Based on the high-performance Toybrick platform, Linaro has released two 96Boards SOM modules

Based on the high-performance Toybrick platform, Linaro has released two 96Boards SOM modules

tenco 2019-04-08

Cambridge, England, April 1, 2019 "open source software engineering organization Linaro based on the ARM ecosystem officially announced today two 96Boards Compute SOM tb-96ai and tb-96aiot based on Rockchip Toybrick platform and a 96Boards SOM expansion board, with a major launch of 96Boards SOM specification version 1.0.96Boards aims to build a globally integrated hardware and software development community to provide cost-effective boards based on ARM technology.

Based on Rockchip high-performance low-power AI AI processing chip RK3399Pro TB - 96, and based on RK1808 AIoT of TB - 96 AIoT platform, can be purchased at t.rock-chips.com.


【TB-96AI SOM】  

Tb-96ai based on RK3399Pro platform, integrated with neural network processing unit NPU, computing power up to 3.0Tops, compatible with various AI architectures.It can be combined with the backboard to form a complete industrial application board, applied to various embedded artificial intelligence fields.SOM design has the following features:

Ultra high performance processor

It is based on the Arm dual-core cortex-a72 and quad-core cortex-a53, the main frequency up to 1.8GHZ, and integrated with the Mali -t860 MP4 quad-core GPU

High-performance AI NPU

Integrated with the AI neural network unit NPU, it supports 8-bit / 16-bit computing, can reach 3.0Tops, and can meet various AI application performance requirements such as visual and audio.

Strong hard solution ability

DP1.2, HDMI2.0, mipi-ds, eDP multiple display output interface, dual or split screen display, 4K VP9, 4K 10bits H265/H264 and 1080P multiple formats (vc-1, mpeg-1/2/4, VP8) video decoding, 1080P (h. 264, VP8 format) video encoding.

Support multiple AI frameworks

Compatible with a variety of AI framework design, support TensorFlowLite/Android NN API, AI software tools support Caffe/TensorFlow models, making it easier for developers to use.

Rich interfaces


Tb-96aiot will adopt 22nm RK1808 AIoT chip, ultra-low power consumption independent NPU, computing power up to 3.0Tops, and provide rich interface and strong scalability.

The 96Boards SOM expansion boards are compatible with tb-96ai/tb-96aiot, suitable for different market needs and better support for multiple different SOM.

The 96Boards SOM specification is out in version 1.0

Linaro has released two specifications for 96Boards SOM, Compute SOM and Wireless SOM.Both specifications provide a highly reliable, cost-effective embedded platform for building end products to meet the needs of engineers for better product development.The specification has been finalized for review by members of the 96Boards technical steering committee.

The Compute SOM specification defines interfaces ranging from generic modules to on-board modules, regardless of the specific SOC selection on the module.Applications include industrial automation, intelligent devices, gateway systems, automobiles, healthcare, robotics, and POS retail systems.

Wireless SOM specification defines a replaceable SOM according to the application, which can support standard or proprietary Wireless standards such as 802.15.4, BLE, WiFi, LoRa, nb-iot, lte-m, etc.Specifications are defined to support multiple products and future innovations in wireless standards.

The release of the 96boards specification means that developers can use a compatible SOM solution across SoC platforms.SOM module currently USES so-dimm connector and MiniModule Plus connector in DRAM application scenarios to expand different functions.96Boards has the advantages of providing a flexible IO interface, robustness of the installation mechanism, and standard form specifications.The definition of the new 96Boards specification is to provide a plug and play and multiple compatibility SOM solution to address the current lack of generic SOM solutions on the market.

天高科技