The edge time series processor can process time series including speech, language, audio, biosignals, RF signals, network traffic, and more at the edge.
The edge time series processor (TSP) from ABR can process large time series data with lower power. It is based on a new algorithm for AI signal processing called the provably optimal Legendre Memory Unit (LMU) architecture. It can process time series including speech, language, audio, biosignals, RF signals, network traffic, and more at the edge. The TSP is capable of providing 10x to 25x cost advantage over CPU / GPU and 10x to 100x power advantage over existing algorithms computed on CPU / GPU. The edge time series processor SP can enable engineers to design devices that saves power and increases response speed, accuracy, and privacy while lowering costs compared to most other existing solutions.
A time series processor refers to a system or device that is capable of handling and analyzing time-based data. It is commonly used in applications such as signal processing, audio and video processing, sensor data analysis, and control systems. A time series processor in electronics may be designed to perform functions such as filtering, smoothing, detecting patterns, and estimating parameters from time series data. The specific design and capabilities of a time series processor in electronics depend on the application and the requirements of the system.
The Signal Processing Chip allows for cloud-sized time-series signal processing AI on a chip. It can reduce cloud cost, moreover, it reduces the central processing unit (CPU) and graphical processing unit (GPU) requirements. The company is also providing full software stack including AI model design and deployment. Thus, reducing the size, cost and power requirement. The TSP can also enhance privacy while lowering costs compared to most other existing solutions.
