In confidential method, the GPU could be paired with any external entity, like a TEE about the host CPU. To allow this pairing, the GPU features a hardware root-of-have faith in (HRoT). NVIDIA provisions the HRoT with a https://one.aistudio.ml