Introduction

Banana Pi BPI-M6 with Synaptics VS680

Banana Pi BPI-M2 Pro S905x3 design

Banana Pi BPI-M5 Amlogic S905X3 Processor

Banana Pi BPI-M4 Realtek RTD1395

Banana Pi BPI-M64 Allwinner A64

Banana Pi BPI-M2+ Allwinner H3/H5/H2+]3

Banana Pi BPI-M6 is the next generation single board computer from Banana Pi in 2022,It is powered by Synaptics VS680 quad-core Cortex-A73 (2.1GHz) and One Cortex-M3 processor,Imagination GE9920 GPU.and NPU Up to 6 .75Tops. Onboard 4GB LPDDR4 memory and 16GB EMMC storage, and supports 4 USB 3.0 interface, a gigabit network port.onboard 1 HDMI-rx port and 1 Hdmi-tx port.

About VideoSmart VS680

VideoSmart™ VS680 solution, an industry-first edge computing SoC that combines a CPU, NPU, and GPU. This new multimodal platform with integrated neural network accelerator is purpose built with perceptive intelligence for applications including smart displays, smart cameras, set-top-boxes and media streamers.The Synaptics VideoSmart VS680 is a multimedia powerhouse that combines a Qdeo 4K-video engine, an audio processor capable of far-field keyword detection and voice recognition, and a proprietary SyNap deep-learning accelerator (DLA). it also integrates a higher-performance Imagination PowerVR Series9 GPU. Another new feature is an ISP with HDR capabilities that can handle two 4K cameras. Previous VideoSmart products target the streaming-video set-top-box (STB) market, but the VS680 aims for a broader range of smart-home devices. It s well suited to the Facebook Portal and other smart displays, which enable video calling. The audio processor can drive a smart speaker or sound bar, but when coupled with the DLA, it handles on-device voice-UI functions in addition to providing the front end to a cloud-based digital assistant. The DLA works with the dual ISP to run neural networks on video streams from front and rear cameras, enabling face ID, object recognition, and security monitoring, among other tasks. The Qdeo engine can drive two displays, allowing the device to power a 1080p touchscreen panel in an STB while simultaneously streaming to a 4K TV.

Key Features

Synaptics VideoSmart VS680 quad-core Cortex-A73 (2.1GHz) and One Cortex-M3 processor
Imagination GE9920 GPU
NPU for AI up to 6 .75Tops
4GB LPDDR4
16GB eMMC flash
M.2 Key E(PCIe + MIPI CSI)
4 USB 3.0
1 GbE ethernet
1 HDMIin and 1 HDMIout

Getting Start

Getting Started with BPI-M6

Hardware

Hardware interface

Hardware spec

HardWare Specification of Banana pi BPI-M6
CPU	Synaptics VS680 quad-core Cortex-A73 (2.1GHz) and One Cortex-M3 processor
GPU	Imagination GE9920 GPU
NPU for AI	Up to 6 .75Tops
Memory	4 GB LPDDR4
Storage	MicroSD slot with support for up to 256GB expansion and 16G eMMC flash with support for up to 64GB
Network	10/100/1000 Mbit/s Ethernet ,Optional WiFi USB dongle
Video	1 x HDMI 2.1 (up to 4K@60Hz with HDR, CEC, EDID) out, and 1 HDMI in Port
Audio Output(s)	1 x HDMI digital output
Display	MIPI DSI interface
M.2 interface	M.2 Key E(PCIe + MIPI CSI)
USB ports	USB 3.0 PORT (x4)
GPIO	40 Pin Header : GPIO (x28) and Power (+5V, +3.3V and GND). GPIO pins can be used for UART, I2C, SPI or PWM
Switches	SPI boot and U-boot
LED	Power Status and Activity status
Power Source	5 volt @3A TYPE C
Size & Weight	92x60mm, 48g
OS	Android and Linux

GPIO PIN define

Software

Source code on github

Documents

Senary VS680(Synaptics SN3680) datasheet:
The VS680 NPU is ranked first on AI benchmark website: https://ai-benchmark.com/ranking_IoT.html
BPI-M6 AI function test: https://www.youtube.com/watch?v=RIO0K_V12D4
BPI-M6 schematic diagram

Baidu Cloud: https://pan.baidu.com/s/18VTAdGtQFb2nu1bku8mHBw?pwd=8888 (pincode:8888)

Google Drive:https://drive.google.com/file/d/1z-o5dZWcYrmTqNH3Wz9ttWUzptveCS6s/view?usp=sharing