Linux

12415 readers

244 users here now

A community for everything relating to the GNU/Linux operating system (except the memes!)

Also, check out:

Original icon base courtesy of lewing@isc.tamu.edu and The GIMP

founded 2 years ago

MODERATORS

Ategon@programming.dev

anzo@programming.dev

dwraf_of_ignorance@programming.dev

PSA for those using DaVinci Resolve with AMD GPUs, don't install ROCM 7.2, stay on 7.1.1 (programming.dev)

submitted 6 days ago* (last edited 6 days ago) by the16bitgamer@programming.dev to c/linux@programming.dev

21 comments fedilink hide all child comments

Been fighting this bug for a while now, but since I updated rocm to 7.2 DaVinci resolve died.

Took me a while to track down that it was rocm until I ran across an endeavour os post about it. But if you’ve noticed that resolve stopped loading and not showing errors, Try downgrading rocm and trying again.

Endevour OS Post: https://forum.endeavouros.com/t/can-t-use-davinci-resolve-studio-crash-to-desktop/77870/11

AUR post which the Endevour OS post was referring to: https://aur.archlinux.org/packages/davinci-resolve-studio#comment-1058336

If you need to downgrade your amdgpu-install package, AMD's repo link is here https://repo.radeon.com/amdgpu-install/

you are viewing a single comment's thread
view the rest of the comments

[–] the16bitgamer@programming.dev 2 points 6 days ago (2 children)

I am curious if it's a Davinci issue or a AMD one. Would love to hear the results.

[–] vikingtons@lemmy.world 3 points 4 days ago (1 children)

have filed this to the ROCm team, hope to get someone looking into this tomorrow.

[–] the16bitgamer@programming.dev 3 points 4 days ago (2 children)

Thank you, hopefully it’s something small

[–] vikingtons@lemmy.world 3 points 3 days ago (1 children)

Good news, seems there's an existing internal ticket covering this defect. It's an application-side issue pertaining to libProResRAW.so.

I believe we're shipping a workaround at the ROCm framework side in 7.2.1

[–] the16bitgamer@programming.dev 3 points 2 days ago (1 children)

That is fantastic to hear!

[–] vikingtons@lemmy.world 1 points 2 days ago (1 children)

Someone on the github issue thread has been asking for clinfo output on affected systems, theorising that client issues are caused by issues with vendor detection.

This is a bit of an ask, but if it's quick to jump forward and back between ROCm releases on your setup, would you be able to pass your clinfo output from 7.2.0 into the ticket linked above? No worries if not

[–] the16bitgamer@programming.dev 2 points 2 days ago (1 children)

More than happy to, though it might be later today thing

[–] vikingtons@lemmy.world 2 points 2 days ago (1 children)

no hurry at all, thanks a bunch!

[–] the16bitgamer@programming.dev 2 points 2 days ago (1 children)

Posted, I also had an error on the output

Memory access fault by GPU node-1 (Agent handle: 0x5a020e3bce10) on address 0x7f86a34eb000. Reason: Page not present or supervisor privilege. Aborted (core dumped)

[–] vikingtons@lemmy.world 1 points 2 days ago* (last edited 2 days ago) (1 children)

can you tell me which device this is with?

edit: nvm i saw you posted it directly into the thread - gfx1103 = 780M iirc

[–] the16bitgamer@programming.dev 2 points 2 days ago

Framework 13 AMD 7640U

[–] vikingtons@lemmy.world 3 points 4 days ago

We'll find out. It's filed to the public issue tracker, which is a nice change of pace. Hope you can stay appraised of the progress from the following link:

https://github.com/ROCm/ROCm/issues/5970

[–] vikingtons@lemmy.world 3 points 6 days ago

I don't suspect it's davinci resolve. This is likely due to rocm-opencl.

It's a long weekend, so I likely won't hear back from my colleagues until Tuesday at the earliest