Hacker News new | ask | show | jobs
by jchw 2181 days ago
It's awesome to see Coreboot on a laptop with discrete graphics.

Sadly, though, this is a bittersweet deal imo, since you will get NVIDIA's trademark Linux experience, including binary blobs that occasionally prevent you from updating to the latest stable kernel and practically being stuck with Xorg for the rest of your life. I understand that NVIDIA simply has the best options (or at least, that would be my guess) but it's hard to still not be a little disappointed since I've been having a great time with AMDGPU on my desktop, running Wayland, up-to-date kernels and having great system stability.
1 comments
sbrother 2181 days ago
In my niche at least (ML, data engineering) one of the main reasons to go with a Linux laptop instead of MacOS is so that you can run GPU accelerated ML workloads (i.e. CUDA) in development without shipping all your data to the cloud. I've been a happy System76 customer for years and I think this use case is fairly common for them -- they even ship a fantastic "out of the box" TensorFlow/CUDA setup (https://support.system76.com/articles/install-tensorflow/) that has saved me days of dev time over my career so far. It's totally reliant on NVIDIA's products of course, but unlike AMD, Nvidia has heavily invested into the deep learning community so there's not really an alternative.
link
jchw 2181 days ago
Yes this is a fair point and understood. I am not a user of ML so it’s irrelevant to me, so I’ll just hold out for hope eventually. I’d probably be okay with an AMD APU, but I have no idea what the Coreboot status is (has AMD made good on their promises?)
link