If you cherry pick with a very select set of hardware and games, you can maybe edge out Windows 11 occasionally. Oh wow. Let's all bail on Windows and never be able to play competitive games ever again!
@@chad_8313 Even when cherry picking, the fact that games made for Windows only run better on an entirely different operating system with a compatibility layer made by open source developers in their free time out of passion is fascinating isn't it? It's sad that some games choose to not include the proton-compatible versions of their anti cheat (looking at Destiny 2, Fortnite, PUBG). Others like Apex Legends, Genshin Impact, The Division 2, didn't used to work but now run beautifully after they've included the proton compatible anti-cheat which is a dead simple process.
@@chad_8313exactly, and half a year ago it was even less, and before that even less, and if we go back further Linux was always a bit behind, further still and it was extremely behind. I can't see why people are optimistic about such cherry picked games
It's not hard to edge out windows. Just keep both systems up to date and watch windows get slower and slower after each update. Even on the rare occasion an update breaks something it's dead simple, and not even 5 minutes to revert to pre update with Linux.
@@Thornskade this was supposed to be a response at another comment, apparently it didn't go through right. Edit: Actually, no. I am referencing the frame times part of the video where it say that arch barely edges out windows, as if comparing fps.
@@entity-bl5og It is comparing fps, the minimum frames per second, so higher is better and lower is worse. It is not showing the time between frames, it is showing frames per second
@@entity-bl5og Yes, but the number they're showing represents the lowest fps they got. Not the highest amount of milliseconds between a frame they got. I'm just repeating myself at this point and I don't know how else to tell you this. I happen to speak German so I visited the website and that's what they did. There isn't much else I can do to convince you but it really, truly is the case and Gardiner evidently did his research
Hmmm? When it comes to frametimes, lower is usually better - not the bigger number. What do they even equalize here into percent? A single number for an avg. frametime is not helpful. You want to see the variation and spikes to get an idea how well frame output is handled and matches a constant fps. We want a smooth flat constant fast output - not big variations and spikes. Therefore we need a graph of frametimes to see how well and stutterfree a picture is produced. Also, the specs of the systems seem not equal and up to date. Arch ran a Kernel 6.6, KDE 5.27 and Mesa 23.2.1, Nobara a Kernel 6.4, KDE 5.27 and Mesa 23.1.6, and Pop!_OS a Kernel 6.5., Gnome 42 and Mesa 23.2.1 again by comparison. My Nobara runs Kernel 6.6.3, KDE 5.27.9 and Mesa 23.3.0 since last weeks update, and was not that far behind before that.
Agreed that 3 games doesn't say much. But this is a very good first step. Curious how gaming on Linux will be in a few years. I'll just install windows or Linux whatever is better. Competition is good for gamers
My theory is that sure, Proton introduces lag and slowdowns because it's a compatibility layer. But Linux itself is so much lighter than Windows that the device has more horsepower to burn.
I assumed that Micro'slop and Bug'thesda deserved each other when I first heard rumors of the acquisition. The accusation of a supposed Linux VKD3D developer that "PC optimized" Starfield improperly implemented DirectX12 and wasted GPU performance was completely believable, even though I know nothing about the Reddit user who posted the claim.
1% lows are much more important that max frame rate as it shows how much the game will stutter. Windows having the much better 1% lows shows that the drivers are most likely better in windows. Also cherry picking 5 games and saying linux is better than windows because of avg frame rates is just misleading
this will depend a lot on hardware wont it, i can’t imagine an nvidia gpu performing better on linux than windows thanks to proprietary drivers, and i’m sure not giving up ray tracing either
And other things, like what binds the game, if the game is CPU bound, linux will probally easly beat it out, GPU bound it depends on the GPU. NVIDIA drivers have recently gotten a bit better on linux due to updates. and which of the proticals the game uses for rendering, Dirrect X older than 11 often performs better on linux due to the GPU manufacturers not optimizing the older ones as much
No really on hardware since the test is relative. More on the quality of the graphical stack. This includes operating system parts, drivers and in this case compatibility layers.
i had some issues with starfield in my ubuntu system it crashed every time in the prolog when the next tunnel gets freed. so i have some games that i play on windows. most of the games run on linux fine. except for raytracing. i got live rendering with raytracing not running but in image processing it workes. i have tested it. i unchecked the hardware raytracing and it took nearly 5 times longer to make the same image
The bias is showing here again. you don’t need to try and prove to anyone that Linux is gaming capable. you’re so excited to show a Linux win that you overlook basic things like frame time being higher is worse. Take your time, do your research, do better.
Because anti cheat isn't an absolute plague that makes games run worse than hacked versions without it. Last I checked game pass games were mostly older anyways, and I can get a lot more games I actually wanted through gog/steam sales for the monthly price.
Why is Pop!_OS, the gaming distro, lagging behind so far? Haha! Also, Arch is just a kernel; what software you use can matter. You could go barebones basic, with the xfce desktop and Pulse Audio, or you could make it prettier than Windows with Wayland KDE and PipeWire . What desktop were they using? What sound server? What other drivers, servers, and daemons were running in the background?
Yes but keep in mind that non of these games were even made with Linux in mind, so they're using no Linux native APIs whatsoever. All the translations that are happening in the background by Wine, DXVK and others are bound to affect performance in new and interesting ways.
Nope, not necessarily. The order in the frametimes graph is actually misleading and the equalization into percent makes no sense (what would be 100% then?). Lower frametimes are better. Frametimes is the "gap" time between each frame. When a game runs at 60 fps, the frametime would be 16.6ms, or 1/60th of a second - presuming the frames would be executed equally at constant pacing. At 144 fps (1/144th of a second) this would mean there’s a frame every 6.9ms. Of course frames are not pumped out at a constant rate - some take longer, some are quick. The less variance, the better - making for a smoother experience. It is when suddenly you get a spike and a frame takes way longer than usual, that you notice - usually in the form of "micro stutters". So we want as little spikes as possible and as low of frametimes as possible. Best would be almost flatline near zero. 1%-lows (and 0.1%-lows) give you an idea how severe such spikes occur, where the framerate drops. Low fps = high frametimes. Again, we prefer a constant almost even output. The lesser the fps cave in, the better - staying at constant low and equal frametimes (according to the possible fps output) at all times. From comparisons between Windows and Linux gameplay in the past (e.g. by Bero Tech), the experience is, that Linux and Vulkan usualy make for a more constant smooth experience, albeit being not necessarily the fastest. So, past experience on Linux was: a little lower avg. FPS compared to Windows, but often better 1%-lows and much smoother and more constant frametimes.
@@breadone_ I'm not saying there's not room for improvement, but realistically gaming on Linux is still a niche market so I'm not surprised the large studios are ignoring it for now. All I was trying to point out was that it can be considered a miracle these games run on Linux at all, let alone perform as well as they do, even if it's worse than on the OS they were actually designed for. And in some cases the experience on Linux is in fact not worse but better than on Windows. It's rare but it does happen.
Linux Mint. Blows Windows out of the water. I'm switching to it today. I'm officially divorcing with Microsoft after a 30+ year relationship. Microsoft is too needy now, need this data, need that data bla bla. F that. I'm out.
Yep, Microsoft is in part to blame for the troubling performance. Bethesda plays it's part with it's choice of obsolete engine functions, bad coding and poor optimization, as we have learned to expect by now. But MS did its part in insisting late in production after take-over the game should be a DX12 title now when it was build from the start with Vulkan AND to insist on the release date still. That's why AMD was asked to help out getting the code transferred - which resulted in a "partnership" only a few short months before release.
Lol imagine getting beaten by a platform running your own game with a compatibility layer.
Devs working on DXVK, VKD3D and WINE deserve a standing ovation for all their hardwork!
If you cherry pick with a very select set of hardware and games, you can maybe edge out Windows 11 occasionally.
Oh wow. Let's all bail on Windows and never be able to play competitive games ever again!
@@chad_8313 Even when cherry picking, the fact that games made for Windows only run better on an entirely different operating system with a compatibility layer made by open source developers in their free time out of passion is fascinating isn't it? It's sad that some games choose to not include the proton-compatible versions of their anti cheat (looking at Destiny 2, Fortnite, PUBG). Others like Apex Legends, Genshin Impact, The Division 2, didn't used to work but now run beautifully after they've included the proton compatible anti-cheat which is a dead simple process.
@@chad_8313exactly, and half a year ago it was even less, and before that even less, and if we go back further Linux was always a bit behind, further still and it was extremely behind. I can't see why people are optimistic about such cherry picked games
It's not hard to edge out windows. Just keep both systems up to date and watch windows get slower and slower after each update. Even on the rare occasion an update breaks something it's dead simple, and not even 5 minutes to revert to pre update with Linux.
Lower frame time is better, it is measured in miliseconds iirc
Here it isn't, they're showing fps, not milliseconds
@@Thornskade this was supposed to be a response at another comment, apparently it didn't go through right.
Edit: Actually, no. I am referencing the frame times part of the video where it say that arch barely edges out windows, as if comparing fps.
@@entity-bl5og It is comparing fps, the minimum frames per second, so higher is better and lower is worse. It is not showing the time between frames, it is showing frames per second
@@Thornskade0:20 is a frame times graph.
@@entity-bl5og Yes, but the number they're showing represents the lowest fps they got. Not the highest amount of milliseconds between a frame they got. I'm just repeating myself at this point and I don't know how else to tell you this. I happen to speak German so I visited the website and that's what they did. There isn't much else I can do to convince you but it really, truly is the case and Gardiner evidently did his research
I run an arch linux pc and I can confirm for at least helldivers (Since i was binging it) it runs so much better with leas powerdraw than on windows
Hmmm? When it comes to frametimes, lower is usually better - not the bigger number. What do they even equalize here into percent?
A single number for an avg. frametime is not helpful. You want to see the variation and spikes to get an idea how well frame output is handled and matches a constant fps. We want a smooth flat constant fast output - not big variations and spikes. Therefore we need a graph of frametimes to see how well and stutterfree a picture is produced.
Also, the specs of the systems seem not equal and up to date.
Arch ran a Kernel 6.6, KDE 5.27 and Mesa 23.2.1,
Nobara a Kernel 6.4, KDE 5.27 and Mesa 23.1.6, and
Pop!_OS a Kernel 6.5., Gnome 42 and Mesa 23.2.1 again by comparison.
My Nobara runs Kernel 6.6.3, KDE 5.27.9 and Mesa 23.3.0 since last weeks update, and was not that far behind before that.
It's FPS, not a percentage
@@mgord9518 I am very sure it says on both performance rating charts for FPS and Frame Times as "Unit: percent". So it is in percentage, not FPS
@@raison7478 it's fps.
@@raison7478Go to their site, it's fps
Imagine thinking a bigger number for frame time is better lol.
Agreed that 3 games doesn't say much.
But this is a very good first step. Curious how gaming on Linux will be in a few years.
I'll just install windows or Linux whatever is better.
Competition is good for gamers
Yeah, just heard it and was thinking wtf? Didn't I hear him just say that?
No biggy, I'm not an all knowing entity myself.
It is better because they're showing it in minimum fps
@@pauldrice1996 .
.
It's not that difficult to outperform Windows 11, considering the amount of bloat and spyware which comes with it.
My theory is that sure, Proton introduces lag and slowdowns because it's a compatibility layer.
But Linux itself is so much lighter than Windows that the device has more horsepower to burn.
I assumed that Micro'slop and Bug'thesda deserved each other when I first heard rumors of the acquisition. The accusation of a supposed Linux VKD3D developer that "PC optimized" Starfield improperly implemented DirectX12 and wasted GPU performance was completely believable, even though I know nothing about the Reddit user who posted the claim.
I saw these graphs in another video somewhere, can someone ELI5 what all of this means? Frametimes, 1% lows, etc
is there a Distro designed specifically for gaming?
ChimeraOS, Bazzite and several others
Nobara and Garuda are the most popular.
@@phishENchimps as for now, CachyOS is the best gaming distro EVER, including it's really cool cachy-bore kernel and v3 level packages
1% lows are much more important that max frame rate as it shows how much the game will stutter. Windows having the much better 1% lows shows that the drivers are most likely better in windows. Also cherry picking 5 games and saying linux is better than windows because of avg frame rates is just misleading
this will depend a lot on hardware wont it, i can’t imagine an nvidia gpu performing better on linux than windows thanks to proprietary drivers, and i’m sure not giving up ray tracing either
And other things, like what binds the game, if the game is CPU bound, linux will probally easly beat it out, GPU bound it depends on the GPU. NVIDIA drivers have recently gotten a bit better on linux due to updates. and which of the proticals the game uses for rendering, Dirrect X older than 11 often performs better on linux due to the GPU manufacturers not optimizing the older ones as much
No really on hardware since the test is relative. More on the quality of the graphical stack.
This includes operating system parts, drivers and in this case compatibility layers.
i had some issues with starfield in my ubuntu system it crashed every time in the prolog when the next tunnel gets freed. so i have some games that i play on windows. most of the games run on linux fine. except for raytracing. i got live rendering with raytracing not running but in image processing it workes. i have tested it. i unchecked the hardware raytracing and it took nearly 5 times longer to make the same image
The bias is showing here again. you don’t need to try and prove to anyone that Linux is gaming capable. you’re so excited to show a Linux win that you overlook basic things like frame time being higher is worse.
Take your time, do your research, do better.
The frametimes is in fps so being higher is indeed better
They probably rigged the results by adding too little ram, and then the obvious result in Linux because it uses less ram
You maybe get 5 more avg fps
But you can kiss the game pass
And a lot of anti cheat games like valorant 💀
Because anti cheat isn't an absolute plague that makes games run worse than hacked versions without it. Last I checked game pass games were mostly older anyways, and I can get a lot more games I actually wanted through gog/steam sales for the monthly price.
Why is Pop!_OS, the gaming distro, lagging behind so far? Haha!
Also, Arch is just a kernel; what software you use can matter. You could go barebones basic, with the xfce desktop and Pulse Audio, or you could make it prettier than Windows with Wayland KDE and PipeWire . What desktop were they using? What sound server? What other drivers, servers, and daemons were running in the background?
@@Monkey_on_Call arch is not “just a kernel”
Higher frame time is not better wtf short form content sucks ass everyone's overconfident and wrong lol
Okay, but isn't frame time more important than overall frame rate? So, the games run faster on Linux, but there are also more stuttery on Linux.
Yes but keep in mind that non of these games were even made with Linux in mind, so they're using no Linux native APIs whatsoever. All the translations that are happening in the background by Wine, DXVK and others are bound to affect performance in new and interesting ways.
Nope, not necessarily. The order in the frametimes graph is actually misleading and the equalization into percent makes no sense (what would be 100% then?).
Lower frametimes are better. Frametimes is the "gap" time between each frame. When a game runs at 60 fps, the frametime would be 16.6ms, or 1/60th of a second - presuming the frames would be executed equally at constant pacing. At 144 fps (1/144th of a second) this would mean there’s a frame every 6.9ms. Of course frames are not pumped out at a constant rate - some take longer, some are quick. The less variance, the better - making for a smoother experience. It is when suddenly you get a spike and a frame takes way longer than usual, that you notice - usually in the form of "micro stutters".
So we want as little spikes as possible and as low of frametimes as possible. Best would be almost flatline near zero.
1%-lows (and 0.1%-lows) give you an idea how severe such spikes occur, where the framerate drops. Low fps = high frametimes.
Again, we prefer a constant almost even output. The lesser the fps cave in, the better - staying at constant low and equal frametimes (according to the possible fps output) at all times.
From comparisons between Windows and Linux gameplay in the past (e.g. by Bero Tech), the experience is, that Linux and Vulkan usualy make for a more constant smooth experience, albeit being not necessarily the fastest.
So, past experience on Linux was: a little lower avg. FPS compared to Windows, but often better 1%-lows and much smoother and more constant frametimes.
@@RudyBleeker so? make excuses all you want but a worse experience is a worse experience
I noticed both lower FPS and more stutters in games like Baldur's Gate 3. But the game was still very playable, just not as smooth as Win 10.
@@breadone_ I'm not saying there's not room for improvement, but realistically gaming on Linux is still a niche market so I'm not surprised the large studios are ignoring it for now. All I was trying to point out was that it can be considered a miracle these games run on Linux at all, let alone perform as well as they do, even if it's worse than on the OS they were actually designed for. And in some cases the experience on Linux is in fact not worse but better than on Windows. It's rare but it does happen.
Catch : they tested all the games that would run
Pretty much every game on steam works on linux now.
Except online games with anticheat almost every game runs in linux
.
oh man watch out, everybody jump on the linux bandwagon linux has such a long way to go before it will even really compete with windows performance
Linux Mint. Blows Windows out of the water. I'm switching to it today. I'm officially divorcing with Microsoft after a 30+ year relationship. Microsoft is too needy now, need this data, need that data bla bla. F that. I'm out.
"Microsoft's own starfield"
We just pretending that they made that now?
Bethesda is a part of the Microsoft gaming division led by Phil Spencer. Microsoft has owned them since 2021.
Yep, Microsoft is in part to blame for the troubling performance. Bethesda plays it's part with it's choice of obsolete engine functions, bad coding and poor optimization, as we have learned to expect by now. But MS did its part in insisting late in production after take-over the game should be a DX12 title now when it was build from the start with Vulkan AND to insist on the release date still. That's why AMD was asked to help out getting the code transferred - which resulted in a "partnership" only a few short months before release.
.
How does Linux vs Windows compare in regards to modern competitive shooters?
Oh wait... you can't play them.
Can't play valorant? I call that a feature
Cs2 runs and being not able to play valorant isnt a big loss
.