https://bugs.kde.org/show_bug.cgi?id=511487
Bryan Gillson <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #10 from Bryan Gillson <[email protected]> --- > So, if I remember correctly: > 1. Install cuda toolkit https://developer.nvidia.com/cuda/toolkit (I have > 13.1, the latest) > 2. Install cuda DNN https://developer.nvidia.com/cudnn > 3. Install Microsoft Redistributable C++ > https://learn.microsoft.com/en-us/cpp/windows/latest-supported-vc- > redist?view=msvc-170#latest-supported-redistributable-version > 4. Set OPENCV_OPENCL_DEVICE to NVIDIA:GPU:0 as system environment variable > (not sure whether this is needed, though). The above steps corrected the issue for me. I had the same issue (OpenCL is not available on this system) despite tests showing it operational through command line tools. GPU is Nvidia 2070 Super, OS is Windows 10 22H2. I upgraded from 8.9.0 to 9.0.0 as part of the process. -- You are receiving this mail because: You are watching all bug changes.
