Munich, Germany 24 February 2022 – Codasip, the leader in processor design automation, today announced the L31 and L11, the latest in its range of low power embedded RISC-V processor cores optimized for customization. With the new cores, customers can more easily customize processor designs using Codasip Studio tools to support challenging tasks such as neural networks (AI/ML) even in the smallest, power-constrained applications – such as IoT edge.
It is very beneficial for AI/ML to run in edge IoT/IIoT devices in order to improve security and power consumption, and to reduce latency for real-time processing. Algorithms for AI/ML are computationally intensive and custom processors are needed to deliver sufficient performance with the limited resources available in such embedded systems. To enable this, the new Codasip embedded cores L31/L11 run Google’s TensorFlowLite for Microcontrollers combined with Codasip Studio tools to customize a new breed of Embedded AI* cores which are ideally suited for IoT applications where both space and power are at a premium.
Codasip CTO, Zdeněk Přikryl commented, “Licensing the CodAL description of a RISC-V core gives Codasip customers a full architecture license enabling both the ISA and microarchitecture to be customized. The new L11/31 cores make it even easier to add features our customers were asking for, such as edge AI, into the smallest, lowest power embedded processor designs.
The ability to customize Codasip cores has always been a cornerstone of its success, and why there are already 2 billion processors using Codasip IP. In addition to making the cores easier to customize to match specific embedded designs, Codasip has also enhanced both of the new cores to support significantly higher frequencies.
AI and ML applications are not well suited to off-the-shelf processors. The data types, the quantization and performance needs of the devices differ significantly from application to application. Codasip’s Design for Differentiation approach means customers using its Studio tools can customize the processor for its specific system, software and application requirements. Similarly, embedded devices in low power IoT applications are extremely resource-constrained: limited in memory and with a limited instruction set. Yet developers of these devices need them to be low power, inherently secure, and able to respond and communicate in real-time.
Custom instructions enabled via Codasip Studio RISC-V design tools are ideally suited to develop processors for AI/ML. TensorFlow Lite for Microcontrollers** (TFLite Micro), RISC-V custom instructions and Codasip processor design tools combine to deliver the benefits of embedded, high-efficiency edge neural network processing, namely: reduced latency, improved security, faster communication, and reduced power consumption. These benefits are essential for emerging IoT and Industrial IoT (IIoT) edge applications where the ability to run real-time AI/ML tasks is rapidly becoming a standard SoC feature.
Codasip’s latest L31 and L11 processor cores are the first to feature TFLite Micro support, but the support is being made available across Codasip’s entire portfolio of RISC-V cores.
With the support for Neural Networks using the TensorFlow Lite AI framework, Codasip RISC-V processor IP is perfectly matched to system developers seeking to embed market leading performance at the core of their AI/ML device. With edge processor capabilities, the Codasip custom designed performance deliver these real-time benefits to mission-critical, embedded IoT applications.
Download the paper
Embedded AI on L-Series cores: Neural networks empowered by custom instructions
To accompany the new Embedded AI cores, Codasip has published a detailed white paper – click here to access and download the document.
Background notes
*Codasip’s Embedded AI is the application of machine and deep learning in embedded software at the device level – enabling small IoT embedded devices to run streamlined AI models at the edge with real-time communication capabilities. From a security perspective this minimizes data transfer time/power costs and avoids the use of communication hardware. This approach is important for critical Industrial Internet of Things (IIoT) infrastructure, where edge-AI algorithms can collect data from various sensors and predict and report system faults in real-time.
**TensorFlow Lite for Microcontrollers is a dedicated AI framework to specifically target embedded systems, addressing their limited memory and power constraints. Its support various microarchitectures make it ideal for vendor-specific optimizations. This fits perfectly with Codasip’s processor design automation tools which simplify domain-specific accelerator development and enable Codasip customers to quickly and easily build application-specific embedded AI/ML devices for IoT.
