I think a temporary fix for Indiana Jones currently is swapping to the "amdvlk" driver rather than using "radv" from what people have said it fixes a lot of the glitches and visual bugs. Not sure if you tried that or not
@@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 would go with something already set up for gaming, if that is what you are looking for. CachyOS, Nobara or Bazzite comes to my mind. There are other distros as well, but what I mentioned, I had a look at those. The reality is that you can choose whichever you like, but you will need make some tweeks you're self, at least for gaming.
@@notaras1985 If you are willing to learn a bit more about linux and take the responsibility for it maintenance, pick cachyos. if you want something a bit easier, Nobara. Both are great. I'm using Nobara at the moment and I'm pretty happy.
Every time i tried Linux it just felt so smooth in comparison, and the 0,1% low and 1%low shows it here. Its are much more stable and enjoyable experience.
IMPORTANT: in the video I mention that I use in CachyOS in Lutris Proton Experimental, but in fact I used ProtonGE, and thus explaining the lower performance in Dragon Age: The Veilguard. If some parts of the captured runs are a bit choppy, it's because I used an aggressive encoding in order to have better image quality.
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.
Damn, the 0.1% and 1% lows are increadible on linux, almost the same as the avg fps. This just shows how much background noise there is on windows causing fps drops. How was it installing the games and running them on linux? Was it hard?
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.
I have not fully watched it, but I wanna say thank you! You promised it and you delivered! Wow I bet that took a while. Comparing Nvidia GPUs and AMD GPUs on Linux and you are not a RUclipsr with 1mio Subs.. so this all costs Money AND Time. THANKS!!! Edit: Okay, after finishing the video, I think it's pretty safe to say: If you have an AMD GPU, Performance is on par with WIndows and sometimes even better! But RayTracing is still better on Windows.
So, what is that "graphics reset" thing? Seems that happen to me when playing Portal with RTX (only at some levels). Here I use Garuda KDE, with an RX 6800 GPU.
6:37 Horizon Forbidden West 4k Does Windows actually utilizes 2 times more CPU time? Would be nice to find a way to represent power consumption comparison for Linux distributions, so we could the practical representation of OS efficiency. Where 7800x3D has plenty of headroom - other CPU will choke under Windows. I have experienced this on my Intel Core Ultra 125H playing DOTA 2 - 4 cores 100% utilized all the time consuming 70-80% of the TDP leaving iGPU with 5-7W On the Linux none of the cores were loaded more than 50%, laptop was cooler and FPS 30% higher. Still not perfect, but night and day difference with Windows.
So, for what I can see what I learned is that RTX sucks on AMD on Linux, everything else is a W for Linux using AMD (except for counted exceptions). More FPS, and more consistent on Linux (better 1% and 0.1% lows) than on Windows.
@ These practices is the reason why i don't game anymore. Now shoving dlss everywhere and call it "optimization" is enough sign that devs are lazy and unwilling to deliver any well made product. rt isn't even that great on nvidia so why bother? I will not buy it as simple as that.
Amazing video. I wish you included linux mint as well. Nevertheless this makes me quite sad about my nvidia 4080. I wish I had amd so I could jump on to linux permanently
@@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'm glad I wasn't imagining things with my 7900XT in Horizon Forbidden West, I was watching clips of the XT in windows and was like but I'm getting better performance??
Wow, check the RAM usage. The difference is huge. Its on average 6Gb of ram more used on Windows. The worst offender is Stalker 2 with 12-14Gb of more RAM used on Windows. Its ridiculous!
When it comes to calculating averages, it is best to use the geometric mean. It will weight all results the same, while the average will weight the higher numbers much higher which is not usefull. Example: System 1: 20 FPS and 100 FPS, average is 60 FPS, geometric mean is 44.7. System 2: 40 FPS and 80 FPS, average again is 60, geometric mean is 56.6. System 2 delivered way better, 20% lower performance is not that much compared to 100% more on the second game. This is clearly reflected by the geometric mean.
@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.
OH! i've been looking forward to these, These bench's are wild! I'm surprised how good nobara pulls through I almost certainly assumed cachyos would be on top in most benchmarks! I am currently questioning why I'm using Nvidia on Linux after this LMAO buttttt tbf that dlss & dlss-fg is always superior in image stability to fsr and that's why I bought it when I did. Nice video! It's also wild to see how far Linux has come. & SO MANY GAMES BEATING WINDOWS ON LINUX ON AMD?
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.
@@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
It's incredible to think that Linux can run games that are not even disegned to be used on it and perform well, just imagine if these games were native, actually, don't, take a look at war thunder, in my case at least, it works amazing, good graphics, responsive, amazing perfomance even in the movie mode. Linux is amazing.
It's almost native if developers decided to implement vulkan/opengl API instead of directX. While maintaining their own libraries instead of windows DLLs, this would make their games run natively on both platforms.
Native isn't always better... it really depends how good the developer is. I have a few Linux native games I still run the Windows version with Proton because its faster than native.
This is how it is reported in Linux, some say that is a bug and in kernel 6.13 it will be fixed. Soon that kernel will be out of testing and we will see.
Thanks, great video. I have similar setup, but I was thinking to go with Bazzite. Do you have any experience with? I never heard for CachyOS or Nobara before.
I've used mint and nobara for dual booting with windows 10 for couple of years and i quite like nobara. But ui cant use it for gaming as both mint and nobara refuse to acknowledge that my TV has freesync. Found no possible solutions and neither got any answers from reddit. This also isnt the new hdmi 2. something problem. The tv has older hdmi and works perfectly in windows.
Perfect case for Linux Gaming: 1. If you are playing older (80s retro to mid 2010s) games without Ray Tracing (promising tech, just not quite there yet) and without stupid anti-cheat (plain BS) multiplayer games... 2. have AMD hardware. The further you get from this the more you will need Windows. I will say it straight: LINUX GAMING > WIN GAMING (FOR YOUR CASE) Here is when you might want to go with Windows: You got a monster intel/nvidia PC and you want to play latest single player games and you really like Raytracing (huge performance downgrade on Linux and a few potential incompatibility) You are a Simulation genre fan and/or VR fan and buy a lot of accessories... WILL PROBABLY WANT WINDOWS FOR BEST COMPATIBILITY AND PERFORMANCE! WORST CASE SCENARIO: You almost exclusively play latest competitive multiplayer games that use anti-cheats... JUST FORGET ABOUT LINUX FOR GAMING!
You are correct, Linux, for the time being works good with an AMD gpu, without RT, HD, and you will need to forget about some competitive games. Let's see if in the future, things will progress. Don't forget that most open source drivers and where Linux is right now is because of a community, not a company. I'm curious what Valve can do in the upcoming years to improve Linux.
We arent superrior yet, but linux gaming development isn't on the same level as windows, but i think very soon we will be in good place for gaming. I do not have any issues with gaming on amd setup in linux.
Very interesting results. Are the gains over Win11 caused solely by better AMD drivers or is it also the OS tweaks? Btw. it'd be helpful to put the game name (and version) somewhere on the screen. Thanks for your content!
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
nice work - as anticipated linux in front on amd by 5-10% but looking more like ~6% with way better frame consistency. i have both nvidia and amd and always knew amd just feels waaaaay smoother. idea for a future vid - amd proprietary vs open source drivers - while we know open source will win - in RT it is likely a different question. not sure how far behind windows they are in RT - but last time i saw it was way better than mesa.
I never tried Bazzite. I know that it is based on Fedora 41 now. I wanted to feature it in my upcoming Arc B580 video, but it's still on Kernel version 6.11 and battalmage requires version 6.12 or newer. CachyOS is good as well, alongside Nobara and I suppose Bazzite should be good. Don't think that there are major difference between distros, at least when it comes to gaming.
Linux is not only good for gaming, but it appears to be objectively a better experience, unless you are addicted to one of these rootkit malware type of anticheat games. And the gap is only going to get bigger.
no HDR still isn't fully implemented, but the necessary protocols have been approved. All we are waiting for now is for the compositors to support it. I would expect it sometime next year tbh
And things are moving so fast. I'm on Fedora (which Nobara and Bazzite are based on) and there is not a week without getting some optimization for the graphic stack, be it a new kernel, a new version of the Wayland compositor (in my case Mutter), of Mesa or Proton. If you are a gamer, you should absolutely use a Fedora or Arch based distro, as they get those updates faster. (If you are a beginner, maybe skip the Arch based distros like CachyOS for now unless you are a very technical person, because things can get messy). Game compatibility and optimization are better now than they were but a month ago, and it just keeps getting better.
@@casuallygamin9 Not that hard to use, I agree, and less so since the archinstall script was implemented, but if you do not read a release note informing you that you need to do a manual intervention before applying an update, it can bork your system. They are quite rare, but I still remember the switch from SysVInit to SystemD. You need to know what you are doing and babysit your system a lot more than with most distros. The community is also a bit less friendly to newcomers, to say the least. That being said, Arch is awesome.
So Linux uses less VRAM, way lesser RAM. And less power draw for the GPU for better average fps numbers in most titles... It's amazing tbh. Thank you for the benchmark.
@@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.
How did you obtain the 1%/.1% low numbers? Unless I missed something these are the same on linux for every single game. This would mean the slowest 1% of frames all have the exact same frame time. That is highly unlikely, I suspect there is an error with measuring or calculating the numbers. While linux performance might be better, this issue makes it look insanely smooth, but that is probably too good to be true
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.
Your windows version is drawing approx 50 more watts than Linux. Fyi AMD fixed power management in kernel 6.13 from what I hear. Otherwise you need to manually set your gpu to gaming or compute mode as it doesn’t boost automatically like in windows. So windows has an unfair advantage.
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
@@casuallygamin9 wall outlet just shows how much the powersupply pulls power not how much the components consume. Due to efficiency standpoint, PSUs always draw more power than what components needs to ensure stable power delivery. So measuring watts from wall outlet doesn't conclude anything.
What I showed in the video is that the power from the wall is the same on both OSs, but I mentioned in the video that this hide small variances. As when it comes to PSU eficience, as the total power pulled from the wall is the same, I would expect that the power used by the whole PC is more or less the same.
yes you can game on linux and it has come a long way and hopefully will soon compete with windows since not everyone will be updating to windows 11. but linux isn'tfor everyone. as a junior penetration tester i work with linux everyday and i game on windows because the fact is while you can game on linux you won't be able to play games with eac like rust and many more. if you don't care about those then yes you can get the distro of your choice and yes you can play some games through vm but not always that easy to make work. yes protondb will show all the games that are working but with updates files break and fixes are done by you not the developer. i have two pc's one is my work machine the other my gaming rig. some people don't have that option so some people say dual boot i wouldn't do it something always goes wrong. now if you have a windows machine with a vm of linux to try it out play with it break it learn from it yes but 2025 is around the corner and windows 10 out the window. linux with a vm of windows 10 you can try it but you will also run into problems since its on a vm machine install tiny11 wouldn't go that route with no security it would make it easy for people to have access to your pc. you have options 1 buy a new rig with windows 11 2 install linux and do research which is best for your needs not being able to game everything 3 install tiny11 and meet people like me on your pc and they will not be a white hat hacker like me if you gonna go with linux i would advice search the 50 basic commands in linux and learn them. install a vm on your machine install linux and learn before you switch linux is great if i wasn't playing games now and then i would only have linux been working with linux almost 3 years and i believe linux is the future awesome stuff.
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
I'm not sure what you mean, it is quite good, if you would give it a shot maybe you would like it. When it comes to Windows, nowdays you pay for Microsoft to gather data and train their AI.
I think a temporary fix for Indiana Jones currently is swapping to the "amdvlk" driver rather than using "radv" from what people have said it fixes a lot of the glitches and visual bugs. Not sure if you tried that or not
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
Using cachyos for gaming and it's great. Especially when I got amd cpu and gpu. I got dual boot, win 10 and cachyos, but I like cachy more.
You did a fantastic job. Thanks for your work! I wish prosperity to your channel!
Thanks for the support
that smooth, smooth frametime and high 1% on linux, i notice that too in most of my games
Yes, the lows are much better on Linux as opposed to Windows 11
@@casuallygamin9 so for new PC builders and gamers which Linux OS is ideal sir?
I would go with something already set up for gaming, if that is what you are looking for. CachyOS, Nobara or Bazzite comes to my mind. There are other distros as well, but what I mentioned, I had a look at those. The reality is that you can choose whichever you like, but you will need make some tweeks you're self, at least for gaming.
@@notaras1985 If you are willing to learn a bit more about linux and take the responsibility for it maintenance, pick cachyos. if you want something a bit easier, Nobara.
Both are great.
I'm using Nobara at the moment and I'm pretty happy.
Every time i tried Linux it just felt so smooth in comparison, and the 0,1% low and 1%low shows it here.
Its are much more stable and enjoyable experience.
@@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.
IMPORTANT: in the video I mention that I use in CachyOS in Lutris Proton Experimental, but in fact I used ProtonGE, and thus explaining the lower performance in Dragon Age: The Veilguard.
If some parts of the captured runs are a bit choppy, it's because I used an aggressive encoding in order to have better image quality.
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 ,🤔
Damn, the 0.1% and 1% lows are increadible on linux, almost the same as the avg fps.
This just shows how much background noise there is on windows causing fps drops.
How was it installing the games and running them on linux? Was it hard?
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.
Incredible. SO nvidia drivers do suck under LInux....... I am quite upset at nvidia looking at these results
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
WOOOO.
gamemode enabled this time!
Good job, man.
My 4080 performs like a 4070ti on linux on some games like wukong and i still dont boot into windows to play anything
same :) windows is gone. hoping that Nvidia will fix their drivers soon
How are you capturing the footage on linux?
If you are using the same pc, you should be using gpu-screen-recorder. That thing is amazing.
I'm using a second mini pc
I have not fully watched it, but I wanna say thank you! You promised it and you delivered! Wow I bet that took a while. Comparing Nvidia GPUs and AMD GPUs on Linux and you are not a RUclipsr with 1mio Subs.. so this all costs Money AND Time.
THANKS!!!
Edit: Okay, after finishing the video, I think it's pretty safe to say: If you have an AMD GPU, Performance is on par with WIndows and sometimes even better! But RayTracing is still better on Windows.
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.
Well done, great work!
Fantastic video! Keep it up! :)
Ty for the benchmarks body!
Thank you for making this.
5:10 Windows looks more choppy than the left one or am I tweaking?
Use frame time graph as I used an aggressive encoder when capturing the video, my bad here. If you see spikes there, then you ges frame drops
So, what is that "graphics reset" thing? Seems that happen to me when playing Portal with RTX (only at some levels). Here I use Garuda KDE, with an RX 6800 GPU.
Age of mythology is interesting.
The performance is better on Linux, but with the same CPU usage and LESS GPU usage ...?
That's so strange
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!
Thank you for making the video :))
6:37 Horizon Forbidden West 4k
Does Windows actually utilizes 2 times more CPU time?
Would be nice to find a way to represent power consumption comparison for Linux distributions, so we could the practical representation of OS efficiency.
Where 7800x3D has plenty of headroom - other CPU will choke under Windows.
I have experienced this on my Intel Core Ultra 125H playing DOTA 2 - 4 cores 100% utilized all the time consuming 70-80% of the TDP leaving iGPU with 5-7W
On the Linux none of the cores were loaded more than 50%, laptop was cooler and FPS 30% higher. Still not perfect, but night and day difference with Windows.
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'm not gonna waste my time trying to fix every little problem on Linux, but thanks I guess
So, for what I can see what I learned is that RTX sucks on AMD on Linux, everything else is a W for Linux using AMD (except for counted exceptions). More FPS, and more consistent on Linux (better 1% and 0.1% lows) than on Windows.
Yes, that is yhe case
Seems like a win win for me. I don't care about rt since it takes out performance anyway.
@@ThePhenoix504FI3D1 Sadly, some games in the future might not allow you to turn RTX off. This could be a potential problem.
@
These practices is the reason why i don't game anymore.
Now shoving dlss everywhere and call it "optimization" is enough sign that devs are lazy and unwilling to deliver any well made product.
rt isn't even that great on nvidia so why bother? I will not buy it as simple as that.
@@ThePhenoix504FI3D1 I've been playing older games (up to 2018ish) and is about when gaming peaked (Y).
You are the best! Thank you so much for your work
Very useful test, without water, only info what we need
Amazing video. I wish you included linux mint as well. Nevertheless this makes me quite sad about my nvidia 4080. I wish I had amd so I could jump on to linux permanently
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
I'm glad I wasn't imagining things with my 7900XT in Horizon Forbidden West, I was watching clips of the XT in windows and was like but I'm getting better performance??
Yea, it performs better than in Windows
Wow, check the RAM usage. The difference is huge.
Its on average 6Gb of ram more used on Windows.
The worst offender is Stalker 2 with 12-14Gb of more RAM used on Windows. Its ridiculous!
Well, Windows has a lot of things. It allocates more than it needs, but it is not efficient due to a lot of bloatware
When it comes to calculating averages, it is best to use the geometric mean. It will weight all results the same, while the average will weight the higher numbers much higher which is not usefull.
Example:
System 1: 20 FPS and 100 FPS, average is 60 FPS, geometric mean is 44.7.
System 2: 40 FPS and 80 FPS, average again is 60, geometric mean is 56.6.
System 2 delivered way better, 20% lower performance is not that much compared to 100% more on the second game. This is clearly reflected by the geometric mean.
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.
OH! i've been looking forward to these,
These bench's are wild! I'm surprised how good nobara pulls through I almost certainly assumed cachyos would be on top in most benchmarks!
I am currently questioning why I'm using Nvidia on Linux after this LMAO buttttt tbf that dlss & dlss-fg is always superior in image stability to fsr and that's why I bought it when I did. Nice video! It's also wild to see how far Linux has come. & SO MANY GAMES BEATING WINDOWS ON LINUX ON AMD?
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
It's incredible to think that Linux can run games that are not even disegned to be used on it and perform well, just imagine if these games were native, actually, don't, take a look at war thunder, in my case at least, it works amazing, good graphics, responsive, amazing perfomance even in the movie mode.
Linux is amazing.
It's almost native if developers decided to implement vulkan/opengl API instead of directX.
While maintaining their own libraries instead of windows DLLs, this would make their games run natively on both platforms.
Native isn't always better... it really depends how good the developer is. I have a few Linux native games I still run the Windows version with Proton because its faster than native.
Great video. Impressive how AMD performs better with Linux compared to NVIDIA on your other video.
Why GPU has 300w limit on Linux?
This is how it is reported in Linux, some say that is a bug and in kernel 6.13 it will be fixed. Soon that kernel will be out of testing and we will see.
Thanks, great video.
I have similar setup, but I was thinking to go with Bazzite. Do you have any experience with?
I never heard for CachyOS or Nobara before.
I never tested Bazzite to be honest so I can share any info regarding it
Bazzite runs really good actually. Its Fedora based like Nobara, but I think I like it a bit better.
I've used mint and nobara for dual booting with windows 10 for couple of years and i quite like nobara. But ui cant use it for gaming as both mint and nobara refuse to acknowledge that my TV has freesync. Found no possible solutions and neither got any answers from reddit. This also isnt the new hdmi 2. something problem. The tv has older hdmi and works perfectly in windows.
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
Take notes Nvidia
Perfect case for Linux Gaming:
1. If you are playing older (80s retro to mid 2010s) games without Ray Tracing (promising tech, just not quite there yet) and without stupid anti-cheat (plain BS) multiplayer games...
2. have AMD hardware. The further you get from this the more you will need Windows. I will say it straight: LINUX GAMING > WIN GAMING (FOR YOUR CASE)
Here is when you might want to go with Windows:
You got a monster intel/nvidia PC and you want to play latest single player games and you really like Raytracing (huge performance downgrade on Linux and a few potential incompatibility)
You are a Simulation genre fan and/or VR fan and buy a lot of accessories... WILL PROBABLY WANT WINDOWS FOR BEST COMPATIBILITY AND PERFORMANCE!
WORST CASE SCENARIO: You almost exclusively play latest competitive multiplayer games that use anti-cheats... JUST FORGET ABOUT LINUX FOR GAMING!
You are correct, Linux, for the time being works good with an AMD gpu, without RT, HD, and you will need to forget about some competitive games. Let's see if in the future, things will progress. Don't forget that most open source drivers and where Linux is right now is because of a community, not a company. I'm curious what Valve can do in the upcoming years to improve Linux.
Very thorough comparison 👍 only one thing: You might want to include the name of the game being shown somewhere in your overlay :)
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
We arent superrior yet, but linux gaming development isn't on the same level as windows, but i think very soon we will be in good place for gaming. I do not have any issues with gaming on amd setup in linux.
Like I said in the video, for aingle player games with RT on AMD GPUs, linux is a bit better.
Very interesting results. Are the gains over Win11 caused solely by better AMD drivers or is it also the OS tweaks? Btw. it'd be helpful to put the game name (and version) somewhere on the screen. Thanks for your content!
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
nice work - as anticipated linux in front on amd by 5-10% but looking more like ~6% with way better frame consistency. i have both nvidia and amd and always knew amd just feels waaaaay smoother. idea for a future vid - amd proprietary vs open source drivers - while we know open source will win - in RT it is likely a different question. not sure how far behind windows they are in RT - but last time i saw it was way better than mesa.
I heard that the AMD proprietary drivers are better for RT then the open source drivers in Linux, will keep that in mind.
So nobara is best for gaming? How is bazzite?
I never tried Bazzite. I know that it is based on Fedora 41 now. I wanted to feature it in my upcoming Arc B580 video, but it's still on Kernel version 6.11 and battalmage requires version 6.12 or newer. CachyOS is good as well, alongside Nobara and I suppose Bazzite should be good. Don't think that there are major difference between distros, at least when it comes to gaming.
@casuallygamin9 in the video nobara had much better lows than catchy
Yes, a bit better
Linux is not only good for gaming, but it appears to be objectively a better experience, unless you are addicted to one of these rootkit malware type of anticheat games. And the gap is only going to get bigger.
Would be interesting to see Nobara 41 you can upgrade to it already, the nobara wiki has a guide for it.
I will cover it in a later video
Does HDR work easily in either distro? That's one feature that I wouldn't want to go without
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
no HDR still isn't fully implemented, but the necessary protocols have been approved. All we are waiting for now is for the compositors to support it. I would expect it sometime next year tbh
And things are moving so fast. I'm on Fedora (which Nobara and Bazzite are based on) and there is not a week without getting some optimization for the graphic stack, be it a new kernel, a new version of the Wayland compositor (in my case Mutter), of Mesa or Proton. If you are a gamer, you should absolutely use a Fedora or Arch based distro, as they get those updates faster. (If you are a beginner, maybe skip the Arch based distros like CachyOS for now unless you are a very technical person, because things can get messy). Game compatibility and optimization are better now than they were but a month ago, and it just keeps getting better.
True, but I must say that Arch is not that hard to use, pacman is nice, but I think Nobara it's a bit easier for new people unfamiliar with Linux
@@casuallygamin9 Not that hard to use, I agree, and less so since the archinstall script was implemented, but if you do not read a release note informing you that you need to do a manual intervention before applying an update, it can bork your system. They are quite rare, but I still remember the switch from SysVInit to SystemD. You need to know what you are doing and babysit your system a lot more than with most distros. The community is also a bit less friendly to newcomers, to say the least. That being said, Arch is awesome.
So Linux uses less VRAM, way lesser RAM. And less power draw for the GPU for better average fps numbers in most titles... It's amazing tbh. Thank you for the benchmark.
@@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
Yeah when you remove all that spyware, and bloatware and telemetry, it becomes better
Why not Bazzite?
I will have a video with Bazzite down the road
@casuallygamin9 great
Большое спасибо за такое интересное видео ❤
Didn't know cachy was said as catchy
It is a catchy distro
How did you obtain the 1%/.1% low numbers? Unless I missed something these are the same on linux for every single game. This would mean the slowest 1% of frames all have the exact same frame time.
That is highly unlikely, I suspect there is an error with measuring or calculating the numbers.
While linux performance might be better, this issue makes it look insanely smooth, but that is probably too good to be true
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.
Your windows version is drawing approx 50 more watts than Linux. Fyi AMD fixed power management in kernel 6.13 from what I hear. Otherwise you need to manually set your gpu to gaming or compute mode as it doesn’t boost automatically like in windows. So windows has an unfair advantage.
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
@@casuallygamin9
wall outlet just shows how much the powersupply pulls power not how much the components consume.
Due to efficiency standpoint, PSUs always draw more power than what components needs to ensure stable power delivery.
So measuring watts from wall outlet doesn't conclude anything.
What I showed in the video is that the power from the wall is the same on both OSs, but I mentioned in the video that this hide small variances. As when it comes to PSU eficience, as the total power pulled from the wall is the same, I would expect that the power used by the whole PC is more or less the same.
@@casuallygamin9 Update to kernel 6.13 and test again, lets see the differences!
Testing was done? Is it stable? I'm using right now the B580, and it's a mess on Linux, a lot of graphical glitches.
yes you can game on linux and it has come a long way and hopefully will soon compete with windows since not everyone will be updating to windows 11. but linux isn'tfor everyone. as a junior penetration tester i work with linux everyday and i game on windows because the fact is while you can game on linux you won't be able to play games with eac like rust and many more. if you don't care about those then yes you can get the distro of your choice and yes you can play some games through vm but not always that easy to make work.
yes protondb will show all the games that are working but with updates files break and fixes are done by you not the developer.
i have two pc's one is my work machine the other my gaming rig. some people don't have that option so some people say dual boot i wouldn't do it something always goes wrong.
now if you have a windows machine with a vm of linux to try it out play with it break it learn from it yes but 2025 is around the corner and windows 10 out the window.
linux with a vm of windows 10 you can try it but you will also run into problems since its on a vm machine
install tiny11 wouldn't go that route with no security it would make it easy for people to have access to your pc.
you have options
1 buy a new rig with windows 11
2 install linux and do research which is best for your needs not being able to game everything
3 install tiny11 and meet people like me on your pc and they will not be a white hat hacker like me
if you gonna go with linux i would advice search the 50 basic commands in linux and learn them.
install a vm on your machine install linux and learn before you switch
linux is great if i wasn't playing games now and then i would only have linux been working with linux almost 3 years
and i believe linux is the future awesome stuff.
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
Nobara has better low 1% than CachyOS. How that? I tought the CachyOS is blazing fast 😂
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
10:20 when I hear a Ukrainian radio I know it's Stalker, lol (didn't recognise the game at first)
You can see the names in the timespans, I didn't included the name of ecah game on the acreen
@@casuallygamin9 I noticed that afterwards, timestamps are not always visible
Would you say it's better to add the game name somewhere on the screen?
@@casuallygamin9 yup
Dragon Age? Seriously?
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.
fake os with fake name shitty Linux, foe poor people who cant afdord windows 🤣🤣🤣🤣🤣
I'm not sure what you mean, it is quite good, if you would give it a shot maybe you would like it. When it comes to Windows, nowdays you pay for Microsoft to gather data and train their AI.
You always get these weirdos. who are anti Linux post On a Linux testing video.
Welll, I don't mind them, I was curious actually if he ever tried Linux