Radeon ROCm 3.5 Released With New Features But Still No Navi Support

[Duve]

You know with the announcement of CDNA, I wonder if we might never see Navi support.

[clintar]

One of my biggest mistakes in my life that i bought Radeon RX 5600 XT..... pitty

I was hoping to get one of these for my birthday. Is it just the compute that is a problem, or is vulkan/desktop use bad? my rx 580 has been great with latest mesa, so I was hoping 5600 xt would be fine, too.

[smitty3268]

Or is it more a matter of upstream adding some special-cased code paths for certain GCN 1.0 devices if enough folks can successfully advocate for such a thing?

Don't get your hopes up. AMD has abandoned that hardware, and it's not coming back. They can't even support their new hardware.

[llukas]

That's not strictly true. They don't support CUDA on some very low-end cards, like GTX 1030.

As far as "anything related to GPU Compute", most of their GPUs top out at OpenCL 1.2, but their Jetson kits don't support it at all.

https://medium.com/@samnco/using-the-nvidia-gt-1030-for-cuda-workloads-on-ubuntu-16-04-4eee72d56791

2GB of RAM doesn't allow to run much but it is definitely supported.

[CochainComplex]

Download source code and compile it? You must be kidding, I haven't had to do that for an Nvidia driver in years.
(I hope you were not hinting at dkms, that would be lame)

It is as lame as saying you have to compile Mesa or the kernel..ever heard of oibaf or xanmod ?

Besides does the Nvidia driver 440.82 "compile" with GCC 10 without passing -ignore-gccthingy ? Does it support the latest kernel 5.7?

Yesterday I have compiled the Xanmod 5.7 kernel. Could have downloaded the Deb package as well but I like to pass march=native to my compiler.

And guess what...it simply works.

[finalzone]

I guess the only thing amdgpu-pro might be good for nowadays is OpenCL. And that only barely. Its OpenGL implementation is hopelessly slow, the oss vulkan drivers are probably at least as fast as the proprietary implementation nowadays.

amdgpu-pro OpenCL is so far better than the current ROCm version on mobile Ryzen powered laptops despite the "unkown AMD GPU" detected status. It seems mobile Ryzen from Raven Ridege to Renoir are still neglected for quite a while on ROCm part.

[birdie]

Indeed this is how you install the Nvidia driver.

Um, what?

Out of arguments how to shit on NVIDIA, now you revert to lying? Oh, wait, your only argument has been "Fuck you NVIDIA" for the past couple of years.

I was hoping to get one of these for my birthday. Is it just the compute that is a problem, or is vulkan/desktop use bad? my rx 580 has been great with latest mesa, so I was hoping 5600 xt would be fine, too.

If you're OK with not having OpenCL support or HW video decoding/encoding acceleration, go ahead.

[oleid]

If you're OK with not having OpenCL support or HW video decoding/encoding acceleration, go ahead.

if non-oss drivers are not an option, what are the alternatives?

[xcom]

Linux is never too good with supporting bleeding edge hardware.
I have rx480 and Vega56 and both are working fantastic, even with Rocm.

For Navi I think you can install opencl-amd 20.10.1048554-1
from AUR or wait a bit more.

However OpenCL is not a very big deal for normal daily use. So it's not a deal breaker.

[CochainComplex]

Um, what?

Out of arguments how to shit on NVIDIA, now you revert to lying? Oh, wait, your only argument has been "Fuck you NVIDIA" for the past couple of years.

For clear linux you have to:
https://docs.01.org/clearlinux/lates...ls/nvidia.html

(and we are not talking about any hybrid quirks ...just plain graphics driver for dGPU)

Here are the "fakenews"

https://github.com/clearlinux/distribution/issues/1835

https://github.com/clearlinux/distribution/issues/1994

https://github.com/clearlinux/distribution/issues/1968

one highlighted comment
https://github.com/clearlinux/distri...ment-637001227

Looking at the bigger picture, though, I want to clarify a few points:

• The section mismatch warning indicates a potential problem between the kernel and the module. Setting this flag doesn't fix the problem, it just makes the warning non-fatal. Disabling it for the whole kernel may mask problems elsewhere.
• The underlying issue is that this particular driver is out-of-tree and doesn't move in lockstep with the upstream kernel itself.
• Problems like this will occur again in the future as long as this driver is out-of-tree.
• The proper solution is for this driver to be upstreamed so it can be maintained by the kernel community.
• We (the Clear Linux team) have no control over that, and can only provide limited support for this driver.

by bwarden member of the Clear Linux Team