Natanael

It’s mostly a way to improve coordination around code which uses multiple processor cores, so it will improve multitasking performance for some programs

This wins lol

I had an old laptop where graphics worked in the LiveCD installer but not once installed.

Tldr it took a bunch of bootloader config changes to make it work again

I have good eyes, the zoom won’t help me because it doesn’t add pixels to the frame (maybe unless you’re downsampling?). I want more physical pixels, and then higher resolution text to make use of it!

More stacking of layers (like 3D cache) and more efficient architectures and nodes will bring it closer.

However, nobody’s ever going to beat TDP. Anything you can fit into a handheld can be multiplied several times over in a stationary console.

That’s incredibly difficult in compact form factors. Modular frames and connectors would also add weight, etc. So far you can basically replace the SSD and spare parts like the joystick. But I’d certainly like to see them try. We’re not going to see them make the CPU or GPU replaceable, but I’ve seen stuff like electrically activated glue for attaching components like screens while keeping it repairable because it’s easy to release with electrical activation again. Would be cool to see Valve make use of stuff like that.

At this stage, it’s likely they’ll move to RISC-V over ARM if they change architecture. Drastically freer licensing would give them more control and fewer headaches.

Proton isn’t the same as an architecture emulator. It wraps the Windows APIs and translate them to Linux system calls. Translating CPU instructions is more complicated because it’s much more latency and overhead sensitive, and slight architectural differences can drastically blow up performance hits to translation. You need hardware based emulation for some instructions.

IIRC the Switch 2 has weaker CPU in exchange for a more powerful GPU compared to the Steam Deck. Considering that console games specifically tend to be more GPU bottlenecked than by CPU, that makes the Switch 2 a bit more future proof since skilled devs will be able to deliver ports with a bit fewer compromises for a bit longer.

But it’s going to take a while before anybody will see the difference in most games, because neither of them will be pushed to their limits in (officially supported) games for a while. Most games that run on the Steam Deck are just running with adjusted graphics settings, not hardware specific tuning.

It’s not that they don’t have any plans. They have plans and are working on it, but they don’t have launch plans. Companies like Valve always have ongoing R&D

You know there’s a funny workaround here - stream the game from your PC to the Deck over Steam Remote Play while keeping that PC monitor on, and then use the Steam Deck only as a controller while looking at the big screen. It will add a bit of latency to controls, but it’s surprisingly not bad

I want a higher resolution screen, for readability. Games don’t need to use a higher resolution in general and can keep targeting something like 720p, but I want text and more to be crisper, etc

https://infosec.pub/comment/17743689

Add these;

Cool to touch
Lightweight

They aren’t going for detachable controllers, most likely. Theoretically they might make a future screen + SoC modular so you can swap a frame (it would kinda make sense from an engineering perspective to build the center similar to a tablet IF you can hit all required specs that way), but even for Valve that’s unlikely. They may prototype it but likely won’t release it.

Would OTOH be nice with a Steam Controller 2 which can fold flat to fit in the same case as the Steam Deck. Also not likely to happen 🤷

Higher resolution. Not for higher resolution games, because at that screen size you won’t pick up everything anyway, but rather higher resolution for text and other visually smaller but important details. I want games to be able to tag a subset of extra important visual details for full resolution rendering, then everything else can keep getting upscaled

https://infosec.pub/comment/17743689

I want a higher resolution screen to make specific games more playable, by rendering text and stuff like that at higher resolution (while the rest of the game could keep targeting lower resolutions and get upscaled)

As described in more detail here; https://infosec.pub/comment/17743689

The main thing I want, besides higher performance, is higher resolution to increase readability. Do something like what Apple did when they introduced their ultra high resolution monitors - present it as a standard resolution monitor to software, but then let the OS handle stuff like font rendering at full resolution and overlay it.

That way you don’t cause a performance hit from games rendering more pixels than what’s necessary for a small screen in 3D scenes, but the detail you do need is there to see. They should work with game engine developers and get the OS side support of it upstreamed to the Linux graphics stack (presumably the game mode window manager Gamescope would be the first place to build it into). It would work in parallel to the upscaling algorithm for the rest of the frame buffer.

Stuff like puzzle games and platformers, etc, could even have game engine support for tagging certain assets and object edges and symbols for higher resolution rendering, not just for fonts, so it’s easier to see the important things. You could even do stuff like render faces specifically at higher resolution and do the rest at low res with upscaling.

You can make Linux load balance over two network connectors, but usually it prioritizes one network adapter for all traffic based on a scoring algorithms (wired and high bandwidth gets most points).

You can manually set a priority too, or route specific traffic (based either on destination, protocol, or source program, etc) to a specific adapter. Some programs (like KTorrent) are capable of using multiple adapters without any specific config (which is why I was able to run torrents one time while literally nothing else worked with an old 3G internet dongle) .

Like that time I got a random no-name action cam’s webcam mode to work on Linux by manually mounting it within seconds of connecting it

Here’s some HTML only tricks to serve a default image plus a fallback image;

https://stackoverflow.com/questions/980855/inputting-a-default-image-in-case-the-src-attribute-of-an-html-img-is-not-vali#980910

This way you can set JPG XL as the default, and a lower quality normal JPG file as the backup (maybe by setting both to the same file size, lol). And then separately give the user a notice (based on feature testing) that their browser doesn’t support JPG XL and that they should request it

“freeing up space on the user’s device”