Eyeq4 Datasheet -

EyeQ4 Datasheet Write-up

• Main oscillator: 25 MHz crystal input (accuracy ±50 ppm over temp)
• Internal PLLs generate: 800 MHz CPU, 600 MHz VMP, 400 MHz ISP

While modern chips like the EyeQ5 or NVIDIA’s Orin offer more raw tera-operations per second (TOPS), the EyeQ4 is a masterclass in efficiency

• EyeQ4-class chips deliver significant TOPS (tera-operations per second) for neural-network inference, enabling multiple camera feeds and complex networks (object detection, segmentation, tracking) to run simultaneously.
• The SoC’s neural accelerators are optimized for common quantized network types used in embedded automotive pipelines, balancing throughput with memory and power constraints.
• Vision-specific hardware blocks accelerate operations like image scaling, warping, feature extraction, and stereo/depth processing — reducing load on programmable processors and lowering end-to-end latency.

The Mobileye EyeQ4 is a high-performance vision system-on-chip (SoC) designed for Advanced Driver Assistance Systems (ADAS) and semi-autonomous driving . It provides approximately 2.5 teraflops of processing power while maintaining a low-power automotive-grade envelope of roughly 3W. Technical Specifications Summary eyeq4 datasheet

Software and Toolchain (Referenced in Datasheet Footnotes)

At the heart of the EyeQ4 is a specialized heterogeneous architecture. Unlike a standard computer processor, the EyeQ4 utilizes a mix of multi-threaded CPU cores vector microcode processors (VMPs) EyeQ4 Datasheet Write-up

Unlike commodity chips (e.g., from TI or STMicroelectronics), Mobileye does not publish the full EyeQ4 datasheet on its public website. To obtain it: Main oscillator: 25 MHz crystal input (accuracy ±50

datasheet lies in its heterogeneous architecture, which uses different types of proprietary accelerators for specific vision tasks: