you are repeating, what you suggested about 14days ago, without answering any of my questions.
I don’t mind issues with the camera porting to older system, if I can diagnose and solve them within a timebox that I can calculate (without plain guessing)… Or if I learn why the camera can’t work with a specific system and most importantly WHY. In that case I can consider the camera less risky choice for the next project…
Again, can you please suggest a way, if possible, to enable more useful diagnostics to analyze the issue?
Or, can you please recommend somebody else from the support team, to help me diagnose the issue in more “systematic” way ?
@pavolk Could you try running a CUDA sample to make sure ? Any sample will do, just to confirm cuda application can run.
If it does, then try running the ZED SDK but with a legacy computer vision (non neural) depth mode, like DEPTH_MODE::ULTRA by changing the InitParameters::depth_mode, you’ll need to compile a sample. It may be a misleading error message but the root cause could be GPU incompatiblity with the AI models.
thanks @adujardin for bringing new ideas. I’ll follow your suggestions tomorrow.
In the meantime, I have got a 3rd host system to test. Lets reference to it as system (c).
c) Windows11/ZedSDK5.1/CUDA12.9 (NVIDIA T1200/4GB)
I’ve installed the CUDA12-SDK and then Zed-SDK 5.1 for CUDA12 as provided by your website. Running the zed-diagnostic tool to verify the installation started fine, but the tool crashed without any error message, when “testing” the camera.
I run the sample-applications to get some more diagnostic help. Please see the screenshot:
I don’t see any HID-devices there. However, reinstalling the NVIDIA-drivers made the camera on system (c) working. I’d still like to know what the “Internal Error…” mean, but today it is gone and everything works fine on system (c).
I verified the CUDA-installation with the “oceanFFT” demo (top left corner). The I tried the zed-sdk samples again. ZED_Camera_Control works find, but the ZED_Depth_Sensing can’t detect the GPU.