Sunshine VS nvidia-patch

> Would the Swift UI also work on an iPad?

Yes, but probably not for the first version.

> Do you have any comparisons with other tools (eg steam streaming, moonlight)

Steam streaming just doesn't really work on linux. Moonlight is somewhat similar in terms of direction, and has an established client base. I know of at least two projects to build servers for the Moonlight protocol[1][2].

The Moonlight protocol is a bit weird, because it's an open-source reverse engineering of a dead NVIDIA project, GeForce now. There are fundamental limitations to the protocol, for example that the cursor must be rendered in-stream or simulated. Using my tool, the cursor is rendered locally, and custom cursor images can actually be pushed to the client, for a seamless experience. This sounds like a minor detail but it matters a lot for subjective latency. I'm also working on employing tricks like hierarchical coding using FEC in the protocol, because I hate VBR encoding for games (it makes text blurry and breaks immersion). Those tricks aren't really possible in Moonlight.

All of the Linux solutions I know about have significantly higher latency compared to Magic Mirror, although I don't have numbers for exactly how much higher. (I have a benchmark to test the latency of my tool, but the others don't.) I'd encourage you to try them out and get a feel for the difference.

Finally, I think Magic Mirror is the easiest to install and get going on the server. It has almost zero runtime library or service dependencies (there's a pesky dynamic link against libxkbcommon which I haven't managed to remove), so you don't need to mess with pipewire or docker or anything - it's completely self-contained.

All that said, the existing tools have the advantage of a larger user and contributor base, whereas Magic Mirror is just me on a mission so far :) So they're likely to be much more stable and usable.

[1]: https://github.com/LizardByte/Sunshine

RDP as a regular or quick solution is actually really decent in this respect.

(1) https://app.lizardbyte.dev/Sunshine

Moonlight + Sunshine for a self hosted solution, works with every OS

server: https://github.com/LizardByte/Sunshine/

client: https://github.com/moonlight-stream

You could use sunshine (https://github.com/LizardByte/Sunshine) + moonlight (https://github.com/moonlight-stream/moonlight-qt). To be honest, at least for me, it works better than most of the RDP/VNC stuff.

EDIT: Just checked again, original was released early 2020, current maintained project started 2022.

On your pc, install Sunshine. It's an open source moonlight server. There's a good walk through on the sunshine github page. Connect your handheld to the wifi running the server & open moonlight. Should work.

I discovered the moonlight client and sunshine server a few months ago. These are open source solutions to provide remote gaming/desktop capabilities with built in input and audio passthrough. I tried NoMachine, but I wasn't able to get audio to work. This looks like a known issue on arch. On sunshine, I didn't have to do any extra tweaking! This allowed me to game on my desktop pc without having to sit at my desk. This was especially helpful while watching my 2nd son. I was really impressed by the performance, I could stream my host's display at high resolutions and frame rates with low latency despite my desktop being in the basement using WiFi. I was getting some instability with WiFi, so I wanted to try connecting my desktop to the router via Ethernet. I decided to go with a headless solution because that gives me more flexibility on the placement of the desktop; I ended up moving my desktop upstairs closer to my router. I figured out a way to stream my hosts display headless by using Nvidia TwinView to create the virtual display. This means I don't need to buy any HDMI/DP dummy plugs. I wrote a Linux Guide for sunshine on how to set this up. If you have any feedback on this guide, let me know! I haven't tried this, but wolf is an interesting docker alternative to sunshine.

This can be patched out. https://github.com/keylase/nvidia-patch

NVIDIA Corporation GA106 [GeForce RTX 3060] and I applied the patch here https://github.com/keylase/nvidia-patch

> CUDA, and pretty much all optimization(hacks) done to run games better

And arbitrary limitations implemented at the driver level to force you to purchase their enterprise GPUs, see https://github.com/keylase/nvidia-patch#nvenc-and-nvfbc-patc...

Nvidia has no motivation to make a consumer card with lots of VRAM, that's basically the only (relevant) separator between the GeForce family and the Quadro lineup.

There are restrictions on NVENC streams with consumer cards, but that has been a solved problem for a while [0].

If they were to make a consumer card with more VRAM, it would immediately undercut their own Quadro/Tesla lineup, which cost substantially more. I don't see a reason for them to do it.

0: https://github.com/keylase/nvidia-patch

I have never had to do the session limit bump thing from the last link. I have a 3090 as well and simply did the initial unlock, which worked fine. I would reinstall fresh drivers from Nvidia, making sure you install the newest one that is supported by the unlock tool (536.40 as of this post, the GitHub for the patch has links to the drivers - https://github.com/keylase/nvidia-patch/tree/master/win)

Aren't these Quadro versions. The patch here. https://github.com/keylase/nvidia-patch supports Quadro versions of you click on the win clickable.

Second, the GPU. The GPU is probably as important as the CPU, and in some cases more important, and when we talk about GPUs we will primarily talk about Nvidia GPUs as they are officialy supported by the Plex team. NVIDIA GPUs are important for Plex hardware transcoding due to their dedicated video encoding/decoding units, superior performance, wide codec support, improved video quality, reduced CPU load, power efficiency. They offer a powerful hardware acceleration solution that can greatly enhance the transcoding capabilities of a Plex server. It's also important to note that Nvidia GPUs require a patch to unlock the number of HW transcoding streams. Dedicated GPUs are large pieces of hardware and have their place in desktop PCs. However, they can also be used with mini-PCs by using an external GPU enclosure.

As there's no NVENC patch available (yet) for the Beta driver branch - referring to this one: https://github.com/keylase/nvidia-patch - which can lift the limits of HW transcoding, I was now wondering a little, as I can see 5 (hw) streams on Plex, which actually shouldn't/cannot be the case no?

I haven’t done this yet but there is a patch on GitHub that removes the limitation for consumer GPUs. Makes lower end cards more attractive for this type of work