[pengaru]

I can get behind the "Android isn't Linux" argument when it comes to claims of how numerous Linux users there are via smartphones. The userspace is quite distinct from anything GNU-like.

But the Steamdeck uses very much a full-blown Arch-derived Linux distro. So I'm not sure it makes sense to categorize their users as anything other than "Linux gamers".

The fact that AMD landed on the Steamdeck vs. NVIDIA or Intel is noteworthy. Their continued investment in mainline Linux support has clearly paid off.

[brucethemoose2]

Its not (just) about software support.

The Deck uses a special, low power (specifically targeting ~9W), graphics heavy AMD SoC. It was actually the first of a new laptop CPU line that AMD seemingly canceled:

https://videocardz.com/newz/amd-ryzen-2021-2022-roadmap-part...

AMD coincidentally had the right CPU at the right time. Intel and Nvidia had nothing comparable for Valve to use. In fact, the successor to the Deck chip is kinda an existential problem, as AMD's CPU-heavy laptop line (including the Z1) is less suitable.

[gpderetta]

Valve also had already spent considerable resources in making proton work well with AMD. even if an appropriate SOC was to be available from NVIDIA, it is possible that Valve would have chosen AMD.

Mind, proton does work well with NVIDIA, but my understanding is that AMD gets the most testing.

[circuit10]

NVIDIA doesn’t have the license to make x86 chips with the modern patented features so they’d need to either have a dedicated GPU with an AMD/Intel CPU or develop, or invest resources into an existing, ARM emulation layer

[brucethemoose2]

Or sell a small die for AMD/Intel to package, ala Vega-M.

Or contract Centaur before they went defunct, maybe?

Both these things would be quite out-of-character for Nvidia.

[xigency]

> AMD coincidentally had the right CPU at the right time.

They have been there pretty consistently, for example they’ve been the SoC provider for a few generations of Xbox and PlayStation consoles, now.

[Uvix]

Those have been desktop parts, not laptop parts.

[kllrnohj]

As others have mentioned they are custom parts, but if you were forced to qualify them as either "laptop" or "desktop" parts they are much, much closer to AMD's laptop lineup than the desktop ones. Monolithic design with integrated GPU is not what AMD's desktop lineup has looked like.

[brucethemoose2]

They are neither, they are both fully custom.

One thing thats under-appreciated is the immense design/tape out cost (9 figures these days, maybe more?) of a fully custom chip. That is a huge flat expense, hence one does not simply make a custom design unless the volume (or margin) is absolutely enormous.

AMD/Nvidia/Intel can't just casually crank out an APU for Valve, as one might think. I'm not even sure MS/Sony can justify a die shrink this generation... Heck, maybe the PS6 will run commodity PC hardware.

[ElectricalUnion]

Compared to what came before, the previous (Modified AMD Jaguar APUs) and the current-generation (Modified AMD Zen 2 APUs) consoles are commodity PC hardware compared to whatever bespoke cost-optimized ISAs consoles used to be built around. So in a way, it's already somewhat commodity PC hardware under very heavy TPM lockdown.

[lhl]

I don't think a next-gen APU for a Steam Deck 2 is going to be that big a problem. The Steam Deck's APU is officially designated: "AMD Custom APU 0405"[1], basically a mix and match of AMD's same-generation parts. With the success of the Steam Deck, a followup custom APU is all but assured, although I don't think it'll show up before next year. If I had to guess, I'd expect a 4C Zen5c and 12 CU RDNA 3.5 on 4nm in late 2024 which should offer a big bump (assuming they can improve memory bandwidth).

Note, we're seeing a lot of Chinese handhelds using 7840s already that while a bit behind at 10W, start beating the Steam Deck's performance (sometimes dramatically) at 15W+. Personally, I'd be pretty excited for a Strix Point handheld next year - AMD seems like they're finally getting serious on the iGPU side of things again (with Meteor Lake looking to be competitive).

[1] https://chipsandcheese.com/2023/03/05/van-gogh-amds-steam-de...

[brucethemoose2]

> basically a mix and match of AMD's same-generation parts.

It is not. Its a distinct die from the Zen2/3 laptop parts.

If AMD was going to continue the Deck APU line, we would have seen Dragon Crest by now... But its not there, and the Z1 is in its place.

Again, I reiterate, taping out a chip is massively expensive. Valve cannot afford a custom die, they are stuck with what AMD has at the moment when they need it... Though if AMD starts tiling their laptop APUs, Valve might be able to tweak the CPU/GPU config.

[lhl]

No, the Steam Deck would have never upgraded to Dragon Crest (2022 on those leaked roadmaps). In case you didn't realize, the Steam Deck wasn't released until Feb 25 2022, and only sold 1M units in 2022 (vs 3M projected in 2023). It's certainly successful enough now (and has enough marketing, like it had just a ridiculous amount of floorspace at TGS) that it's all but guaranteed there will be a Steam Deck 2 (but probably not until 2025).

Since all the details are NDA'd, I guess we'll have to disagree on the economics, but "tweaking CPU/GPU config" is the whole point of AMD's Semi-Custom Solutions group and if they can't handle delivering a solution for a product that has at least $2B in sales, then I don't know what customers they're supposed to serve (tape-out costs are expensive, but not nearly as expensive as you're imagining I suspect, when reusing existing IP blocks). But we can revisit in a few years and see.

[forrestthewoods]

> So I'm not sure it makes sense to categorize their users as anything other than "Linux gamers".

It depends on what you’re using this data for.

If you are a game developer deciding what platforms to support then Steamdeck is fully distinct from Linux, imho. Support Steamdeck, it’s likely worth it (depends on type of game)!

However supporting Steamdeck may not require a native Linux port. It turns out the best way to support Linux May infact be to simply use the Win32 API!

And even if you do support Steamdeck with a native Linux port it may not be worth your time to try and support Ubuntu and a billion flavors of Linux that are each broken in different ways.

Supporting Linux clients beyond Steamdeck is likely not worth it for most games.

Source: have shipped games with Linux support. Was extremely painful and not worth it.

[pengaru]

"Linux Gamers" all have access to Proton/WINE. It's not some Steamdeck exclusive capability...

[forrestthewoods]

Steamdeck is a device with one hardware configuration, one set of drivers, one operating system, and one local environment. "Linux" is an infinite number of combinations derived from an a large and unbounded set of hardware, driver, OS, and environment choices.

The reason that "supporting Linux is hard" is the combinatorial matrix of broken ass shit. Supporting a single configuration is easy.

Proton/WINE works well on Steamdeck. It gets updated regularly by Valve for specific games when it doesn't. It is not as reliable for random gamer's random ass frankenstein setup.

[vcxy]

It's a funny thing. I think what you're saying is exactly right from the standpoint of a dev. As a mere consumer, if you support Steamdeck via Proton, then it sure feels to me like you're supporting Linux, but I get why you wouldn't officially say that.

[forrestthewoods]

Yup. I think you get it.

If I were shipping a game today I would say “I support Steamdeck”. If any users complained about it not working on their particular Linux machine I would say “you’re on your own, good luck!”.

I would proudly advertise “Steamdeck support” and I would definitely never claim to “support Linux”.

IIRC for my Linux project something like 40% of support tickets were from the 1% of Linux users. Give or take. Never again!

[rowanG077]

Damn, so you are saying you get free testers from the small share the linux users are? That's actually a great incentive by itself.

[PaulDavisThe1st]

> "Linux" is an infinite number of combinations derived from an a large and unbounded set of hardware, driver, OS, and environment choices.

So, sort of like Windows?

[forrestthewoods]

Not really, no. In practice Linux is radically more fragile and roughly an order of magnitude more expensive to support for two orders of magnitude fewer users.

[PaulDavisThe1st]

The opposite of my experience (I write, build, maintain and distribute a cross-platform DAW).

[Scramblejams]

Just curious, what engine did you use for the games you shipped on Linux? And any differences in how well they did(n’t) work that corresponded to which store you shipped on?

[forrestthewoods]

Custom engine. Store made no difference.

FWIW Linux is easy to support if all you want to do is run a headless server on a single distro. Supporting more distros may require a little bit of dependency hell bullshit, but it's doable.

What's a bloody nightmare is graphics and sound and the infinitely large matrix of janky environments gamers have.

[Scramblejams]

My current plans are Godot, Steam, and for Linux packaging the client to run in Steam’s container environment (“sniper”). Will be interesting to see how many problems that doesn’t solve.

[adgjlsfhk1]

The good news is you mostly don't have to support the graphics/sound/environments variations. If you test it on stock Ubuntu (and maybe SteamOS), it will probably work for almost all the noobs and everyone else can probably figure it out without much help.

[forrestthewoods]

My experience is that this is not the case.

[throwaway2037]

    the infinitely large matrix of janky environments gamers have.

Raymond Chen ("The Old New Thing" blogger) would agree with this from the perspective of Win32 API!

[Zambyte]

> I can get behind the "Android isn't Linux" argument when it comes to claims of how numerous Linux users there are via smartphones. The userspace is quite distinct from anything GNU-like.

But why does having anything "GNU-like" determine whether or not something is Linux? Surely the fact that it is literally running Linux makes it Linux more than some related software (GNU) not being used.

[xethos]

Most people say linux, but mean GNU+Linux. Clearly, being a world-renowned software dev, and being in-your-face pedantic for literal decades isn't enough to get the point across for some people. Add whichever tooling you want (I've heard KDE+Linux before, and without looking it up, suspect Gnome+Linux wouldn't be wrong), but don't pretend Android doesn't use the Linux kernel.

I'll take my downvotes here for being an ass, but on a forum that loves being pedantic, I legitimately expect better. Android is Linux. Typically with an outdated kernel, without root, with a locked bootloader, and without the GNU tooling, but it's pretty hard to argue Android's not Linux.

[adgjlsfhk1]

GNU is a distraction here. In a decade or two there will probably be mainstream linux distros where all the GNU parts have been replaced by newer and better libraries (probably written in Rust). People say linux, but mean a linux system to which they have root access (or can phone up the sysadmin who does).

[xethos]

> but mean a linux system to which they have root access

Ah, like my Pixel 2. Respectfully, you might be on the wrong forum (and possibly the worst one outside XDA) to say "Most people here do not, and never did, have root access on their Android handsets".

If your distinction for what makes a Linux system a Linux system, is root access, recognize that many, many people here have met and cleared that bar.

[smabie]

Have most people on HN? Highly doubt it