Asahi Linux May Have OpenGL 2.1 For Apple M1/M2 By Year's End

[phoronix]

Phoronix: Asahi Linux May Have OpenGL 2.1 For Apple M1/M2 By Year's End

Alyssa Rosenzweig who is known for her work on the Panfrost open-source, reverse-engineered Arm Mali driver and has been spending nearly two years now involved with the Asahi Linux crew working on reverse-engineered Apple M1/M2 graphics support has shared a new status update...

https://www.phoronix.com/news/Asahi-Linux-GL-2.1-EOY2022

[skeevy420]

It's funny. This doesn't affect me one way or the other since I won't ever buy the hardware to use this software but I still think that this project is one of the great highlights as to why Linux kicks ass and displays the better side of humanity by showing how even one passionate individual can make things better for the rest of us.

Thanks, Alyssa.

[aorth]

Awesome work! Cheering her on from the sidelines. Is she employed, or does she have a Patreon? We should support this work.

Edit: from the post, "I work full-time at Collabora on my baby, the open source Panfrost driver for Mali GPUs."

[JacekJagosz]

I hope opensource Vulkan driver from Imagination will be of help for them, even a bit.

[ALRBP]

When it comes to GPUs, history have proven (Nvidia) you can not do anything with a non-cooperative vendor. Running Linux on Apple hardware will never be viable in production (and it's not just the GPU). Apple is not cooperative at all (not to mention hardware is overpriced). If you want Linux on an ARM64 computer, you should look at other manufacturers. Yes, the hardware you want may only become available in a few years, but Apple Linux support will not be working for a few years too, and supported hardware will be obsolete by this time (Linux won’t run properly on newer). Even if ARM64 eventually kills AMD64 (highly speculative), Linux will never run (outside experiments) on Apple Silicon, but other ARM64 systems (or maybe it will be more like RISC killing CISC and RISC-V will rule the Linux world). It's not me who is negative, it's just how things are. There are much more interesting/useful/viable projects for talented programmers to work on.

[risho]

this might be a stupid question, but doesn't the existence of zink make developing drivers for opengl kind of a waste of time? wouldn't it be better to just implement vulkan and then just pick up opengl via zink for free?

[user1]

this might be a stupid question, but doesn't the existence of zink make developing drivers for opengl kind of a waste of time? wouldn't it be better to just implement vulkan and then just pick up opengl via zink for free?

Yeah, I thought the same. It's one thing if you already have a perfectly working, super optimized GL driver like RadeonSi - then replacing it with Zink is pointless. But it's another thing when you develop new drivers for completely new hardware. In this case I think they should just develop a Vulkan driver and use Zink for OpenGL. If they develop a native GL driver, it will take a long time to mature anyway.

[mirmirmir]

When it comes to GPUs, history have proven (Nvidia) you can not do anything with a non-cooperative vendor. Running Linux on Apple hardware will never be viable in production (and it's not just the GPU). Apple is not cooperative at all (not to mention hardware is overpriced). If you want Linux on an ARM64 computer, you should look at other manufacturers. Yes, the hardware you want may only become available in a few years, but Apple Linux support will not be working for a few years too, and supported hardware will be obsolete by this time (Linux won’t run properly on newer). Even if ARM64 eventually kills AMD64 (highly speculative), Linux will never run (outside experiments) on Apple Silicon, but other ARM64 systems (or maybe it will be more like RISC killing CISC and RISC-V will rule the Linux world). It's not me who is negative, it's just how things are. There are much more interesting/useful/viable projects for talented programmers to work on.

Chill, the dude who's working for Linux support had old PlayStation working properly with Linux. Not even a computer in a traditional sense made into one.

It's safe to assume the fact that the dude is taking it at all means it's doable. I mean, someone like him has the knowledge whether something is worth doing or not.

[slalomsk8er]

Yeah, I thought the same. It's one thing if you already have a perfectly working, super optimized GL driver like RadeonSi - then replacing it with Zink is pointless. But it's another thing when you develop new drivers for completely new hardware. In this case I think they should just develop a Vulkan driver and use Zink for OpenGL. If they develop a native GL driver, it will take a long time to mature anyway.

Maybe but Apple uses Metal in place of Vulkan, in the role as low level graphics API. I don't know how hard it would be to write a Vulkan driver for a Metal GPU.
So, maybe Zink -> MoltenVK -> Metal?

[user1]

Maybe but Apple uses Metal in place of Vulkan, in the role as low level graphics API. I don't know how hard it would be to write a Vulkan driver for a Metal GPU.
So, maybe Zink -> MoltenVK -> Metal?

But Metal only works on MacOS and it's closed source..