In NVIDIA's autonomous vehicle platforms (e.g., DRIVE AGX), GPUs, CPUs, and DPUs (Data Processing Units like BlueField) work synergistically. GPUs excel at parallel processing for object detection algorithms (e.g., CNNs), delivering the high compute power needed for real-time performance. CPUs handle decision- making logic, such as path planning or control, leveraging their sequential processing strengths. DPUs offload network and storage tasks (e.g., sensor data ingestion), reducing the burden on GPUs and CPUs, enhancing overall system efficiency.
Option B is incorrect-CPUs lack the parallelization for efficient object detection. Option C underestimates the CPU's role, which is critical for decision-making. Option D ignores the DPU's contribution, which NVIDIA emphasizes for I/O optimization in DRIVE systems. Option A aligns with NVIDIA's documented architecture for autonomous driving.