Kneron KL720 AI SoC (System on a Chip) offers an ideal balance of performance, power saving, and cost for hardware makers who are looking to benefit from on-device edge Artificial Intelligence (AI). Powered by Kneron’s Neural Processing Unit (NPU) that accelerates neural network models, the KL720 SoC makes possible endless AI applications for smart devices.
The NPU core is of the KL720 is designed to accelerate major computing layers inside the Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) to off-load heavy computing from the traditional CPU or GPU structure.
Features Highlights
- Low-power design enabled by ARM Cortex M4 CPU
- Power to process 4K images, Full HD videos, and 3D sensing for fool-proof facial recognition and to enable innovative gesture control for gaming, shopping kiosks, and more
- Ideal for any high-end IP Cams, Smart TVs, AI glasses and headsets, and AIoT Gateways
- Power to drive natural language processing (NLP) for translators and AI assistants
- All the above and more can be processed at the same on the KL720
Features
- NPU
- 696MHz maximum frequency
- GOPS, 1024MAC/cycle peak throughput of 8-bit mode:
- CPU
- Arm® Cortex®-M4 @ 400MHz for system control
- DSP
- Tensilica DSP @ 500Mhz for AI assistance
- Memory
- 128MB SIP SDRAM, 32-bit LPDDR2-2133
- Up to 128MB SPI NOR or NAND Flash
- OS Support
- Power
- 1.725W average power consumption @ Yolov3_608
- 0.9V core voltage
- 1.8V or 3.3V I/O voltage, grouped by I/O banks
- Video in interface
- Video out interface
- Audio Interface
- Peripheral Interface
- I2C
- SPI
- UART
- USB 2.0 OTG
- PWM
- GPIO
- SDIO
Block Diagram
more information: https://www.kneron.com/solutions/soc/
