Schedulers really don’t matter much for gaming workloads.
Ntsync is a bit more exciting, I’ll give you that. Looks like it’s in testing, so you can update to trixie to get it. Running on the testing branch is usually fine, and it’ll be pretty up to date until release freeze time, at which point it’s best to sit tight until the new testing branch settles down (a point release or two is usually plenty).
manual tweaking
Tweaking is usually not worth it. The most I really do is up sending like the mmap limit if a game is struggling.
comparatively niche game-related problems.
I’m not really sure what types of issues you’re referring to. Either you’ll have OS questions, in which case the generic help is ideal, or you’ll have game specific issues, in which case most distros will be extremely similar (e.g. proton db for game specific workarounds).
For people coming to Linux, I’ll usually recommend sticking with Steam verified or playable games as well, since those should largely just work. Yeah, you might be able to eek out a few FPS or, more likely, framerate stability with some tweaks, but that’s honestly not worth it to most.
If you want to tweak, you’re probably also the type to want the “advanced” distros like Arch and will ignore my advice anyway. And that’s fine, I use something else as well, but I’m going to stick to a more predictable experience for new users.
Sorry, yeah. Bookworm. I’m terrible at remembering codenames.
Schedulers really don’t matter much for gaming workloads.
sched_ext is a bit more interesting. It’s not a scheduler with a new set of knobs to turn, but actually more of a BPF-powered scheduler framework. Being able to swap out the scheduler depending on the game’s needs could be a huge win, and there’s already a couple of schedulers like LAVD (phoronix article) that are designed for improving the Linux gaming experience.
Tweaking is usually not worth it. The most I really do is up sending like the mmap limit if a game is struggling.
Trying to squeeze every bit of performance out, yeah. I don’t want to waste my time trying to get a 1 FPS improvement either.
My concerns with tweaking are more along the lines of things like this happening. User reads about how to get better FPS, sees that some system package is really old, then tries to replace it with a newer one without knowing the risks.
That guy actually asked around and was convinced to just use Flatpak instead. But, suppose they didn’t ask around and instead followed a guide or YouTube video, breaking something in the process. The typical self-proclaimed “gamer” isn’t any more technologically savy than your average consumer, and probably isn’t going to spend the time learning how to fix their mistakes. It’s a more likely that they are just going to walk away with the impression that Linux sucks, and tell everyone else about how bad it was. And unfortunately, first impressions matter a lot when convincing other people to try things like ditching Windows for Linux. Avoiding stable distros is mitigation for that. If the package is already up to date or doesn’t require going off the happy path to update, it leads to less bad choices.
My overall perspective is that if you give gamers a gift-wrapped install that immediately just works, they won’t have a reason to improve or fix things until they’re actually ready to. If 19/20 of their games work out of the box, they’ll play those 19 games until they find the courage to try and learn how to make game #20 work. Or, maybe someone else will fix it for them first.
If 0 out of 20 games work and they need to follow a guide telling them to run various commands to install stuff and get started, they’re going to be overconfident and underprepared when they step outside of the guardrails and do at-your-own-risk stuff they find online.
Either you’ll have OS questions, in which case the generic help is ideal, or you’ll have game specific issues, in which case most distros will be extremely similar (e.g. proton db for game specific workarounds).
I meant game-specific issues, yeah.
Even with ProtonDB, I don’t think the typical coming-from-Windows gamer is going to end up having a simple experience.
With Windows, practically everyone is running the same system software. If some game is broken, they Google “windows 11 Cyberpunk crash” and get a bunch of suggestions and solutions relevant to them. I don’t like Windows, but it does have the fact that there’s a massive community of gamers running the same “distro” (or in this case, Windows version) going for it.
Even with ProtonDB, you have a bunch of people running different distros, different graphics driver implementations for a single graphics card vendor, and just a bunch of software fragmentation… A stable distro would probably be fine for newcomers if was tailored for gaming and had a critical mass of users, I’m sure. But in the current landscape, and for the reasons I listed in rant above, I simply don’t think it would be a good idea to put someone on a stable distro when they’re going to be surrounded with solutions from people on other distros describing how they fixed the problem by updating something or another.
For people coming to Linux, I’ll usually recommend sticking with Steam verified or playable games as well, since those should largely just work.
Absolutely. If it Just Works™, it leads to a better first impressions.
Bookworm?
Schedulers really don’t matter much for gaming workloads.
Ntsync is a bit more exciting, I’ll give you that. Looks like it’s in testing, so you can update to trixie to get it. Running on the testing branch is usually fine, and it’ll be pretty up to date until release freeze time, at which point it’s best to sit tight until the new testing branch settles down (a point release or two is usually plenty).
Tweaking is usually not worth it. The most I really do is up sending like the mmap limit if a game is struggling.
I’m not really sure what types of issues you’re referring to. Either you’ll have OS questions, in which case the generic help is ideal, or you’ll have game specific issues, in which case most distros will be extremely similar (e.g. proton db for game specific workarounds).
For people coming to Linux, I’ll usually recommend sticking with Steam verified or playable games as well, since those should largely just work. Yeah, you might be able to eek out a few FPS or, more likely, framerate stability with some tweaks, but that’s honestly not worth it to most.
If you want to tweak, you’re probably also the type to want the “advanced” distros like Arch and will ignore my advice anyway. And that’s fine, I use something else as well, but I’m going to stick to a more predictable experience for new users.
Sorry, yeah. Bookworm. I’m terrible at remembering codenames.
sched_extis a bit more interesting. It’s not a scheduler with a new set of knobs to turn, but actually more of a BPF-powered scheduler framework. Being able to swap out the scheduler depending on the game’s needs could be a huge win, and there’s already a couple of schedulers like LAVD (phoronix article) that are designed for improving the Linux gaming experience.Trying to squeeze every bit of performance out, yeah. I don’t want to waste my time trying to get a 1 FPS improvement either.
My concerns with tweaking are more along the lines of things like this happening. User reads about how to get better FPS, sees that some system package is really old, then tries to replace it with a newer one without knowing the risks.
That guy actually asked around and was convinced to just use Flatpak instead. But, suppose they didn’t ask around and instead followed a guide or YouTube video, breaking something in the process. The typical self-proclaimed “gamer” isn’t any more technologically savy than your average consumer, and probably isn’t going to spend the time learning how to fix their mistakes. It’s a more likely that they are just going to walk away with the impression that Linux sucks, and tell everyone else about how bad it was. And unfortunately, first impressions matter a lot when convincing other people to try things like ditching Windows for Linux. Avoiding stable distros is mitigation for that. If the package is already up to date or doesn’t require going off the happy path to update, it leads to less bad choices.
My overall perspective is that if you give gamers a gift-wrapped install that immediately just works, they won’t have a reason to improve or fix things until they’re actually ready to. If 19/20 of their games work out of the box, they’ll play those 19 games until they find the courage to try and learn how to make game #20 work. Or, maybe someone else will fix it for them first.
If 0 out of 20 games work and they need to follow a guide telling them to run various commands to install stuff and get started, they’re going to be overconfident and underprepared when they step outside of the guardrails and do at-your-own-risk stuff they find online.
I meant game-specific issues, yeah.
Even with ProtonDB, I don’t think the typical coming-from-Windows gamer is going to end up having a simple experience.
With Windows, practically everyone is running the same system software. If some game is broken, they Google “windows 11 Cyberpunk crash” and get a bunch of suggestions and solutions relevant to them. I don’t like Windows, but it does have the fact that there’s a massive community of gamers running the same “distro” (or in this case, Windows version) going for it.
Even with ProtonDB, you have a bunch of people running different distros, different graphics driver implementations for a single graphics card vendor, and just a bunch of software fragmentation… A stable distro would probably be fine for newcomers if was tailored for gaming and had a critical mass of users, I’m sure. But in the current landscape, and for the reasons I listed in rant above, I simply don’t think it would be a good idea to put someone on a stable distro when they’re going to be surrounded with solutions from people on other distros describing how they fixed the problem by updating something or another.
Absolutely. If it Just Works™, it leads to a better first impressions.