Hacker News Reader

Steam and wine has definitely pushed Linux into “good enough” territory for Linux gaming.

I, like you, switched from windows to Linux for gaming. This is the first year I haven’t owned a windows machine. End of an era.

I don’t miss it at all.

Basically the state is getting to be that almost every game exluding some, which require specific anti-cheat engines, will work. In some cases it is just a matter of enabling EAC linux support by game publishers.

Future is here!

That’s a pretty big “just”, tbh. It would absolutely require the sort of proprietary kernel drivers most distros (understandably!) consider anathema, and I’m not sure game publishers want to shoulder that burden entirely themselves.

In one sense, the problem is "centralized".

Instead that there is a need to fix something special in many games, we can make many games working by fixing selected few entities.

EAC has already support for linux, so it should not be impossible.

There's definitely a lot games that don't run as well as they do on windows despite being generally playable on linux. It's nice to have a dual boot for games like that even if you play the vast majority of games on the linux side.

If you’re going to be investing much time and energy into AI/ML on your GPU, I urge you to consider going the nvidia route rather than an AMD GPU. (CPU-wise, either AMD or Intel is just fine.) Using ROCm instead of CUDA may be possible, but it’s unloved, unsupported on consumer hardware, doesn’t even build with PyTorch at its current minor version, and generally a frustrating time-suck.

The proprietary nvidia driver situation is unfortunate, and if you’re sufficiently driven off by it, ROCm can work, sometimes, eventually. But do consider the effort and time it’ll cost.

I think Arubis didn't push the point hard enough. Nvidia is a long way ahead of AMD in AI/ML hardware capability, from my very limited experience.

I got a Stable Diffusion thing up and running locally on a relatively old AMD RX 580 with 8GB of memory, and default settings of 512x512 with 10 (I think) passes, and it takes about 10 minutes to generate an image. On a similarly aged Nvidia 1060 with 6GB of memory that a colleague of mine uses for their local Stable Diffusion instance, it takes about 20 seconds to generate an image.

I don't know how apples and oranges this comparison is, my OS is Linux and his is Windows, for example, but 10 minutes versus 20 seconds is, for me, a massively disappointing difference.

I'm sure the gap has decreased in the intervening 6-7 years, but I don't know how much - such that I'd be opting for a second hand Nvidia over a new AMD (and I philosophically prefer AMD).

This may be a worthwhile read if you're looking to squeeze the most performance: https://timdettmers.com/2023/01/30/which-gpu-for-deep-learni...

That doesn’t seem like a difference caused by the chip, but by whatever integration is used.

To clarify this. On my RX5700 XT, with ROCm, Stable Diffusion generates images on the default sampler at 512x512 20 passes in about 2 seconds. It's definitively slower than a similar NVIDIA card, but I've no concrete numbers.

It's still a pain in other ways. AMD does not support ROCm on consumer-grade GPUs, it only "happens to work". I had to explicitly install the 3.10 python interpreter and use an older torch version, because seemingly, the newer ROCm releases do not support the RX5700 XT. Also ROCm only works on Linux.

When you say "unsupported on consumer hardware", make it clear that it does work it just isn't supported if you run into issues.

My experience with a RX6800 was that compiling ROCm for myself for use with Pytorch just worked, though from reading online it's clear that's a rarity. Current gen cards apparently don't even work.

That's the plan. AMD CPUs and Nvidia GPUs.

change your plans. Rusticl will make your life less painful. Avoid Nvidia until NVK matures.

Now that Mesa supports rusticl, your concerns are irrelevant. ROCm is no longer needed.

Microsoft is losing this particular market, but they don't seem to care.

I would pay good money for a Windows version that had a more professional and consistent interface (Win 2K comes to mind, but Win 7 interface was still OK) no advertisement inside the OS, no telemetry, automatic update notification but manual installation, no unexplained HDD usage when idle that goes away as soon as I open the system monitor, WSL support.

Closest thing is LTSC but it is discouraged and hard to legally buy.

Linux can do all of this. Compatibility with games is good enough and better in some edge cases. KVM performance running a VM is good enough for running MS Office for work, and still a better experience than in bare metal as you can limit core usage and the computer will run cooler and use less power.

Ever considered cloud gaming?

I built dozens of gaming desktops as a kid and young adult, but lately the time demands of adulthood means it's hard to keep up with the latest trends. Like you I got sick of fighting Windows just to launch Steam. But I also had numerous bad experiences with desktop Linux, and with not wanting to deal with Proton and Wine. When I want to relax I just want to get in a quick game for an hour or two, not go through a ton of driver updates etc.

To that end, GeForce Now has been... no pun intended... game changing. It works miraculously well, and has almost top of the line hardware. Paired with my ultrawide QHD screen, most games run beautifully. Shooters are out due to the input lag (small but definitely noticeable) but all the other genres feel fantastic. No fan noise, no heat, no giant tower taking up space, no worrying about upgrades... It's an amazing service at a terrific price point. I have a Mac but just plug in a Windows keyboard and mouse and it works flawlessly. Same with an Xbox controller.

People pooh poohed cloud gaming for so long, but these days it's so easy and seamless I'm honestly worried about the future of the gaming PC market. Nvidia seems well on their way to providing data center and AI cards and just renting games as a service. (Honestly, that's fine with me...).

(I'm not suggesting anyone be like me, just chiming in!)

I feel like a combination of a few comments here. Itching to put together something that can bring justice to Zelda and tinker with AI

Apparently there is a way to stream games from your own machine similar to GeForce now. It's built into their GPUs -- https://arstechnica.com/gaming/2023/04/nvidias-gamestream-is...

Yeah, you've been able to stream from your hardware doe a while now. It never works quite as well as just plugging in a HDMI cord though.

There are also services like Shadow that just rent you a GPU enabled Windows box, and you can run whatever emulator on there you want. It's cheaper than renting GPU hours from the enterprise providers (who mostly target AI workloads, I guess?)

I wish there was a plug-and-play cloud gaming box I could just plug into my TV. I’m willing to pay a monthly subscription but not hundreds of dollars to buy into a console ecosystem.

Stadia seemed perfect for a casual gamer like me, but it had so few games at launch and I knew better than to buy into a Google consumer product. I have a MacBook, but using it to game on my TV would be a wonky setup.

Nvidia Shield would work for that with the Geforce Now service. Pros: don't have to pay unless stuck in queue and want priority or higher quality. Cons: have to own your games elsewhere for Geforce Now to be able to stream it (for license reasons), and it's about the same price as an Apple TV.

Game Pass Cloud does work on some smart TVs, mostly the latest ones from Samsung so if you got one of those then you don't even need a box. Amazon Luna is the same. Samsung TV.

You can easily do this with a Google TV (like $40?) plus an Xbox controller. Or just hook up any old Android phone or PC laptop.

Since all the actual gaming hardware is in the cloud, all you need is a device that can output the frame rate and resolution (and maybe HDR) you want, which should be easy enough to find.

Enshittification has three stages:

1. Provide great value to the user while you are weak. Provide them a great deal and a disproportionate amount of the surplus value of the exchange.

2. Provide great value to your corporate partners once you become stronger than your consumers can organise against. Provide your corporate partners a huge amount of the surplus value of the exchange.

3. Take all surplus value for yourself once you're stronger than your corporate partners.

Cloud gaming is at stage 1.

I have seen this too many times at this point of my life to ever trust a good deal like this again.

Investing in things you own is an investment in yourself.

Subscriptions are signing up to be a tenant.

I agree that this process happens. That doesn't mean you shouldn't take advantage when situation 1 is occurring. It means you should evaluate periodically if you are still in a situation 1 for these types of things. Amazon prime and Netflix used to be situation 1s. They aren't anymore. If you're willing to give them up I don't see the problem.

Your data is valuable.

There are network effects help the companies from phase 1 to phase 2.

Ultimately there's an inertia in human decision making that means we almost always stick around longer than we should when we are being taken advantage of.

I would say that it's not worth the hassle of creating the hassle of escaping.

To each their own.

Shrug. It's just games. If I lost all my games, as I have before... life goes on? What does it mean to own a digital copy of a commoditized good? The hardware is even worse. What costs $4000 today will be $200 in a few years and worthless soon after that.

These are experiences that you go through and move on from, not heirlooms you keep. If you absolutely must play old games, they'll likely become either abandonware for emulators or remastered in 16k a few years from now.

You can go to great lengths to horde hardware and your own DRM free copies of games if that's your thing. Me, I'd rather just rent them from Steam while they're on sale for a few bucks, play them on hardware someone else maintains, and then forget about them. I'll never be able to keep up with all the great new releases in a lifetime anyway.

Exactly what I did 6 months ago, abandoning windows was an amazing experience. Some games don't work though, but it's a price I am willing to pay.

I left for NixOS instead of upgrading to Windows 11.

Honestly it can play anything I can throw at it. Including FFXIV which has zero official Linux support. And it might play even better than it does on Windows.

I wish I had migrated way earlier.

Don't buy nvidia. The driver situation is terrible.

You might need a nvidia GPU for your LLM training case tbh.