Is this your business? Claim it to manage your IP and profile
The Akida neural processor IP is a groundbreaking solution, touted as BrainChip's first-to-market digital neuromorphic processor. It emulates the human brain, processing only essential sensor inputs at the point of acquisition. This technology remarkably reduces latency, boosts precision, and significantly lowers power consumption compared to traditional methods. One of the IP's most compelling features is its ability to keep AI/ML tasks local, eliminating the need for cloud dependency while enhancing data security and privacy. Designed for both inference and on-chip learning at the edge, Akida offers a fully customizable event-based AI neural processor. Its scalable architecture supports up to 256 nodes connected via a mesh network, and each node comprises four Neural Network Layer Engines with configurable SRAM. This allows the processor to adjust dynamically for either convolutional or fully connected operations. Moreover, the processor is event-driven, meaning it utilizes data sparsity in activations and weights, drastically cutting down operations. This nature allows for efficient processing across various applications without the energy demands of a typical neural network accelerator, making it an optimal choice for high-speed, low-power computing needs.
BrainChip's Akida 2nd Generation is an enhanced version of their pioneering neuromorphic technology, designed to support a wider range of complex network models through advanced event-based processing. This iteration adds significant features such as 8-bit support for weights and activations, improving both energy efficiency and computational accuracy. The 2nd Generation builds upon the foundational Akida technology by expanding its capabilities for more sophisticated applications without increasing the need for cloud reliance. It is set to bring remarkable improvements in the deployment of intelligent applications across a variety of edge devices, supporting crucial functionalities like vision transformers and temporal event-based neural networks (TENNs). This platform is versatile and scalable, enabling processing across 1 to 128 nodes. Its design prioritizes privacy and security by ensuring that sensitive data remains on-device. Its unique advantage lies in reducing model storage requirements and offers increased operational efficiency, catering to the diverse needs of industries like automotive, healthcare, and industrial IoT.
MetaTF is a comprehensive development environment for neural networks, perfectly aligned with BrainChip's vision of enhancing Edge AI. It simplifies the creation, training, and testing of neural networks for use on BrainChip’s Akida neural processor by supporting the automatic conversion of TensorFlow models. This tool leverages the Python programming language and popular libraries like NumPy and Jupyter notebooks, allowing developers to work within a familiar ecosystem. A standout feature of MetaTF is its ability to transform Convolutional Neural Networks (CNNs) into Spiking Neural Networks (SNNs), optimizing them for low-latency, energy-efficient applications at the edge without necessitating new framework learning. In addition to the conversion capabilities, MetaTF offers a range of pre-created network models in their model zoo. This rich repository provides a variety of quantized Keras models converted using CNN2SNN, giving developers a head start in building efficient, performant solutions. Its integration with BrainChip's Akida explores true on-device intelligence by allowing developers to innovate without limitations traditionally posed by less efficient networks.
Join the world's most advanced semiconductor IP marketplace!
It's free, and you'll get all the tools you need to evaluate IP, download trial versions and datasheets, and manage your evaluation workflow!
To evaluate IP you need to be logged into a buyer profile. Select a profile below, or create a new buyer profile for your company.