Fedora Continues Working On Better NVIDIA Support, PipeWire Could Replace PulseAudio

[pal666]

2 years ago, I upgraded to the Nvidia 770, because all the games Feral and other were porting, were only working with Nvidia cards

they didn't have official support, though they all work with radeonsi, usually slower initially

[pal666]

Will the rt patchset ever be merged upstream?

(parts of)it is being merged continuously

[pal666]

I really need NVIDIA for CUDA related stuff.

no, you really don't. cuda is proprietary shit and amd has cuda compiler

[microcode]

Oh look, RedHat forcing pre-alpha quality NIH software down our throats again. Lets break ALLTHETHINGSAGAIN just so Redhat can be in better control of the linux. Pipewire is made by a Redhatter so vain, he made a wikipedia page for himself. https://en.wikipedia.org/wiki/Wim_Taymans

He did not write the Wikipedia page about himself, Christian F.K. Schaller did. I eagerly await dh04000's fan-written Wikipedia page.
https://en.wikipedia.org/w/index.php...action=history

[polarathene]

We are talking with NVidia about this, our hope is that once we remove the obstacles to make hybrid graphics work without things like Bumblebee they have more of a reason and motivation to work with us on finding a way to allow proper hybrid mode between the Intel and NVidia driver.

Hey thanks for making this happen I own both an optimus laptop and a desktop with nvidia drivers and both have made trying certain distros a bit more painful. I think when I tried Fedora out my main issue for not using it was similar to openSUSE where having nvidia drivers was a bit of a pain to get setup(I came across several articles/guides, some were outdated I think or wanting to you to use different package/repos) and that updating the drivers/kernel could cause problems. Not quite sure how Arch does it with their dkms package but it was less troublesome. Getting a seamless switch without bumblebee would also be good(prior to Arch I also had difficulty on other distros getting this to work too).

Out of curiousity, will these improvements possibly pay off for virtualization too? Optimus laptops can pass through the nvidia GPU with VFIO but the hardware topology is different from a desktop where the dGPU uses the Intel iGPU framebuffer from what I can tell, no idea if these improvements can help work around that or not.

[polarathene]

no, you really don't. cuda is proprietary shit and amd has cuda compiler

And if you use software that uses CUDA to do it's thing, what then? I'm assuming the compiler is only useful for your own projects where you can provide it the cuda code to convert. I'm not sure how well that plays with CUDA based libraries/frameworks either where you might write most of your source with say Python using an API that handles the CUDA part for you. I have an nvidia gpu for these reasons mostly, but looking to get AMD as my next GPU.

[theghost]

my kneejerk reaction to PipeWire is XKCD's standards - https://xkcd.com/927/

Yeah, the audio stack grows even further

[srakitnican]

There is still an issue with Gnome on Wayland that is a deal breaker for me - input lag under load. Last news are that it seems that there is still nobody working on it. https://bugzilla.gnome.org/show_bug.cgi?id=745032

[theghost]

There is still an issue with Gnome on Wayland that is a deal breaker for me - input lag under load. Last news are that it seems that there is still nobody working on it. https://bugzilla.gnome.org/show_bug.cgi?id=745032

Also had this issue last time when I tested Gnome on Wayland. Strange that this is still an issue at this stage of Wayland adoption....

[bug77]

Wow it's so nice using AMD hardware with drivers that just work out of the box.

After about a decade of waiting about 6 months for AMD to support a new kernel or X, you're faulting Nvidia for taking a week or two to do it. So not hypocritical.