Hi people, thanks for this wonderful light distro based on Arch (got back to something closer to arch after three years using Manjaro as my main distro - and many other distros since late 90’s)!
I got surprised by how well the latest kernel 5.8 with nouveau + intel is working perfectly fine on my Dell G5 5590 (2019) with GeForce 2060 + Intel Skylake.
It runs gdm+gnome on wayland, boots really fast, suspends and resumes in a second without delays unlike when I was using nvidia drivers.
It detects automagically my external monitors on DP and HDMI (hardwired to geforce card), unlike last year that I needed to switch on X11 manually to a nvidia-only configuration or nvidia-prime configuration (the second one if I wanted more juice from battery and less heat).
vkcube automatically matches my displays frequency. It runs with 144 FPS on my eDP monitor with 144Hz and 60 FPS on the 60 Hz external monitor with vsync and no tearing (unlike using nvidia prime on X11, it matched randomly one or another with heavy tearing.
I was wondering - and having hard time to find this on Arch Wiki and googling - how can I run things like glxinfo, glxgears, vulkan on wayland to test if my nouveau output is really working.
With nvidia-prime and X11 it was a matter of using _NV_PRIME_RENDER_OFFLOAD=1 __GLX_VENDOR_LIBRARY_NAME and such to get rendering from nvidia card. For nouveau it seems to rely on DRI_PRIME=1 but it always seems to output using Intel.
Any tips for this? All docs around the web seem focused on nvidia proprietary driver plus X11.
I want instantaneous suspend + resume like I have on the preinstalled Windows 10, and the latest developments of kernel + nouveau + wayland are providing that, X11 + nvidia are not.
Thanks community!
Hello and welcome to the forum. I have to ask this question: are you just interested in specifications or just enjoying computing? Speed is not everything but taking time out to enjoy the finer things in life might be worth it. Flame me if you like, I care not. 
I think both. I have not much time for solving finer details like 10 years ago, but was a pleasure to dig nouveau and wayland docs to know better how this works. But it IS troublesome to get all pieces together.
Linux have a special place in my heart, so I spend a lot of time customizing things. I was a kernel hacker in the past developing some code for networking modules, but then I hated this job and went high level with web development, Ruby, Python. For some time I used Slackware in deep past but regreted a week later. Used gentoo and compiled a lot of customized things in the Pentium 4 era.
For instance, right now, I need CUDA SOMETIMES to use pytorch, then I was thinking to hack a script to blacklist nouveau and activate nvidia. But then I discovered that keeping nvidia installed my GDM is not booting with nouveau. Then I found some random arch forum post about libGL paths, but didn’t had time to investigate further.
Now I finding easier to either boot my windows installation and have all python/pytorch/cuda things again, or using another distro (a more “stable”/“lazy”/not-so-rolling one, like Fedora), just for CUDA/ML setup.
This is just because I can’t figure alone how this sh** work together, and I exausted my google fu.
If I can get more information on how to hack this setup, specially more interested people to hack together, it will be awesome.
It will be also nicer to get proper nouveau working for 3D/HW acceleration things and discovering more about how to get most from this piece of hardware. When you have a discrete card it’s easier, but I’m using this PRIME setup right now and for the next 5 years at least, like my last computer.
Hmm…using integrated graphics certainly makes things trickier for you! Fedora might be good, lots of new things get used there. Sorry I can’t help, I stay away from nvidia, it gave me too much trouble, all my set ups are now AMD or Intel integrated. Good luck! 