I'm assuming that Intel and nVidia haven't even tried adding open source HDMI 2.1 on Linux.

After AMD wants to go public all the time (no R&D money), I bet the other 80 companies after AMD, Intel and Nvidia didn't really push for open source either.

fantaskarsef:

After AMD wants to go public all the time (no R&D money), I bet the other 80 companies after AMD, Intel and Nvidia didn't really push for open source either.

NV and Intel didn't have to, because they already support it through firmware. AMD opted to not implement it in firmware, so now they're stuck. They could add it to their firmware blobs, but it's more likely they will simply not support it at all with their current cards and only do this for future GPUs.

On other hand HDMI is only usefull on mainstream material; TB, DP and USB are more for us.

heffeque:

I'm assuming that Intel and nVidia haven't even tried adding open source HDMI 2.1 on Linux.

Intel as far as I'm aware has HDMI 2.1 working; they only offer open-source drivers for Linux. Nvidia contributes practically nothing toward open-source drivers, so they have no such issues.

It's not about the HDMI working at all, but for it being able to successfully output e.g. 4k 144Hz.

schmidtbag:

... Nvidia contributes practically nothing toward open-source drivers, so they have no such issues.

With each new driver release, NVIDIA publishes a snapshot of the source code on the NVIDIA/open-gpu-kernel-modules GitHub page. Community submitted patches are reviewed and if approved, integrated into a future driver release.

https://developer.nvidia.com/blog/nvidia-releases-open-source-gpu-kernel-modules/ https://github.com/NVIDIA/open-gpu-kernel-modules ...and of course Nvidia open source projects - much, much more extensive than AMD (https://www.amd.com/en/corporate/research/open-source.html) - you can find it also there: https://developer.nvidia.com/open-source Looking at AMD list of open source projects I can almost feel their pain....they simply don't count in the open source community.

barbacot:

https://developer.nvidia.com/blog/nvidia-releases-open-source-gpu-kernel-modules/ https://github.com/NVIDIA/open-gpu-kernel-modules

The kernel modules don't mean much, as the userspace drivers (OpenGL, Vulkan, OpenCL, CUDA, etc) are all remaining closed, and that's the stuff we actually care the most about. As far as I understand, the firmware is still closed too, which controls reclocking, fan control, and presumably (but maybe not) things like the HDMI spec. EDIT: Seems my suspicion about the firmware controlling the HDMI spec was true, and because of that, HDMI 2.1 won't be an issue for Linux Nvidia users: https://www.phoronix.com/news/NVIDIA-Firmware-Blobs-HDMI-2.1 /EDIT It's also worth pointing out that drivers like Nouveau (OpenGL) and NVK (Vulkan) are mostly driven by 3rd parties and volunteers. On top of all that, their kernel drivers only apply to Turing and newer. Sure, their kernel modules were a welcome bone thrown to the open-source community, and it has already made a positive difference. But, it's kind of the bare minimum.

...and of course Nvidia open source projects - much, much more extensive than AMD (https://www.amd.com/en/corporate/research/open-source.html) - you can find it also there:

I never said Nvidia doesn't make open-source software. I specifically mentioned drivers. If you haven't noticed, much of what's on that page is just templates, demos, and development tools for their own [often closed-source] platforms. Nobody would adopt things like CUDA or NVDLA if they had to pay just to learn how to develop for it. Nvidia dominates in markets like GPU compute and AI because of their excellent and free examples, libraries, and documentation. They may be dubbed "ngreedia" but they know that they will return more profits when they make their platforms more accessible.

schmidtbag:

The kernel modules don't mean much, as the userspace drivers (OpenGL, Vulkan, OpenCL, CUDA, etc) are all remaining closed, and that's the stuff we actually care the most about. As far as I understand, the firmware is still closed too, which controls reclocking, fan control, and presumably (but maybe not) things like the HDMI spec. EDIT: Seems my suspicion about the firmware controlling the HDMI spec was true, and because of that, HDMI 2.1 won't be an issue for Linux Nvidia users: https://www.phoronix.com/news/NVIDIA-Firmware-Blobs-HDMI-2.1 /EDIT It's also worth pointing out that drivers like Nouveau (OpenGL) and NVK (Vulkan) are mostly driven by 3rd parties and volunteers. On top of all that, their kernel drivers only apply to Turing and newer. Sure, their kernel modules were a welcome bone thrown to the open-source community, and it has already made a positive difference. But, it's kind of the bare minimum. I never said Nvidia doesn't make open-source software. I specifically mentioned drivers. If you haven't noticed, much of what's on that page is just templates, demos, and development tools for their own [often closed-source] platforms. Nobody would adopt things like CUDA or NVDLA if they had to pay just to learn how to develop for it. Nvidia dominates in markets like GPU compute and AI because of their excellent and free examples, libraries, and documentation. They may be dubbed "ngreedia" but they know that they will return more profits when they make their platforms more accessible.

You're right that it's not entirely open source but still their commitment at least to scientific community is something to applaud - I can't even imagine how would we conduct our research without CUDA for particle interactions and processes simulations. TensorFlow is again something that we use a lot and Nvidia made a big commitment to it.

barbacot:

You're right that it's not entirely open source but still their commitment at least to scientific community is something to applaud - I can't even imagine how would we conduct our research without CUDA for particle interactions and processes simulations. TensorFlow is again something that we use a lot and Nvidia made a big commitment to it.

Agreed - as much as I don't like to admit it, Nvidia deserves the revenue they got for the effort they put into making CUDA easy to adopt. Unlike anybody else, Nvidia actually cared about it being successful. I've mentioned a while back how I once tried to do a Python robotics project involving OpenCL. I put tens of hours into just trying to figure out how to get it to do anything at all. I gave CUDA a shot and pretty much got the desired result within an hour. There's no fundamental reason why OpenCL couldn't have worked for me, it's just that Nvidia did all the hard lifting for CUDA.

heffeque:

I'm assuming that Intel and nVidia haven't even tried adding open source HDMI 2.1 on Linux.

Nvidia has no need to as their driver blobs are still blackbox, and Intel doesn't have native HDMI at all.