Coral System-on-Module (SoM) - 4GB RAM Version
Edge

Coral System-on-Module (SoM) - 4GB RAM Version

Seeed Studio

Overview

Coral System-on-Module (SoM) - 4GB RAM Version is a Google Coral Edge TPU device sold by Seeed Studio, rated by Seeed at 4 TOPS. Its GPU is NXP i.MX 8M SoC Quad-core ARM Cortex-A53, plus Cortex-M4F 2D/3D Vivante GC7000 Lite GPU and VPU. Seeed publishes a fixed price for this configuration.

Key Features

  • Coral System-on-Module (SoM) - 4GB RAM Version rated at 4 TOPS
  • Arm Cortex-A53 MPCore platform: Quad symmetric Cortex-A53 processors: 32 KB L1 Instruction Cache 32 KB L1 Data Cache Support L1 cache RAMs protection w
  • Arm Cortex-M4 core platform: 16 KB L1 Instruction Cache 16 KB L1 Data Cache 256 KB tightly coupled memory (TCM)
  • I/O connectivity: 2x USB 3.0/2.0 controllers with integrated PHY interfaces 1x Ultra Secure Digital Host Controller (uSDHC) interfaces 1x
  • On-chip memory: Boot ROM (128 KB) On-chip RAM (128 KB + 32 KB)
  • External memory: 32/16-bit DRAM interface: LPDDR4-3200, DDR4-2400, DDR3L-1600 8-bit NAND-Flash eMMC 5.0 Flash SPI NOR Flash QuadSPI Flash

Ideal For

developers running quantised TensorFlow Lite models at the edge on an Edge TPU rather than a GPU

Feature

Details

Arm Cortex-A53 MPCore platform

Quad symmetric Cortex-A53 processors: 32 KB L1 Instruction Cache 32 KB L1 Data Cache Support L1 cache RAMs protection with parity/ECC Support of 64-bit Armv8-A architecture: 1 MB unified L2 cache Support L2 cache RAMs protection with ECC Frequency of 1.5 GHz

Arm Cortex-M4 core platform

16 KB L1 Instruction Cache 16 KB L1 Data Cache 256 KB tightly coupled memory (TCM)

Graphic Processing Unit (GPU)

Vivante GC7000Lite 4 shaders 267 million triangles/sec 1.6 Gigapixel/sec 32 GFLOPs 32-bit or 64 GFLOPs 16-bit Supports OpenGL ES 1.1, 2.0, 3.0, 3.1, Open CL 1.2, and Vulkan

Video Processing Unit (VPU)

4Kp60 HEVC/H.265 main, and main 10 decoder 4Kp60 VP9 and 4Kp30 AVC/H.264 decoder (requires full system resources) 1080p60 MPEG-2, MPEG-4p2, VC-1, VP8, RV9, AVS, MJPEG, H.263 decoder

I/O connectivity

2x USB 3.0/2.0 controllers with integrated PHY interfaces 1x Ultra Secure Digital Host Controller (uSDHC) interfaces 1x Gigabit Ethernet controller with support for EEE, Ethernet AVB, and IEEE 1588 2x UART modules 2x I2C modules 2x SPI modules 16x GPIO lines with interrupt capability 4x PWM lines Input/output multiplexing controller (IOMUXC) to provide centralized pad control Note: The list above is the number of signals available to the baseboard (after considering SoC signals used by the SoM).

On-chip memory

Boot ROM (128 KB) On-chip RAM (128 KB + 32 KB)

External memory

32/16-bit DRAM interface: LPDDR4-3200, DDR4-2400, DDR3L-1600 8-bit NAND-Flash eMMC 5.0 Flash SPI NOR Flash QuadSPI Flash with support for XIP

Display

HDMI Display Interface: HDMI 2.0a supporting one display up to 1080p Upscale and downscale between 4K and HD video (requires full system resources) 20+ Audio interfaces 32-bit @ 384 kHz fs, with Time Division Multiplexing (TDM) support SPDIF input and output Audio Return Channel (ARC) on HDMI MIPI-DSI Display Interface: MIPI-DSI 4 channels supporting one display, resolution up to 1920 x 1080 at 60 Hz LCDIF display controller Output can be LCDIF output or DC display controller output

Audio

1x SPDIF input and output 2x synchronous audio interface (SAI) modules supporting I2S, AC97, TDM, and codec/DSP interfaces 1x SAI for 8 Tx channels for HDMI output audio 1x SPDIF input for HDMI ARC input

Camera

2x MIPI-CSI2 camera inputs (4-lane each)

Security

Resource Domain Controller (RDC) supports four domains and up to eight regions Arm TrustZone (TZ) architecture On-chip RAM (OCRAM) secure region protection using OCRAM controller High Assurance Boot (HAB) Cryptographic acceleration and assurance (CAAM) module Secure non-volatile storage (SNVS): Secure real-time clock (RTC) Secure JTAG controller (SJC)

Edge TPU coprocessor

ASIC designed by Google that provides high-performance ML inferencing for TensorFlow Lite models Uses PCIe and I2C/GPIO to interface with the iMX8M SoC 4 trillion operations per second (TOPS) 2 TOPS per watt

Random-access memory (SDRAM)

4 GB LPDDR4 SDRAM (4-channel, 32-bit bus width) 1600 MHz maximum DDR clock Interfaces directly to the iMX8M build-in DDR controller

Flash memory (eMMC)

8 GB NAND eMMC flash memory 8-bits MMC mode Conforms to JEDEC version 5.0 and 5.1

Expandable flash (MicroSD)

Meets SD/SDIO 3.0 standard Runs at 4-bits SDIO mode Supports system boot from SD card

Ethernet

10/100/1000 Mbps Ethernet/IEEE 802.3 networks Reduced gigabit media-independent interface (RGMII)

Wi-Fi

Murata LBEE5U91CQ module: Wi-Fi 2x2 MIMO (802.11a/b/g/n/ac 2.4/5 GHz) Supports PCIe host interface for W-LAN

Bluetooth

Murata LBEE5U91CQ module: Bluetooth 4.2 (supports Bluetooth low-energy) Supports UART interface

Cryptographic coprocessor

Microchip ATECC608A cryptographic coprocessor: Asymmetric (public/private) key cryptographic signature solution based on Elliptic Curve Cryptography and ECDSA signature protocols

Baseboard connectors

3x 100-pin connectors (Hirose DF40C-100DP-0.4V)

Antenna connectors

2x coaxial cable connectors (Murata MM8930-2600)

Provides a complete system

The Coral SoM is a fully-integrated Linux system that includes NXP's iMX8M system-on-chip (SoC), eMMC memory, LPDDR4 RAM, Wi-Fi, and Bluetooth, and the Edge TPU coprocessor for ML acceleration. It runs a derivative of Debian Linux we call Mendel.

Performs high-speed ML inferencing

The onboard Edge TPU coprocessor is capable of performing 4 trillion operations (tera-operations) per second (TOPS), using 0.5 watts for each TOPS (2 TOPS per watt). For example, it can execute state-of-the-art mobile vision models such as MobileNet v2 at 400 FPS, in a power-efficient manner.

Integrates with your custom hardware

The SoM connects to your own baseboard hardware with three 100-pin connectors.

Supports TensorFlow Lite

No need to build models from the ground up. TensorFlow Lite models can be compiled to run on the Edge TPU.

Supports AutoML Vision Edge

Easily build and deploy fast, high-accuracy custom image classification models to your device with AutoML Vision Edge.

Feature

High-speed and low-power ML inferencing (4 TOPS @2 W)

Feature

A complete Linux system (running Mendel, a Debian derivative)

Feature

Small footprint (40 x 48 mm)

Feature

NXP i.MX 8M SoC Quad-core ARM Cortex-A53, plus Cortex-M4F 2D/3D Vivante GC7000 Lite GPU and VPU

Feature

Quad-core ARM Cortex-A53, plus Cortex-M4F

Feature

2D/3D Vivante GC7000 Lite GPU and VPU

Feature

Google Edge TPU ML accelerator

Feature

Cryptographic coprocessor

Feature

Wi-Fi 2x2 MIMO (802.11b/g/n/ac 2.4/5 GHz)

Feature

Bluetooth 4.2

Feature

Gigabit Ethernet

Feature

HDMI and MIPI-DSI

Feature

Up to 95x GPIO (including SPI, I2C, PWM, UART, SAI, and SDIO)

Feature

32 KB L1 Instruction Cache

Feature

32 KB L1 Data Cache

Feature

Support L1 cache RAMs protection with parity/ECC

Feature

1 MB unified L2 cache

Feature

Support L2 cache RAMs protection with ECC

Feature

Frequency of 1.5 GHz

Feature

16 KB L1 Instruction Cache

Feature

16 KB L1 Data Cache

Feature

256 KB tightly coupled memory (TCM)

Feature

Vivante GC7000Lite

Feature

267 million triangles/sec

Feature

1.6 Gigapixel/sec

Feature

32 GFLOPs 32-bit or 64 GFLOPs 16-bit

Feature

Supports OpenGL ES 1.1, 2.0, 3.0, 3.1, Open CL 1.2, and Vulkan

Feature

4Kp60 HEVC/H.265 main, and main 10 decoder

Feature

4Kp60 VP9 and 4Kp30 AVC/H.264 decoder (requires full system resources)

Feature

1080p60 MPEG-2, MPEG-4p2, VC-1, VP8, RV9, AVS, MJPEG, H.263 decoder

Feature

2x USB 3.0/2.0 controllers with integrated PHY interfaces

Feature

1x Ultra Secure Digital Host Controller (uSDHC) interfaces

Feature

1x Gigabit Ethernet controller with support for EEE, Ethernet AVB, and IEEE 1588

Feature

2x UART modules

Feature

2x I2C modules

Feature

2x SPI modules

Feature

16x GPIO lines with interrupt capability

Feature

4x PWM lines

Feature

Input/output multiplexing controller (IOMUXC) to provide centralized pad control

Feature

Boot ROM (128 KB)

Feature

On-chip RAM (128 KB + 32 KB)

32/16-bit DRAM interface

LPDDR4-3200, DDR4-2400, DDR3L-1600

Feature

8-bit NAND-Flash

Feature

eMMC 5.0 Flash

Feature

SPI NOR Flash

Feature

QuadSPI Flash with support for XIP

Feature

HDMI 2.0a supporting one display up to 1080p

Feature

Upscale and downscale between 4K and HD video (requires full system resources)

Feature

20+ Audio interfaces 32-bit @ 384 kHz fs, with Time Division Multiplexing (TDM) support

Feature

SPDIF input and output

Feature

Audio Return Channel (ARC) on HDMI

Feature

MIPI-DSI 4 channels supporting one display, resolution up to 1920 x 1080 at 60 Hz

Feature

LCDIF display controller

Feature

Output can be LCDIF output or DC display controller output

Feature

1x SPDIF input and output

Feature

2x synchronous audio interface (SAI) modules supporting I2S, AC97, TDM, and codec/DSP interfaces

Feature

1x SAI for 8 Tx channels for HDMI output audio

Feature

1x SPDIF input for HDMI ARC input

Feature

2x MIPI-CSI2 camera inputs (4-lane each)

Feature

Resource Domain Controller (RDC) supports four domains and up to eight regions

Feature

Arm TrustZone (TZ) architecture

Feature

On-chip RAM (OCRAM) secure region protection using OCRAM controller

Feature

High Assurance Boot (HAB)

Feature

Cryptographic acceleration and assurance (CAAM) module

Secure non-volatile storage (SNVS)

Secure real-time clock (RTC)

Feature

Secure JTAG controller (SJC)

Feature

ASIC designed by Google that provides high-performance ML inferencing for TensorFlow Lite models

Feature

Uses PCIe and I2C/GPIO to interface with the iMX8M SoC

Feature

4 trillion operations per second (TOPS)

Feature

2 TOPS per watt

Feature

4 GB LPDDR4 SDRAM (4-channel, 32-bit bus width)

Feature

1600 MHz maximum DDR clock

Feature

Interfaces directly to the iMX8M build-in DDR controller

Feature

8 GB NAND eMMC flash memory

Feature

8-bits MMC mode

Feature

Conforms to JEDEC version 5.0 and 5.1

Feature

Meets SD/SDIO 3.0 standard

Feature

Runs at 4-bits SDIO mode

Feature

Supports system boot from SD card

Feature

10/100/1000 Mbps Ethernet/IEEE 802.3 networks

Feature

Reduced gigabit media-independent interface (RGMII)

Feature

Wi-Fi 2x2 MIMO (802.11a/b/g/n/ac 2.4/5 GHz)

Feature

Supports PCIe host interface for W-LAN

Feature

Bluetooth 4.2 (supports Bluetooth low-energy)

Feature

Supports UART interface

Feature

Asymmetric (public/private) key cryptographic signature solution based on Elliptic Curve Cryptography and ECDSA signature protocols

$134.99

Prices may vary. Verify on vendor site.

Quick Specs

Feature
Asymmetric (public/private) key cryptographic signature solution based on Elliptic Curve Cryptography and ECDSA signature protocols
Arm Cortex-A53 MPCore platform
Quad symmetric Cortex-A53 processors: 32 KB L1 Instruction Cache 32 KB L1 Data Cache Support L1 cache RAMs protection with parity/ECC Support of 64-bit Armv8-A architecture: 1 MB unified L2 cache Support L2 cache RAMs protection with ECC Frequency of 1.5 GHz
Arm Cortex-M4 core platform
16 KB L1 Instruction Cache 16 KB L1 Data Cache 256 KB tightly coupled memory (TCM)
Graphic Processing Unit (GPU)
Vivante GC7000Lite 4 shaders 267 million triangles/sec 1.6 Gigapixel/sec 32 GFLOPs 32-bit or 64 GFLOPs 16-bit Supports OpenGL ES 1.1, 2.0, 3.0, 3.1, Open CL 1.2, and Vulkan
Video Processing Unit (VPU)
4Kp60 HEVC/H.265 main, and main 10 decoder 4Kp60 VP9 and 4Kp30 AVC/H.264 decoder (requires full system resources) 1080p60 MPEG-2, MPEG-4p2, VC-1, VP8, RV9, AVS, MJPEG, H.263 decoder
I/O connectivity
2x USB 3.0/2.0 controllers with integrated PHY interfaces 1x Ultra Secure Digital Host Controller (uSDHC) interfaces 1x Gigabit Ethernet controller with support for EEE, Ethernet AVB, and IEEE 1588 2x UART modules 2x I2C modules 2x SPI modules 16x GPIO lines with interrupt capability 4x PWM lines Input/output multiplexing controller (IOMUXC) to provide centralized pad control Note: The list above is the number of signals available to the baseboard (after considering SoC signals used by the SoM).

Tags

edge-inferencecomputer-visionrobotics