Hi all,
I’m in the market for a new big desktop replacement gaming laptop, and looking at the market there are almost exclusively Nvidia powered.
I was wondering about the state of their new open-source driver. Can I run a plain vanilla kernel with only open source / upstream packages and drivers and expect to get a good experience? How is battery life, performance? Does DRI Prime and Vulkan based GPU selection “just work”?
The only alternative new for my market is a device with an Intel Arc A730M, which I currently think is going to be the one I end up buying.
I have had so many issues with Nvidia drivers, especially on laptops with Optimus. Black screens after booting, random breakage when updating, having to fuck around with OpenGL libraries all the time when you have integrated Intel graphics and Nvidia graphics on the same system. It’s just a pain for me on laptops.
Wouldn’t be such a big issue on a desktop, but I’ve had a work-provided workstation with an Nvidia and 99% of the time if something broke on that machine, it was because Nvidia wasn’t compatible with some updated kernel or libraries.
Intel and AMD have both provided us with a painless driver experience that just works out of the box all the time and is integrated in all the open source things (mainly the Linux kernel and the Mesa libraries for OpenGL & Vulkan). With Nvidia, you need to throw all that out and use their proprietary blobs for OpenGL and Vulkan.
Also, I just think Nvidia is a scumbag company, trying to force single-vendor proprietary solutions on the market by abusing their dominant position (pushing CUDA while refusing to implement any new OpenCL version for over a decade, so software vendors couldn’t just pick a competitive open alternative is one example, the original G-Sync is another). I prefer not to give them any money if I can help it.
I’ve had all those issues back in like 2014.
Nvidia Optimus has come a long way on Linux. Manjaro and Mint have utilities to enable it out of the box.
THAT SAID
We still have to prepend all programs we want to use the Nvidia GPU with
prime-run. I’m not sure if mobile AMD users have to do the same, but this is legitimately annoying as hell this many years later and would actually be a good reason to pick AMD over Nvidia.No we don’t. Mesa and the kernel automatically decide to use the dGPU for intensive tasks. It is only on rare ocassions that I have to use the DRI_PRIME=1 to force the use of the dGPU. It has months since I did it
Thanks. I’ve been curious about that.
Gonna start sharing it as another reason why I would choose AMD over Nvidia, in addition to the drivers being open source.