The team at Gamers Nexus did a lot of work on analysing and benchmarking on Linux albeit with a lot of caveats. I recommend watching the twenty-minutes introduction where Steve discusses the...
The team at Gamers Nexus did a lot of work on analysing and benchmarking on Linux albeit with a lot of caveats. I recommend watching the twenty-minutes introduction where Steve discusses the complexity of benchmarking on Linux and how they approached it.
What was really interesting is how inconstent Nvidia is on Linux and huge 1% low drops versus AMD. The definitive advantage that Nvidia has over AMD on Windows doesn't seem to apply on Linux where 9070 XT seems very good and far cheaper than the alternatives.
It'll be interesting to see what else they do to improve their drivers. I think we'll see a lot of devs start optimising their games for AMD since that's what the steam machine uses, not to...
It'll be interesting to see what else they do to improve their drivers.
I think we'll see a lot of devs start optimising their games for AMD since that's what the steam machine uses, not to mention Valve has a vested interest in pushing for AMD, as does AMD themselves.
Meanwhile only Nvidia will be really pushing to improve Nvidia stuff.
I watched this video yesterday, and while I do understand most of their rationale for choosing bazzite, I do wish they chose CachyOS for two reasons: I use it I think it would be meaningful to see...
I watched this video yesterday, and while I do understand most of their rationale for choosing bazzite, I do wish they chose CachyOS for two reasons:
I use it
I think it would be meaningful to see how often rolling release and the latest updates affect performance as well. Maybe if they did a video that like even twice a year it would be good content for those who would be interested.
Why CachyOS over Arch? In every benchmark I've seen it "outperforms" Arch by 1-2% in exchange for using performance patches with 1/1000th of the visibility that Arch has. In my opinion that's just...
Why CachyOS over Arch?
In every benchmark I've seen it "outperforms" Arch by 1-2% in exchange for using performance patches with 1/1000th of the visibility that Arch has. In my opinion that's just a recipe for a bad experience down the line.
I think they made a good argument against rolling release for benchmarking -- you want shelf stable benchmarks you can append to for at least 6 months. They could use Arch and just not update it, but you do risk landing on a broken package.
I don't have enough knowledge about performance patches and visibility - I just used it because it seemed wildly recommended by others on the internet and would probably be easier for a simpleton...
I don't have enough knowledge about performance patches and visibility - I just used it because it seemed wildly recommended by others on the internet and would probably be easier for a simpleton like me to setup.
I think they made a good argument against rolling release for benchmarking
Unfortunately I think there is a lot of cargo culting in some of what I'll call the pop-Linux community. You can tell a lot of it has been filtered through layers of retellings and while well...
I just used it because it seemed wildly recommended by others on the internet
Unfortunately I think there is a lot of cargo culting in some of what I'll call the pop-Linux community. You can tell a lot of it has been filtered through layers of retellings and while well intentioned, has been stripped of a lot of nuance.
I'm not trying to make you feel bad about your choice or anything, there's nothing wrong with CachyOS if you know what you're getting. I'm just surprised that's what a lot of people are reaching for first.
Most of what CachyOS does is apply a bunch of configuration options and patches to the linux kernel, include some kernel modules you may or may not need, and then re-compile all the Arch packages with various compiler flags for optimization. Which sounds nice, but when a new kernel releases it probably hasn't been seen by very many people running the same weird subset of patches that you're running which could lead to problems.
If you want an out-of-the-box Arch experience, the best option is probably EndeavorOS, which IIRC just uses Arch packages directly. Then if you want to experiment with more performance just install the Zen kernel which does a lot of the same things. If it ever breaks, just boot into the normal or -lts kernel.
At this point, with dxvk, vkd3d, Vulkan and the sheer amount of work Valve have put into both SteamOS and Proton, the only true hurdle preventing gamers from switching en masse to Linux is......
At this point, with dxvk, vkd3d, Vulkan and the sheer amount of work Valve have put into both SteamOS and Proton, the only true hurdle preventing gamers from switching en masse to Linux is... kernel level anticheat.
League of Legends, Fortnite, Valorant, Phantasy Star Online 2, MapleStory, Battlefield 6, Helldivers 2, Blade & Soul and any other game that requires kernel level anti-cheat is unplayable on Linux. Even Valve's own offerings aren't 100% fully compatible, with many DOTA 2 and CS2 players using FACEIT, a third-party esports matchmaking platform that requires... you guessed it... kernel level anti-cheat.
The only kernel level solution that actually has Proton/Linux support is BattlEye and many developers choose to opt out of Linux support because of how much harder it is to detect cheats on that OS.
Valve would be wise to sell their upcoming Steam Machine at a loss to bring as many players to their ecosystem as possible, but they've already confirmed they won't do that.
On that point I dropped my favourite and most-played game on Windows when they introduced a kernel level anti-cheat: League of Legends. I realise that it doesn't give them access to much more than...
On that point I dropped my favourite and most-played game on Windows when they introduced a kernel level anti-cheat: League of Legends.
I realise that it doesn't give them access to much more than a regular administrator process but considering it loads before almost anything else it's the perfect place for a malware infection to hide. I wouldn't trust the security practices of Riot to be robust enough to risk such an attack vector.
So whilst it's unfortunate that these games block Linux I think it's an acceptably small number of games to skip to avoid the risk of additional exploit channels.
Yeah while I understand not everyone will want to make the same choices, and it was easier for me since most of my multiplayer days are behind me, but I decided that making the tradeoff of missing...
Yeah while I understand not everyone will want to make the same choices, and it was easier for me since most of my multiplayer days are behind me, but I decided that making the tradeoff of missing out on some games, even potentially ones I would have some level of interest in, was worth it to abandon Windows, and I have never regretted the decision. Turns out most of the games that don't work are games I would never want to play anyway for any number of reasons, whether it be deep anti-cheat PC access or otherwise and nearly everything I have tried and wanted to play has worked out of the box or with slight tweaking
For certain games sure, there is no need to trust the client at all. Kind of hard to do in the genres that have anti-cheat though. There is a big difference between translating game objects into...
If kernel anti cheat was good enough they'd roll your loot on your client and the server would accept it, but they're obviously well aware that doesn't make any sense to do in practice.
For certain games sure, there is no need to trust the client at all. Kind of hard to do in the genres that have anti-cheat though. There is a big difference between translating game objects into visible pixels for every person in the game vs. generating a random number and recording the result. And also when client inputs are inherently skill-based (e.g. mouse input in a FPS game) trusting the client is a hard requirement of the game.
The perception has already been shifting toward blaming the devs that deliberately make restrictive decisions rather than blaming the platform. I think it puts that whole model of Windows exclusive anti cheat on a clock.
I don't think this is true either. FWIW I'm a Linux-stan who's been on Linux for 15+ years and anti-cheat on Linux is a growing problem, not a shrinking one. As more people adopt Linux this problem will get worse and not better. The fact is that anti-cheat on Linux is fundamentally broken until Valve or somebody else offers a locked down and signed kernel that requires secure boot/TPM attestation that anti-cheat developers could reasonably develop on top of.
I don't think many networked systems other than games are concerned with the concept of fairness. Yes, you can do a lot server side but certainly not everything. For instance if I train my aimbot...
They can and should use server side validation to flag suspicious behaviors of clients just as other networked systems do. Other networked systems don't try to pretend that their problems go away just by trying to install something onto the client.
I don't think many networked systems other than games are concerned with the concept of fairness. Yes, you can do a lot server side but certainly not everything. For instance if I train my aimbot to operate within the 95-99th percentile of human performance you aren't detecting me on the server without a lot of false positives.
It's about compromise. If in a game like Valorant or CS:GO the server determined the player's positions and lines of sight, the game would be unplayable. That's OK in like, league of legends. It's...
If kernel anti cheat was good enough they'd roll your loot on your client and the server would accept it
It's about compromise. If in a game like Valorant or CS:GO the server determined the player's positions and lines of sight, the game would be unplayable. That's OK in like, league of legends.
It's not about right or wrong, even just the ping from the speed of light would make it impossible, let alone an actual noisy and slow network connection.
I find this to be a particularly unconvincing argument. a) The idea that every single live service developer decided that they could do server-side detection, but instead decide to invest time,...
I find this to be a particularly unconvincing argument.
a) The idea that every single live service developer decided that they could do server-side detection, but instead decide to invest time, money, and serious reputational damage on creating a kernel-level anti cheat does not add up to me. I do not see why a game developer would go this route, if another route existed.
b) Valve has, for the past 8 years attempted to do ML-based server side detection, along with ML-based user classification based on non-game activity. They also have the greatest incentive to make this work. Yet, despite this, the system is largely a failure. Cheaters are rampant at higher ranks, and the leaderboard has at many points served as a mechanism for cheaters to boast. Also, there was a point you could get banned for spinning your mouse too fast.
c) For a Linux CS2 player, there is no viable way for them to become a professional, and qualify for a Valve-sponsored event, and get their signature into the game. Every step beyond in-game matchmaking requires a kernel-level anti cheat, whether it is for the external MM service used by pros, FACEIT (or any of their competitors like GamersClub or ESPORTAL), or to play in online competitive tournaments. Prior to Valve abandoning direct qualifiers, all qualifiers hosted for Major qualifications required a kernel-level anti cheat.
Disclaimer, I don't have a strong opinion on where cheat detection should be done. But, I did see some of your points and feel like I can respond to those from a more general perspective. For...
Disclaimer, I don't have a strong opinion on where cheat detection should be done. But, I did see some of your points and feel like I can respond to those from a more general perspective.
The idea that every single live service developer decided that they could do server-side detection, but instead decide to invest time, money, and serious reputational damage on creating a kernel-level anti cheat does not add up to me. I do not see why a game developer would go this route, if another route existed.
For starters, it isn't developers making these calls. Companies (management) are deciding on this and they often decides things not purely on technical merits.
From that perspective there are a variety of potential reasons for this.
One of them is the same reason why on the web these days, websites are so front-end heavy. A lot of the modern frameworks are doing client side rendering and generation of all sorts of things. This offloads a lot of the server side requirements to the clients, which at scale can be a huge cost saving.
In a similar sense, expanding anti-cheat detection on the server means that you need to maintain a lot more of the game state and situation on the server. Not only that, you need to then also do a lot of extra calculations on the server as well.
In addition to that, from a cynical cost benefit calculation, kernel level anti-cheat is currently accepted enough to just invest in that area. Combined with the fact that I believe it likely also is cheaper, I think the risk for reputation damage is something most companies are willing to accept.
In fact, the comment you linked on reddit basically states the same thing but framed slightly different. The comment might be right that at is not feasible to keep everything server-side from a cost perspective. Or it simply might be a way of saying "it could be done, but nobody wants to cut in the margins because of it".
I should also note that this comment is over 8 years old. It merely comments on the state back then. I am not entirely sure if Valve actually invested all that much in anti-cheat detection since then. Again, a bit cynical, but it just needs to be "good enough" for people to not walk away from the game.
Also, don't discount inertia in big companies to simply keep doing things that they are already doing. Moving everything server side requires different expertise which needs to be built up. Not just about server side cheat detection, but also net code and all that.
Companies, even if willing, are slow to adjust in these areas. Going back to the cynical cost benefit calculation, the "solution" they currently invested in is accepted by most customers and sales don't seem to hurt.
I mean, I would love it if all detection was server-side and effective. I just don't think that this is at all feasible. (Also, one point I didn't mention: none of this would cover informational...
Disclaimer, I don't have a strong opinion on where cheat detection should be done.
I mean, I would love it if all detection was server-side and effective. I just don't think that this is at all feasible.
(Also, one point I didn't mention: none of this would cover informational cheats, like wallhacks. Developers should do the basics here, but there is a limit, as at some point the client has to know where the enemy is. If you make that point the literal moment they are visible, the pop-in is really bad.)
In a similar sense, expanding anti-cheat detection on the server means that you need to maintain a lot more of the game state and situation on the server. Not only that, you need to then also do a lot of extra calculations on the server as well.
So, maintaining the game state actually provides user-value in replay solutions, something which the Counter-Strike series has had since the 90s, and VALORANT is getting soon due to user-demand.
Like, fair, there is a cost to observing and analysing behaviour, but I'm still not convinced that cost is more than having to maintain a kernel-level solution, buying and observing cheats to develop signatures for detection, etc.
It merely comments on the state back then. I am not entirely sure if Valve actually invested all that much in anti-cheat detection since then. Again, a bit cynical, but it just needs to be "good enough" for people to not walk away from the game.
GDC talk from 2018 that goes more into depth about their solution. Data mining indicates they've even recently acquired NVIDIA H200s, which is frontier tier tech. Not to mention their active branding of the solution as VAC Live for the launch of CS2. I'd say it is pretty clear they are actively investing into it.
Also, don't discount inertia in big companies to simply keep doing things that they are already doing. Moving everything server side requires different expertise which needs to be built up.
But there has been a change here. For Battlefield, they used to use PunkBuster (a third-party non-kernel anti-cheat), before adding FairFight (a third party behavioural anti-cheat), before abandoning all of those to create an in-house team to develop EA AntiCheat. Similarly, CoD migrated from PunkBuster to VAC, to I think an in-house non-kernel solution, and finally to developing their own in-house solution.
These companies historically contracted out, but are now developing internal capabilities for securing their games.
Valve can't even roll out an effective anti-cheat in their own games, which is why everybody who plays Counter Strike even remotely seriously plays on ESEA or FACEIT. VAC Live has genuinely been a...
Valve can't even roll out an effective anti-cheat in their own games, which is why everybody who plays Counter Strike even remotely seriously plays on ESEA or FACEIT.
VAC Live has genuinely been a total shitshow, and I think a big reason for that is because they decided not to go down the kernel level route.
I got bit by Rockstar implementing BattlEye for Grand Theft Auto Online and refusing to implement the Linux version. I have a character with nearly every property asset in the game, tons of...
I got bit by Rockstar implementing BattlEye for Grand Theft Auto Online and refusing to implement the Linux version. I have a character with nearly every property asset in the game, tons of achievements, and $40 million in the bank just sitting there, untouchable due to not having Windows.
The most telling thing for me is that every game I have with these Windows only kernel level anti-cheats still have blatant cheating. It's been a few years since I played Fortnite, but when I did...
The most telling thing for me is that every game I have with these Windows only kernel level anti-cheats still have blatant cheating. It's been a few years since I played Fortnite, but when I did play it was filled with people using auto-aiming, auto-build macros, etc. So either these companies are grossly misinformed, or they're not resisting Linux because it would increase cheating.
For Fortnite, I honestly believe they won't enable Linux support because they can't support Epic Games Store on it. If they didn't have Fortnite tied to EGS, they would have to show what a colossal failure it has been.
Hoyoverse games (genshin, honkai, zenless, etc) all use kernel level anti cheat that doesn't run on Linux and have pretty massive playerbases, so people who play those probably won't make the jump...
Hoyoverse games (genshin, honkai, zenless, etc) all use kernel level anti cheat that doesn't run on Linux and have pretty massive playerbases, so people who play those probably won't make the jump any time soon either.
In case anyone wants to check specific games in regards to anti-cheat on Linux: https://areweanticheatyet.com Quite a few multiplayer-games work fine on Linux, e.g. the recently released Arc...
I don’t really see how you can square that circle. Personally kernel level anticheat is table stakes in a serious competitive game. The fact that CSGO doesn’t have it is a serious demerit (and why...
I don’t really see how you can square that circle. Personally kernel level anticheat is table stakes in a serious competitive game. The fact that CSGO doesn’t have it is a serious demerit (and why face it and ESEA exists). But Linux just isn’t architectures in a way where having kernel level anticheat makes sense.
There is only one way to square that circle: Valve's Steam Machines using a custom signed kernel + TPM. Doesn't really help anyone who does not have a Steam Machine but that's the only way you'd...
There is only one way to square that circle: Valve's Steam Machines using a custom signed kernel + TPM. Doesn't really help anyone who does not have a Steam Machine but that's the only way you'd be able to play these games on "Linux" (really Steam Machines only).
Apparently not all anticheat is the same. I've been playing Helldivers 2 on Linux almost since release (I started a couple weeks later) and I've had very few issues. Most of the problems I have...
Apparently not all anticheat is the same. I've been playing Helldivers 2 on Linux almost since release (I started a couple weeks later) and I've had very few issues. Most of the problems I have encountered have been unrelated to my OS and were problems other players complained about regardless of platform
Ideally, Valve will come up with an anti-cheat Linux solution with controls in place that make it a little bit less of a rootkit, keylogger security risk. Maybe a "game mode" minimal kernel in...
Ideally, Valve will come up with an anti-cheat Linux solution with controls in place that make it a little bit less of a rootkit, keylogger security risk. Maybe a "game mode" minimal kernel in steamOS with way more restrictions, which would basically just be a console?
I made the switch to Bazzite on my gaming laptop a few months ago and it's been surprisingly painless. Especially compared to my attempt to switch to Linux Mint on the same laptop earlier this...
I made the switch to Bazzite on my gaming laptop a few months ago and it's been surprisingly painless. Especially compared to my attempt to switch to Linux Mint on the same laptop earlier this year. I'm not always playing the latest and greatest (I think Baldur's Gate 3 is my most recent game?), but it suits my needs nicely. For everything else there's my oft neglected xbox series S and PS4.
I should try Bazzite. I liked Mint well enough on my laptop, but had these little issues consistently and one which I felt was a major issue (fans not spinning up during one particular game) and...
I should try Bazzite. I liked Mint well enough on my laptop, but had these little issues consistently and one which I felt was a major issue (fans not spinning up during one particular game) and so I eventually went back to Windows, since I was just constantly investigating and solving issues, rather than playing games.
Love to finally see some professional level game benchmarks on Linux. It pretty clearly confirms what we guessed about the general advantages and disadvantages depending on the hardware used....
Love to finally see some professional level game benchmarks on Linux. It pretty clearly confirms what we guessed about the general advantages and disadvantages depending on the hardware used.
Valve says they are working to help get raytracing up to par on AMD, will be interesting to see how far they can take it.
While I have had a good experience on Linux with Nvidia, I am curious if the frame-time consistency and 1%/0.1% lows issues are things I have ever encountered, because they are very apparent in...
While I have had a good experience on Linux with Nvidia, I am curious if the frame-time consistency and 1%/0.1% lows issues are things I have ever encountered, because they are very apparent in these results. That said I am not on Bazzite or a similar Arch/etc base
As GN admits, the ecosystem is so complex and rapidly changing on Linux that it's possible some users may never have the same experience as the benchmark results show, because of the interwoven complexity between the OS, drivers, kernels, hardware, etc
I really appreciate the work Steve and all the GN staff have done to help make it easier for the average person to transition to Linux. Good on them. Also, their investigative journalism is beyond...
I really appreciate the work Steve and all the GN staff have done to help make it easier for the average person to transition to Linux. Good on them.
Also, their investigative journalism is beyond world class. Just amazing.
The gist of it seems to be, an enormous amount of progress has been made but there's still a long way to go. That matches my roughly ten days of experience trying to game on Linux as well. I am...
The gist of it seems to be, an enormous amount of progress has been made but there's still a long way to go. That matches my roughly ten days of experience trying to game on Linux as well.
I am running Arch with Hyprland and despite tiling window managers not playing so nice with games in general, I've had less issues with it than CachyOS (Gnome). I tried CS2, ARC Raiders, and some indie titles like Megabonk and BALL X PIT, and all of them worked from OK to great. The game I had issues with the most was, surprisingly, Valve's own CS2. I had issues where the console wouldn't accept my text input, my cursor would escape to my second monitor (despite using gamescope's --force-grab-cursor) and the performance was terrible despite running it at a very low resolution (1280x960). For some reason, re-installing it solved all my issues. My FPS was limited to 400 and I couldn't set it to unlimited but my primary monitor is 360hz so it wasn't a big deal. The game was fluid enough.
I agree with @Bullmaestro that the biggest hurdle, from my not-so-technically-minded point of view is anti-cheat. Some games that require anti-cheat does work. Helldivers 2 has already been mentioned and I'll add that ARC Raiders work just as fine too. But that's not good enough. I don't know how suitable Linux is for the current kernel level anti-cheats, but Valve either need to prove that they can beat cheating just as easily without invasive techniques, or come up with another solution fully compatible with Linux.
My guess is that Valve does not care about these games all that much, at least yet. They're a patient company, they like to take their time. They will likely focus on the experience of couch co-ops, platformers and the likes to increase their share of the pie, as well as Linux's, to make a more compelling case that Linux is indeed a viable option to Windows.
Most of the popular anti-cheat systems already have Linux or Proton compatibility, but it requires the game developers to enable it. How effective they are without a locked kernel is as debatable...
or come up with another solution fully compatible with Linux
Most of the popular anti-cheat systems already have Linux or Proton compatibility, but it requires the game developers to enable it. How effective they are without a locked kernel is as debatable as how effective kernel-level anti-cheat on windows is anyway.
I've started suggesting to anyone that can't switch because of one or two games with anti-cheat that they just play those titles on xbox or ps5 instead since most of the PC advantages of customizing or modding wont apply to those titles anyway. Not something that goes over well with the PC-only crowd though.
I don't know how effective Easy, BattlEye and the likes are, but I'm more familiar with the Faceit's own AC and it's pretty good. You can still get past it but it requires hardware that costs...
I don't know how effective Easy, BattlEye and the likes are, but I'm more familiar with the Faceit's own AC and it's pretty good. You can still get past it but it requires hardware that costs thousands of dollars. I don't know what they're doing differently than others, I'm not technically literate enough to differentiate them, but it does work.
If people already have consoles, I guess that's an option. I have a console myself but I also have a beefy PC and I just don't feel like turning on my console. I'm rather old school though, I simply cannot imagine playing a shooter on a gamepad.
It doesn't and neither does any other anticheat. All client-side ACs have working software cheats. I would guess that there are probably hardware solutions that are way cheaper than thousands of...
I don't know how effective Easy, BattlEye and the likes are, but I'm more familiar with the Faceit's own AC and it's pretty good. You can still get past it but it requires hardware that costs thousands of dollars.
It doesn't and neither does any other anticheat. All client-side ACs have working software cheats. I would guess that there are probably hardware solutions that are way cheaper than thousands of dollars as well.
The main purpose of AC is to maintain the appearance of competitive integrity to keep players engaged in the game. Completely stopping cheating is out of scope, what's important is blocking obvious cheating such that cheaters have to resort to looking like skilled players.
Rootkit anti cheat is disgusting and the only way I'd countenance such a thing is if a system was built entirely and only for gaming. Why in the ever loving Arkleseizure would I do anything,...
Rootkit anti cheat is disgusting and the only way I'd countenance such a thing is if a system was built entirely and only for gaming. Why in the ever loving Arkleseizure would I do anything, anything at all, like my banking or use any other credential at all, when there's something as invasive and creepy as that installed on the computer. The sheer hubris of these companies thinking they have the right to such access.
So, a console then, pretty much.
Of course, I built a system based on Endeavor OS (Arch), store it out of sight and use WOL and Sunshine/Moonlight to play it. All the games that want to creep on my activity can get in the bin.
Faceit has a trust and safety team that monitors potential cheaters based on the system they themselves have implemented and from the reports they receive. Me saying "it does work" doesn't mean it...
Faceit has a trust and safety team that monitors potential cheaters based on the system they themselves have implemented and from the reports they receive. Me saying "it does work" doesn't mean it cannot be bypassed either by software or hardware, but software based solutions are relatively easy to detect and it's only a matter of time before they take an action against it, whereas hardware ones especially the DMA ones, are extremely hard to detect. (If the cheater is a seasoned player who knows how to act innocent, good luck.)
It is, as you said, a simple game of economics and increasing the barrier to entry. If someone's willing to buy themselves a second computer and a special hardware from some shady dealer that costs thousands of dollars, you're going to have a harder time catching them than to catch some kid in his basement who downloaded an .exe file from "csaimbot.net" or whatever.
I rarely play CS2 matches anymore (despite playing CS2 almost every day) but whenever I have a Premier game, most headshots feel suspicious, whereas I don't have that doubt with Faceit. Does that mean it can't happen? No, but it's extremely unlikely. That's the point. Like it or not, and I absolutely don't like it, kernel level ACs do a better job.
The team at Gamers Nexus did a lot of work on analysing and benchmarking on Linux albeit with a lot of caveats. I recommend watching the twenty-minutes introduction where Steve discusses the complexity of benchmarking on Linux and how they approached it.
What was really interesting is how inconstent Nvidia is on Linux and huge 1% low drops versus AMD. The definitive advantage that Nvidia has over AMD on Windows doesn't seem to apply on Linux where 9070 XT seems very good and far cheaper than the alternatives.
It makes sense to me since Valve has been pushing a lot of the Linux gaming scene forward and their hardware all uses AMD chips.
It will be interesting to see if the Nvidia situation improves as the Nova module begins to reach maturity
I think so, it seems that Nvidia is finally investing and invested in the Linux Desktop.
It'll be interesting to see what else they do to improve their drivers.
I think we'll see a lot of devs start optimising their games for AMD since that's what the steam machine uses, not to mention Valve has a vested interest in pushing for AMD, as does AMD themselves.
Meanwhile only Nvidia will be really pushing to improve Nvidia stuff.
I watched this video yesterday, and while I do understand most of their rationale for choosing bazzite, I do wish they chose CachyOS for two reasons:
Why CachyOS over Arch?
In every benchmark I've seen it "outperforms" Arch by 1-2% in exchange for using performance patches with 1/1000th of the visibility that Arch has. In my opinion that's just a recipe for a bad experience down the line.
I think they made a good argument against rolling release for benchmarking -- you want shelf stable benchmarks you can append to for at least 6 months. They could use Arch and just not update it, but you do risk landing on a broken package.
I don't have enough knowledge about performance patches and visibility - I just used it because it seemed wildly recommended by others on the internet and would probably be easier for a simpleton like me to setup.
I never said they did not.
Unfortunately I think there is a lot of cargo culting in some of what I'll call the pop-Linux community. You can tell a lot of it has been filtered through layers of retellings and while well intentioned, has been stripped of a lot of nuance.
I'm not trying to make you feel bad about your choice or anything, there's nothing wrong with CachyOS if you know what you're getting. I'm just surprised that's what a lot of people are reaching for first.
Most of what CachyOS does is apply a bunch of configuration options and patches to the linux kernel, include some kernel modules you may or may not need, and then re-compile all the Arch packages with various compiler flags for optimization. Which sounds nice, but when a new kernel releases it probably hasn't been seen by very many people running the same weird subset of patches that you're running which could lead to problems.
If you want an out-of-the-box Arch experience, the best option is probably EndeavorOS, which IIRC just uses Arch packages directly. Then if you want to experiment with more performance just install the Zen kernel which does a lot of the same things. If it ever breaks, just boot into the normal or -lts kernel.
At this point, with dxvk, vkd3d, Vulkan and the sheer amount of work Valve have put into both SteamOS and Proton, the only true hurdle preventing gamers from switching en masse to Linux is... kernel level anticheat.
League of Legends, Fortnite, Valorant, Phantasy Star Online 2, MapleStory, Battlefield 6, Helldivers 2, Blade & Soul and any other game that requires kernel level anti-cheat is unplayable on Linux. Even Valve's own offerings aren't 100% fully compatible, with many DOTA 2 and CS2 players using FACEIT, a third-party esports matchmaking platform that requires... you guessed it... kernel level anti-cheat.
The only kernel level solution that actually has Proton/Linux support is BattlEye and many developers choose to opt out of Linux support because of how much harder it is to detect cheats on that OS.
Valve would be wise to sell their upcoming Steam Machine at a loss to bring as many players to their ecosystem as possible, but they've already confirmed they won't do that.
On that point I dropped my favourite and most-played game on Windows when they introduced a kernel level anti-cheat: League of Legends.
I realise that it doesn't give them access to much more than a regular administrator process but considering it loads before almost anything else it's the perfect place for a malware infection to hide. I wouldn't trust the security practices of Riot to be robust enough to risk such an attack vector.
So whilst it's unfortunate that these games block Linux I think it's an acceptably small number of games to skip to avoid the risk of additional exploit channels.
Yeah while I understand not everyone will want to make the same choices, and it was easier for me since most of my multiplayer days are behind me, but I decided that making the tradeoff of missing out on some games, even potentially ones I would have some level of interest in, was worth it to abandon Windows, and I have never regretted the decision. Turns out most of the games that don't work are games I would never want to play anyway for any number of reasons, whether it be deep anti-cheat PC access or otherwise and nearly everything I have tried and wanted to play has worked out of the box or with slight tweaking
For certain games sure, there is no need to trust the client at all. Kind of hard to do in the genres that have anti-cheat though. There is a big difference between translating game objects into visible pixels for every person in the game vs. generating a random number and recording the result. And also when client inputs are inherently skill-based (e.g. mouse input in a FPS game) trusting the client is a hard requirement of the game.
I don't think this is true either. FWIW I'm a Linux-stan who's been on Linux for 15+ years and anti-cheat on Linux is a growing problem, not a shrinking one. As more people adopt Linux this problem will get worse and not better. The fact is that anti-cheat on Linux is fundamentally broken until Valve or somebody else offers a locked down and signed kernel that requires secure boot/TPM attestation that anti-cheat developers could reasonably develop on top of.
I don't think many networked systems other than games are concerned with the concept of fairness. Yes, you can do a lot server side but certainly not everything. For instance if I train my aimbot to operate within the 95-99th percentile of human performance you aren't detecting me on the server without a lot of false positives.
Yea I'll admit that's a pretty good argument against client side anti-cheat or at least their limitations.
It's about compromise. If in a game like Valorant or CS:GO the server determined the player's positions and lines of sight, the game would be unplayable. That's OK in like, league of legends.
It's not about right or wrong, even just the ping from the speed of light would make it impossible, let alone an actual noisy and slow network connection.
I find this to be a particularly unconvincing argument.
a) The idea that every single live service developer decided that they could do server-side detection, but instead decide to invest time, money, and serious reputational damage on creating a kernel-level anti cheat does not add up to me. I do not see why a game developer would go this route, if another route existed.
b) Valve has, for the past 8 years attempted to do ML-based server side detection, along with ML-based user classification based on non-game activity. They also have the greatest incentive to make this work. Yet, despite this, the system is largely a failure. Cheaters are rampant at higher ranks, and the leaderboard has at many points served as a mechanism for cheaters to boast. Also, there was a point you could get banned for spinning your mouse too fast.
c) For a Linux CS2 player, there is no viable way for them to become a professional, and qualify for a Valve-sponsored event, and get their signature into the game. Every step beyond in-game matchmaking requires a kernel-level anti cheat, whether it is for the external MM service used by pros, FACEIT (or any of their competitors like GamersClub or ESPORTAL), or to play in online competitive tournaments. Prior to Valve abandoning direct qualifiers, all qualifiers hosted for Major qualifications required a kernel-level anti cheat.
Disclaimer, I don't have a strong opinion on where cheat detection should be done. But, I did see some of your points and feel like I can respond to those from a more general perspective.
For starters, it isn't developers making these calls. Companies (management) are deciding on this and they often decides things not purely on technical merits.
From that perspective there are a variety of potential reasons for this.
One of them is the same reason why on the web these days, websites are so front-end heavy. A lot of the modern frameworks are doing client side rendering and generation of all sorts of things. This offloads a lot of the server side requirements to the clients, which at scale can be a huge cost saving.
In a similar sense, expanding anti-cheat detection on the server means that you need to maintain a lot more of the game state and situation on the server. Not only that, you need to then also do a lot of extra calculations on the server as well.
In addition to that, from a cynical cost benefit calculation, kernel level anti-cheat is currently accepted enough to just invest in that area. Combined with the fact that I believe it likely also is cheaper, I think the risk for reputation damage is something most companies are willing to accept.
In fact, the comment you linked on reddit basically states the same thing but framed slightly different. The comment might be right that at is not feasible to keep everything server-side from a cost perspective. Or it simply might be a way of saying "it could be done, but nobody wants to cut in the margins because of it".
I should also note that this comment is over 8 years old. It merely comments on the state back then. I am not entirely sure if Valve actually invested all that much in anti-cheat detection since then. Again, a bit cynical, but it just needs to be "good enough" for people to not walk away from the game.
Also, don't discount inertia in big companies to simply keep doing things that they are already doing. Moving everything server side requires different expertise which needs to be built up. Not just about server side cheat detection, but also net code and all that.
Companies, even if willing, are slow to adjust in these areas. Going back to the cynical cost benefit calculation, the "solution" they currently invested in is accepted by most customers and sales don't seem to hurt.
I mean, I would love it if all detection was server-side and effective. I just don't think that this is at all feasible.
(Also, one point I didn't mention: none of this would cover informational cheats, like wallhacks. Developers should do the basics here, but there is a limit, as at some point the client has to know where the enemy is. If you make that point the literal moment they are visible, the pop-in is really bad.)
So, maintaining the game state actually provides user-value in replay solutions, something which the Counter-Strike series has had since the 90s, and VALORANT is getting soon due to user-demand.
Like, fair, there is a cost to observing and analysing behaviour, but I'm still not convinced that cost is more than having to maintain a kernel-level solution, buying and observing cheats to develop signatures for detection, etc.
GDC talk from 2018 that goes more into depth about their solution. Data mining indicates they've even recently acquired NVIDIA H200s, which is frontier tier tech. Not to mention their active branding of the solution as VAC Live for the launch of CS2. I'd say it is pretty clear they are actively investing into it.
But there has been a change here. For Battlefield, they used to use PunkBuster (a third-party non-kernel anti-cheat), before adding FairFight (a third party behavioural anti-cheat), before abandoning all of those to create an in-house team to develop EA AntiCheat. Similarly, CoD migrated from PunkBuster to VAC, to I think an in-house non-kernel solution, and finally to developing their own in-house solution.
These companies historically contracted out, but are now developing internal capabilities for securing their games.
Valve can't even roll out an effective anti-cheat in their own games, which is why everybody who plays Counter Strike even remotely seriously plays on ESEA or FACEIT.
VAC Live has genuinely been a total shitshow, and I think a big reason for that is because they decided not to go down the kernel level route.
Not super familiar with a lot of the games you list, but Helldivers II runs great on Linux for what it's worth, no issues with anti cheat
I concur, running Fedora with Nvidia GPU, Helldivers 2 is pretty great.
I got bit by Rockstar implementing BattlEye for Grand Theft Auto Online and refusing to implement the Linux version. I have a character with nearly every property asset in the game, tons of achievements, and $40 million in the bank just sitting there, untouchable due to not having Windows.
The most telling thing for me is that every game I have with these Windows only kernel level anti-cheats still have blatant cheating. It's been a few years since I played Fortnite, but when I did play it was filled with people using auto-aiming, auto-build macros, etc. So either these companies are grossly misinformed, or they're not resisting Linux because it would increase cheating.
For Fortnite, I honestly believe they won't enable Linux support because they can't support Epic Games Store on it. If they didn't have Fortnite tied to EGS, they would have to show what a colossal failure it has been.
Hoyoverse games (genshin, honkai, zenless, etc) all use kernel level anti cheat that doesn't run on Linux and have pretty massive playerbases, so people who play those probably won't make the jump any time soon either.
It has indeed intermittently worked, so it may still be possible for it to stop working again, and afaik Hoyoverse has never commented on the matter.
In case anyone wants to check specific games in regards to anti-cheat on Linux: https://areweanticheatyet.com
Quite a few multiplayer-games work fine on Linux, e.g. the recently released Arc Raiders does.
I don’t really see how you can square that circle. Personally kernel level anticheat is table stakes in a serious competitive game. The fact that CSGO doesn’t have it is a serious demerit (and why face it and ESEA exists). But Linux just isn’t architectures in a way where having kernel level anticheat makes sense.
There is only one way to square that circle: Valve's Steam Machines using a custom signed kernel + TPM. Doesn't really help anyone who does not have a Steam Machine but that's the only way you'd be able to play these games on "Linux" (really Steam Machines only).
Apparently not all anticheat is the same. I've been playing Helldivers 2 on Linux almost since release (I started a couple weeks later) and I've had very few issues. Most of the problems I have encountered have been unrelated to my OS and were problems other players complained about regardless of platform
Ideally, Valve will come up with an anti-cheat Linux solution with controls in place that make it a little bit less of a rootkit, keylogger security risk. Maybe a "game mode" minimal kernel in steamOS with way more restrictions, which would basically just be a console?
I made the switch to Bazzite on my gaming laptop a few months ago and it's been surprisingly painless. Especially compared to my attempt to switch to Linux Mint on the same laptop earlier this year. I'm not always playing the latest and greatest (I think Baldur's Gate 3 is my most recent game?), but it suits my needs nicely. For everything else there's my oft neglected xbox series S and PS4.
I should try Bazzite. I liked Mint well enough on my laptop, but had these little issues consistently and one which I felt was a major issue (fans not spinning up during one particular game) and so I eventually went back to Windows, since I was just constantly investigating and solving issues, rather than playing games.
Love to finally see some professional level game benchmarks on Linux. It pretty clearly confirms what we guessed about the general advantages and disadvantages depending on the hardware used.
Valve says they are working to help get raytracing up to par on AMD, will be interesting to see how far they can take it.
While I have had a good experience on Linux with Nvidia, I am curious if the frame-time consistency and 1%/0.1% lows issues are things I have ever encountered, because they are very apparent in these results. That said I am not on Bazzite or a similar Arch/etc base
As GN admits, the ecosystem is so complex and rapidly changing on Linux that it's possible some users may never have the same experience as the benchmark results show, because of the interwoven complexity between the OS, drivers, kernels, hardware, etc
I really appreciate the work Steve and all the GN staff have done to help make it easier for the average person to transition to Linux. Good on them.
Also, their investigative journalism is beyond world class. Just amazing.
The gist of it seems to be, an enormous amount of progress has been made but there's still a long way to go. That matches my roughly ten days of experience trying to game on Linux as well.
I am running Arch with Hyprland and despite tiling window managers not playing so nice with games in general, I've had less issues with it than CachyOS (Gnome). I tried CS2, ARC Raiders, and some indie titles like Megabonk and BALL X PIT, and all of them worked from OK to great. The game I had issues with the most was, surprisingly, Valve's own CS2. I had issues where the console wouldn't accept my text input, my cursor would escape to my second monitor (despite using gamescope's
--force-grab-cursor) and the performance was terrible despite running it at a very low resolution (1280x960). For some reason, re-installing it solved all my issues. My FPS was limited to 400 and I couldn't set it to unlimited but my primary monitor is 360hz so it wasn't a big deal. The game was fluid enough.I agree with @Bullmaestro that the biggest hurdle, from my not-so-technically-minded point of view is anti-cheat. Some games that require anti-cheat does work. Helldivers 2 has already been mentioned and I'll add that ARC Raiders work just as fine too. But that's not good enough. I don't know how suitable Linux is for the current kernel level anti-cheats, but Valve either need to prove that they can beat cheating just as easily without invasive techniques, or come up with another solution fully compatible with Linux.
My guess is that Valve does not care about these games all that much, at least yet. They're a patient company, they like to take their time. They will likely focus on the experience of couch co-ops, platformers and the likes to increase their share of the pie, as well as Linux's, to make a more compelling case that Linux is indeed a viable option to Windows.
Most of the popular anti-cheat systems already have Linux or Proton compatibility, but it requires the game developers to enable it. How effective they are without a locked kernel is as debatable as how effective kernel-level anti-cheat on windows is anyway.
I've started suggesting to anyone that can't switch because of one or two games with anti-cheat that they just play those titles on xbox or ps5 instead since most of the PC advantages of customizing or modding wont apply to those titles anyway. Not something that goes over well with the PC-only crowd though.
I don't know how effective Easy, BattlEye and the likes are, but I'm more familiar with the Faceit's own AC and it's pretty good. You can still get past it but it requires hardware that costs thousands of dollars. I don't know what they're doing differently than others, I'm not technically literate enough to differentiate them, but it does work.
If people already have consoles, I guess that's an option. I have a console myself but I also have a beefy PC and I just don't feel like turning on my console. I'm rather old school though, I simply cannot imagine playing a shooter on a gamepad.
It doesn't and neither does any other anticheat. All client-side ACs have working software cheats. I would guess that there are probably hardware solutions that are way cheaper than thousands of dollars as well.
The main purpose of AC is to maintain the appearance of competitive integrity to keep players engaged in the game. Completely stopping cheating is out of scope, what's important is blocking obvious cheating such that cheaters have to resort to looking like skilled players.
Rootkit anti cheat is disgusting and the only way I'd countenance such a thing is if a system was built entirely and only for gaming. Why in the ever loving Arkleseizure would I do anything, anything at all, like my banking or use any other credential at all, when there's something as invasive and creepy as that installed on the computer. The sheer hubris of these companies thinking they have the right to such access.
So, a console then, pretty much.
Of course, I built a system based on Endeavor OS (Arch), store it out of sight and use WOL and Sunshine/Moonlight to play it. All the games that want to creep on my activity can get in the bin.
Faceit has a trust and safety team that monitors potential cheaters based on the system they themselves have implemented and from the reports they receive. Me saying "it does work" doesn't mean it cannot be bypassed either by software or hardware, but software based solutions are relatively easy to detect and it's only a matter of time before they take an action against it, whereas hardware ones especially the DMA ones, are extremely hard to detect. (If the cheater is a seasoned player who knows how to act innocent, good luck.)
It is, as you said, a simple game of economics and increasing the barrier to entry. If someone's willing to buy themselves a second computer and a special hardware from some shady dealer that costs thousands of dollars, you're going to have a harder time catching them than to catch some kid in his basement who downloaded an
.exefile from "csaimbot.net" or whatever.I rarely play CS2 matches anymore (despite playing CS2 almost every day) but whenever I have a Premier game, most headshots feel suspicious, whereas I don't have that doubt with Faceit. Does that mean it can't happen? No, but it's extremely unlikely. That's the point. Like it or not, and I absolutely don't like it, kernel level ACs do a better job.