EyeQ5 from Mobileye, an Intel company, offers better deep learning performance efficiency than NVIDIA’s Xavier, says Intel.
The EyeQ5 is the Mobileye’s fifth-generation SoC for fully autonomous (L4/5) driving vehicles. First silicon for the EyeQ5 will be in 2018 and volume production is expected by 2020.
The first design of EyeQ5 was announced by Mobileye in 206, prior to its acquisition by Intel. The design was later improved by Intel.
Intel expects the Mobileye EyeQ5 to deliver 2.4 DL TOPS per Watt – 2.4 times more efficient than what NVIDIA has publicly claimed for Xavier.
The product implements a smarter approach in Artificial Intelligence (AI), focusing on the efficient handling of the workloads.
The limited real estate inside vehicles offers several limitations for the operation of high-performance computers, so delivering a high performance system while meeting stringent cooling requirements is the ultimate goal, Intel said.
Tera Operations Per Second (TOPS) is a common performance metric used for high-performance SoCs; TOPS per watt extends that measurement to describe performance efficiency. The higher the TOPS per watt, the better because it will result in less heat and thereby brings fuel efficiency and cost reduction. \
The greater efficiency of EyeQ5 thus translates into lower vehicle costs as well as savings in operational expenses. It will also result in more design choices and greater flexibility for installing the compute box inside the vehicle.