Neural Network Accelerator (NNA)
What is NNA?
A neural network accelerator (NNA) is a specialised processor that is optimised to handle specific neural network workloads. Neural networks are the algorithms that power the clustering and classification of data for machine learning and are commonly used for AI-related tasks, such as face identification, object detection and image classification. A dedicated neural network accelerator can perform these essential tasks much faster than either a CPU or GPU, at a fraction of the power consumption and silicon area size.
While training the neural network models is typically performed on high-powered systems, an NNA is typically used for inference in edge devices, be it a smartphone, a security camera or a vehicle. In the latter, processing speed is of particular importance, where the vehicle must absorb and process the data from sensors, (camera, radar, lidar etc) and then sent to the neural network to respond to input in real-time on a dedicated device such as a neural network accelerator.
PowerVR Vision and AI IP cores bring high-performance compute processing capabilities to devices that need comprehensive real-time, in-system intelligence.
IMG Series4 NNA
The edge platform of choice for assisted driving and full self-driving cars.
Tensor Tiling Whitepaper
Our white paper, Imagination Tensor Tiling, will help you learn more about this critical bandwidth-saving technology in-depth, and provide insight into how it provides real-world benefits in our IMG Series4 NNA with key neural network models.