XMOS claims to lower AI cost and raise the performance bar
Artificial intelligence (AI), digital signal processing (DSP), control and I/O are delivered in a single device, the xcoer.ai, which XMOS claims is the world’s lowest cost, most flexible AI processor.
Single device prices for the xcore.ai crossover processor start from $1.It is architected to deliver real-time inferencing and decision making at the edge, as well as signal processing, control and communications. It can be used by electronics manufacturers to integrate high-performance processing and intelligence economically into products.
According to XMOS, xcore.ai is a new generation of embedded platform. It has fast processing and neural network capabilities to enable data to be processed locally and actions taken on-device within nanoseconds. It can interpret data without communication with the cloud and delivers the performance of an applications processor with the ease-of-use of a microcontroller, enabling embedded software engineers to deploy every different class of processing workload on a single multi-core crossover processor, says XMOS.
For example, an xcore.ai embedded smoke detector can build an intelligent picture of an emergency which can be fed directly to emergency services to improve accuracy and speed of response. For example, a smoke detector could use radar and imaging to identify whether there are people in a building and, if so, determine how many, where they are located and use voice interfaces to communicate with those inside, while vital sign detection could identify whether they are breathing.
It is fully programmable in C, with specific features such as DSP and machine learning accessible through optimised c-libraries. It supports the FreeRTOS real-time operating system and the TensorFlow Lite to xcore.ai converter, allows easy prototyping and deployment of neural network models.
For connectivity, there are up to 128 pins of flexible IO (programmable in software) and integrated hardware USB 2.0 PHY and MIPI interfaces for collection and processing of data from sensors.
The xcore.ai employs deep neural networks using binary values for activations and weights instead of full precision values, dramatically reducing execution time.
By using binary neural networks, xcore.ai delivers 2.6 to four times more efficiency than its eight-bit counterpart, XMOS reports.
Product demos of the xcore.ai will be available from June 2020.
