14 comments

[ 4.3 ms ] story [ 51.8 ms ] thread
To be fair... AMD also makes CPUs and APUs. In addition, the community around AMD was smaller for quite a long time.

Not saying it's not good to contribute (it is), but perspective.

Should be somewhat careful with using LOC as a decent metric for code
But in real terms, AMD is migrating their proprietary user space drivers to run on fully open source kernel drivers. It'll be an awfully chilly day in hell when Nvidia decides to do something similar.
I'm running a Radeon RX560 (replaced an Nvidia card), using the open source AMDGPU kernel driver and Mesa. It's seamless and works at least equal to (probably better than) the proprietary Nvidia drivers.
RX460 and the same, used it since June 2018 with not a single problem.
I went with Radeon as well because their open source drivers. I am a very happy user. No more bad surprises when updating kernel, which has saved me more time than I'd like to admit, since I used Nvidia because of performance.
AFAIK AMD also goes out of their way to provide support. NVIDIA I'm not so sure e.g. EGLStream and Wayland, Optimus, G-SYNC
Well it's not perfect, but can you give some idea. Some times removing code and adding back a small chunk of code can be helpful. That would give only a negative LOC, so yes it has problems. However, it's definitely one those things that I don't think will ever have perfect generalize-able metric.
Yeah, remember that time AMD dropped 100K LOC of garbage on Kernel maintainers?
Using nvidia card on Linux is a pain.