In confidential method, the GPU is often paired with any external entity, such as a TEE to the host CPU. To empower this pairing, the GPU features a hardware root-of-rely on (HRoT). NVIDIA provisions the HRoT with a https://one.aistudio.ml