Discussion: Do you think the next steamdeck will be x86 or ARM?

My understanding is Apple did this by having the more "expensive" to emulate commands be actual hardware, so some commands are emulated/translated, but others are just implemented in hardware on the processor. They could do this because they control the hardware and OS. Microsoft or anyone else trying to get that done is much more of a challenge.

Translation layers are possibile, but from what I can see, are quite unreliable for games (especially more graphically-demanding), which is not of high importance for Apple

RISC-V could be a lot better supported then. But I don't think a lifetime this long would work for the Deck. 7 years is nearly as long as the Switch 1, but that device had the benefit of being a platform in itself with no alternative (as in there are no other switch-compatible-devices). This forces the devs to target it, no matter what performance or fidelity they might wish for.

The Steam Deck might feel a lot like a console, but in the end it is just a PC and the PC gaming world isn't going to wait for Valves next device. The game-tech will just move on past the steam decks capabilities and a lot of gamers will leave it behind and move to other SteamOS (or windows) compatible hardware. The Deck would still have a lot of value as an indie gaming machine, though.

I’m of the opinion that Valve should keep the og steam deck around for 7 years to keep squeezing more value out of the hardware with software updates and to keep the baseline simple for developers to ease the transition to the linux and mobile technologies then release the next iteration of hardware in 2029, RISC-V would be much more developed at that point especially with Valve’s full backing.

Absolutely 0 chance as current RISC-V chips are dog slow and inefficient. Currently RISC-V is only really used in microcontrollers and everything else is *highly* experimental.

GamingOnLinux is a shitty spam blog and not a reliable source. We saw Proton-ARM and Waydroid in SteamDB and those could just be internal research that might never see the light of day. That's it. Everything else is conjecture.

I'd recommend looking at the articles I linked, I probably should have refreshed my memory on them better before commenting.

In addition to knowing that valve is working on compatibility layers for running x86 on arm devices, there was also a steamVR update 9 months ago contains files for an ARM device code named deckard. There's probably more relevant leaks too, I think some renders of deckard controllers got leaked at some point as well.

We do know that Valve is working on a new VR headset that will be ARM based though

is enough for most people to assume that an ARM VR headset is coming

An assumption is not knowing a fact.

It would be stupid for Valve to not at least to explore possibilities but to say that we know that Valve is working on an ARM VR headset is not truthful.

There were previous leaks that were covered here about it, here are two articles on it:

• https://www.gamingonlinux.com/2024/09/valve-appear-to-be-testing-arm64-and-android-support-for-steam-on-linux/

• https://www.gamingonlinux.com/2021/09/deckard-is-apparently-the-codename-for-a-possible-standalone-valve-vr-headset/

As with anything like this, it's not 100%, but it's pretty much confirmed that valve is working on getting x86 vr games to run for ARM hardware, which is enough for most people to assume that an ARM VR headset is coming.

What's the source on the VR headset being ARM based? I must have missed that

On the contrary, my arms are quite ready for gaming, thank you very much!

ARM boards with slotted RAM use the same type as x86 (although mostly LPDDR, as found in laptops), so I assume there isn't any difference that is related to the CPU architecture.

Is ram that's used on arm chips usually lower power than x86 or can they be interchangeable?

Right, so basically mobile games

It also seems likely that the perf hit from x86_64 -> ARM64 emulation, even if the emulation is really excellent, might very well eat up any battery savings from the more efficient ARM chip.

My understanding is that ARM chips work fine for gaming when games are built for ARM, but that is far from the norm at the moment.

Sounds unlikely to me.

Valve would be taking on a huge compatibility problem in exchange for maybe better battery life in scenarios that already have good battery life.

arms not ready yet for gaming. id expect at least 1 more generation of x86 hardware. especially if AMDs 1W optimized efficiency cores end up becoming an option.

I would guess steam deck 2 will be x86, I'm assuming that valve is already working on hardware for it. We do know that Valve is working on a new VR headset that will be ARM based though, and that they're working on an ARM compatibility layer for it. If the early testing of the VR headset is promising enough I could imagine valve pivoting to make the next Deck ARM based, but that will probably cause a longer delay before we get a Deck 2.

I don't think so, Valve already have their hands full with one translation layer and probably don't want to add another layer on top of it.

I also wouldn't be surprised if battery life ends up being worse through the hypothetical vm. It still has to do the same amount of work after all.

Standby time is 99% a function of "how much power the ram takes". If you have less ram, it takes less power to refresh.

It will be there eventually, but on Valve time. There isn't any huge issue with what AMD is releasing

Hopefully they go straight into RISC-V!

Thit seem that the problem with arm is the device tree, from what i understood, it can't be discoverable and you have to do a list for each devices.