nVidia release versions compatibility

Sorry, but I don’t believe this answers my question. My question was about 3-way release compatibility between TensorRT, CUDA and TensorRT Docker image, specifically when applied to v8.5.2 of TensorRT.

Your answer is about ONNX operations compatibility in TensorRT 8.6.

Did I miss some part of it which answers my question?