DxO 9.3 AI causing an Internal Error

DxO PhotoLab
, ,
1

Hi,

I have a Windows machine with a GPU AMD RX7600 and CPU AMD Ryzen 7 9700X. With the new DxO 9 (I’m running on 9.3) when I’m try to use the AI, the app causing an Internal Error and doesn’t let me to use it. Are any others using a AMD GPU having the same problem?

Thank you

2

I have an older CPU and an RTX 5060 16GB and everything runs fine. But why not update to the latest version, PL 9.4 Maybe that will fix your problem…

3

GPU AMD RX7600

I guess the GPU is 8GB VRAM. Note: the VRAM amount what really counts.

Regarding ‘AI’: I guess the ‘Pre-defined’ (keyword) AI masking (like: People, Sky) raise your problem. I suggest to try first the ‘Manual’ AI masks like, Selection, Area, Auto-Brush to how they works. With 8GB VRAM GPU i also really suggest to quit all other applications as possible, like Photoshop, Web browsers, etc. If ‘DeepPrime rendering’ enabled in preferences, than disable it. Other things may also can count, like its happen before Exports or after any Export.

p.s.: and try the latest PL 9.4.

4

I also have this issue with the iGPU on my Ryzen 5800H system. Worked just fine before, just slowly. Not sure why people insist “not enough vram”

5

I’m having the same issue with an RX 9600XT with 16GB of VRAM on PL 9.4. How much system RAM do you have? I have 16GB and am wondering if that’s maybe causing the issue, even though it’s above the minimum requirements published by DXO.

6

Hi,

This is the answer from DxO support to this problem:

At this stage, we have identified that the root cause of the issue you are experiencing with AI features in DxO PhotoLab 9 is linked to the current behavior of AMD GPU drivers. We have encountered a very similar situation in the past with NVIDIA drivers as well. In that case, NVIDIA was significantly more reactive on their side, and several driver updates did improve the situation — even though, to be completely honest, the issue was not resolved for 100% of affected users.
*
With AMD, despite our ongoing exchanges and the detailed diagnostics we have provided, we are currently not seeing the same level of progress or reactivity. As a result, we do not have any visibility on when — or if — a corrective driver update will be released on their side.*

We have to wait :frowning: