchy distro

No, I haven't. I will pin your comment in case others are looking for a fix.

@@GhostMirror3567 amdvlk is the same source that is used for the windows vulkan driver so it does make sense that it fixes graphical glitches because it's more focusing on correctness than performance

I heard that the AMD proprietary drivers are better for RT then the open source drivers in Linux, will keep that in mind.

What's wrong with that one? It is one of the best optimized games released in some time, no DRM, works good.

@@casuallygamin9 Perfectly reasonable benchmark. Don't know what this guy's deal is.

I added the games in the time stamp, I thought that is enough to be honest

@casuallygamin9 ah ok, didn't see that

No worries

This is what is reporting, so I'm not actually sure. It looks like Windows 11 uses a lot more the CPU in Windows as opposed to Linux

I didn't try to be honest

HDR is still pain on Linux right now but Wayland should address HDR in the future. Compositor extensions for HDR are actively being worked on

Those vales are displayed by GOverlay, I just use those, and it's true, in most cases the 0.1%Lows = 1% Lows in Linux die to the fact that there are less stutters and it's a bit smother

@casuallygamin9 Then GOverlay does not show correct values. These values make no sense. I mean just imagine what the numbers mean. The 1% lows are different to the avg framerate. That means there is a variation in frametimes, which is always the case in GPU bound performance tests. But then suddenly the slowest 1% of frames all have the VERY SAME FRAMETIME? Why would they? They should still show some variation in frametime. I mean the calculations the GPU change ever so slightly with every different frame, why would there be 'upper bounds' for frametime? Also no performance test I have ever seen shows this behaviour. As you say the experience in Linux is a bit smoother, but your numbers say it is out-of-this-world smooth. Which I think it most likely is not. Your test is great otherwise, but those numbers are quite misleadings. The only way to reach identical 1% and .1% lows is probably to limit frame rate to something that creates very low load on CPU and GPU. And even then there might still be some variation.

I know what you are saying, but if you look closely you will see a small variation in some games. The issue is that there is no other way to measure those results. When the 0.1% lows are equall to the 1% Lows this means that there is no stuttering. Also, don't forget that GOverlay's final result is truncated to the last 2 digits. I'm not sure if it will be the same if 5 digits are included, maybe there could be some small variation there.

@@casuallygamin9 sorry but this is just not right. It does not just mean there is 'no stuttering'. It means that all the frametimes of the slowest 1% of frames are the exact same. That is something much more. Even on very smooth gameplay experiences there are usually some frames, that take way longer to calculate than most others. This might be due to some asset being loaded in, many effects happening at the same time or some other technical reason that causes more work for either CPU or GPU. In a smooth gameplay experience this is not usually an issue as these are not really noticable and do not happen often. But even if there are very few slower frames, they will impact the .1% lows way more than the 1% lows (because there they are 'averaged out' a bit more). Your numbers basically say, that those slower frames do not exist on Linux. Games that show this behaviour on Windows suddenly do not show it with a virtualization layer below? How would Proton or Linux get around the behavior of the game it is running? How would it circumvent loading something or having to calculate many things in one specific frame? And thus having one slower frame, that greatly impacts the .1% lows and slightly less also the 1% lows (separating the two numbers in the process). Are all the games suddenly perfectly optimized once you run them via Proton? No, the numbers are just wrong.

I understand what you are saying and I agree with you on some degree. Think of this scenario. Windows does a lot in the background and because of that some will have lower 1% Lows and .1 Lows. This doesn't mean that I can get 1 run from the 3 that I do, without any stutters, and have close to the same 1%low and .low values. You are correct about what the mean, but if you run a game on your machine and make a 10 second run, you could get a clean run. With that said, what I can see on Linux is that games run much smoother, and that the .0 are close to the 1% lows, but not always the same. Don't froget that the 0.1 lows are exactly that, the slowest frames, for example from 10000 grames, it will agregate the top 10 slowest. And if you have, for example a value of 20.0716354 FPS 0.1 values, you could have a value of 20.074983 FPS value for the 1% Lows. This would mean that the drop is there, but ther is no stuttering. Keep in mind also that on Windows the polling is 1000 ms while on Linux I can't go above 500ms. This may impact a bit the results. Is it possible that the results for the 0.1% lows are not accurate enough? I wouldn't know as I don't have a way to measure that.

I'm not sure ehere the gains come from. I expect that it's because Linux is more tweaked and the AMD open source drivers are par with Windows drivers. I added the games in the timestamp, but yhe version I never thought. To be honest, some games have some really long version so I'm not sure where to add it on the screen.

system overhead when people say windows is bloated they weren't lieing and the overhead is only getting worse on windows heck win10 still games better then 11 thanks to the bloat that comes with 11 the reasons windows sometimes work better is due to a optimization which linux can eventually also get which likely results in again the lower overhead running the game faster

I don't think that is the case. The 1% Lows and the 0.1% Lows will show stutters, while uaing your example would not.

@casuallygamin9 i didn't refer on 1% lows, taking those and making a geometric mean would be even better. I talked about summarising many results, instead of calculating the simple average you could use Geo means which are better in representing all values equally No Matter how high/Low they were

I understand what you mean, the reality is that I'm using what MSI Afterburner and GOverlay are reporting, as those calculate the averages somehow. What I did, is create a python script and pull the results per game from each tool.

@@casuallygamin9 oh, okay... At 17:20 you show the averages of the games, there you *could* use the geometric mean. But that is just a sidenote, doesn't change anything on you review/comparison, it was still all valid and fine, just my inner monk that saw some possible, tiny improvement... Maybe I also forgot to mention that this video was awesome, thank you for the work you did there, very nice to see and Linux did really good in the games you selected, nice!!!

Those values are optained by using a JS script that pulls the average values from all games at a given resolution and averages the results. I know what you mean now, it would increase the gap between Windows and Linux.

Like I said in the video, for aingle player games with RT on AMD GPUs, linux is a bit better.

Thanks for the support

As long as you are on steam it is the same as on windows. Just click Install. If you need a compatibility layer like proton you can enable that directly in steam in game options. For the EA and epic launcher you have to download a program to get it working. Like lutris or heroic. And that you can download your games and start them. Simple as that. Most games work without proton and run native in my experience. Only games with a specific form of Anti-Cheat are impossible to start. Like PUBG.

@@KaAyYj thank you!

It is like I showed in the video. You can put the translation layer(proton experimental, or proton ge) and after just click install, really easy.

Yes, the lows are much better on Linux as opposed to Windows 11

Yes, that is yhe case

Both are good

@casuallygamin9 my Arch Linux with my Packages with Custom.Kernel are better, but Its only for me! 🥰

I believe you as you probably complied and tuned everything for your setup

@@Rekkeni yes, the Focus shouldn't be in average FPS but on 1% lows. I dont mind 0.1% Low as much but it is still more important than average.

It cost money for sure. I just bought a B580 as eell, to see how intel is doing. Thanks for the support, it means a loot.

Tiny11 is obsolete when W11 Enterprise IoT LTSC exists

Meh, I play on linux and there are no issues at all except for multiplayer games. However, getting people to understand linux is trying to make people to be responsible and knowledgeable to the very pc they own. People hardly wants to learn anything nowadays.

i used to have that problem and it is solved by not playing fortnite. changed to full linux almost 3 years ago during the tpm fiasco and i don't regret one bit. if the devs don't support linux i don't want to play their games. i mean could play on my ps5 if i wanted but all EAC games are cancer and filled with 11 year old tryhards anyways

@@InnerFire6213 These days, game devs don't even have to support Linux - they just have to no actively block their game from running on it.

@@DanIsNotHome yes my friend you are right if you have compatible hardware for windows 11 and want to go tiny

I did feature Linux Mint in a passed video, when I was testing with the 4080 Super. I plan to feature Linix Mint in a future video as well.

@@casuallygamin9 You shouldn't put Linux Mint in a row with CachyOS and Nobara, first it's better to compare Linux Mint with all other big distributions like Ubuntu, Fedora, OpenSUSE. And then include the winner in a row with distributions aimed at gaming - CachyOS, Nobara, PikaOS (basically debian based CachyOS with high compability)

I will try that approach down the road

Yea, AMD cards are performing quite nice in Linux. Nobara is quite well optimized, after Glorious Eggroll knows what he is doing and he tweaked the distro especially for gaming.

@@casuallygamin9 yah for sure, idk why were not recommending nobara by default to linux curious gamers

This is why I feature it as well, to bring visibility to it alongside CachyOS

@@niamhrz because rpm sucks.

@@DJDocsVideos well we reccomend bazzite which is still rpm and i dont like that one so im just confused why we don't just go with nobara as reccomended

Yea, it performs better than in Windows

In 1440p at least, in 4k they do hover at 100%

I didn't dig in why CachyOS shines in this situation. I will check this in a future video, where I will focus more on CPU performance.

@@casuallygamin9 thank you!

Yes, Nvidia needs to work on their Linux drivers. Hopefully it will not take years to be at least on par with Windows

What makes you think that in this video ? He doesn't talk about nvidia

@@xenio8736 I am comparing to the last video

I did a video about Nvidia GPUs 2 weeks ago, where I tested a few distros using the 4080 Super.

@@casuallygamin9 Oh yes okay, I even saw and commented that video haha. Yeah it's sad, because dlss is so awesome

Dragon Age: The Veilguard is the low perfomance

Yes, I think that is the cause why CachyOS performs worse in Dragon Age, I used by mistake Proton GE

something that you messed up here is the mesa. prolly why nobara did pull ahead. saw that you wrote mesa 24.3 on em all, but mangohud is showing 25.0 devel bleeding edge. Cachyos is pronounced like cashyos, like money

I'm not sure to be honest. I added the info that I saw when I used the command line command to see the driver version, I used the same for both Linux distros. That is the info I added in the chart.

...maybe I applied updates after I used that command and I captured that video ,🤔

Yes, I believe this is having a toll on people considering Linux for gaming.

Better performance than this is a bigger issue for me. I'm not throwing the performance of my 4090 down the toilet for Linux.

I would recommend to try a few and see which one you like the most. I like Arch based diatros as those ahve the latest patches and improvements available. It depends a lot of the hardware you have, so try and see which one is best for you.

@@casuallygamin9 Currently I'm still running GTX 1080 with 5900x, weird combo I know. I'll be definitely doing some distro hopping after getting new SSD to back up my files etc. Obviously gonna have to still dual boot for games that have anti-cheat measures so ugly that only mother could love, but that's just how it goes.

I'm editing the video, as we speak, were I use the 7900 XTX

super excited to see that one, thanks for all your hard work!

Editing as we speak. I think I will have it by Tuesday the latest

The best cooling, from the 3 mentioned, is on the merc card, but it can get loud because of the aggressive fan curve. I would get that one if you don't mind twerking the fun curve

@casuallygamin9 but i kinda like the sapphire aesthetics, should I take it insted of the other 2 just for that?

It's a great GPU, it's the one I'm rocking right now and I'm gonna release a video comparing Windows 11 to Linux for gaming.