Coral Dev Board Mini
Download Coral Mini Dev Board Datasheet
Getting Started with Coral Mini Dev Board
Coral Mini Dev Board Schematics
The Coral Dev Board Mini is ideal for quick on-device ML product prototyping. This board offers fast Machine Learning (ML) inferencing. Evaluation for the Accelerator Module is the primary function of Coral Dev Mini. An Accelerator Module is surface-mounted hardware that offers the Edge TPU. In addition, Coral Dev Mini is an embedded system with fully-functional capabilities for many on-device ML prototyping projects.
Fast Machine Learning (ML) Inferencing
Coral Dev Board Mini features an Edge TPU Coprocessor that can perform Tera Operations per Second (TOPS). Simply speaking, Tera Operations are 4 trillion operations. The Coprocessor performs one TOPS consuming just 0.5 watts, i.e., 2x TOPS/watt. As a practical example, this coprocessor can execute MobuleNet v2 or other high-end mobile vision models at approximately 400 FPS with efficient power consumption.
A Complete System
Coral Dev Board Mini is a complete system that combines SoC, Machine Learning (Edge TPU), Wireless Connectivity, and TensorFlow Lite support. The board is compatible with Mendel (a variation of Debian Linux) allowing you to run Linux tools you need for your projects.
TensorFlow Lite Support
With TensorFlow Lite support, you don’t have to build models from scratch. You can compile TensorFlow Lite models to run on the Edge TPU.
Video Presentation of the Product
Specifications of Coral Dev Board Mini
- System on Chip (SoC): Media Tek MT8167s
- Processor: ARM Cortex-A35 | Quad-Core Armv8-A 64-bit architecture | Up to 1.5 GHz
- Cache: L1 Data 32 KB | L1 Instructions 32KB | L2 Unified 512 KB
- Graphics Processing Unit (GPU): Imagination PowerVR GE8300 | Open GL ES 1.x, 2.0, 3.0, 3.2 3D Support | OpenCL 1.2 Support | Vulkan 1.0 Profile Support | 1.6 gigapixel / sec. at 600MHz | 150 million triangles / sec.
- SDRAM: 2GB 4-Channel 32-bit LPDDR3 | DDR Clock up to 1600 MHz | Direct interface to the built-in SoC DDR controller
- Flash Memory (eMMC): 8GB eMMC Flash Memory NAND | JEDEC Version 5.0 and 5.1 compatible | 8-bit MMC mode
- Expandable Flash: SD / SDIO 3.0 Micro SD Card Slot | MMC / SD / SDHC compatible | Operates at 4-bit SDIO mode | Supports up to 32 GiB Card (Card is NOT included)
- Wi-Fi: MT7658 MediaTek module | Wi-Fi 5 (802.11 a / b / g / n / ac 2.4 / 5 GHz) | SDIO3.0 W-LAN interface support
- Bluetooth: MT7658 MediaTek Module | Bluetooth BT 5.0 | Bluetooth Low Energy (BLE) Support | UART interface support
- Video Processing Unit (VPU): HEVC main profile decoder 1080p @ 60 fps | MPEG4, constraint baseline profile, main decoder 1080p @ 60 fps | MPEG-1, 2, H.364, H.263 decoder 720p @ 30 fps
- On-chip Memory: RAM On-chip 256 KB + 64 KB | 96 KB Boot ROM
- Audio: SPDIF, TDM, I2S, DAI Interface Support |Supports Mono, Stereo, 24-bit, 16-bit sample formats | IIR Compensation Filter 4-Band (for loudspeaker response enhancement)
- HDMI Display Output: HDMI 1.4a | 1920 x 1080 pixels panel resolution support
- MIPI DSI Display Output: CSI connector 24-Pin | 4 Data Lanes | 1 Gb/s Maximum Bit Speed
- MIPI CSI2 Camera Input: CSI Connector 24-Pin | 4 Data Lanes | Up to 8 MP Camera Support
- Security: On-Board Cryptochip A71CH Root of Trust | Chip-to-Cloud Security with ARM TrustZone
- ML Accelerator: Edge TPU Coprocessor | High-Performance ML Inferencing for TensorFlow Lite Models | MT8167s SoC with USB 2.0 Interface | 4 Trillion Operations Per Second (TOPS) at 2 TOPS per Watt
- Weight: 22.5 Grams
- Dimensions: 6.4cm x 4.8cm x 1.46cm | 64mm x 38mm x 14.6mm
Connectors and Interfaces
- 40-Pin GPIO Header
- USB Type-C Port
- USB 2.0 Type-C OTG Port
- Mono Speaker 2-Pin Terminal
- 5mm Headphone Jack
- Micro HDMI
- MicroSD Card Slot (Half Size Reader)
- MIPI CSI Camera Connector for 24-Pin FFC
- MIPI DSI Display Connector for 24-Pin FFC
GPIO Pin Default States
Items Included
- Coral Dev Board Mini (x1)
