Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Aah that’s a shame... but not surprising given Nvidias track record with Linux drivers



My, how the times have changed.


Its been like this for about the past 5 years since AMDGPU was added to the kernel. AMD cards are better for everything except compute where they failed to support the 5700 series.


Compute is pretty bad all around no matter what card, but it is egregious that their flagships aren't even supported by ROCm to begin with.


>AMD cards are better for everything except compute

Ive heard this, but Im not convinced. Don't they have more memory, so while the processing is slower there is less fetching of memory outside of the card?

Of course this would require custom compute kernels for same throughput like matrix multiplication, but still.


it is a driver and software problem with compute, the hardware is fine. things just don't work right


What does that mean? I have an RX5700 card in my Linux computer. I'm not very interested in benchmarks but it works well so far. I mostly play simple games like Dead Cells, Caves of Qud and such.


It means nothing at all for gaming. The 5700/xt is a very good card for gaming on linux.

Where it fails is things that use the compute functions (not vulkan) like blender and many neural network tools. Basically AMDs version of cuda called ROCm. They only support ROCm on a few of their cards while nvidia has cuda on everything on day one.


How I would like to use the OpenGL 4.1 support of fxglr instead of "here OpenGL 3.3 and not complain" from the open source driver.


radeonsi supports OpenGL 4.6 just fine.


Next ROCm is supposed to support RDNA1/2.

It's supposed to be out, and still isn't. ¯\_(ツ)_/¯


Compute sucks on AMD no matter the OS. CUDA is so far ahead.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: