Supporting real-time 3D depth map and point cloud processing for 3D scanning, CEVA-XM4 aids implementation of energy-efficient, human-like vision and visual perception capabilities in embedded systems. Programmable wide-vector architecture features fixed- and floating-point processing, multiple simultaneous scalar units, and vision-oriented instruction set. Parallel random memory access and 2D data processing scheme enable 4,096-bit processing in single cycle with memory bandwidth <512 bits.