Any chance for Unity to support for ARM Linux?

Lumin OS is basically Android running a custom Linux kernel.

https://developer.magicleap.com/learn/guides/lumin-os-overview

 

That's a very interesting thought. It looks like no one tried. I expect it might be difficult to get working though, since who knows what kind of modifications Lumin OS has that a plain Linux install might not have?

 

Yeah well, not exactly the same capabilities between Intel HD 400 and Kepler GPU with 192 CUDA cores... Just saying...

Btw there are a lot of automotive R&D ongoing right now using Nvidia embedded systems and AI dedicated hardware that would definitively need some powerful and qualitative interfaces to come along. As Unity3D tends to conquer automotive market, that's a +1000000 reasons to finally deliver this ARM build to Unity!

EDIT : I'm currently running a Unity build on my TK1 thanks to an dedicated Android OS made by the community (Jedroid). That's nice but not everything is functional and that's quite a pain because Jedroid is no more supported

 

Any updates on this?

 

With the release of various Linux smart phones, notably the Pinephone Pro, not having unity support on Arm Linux is making me reconsider game dev practices. The raspberry pi has also sold over 36 million units.

 

Also interested, we had a client that wanted to port a Unity android app to the Nvidia Jetson which is linux, but ARM.

 

The Raspberry Pi is also not a $400 phone. Be careful you don't ascribe it's success to the wrong reasons.

 

In addition to this, "various Linux smart phones" doesn't mean anyone is going to own those phones in any real numbers. In addition to that, "36 million units" doesn't really mean much when the rPi generally falls in very specific, very limited use cases.

 

I would say though, I have run a unity android build on a $40 firetv usb thumb stick, I ended up buying the $100 one but it did run on the little thumb stick which I thought was amazing tbh lol.

 

I mentioned the pi as a use case, I mentioned the phone as that's what I want to run games on!

'In addition to this, "various Linux smart phones" doesn't mean anyone is going to own those phones in any real numbers. In addition to that, "36 million units" doesn't really mean much when the rPi generally falls in very specific, very limited use cases.'

A lot of people have them! They are generally great, if I could run games on my Pi I would! Why not? I'm sure many others would too. This would still be in the millions which is a lot of potential market space.

'I would say though, I have run a unity android build on a $40 firetv usb thumb stick, I ended up buying the $100 one but it did run on the little thumb stick which I thought was amazing tbh lol.'

That's right, arm processors have a lot of potential in the future. Games don't have to just be on x86!

 

Yes, because it's a $35 computer. That it is ARM-based and running Linux are merely bonuses for most people.

 

I have five of them. They are not game capable machines almost entirely because of their hardware. They struggle to run a lot of stuff more complicated than enhanced Quake ports.

 

I agree with the logic. But why can Godot run on it but not Unity. How many external libraries does Unity rely on that make compiling for it harder. I guess these things aren't clear to me as an end user of Unity Engine. Unity is worth billions and used most by game developers, advertises as democratizing game development, it still seems like a worthy investment compared to some of the tools they have purchased which are more niche.

It is also worthy in how it would benefit humanity, there should always be a certain amount of riskier longer term investment based on long term payoffs. There's also the case that making the engine more widely available would increase future worth of the product as more people using it. Again, I lack analytics on Unity Engine (both for user analytics and revenue streams) so it's impossible for an outsider to make such speculation.

As for selling games, they can be done anywhere. Mojang sold Minecraft on their website, this even had a pi version. As for performance claims, I don't agree, as I'm a big fan of ECS and most games I see on mobile with minimal functionality would run easily with the right optimizations. But I do agree that it isn't popular to game on a low end machine. And that people using it won't have the budget for games and would most likely go with old emulated games.

I also think you could easily make shop setups, arcade systems, using arm 64. As retail is plummeting. I grew up with net cafes and I still see a potential in them Using cheaper parts could lower hourly costs, not to mention electricity prices. I don't have the money to invest in such an endeavor, but I do see opportunity there due to the changing market. There was a rise of arcade like VR spaces, but costs for high end VR machines is a lot higher.

 

Because it is written in C++ with source code provided. Meaning in theory you can rebuild it for anything that has C++ compiler. And if t he game is 2d, in theory it can be rebuild to run on software rendering.

For example, it is possible to run unreal 2004 in linux terminal.

Spoiler

This is basically done by forcing the game run in software rendering mode and swapping libsdl with the version that uses libcaca for output.

 

So the issue is just a matter of open vs closed source? That is interesting. It uses OpenGL 2.0 which is pretty old and I heard the drivers for GPU under the hood due to more licensing issues were reverse engineered on the pinephone pro. Not sure on the pi's graphics libraries though, but I am only talking about 3D and using the integrated graphics cards in the arm chips. Wait I just checked and most use MESA 3D for graphics drivers.
https://upload.wikimedia.org/wikipedia/commons/2/2d/The_Linux_Graphics_Stack_and_glamor.svg

Yeah the business costs of testing and maintaining. I don't see the actual tests running being too costly though. Based on the cost of a virtual machine running for a day that I recall from amazon and other services. I can see developer resources being used up though, which adds up to be significant enough to be not worth it. I guess all this is easier with more open source development due to the community's power. Thank you for the information.

 

Godot 3 uses OpenGL ES 2.0 which is based off of OpenGL 2.0 but has the improvements introduced by OpenGL 3.0/3.1. Like the blog below says Godot 4 will no longer use OpenGL ES 2.0 having replaced it with OpenGL ES 3.0 which is essentially OpenGL 4.3.

https://godotengine.org/article/about-godot4-vulkan-gles3-and-gles2

This won't have a huge impact since most people have moved on but it does mean that anything older than a decade will no longer work. For example the Raspberry Pi 4 can run OpenGL ES 3.0 but the older models use OpenGL ES 2.0. The Pinephone Pro supports OpenGL ES 3.0 as you would expect.

http://rockchip.wikidot.com/rk3399

It's more that someone wanting to support an unofficially supported platform for their game may decide that as long as the game is playable it doesn't matter if it perfectly passes all of its tests. Unity does not have this luxury.

 

That's not right.

Supporting or not supporting a platform comes to someone's decision to put effort into it.
If the platform is proprietary, the decision is made by its owner.
If it comes with source code, the decision can be made by you, where you'll herioically sacrifice your time without getting paid for it.

So while not having source code reduces your chances of running the engine on strange hardware, saying that the hardware is not supported because the code is not opensource is incorrect.

Practically speaking, adding a new platform means CONTINUOUS support of the feature from now on till the end of time. Or till the end of the engine. And realistically, if you port it yourself to a new platform, most of the time you will not continue to support it and will abandon it fairly quickly.