Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

The UK education charity Raspberry Pi Foundation launched its new generation product in June this year:Raspberry Pi 4,with a CPU performance increase of 3 times, comparable to mainstream PCs, it has sparked heated discussions on various technical forums, with some believing that this iteration has surpassed the category of a laptop, creating a new golden standard for single-board computers, while others still consider it a dust collector, issuing dust warnings. Regarding the official announcement of entering the industrial sector for the first time, positioning the industrial IoT as the key promotional market forRaspberry Pi 4, we at Taobao IoT have conducted a detailed assessment from a technical perspective on whetherRaspberry Pi 4can be applied to industrial products, let’s take a look together!

Part 1 Hardware Layout

Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

The layout style remains consistent, with the power supply PMIC using a Type-C scheme, USB and RJ45 mirroring (analysis indicates support for USB 3.0 and Gigabit Ethernet, increasing wiring difficulty, requiring hardware signal integrity), and the USB-HUB chip upgraded to support USB 3.0.

Part 2 Performance Comparison

Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

Analysis Conclusion

  • In terms of CPU: the single-core A72 frequency performance is about 1.9 times that of A53, with a significant performance increase; the die area of one A72 core is equal to that of four A53 cores, so the cost of using A72 chips is higher; in terms of power consumption, considering chip process optimization, the BCM2711 is produced using a 28nm process, theoretically allowing for some optimization, but due to the BCM2837 (40nm) A53’s weaker performance (for energy saving), actual power consumption needs to be measured and compared.

  • In terms of memory: LPDDR4-2400 (maximum bandwidth 25.6GB/S) shows a significant improvement over LPDDR2-900 (maximum bandwidth 6.4GB/S), but considering the current market price of LPDDR4, Raspberry Pi 4 has set three configurations and tiered pricing based on memory capacity: 1GB ($35), 2GB ($45), 4GB ($55).

  • For the GPU: there is no significant improvement in hardware specifications, supporting OpenGL ES 3.0.

  • Encode: primarily used for camera encoding, 1080P@30 can meet most application scenarios, considering the design difficulty of the Encode core, area cost, and market research of application scenarios, this version remains unchanged.

  • Decode: as market demands for display output resolution and frame rate increase, 4K@60 hardware decoding (H265+HEVC) has been added to meet market trend requirements, and the HDMI IP core has been upgraded to HDMI2.0 version (4K@60fps).

Part 3 Interface Comparison

Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

Analysis Conclusion

  • The output interfaces HDMI and DSI both support two paths, which can be used for dual-screen applications.

  • The Camera CSI also supports two paths, which can be used for dual-camera input applications.

  • The network port supports 1000M, meeting the requirements for high bitrate video network transmission applications.

  • Bluetooth supports up to 5.0, whether it supports Mesh functionality needs further confirmation.

  • WiFi continues the dual-mode design (WiFi 5 standard), which can meet the requirements for industrial applications (WiFi 6 standard has already been showcased on laptops).

  • USB supports two paths of 3.0 standard, which can be used for high bandwidth peripherals, such as industrial cameras.

  • The power interface uses Type-C, indicating that overall power consumption has increased compared to before, which is also a requirement of the evolution of USB technology.

  • GPIO reusable functions: 6 * UART, 6 * I2C, 5 * SPI, 1 * SDIO, 1 * DPI (Parallel RGB Display), 1 * PCM (audio), 2 * PWM channels, 3 * GPCLK outputs, can meet most IoT device applications, including industrial applications.

Part 4 Measured Comparison

Test Item

Rpi 3B+

Rpi 4

Change

Boot Time

39.9

41.7

-4.50%

Idle Power(Amps)

0.505

0.684

-35.40%

Peak Power(Amps)

1.14

1.12

1.80%

CPU-sysbench primes – thread

317.7

250.4

21.20%

CPU-sysbench primes – 4threads

86.2

62.8

27.10%

RAM Bandwidth – mbw

1420

2983

110.10%

OpenGL – videogl32

30.9

35.8

15.90%

Ethernet – iperf3(Mbps)

332

933

181.00%

WiFi 2.4Ghz – iperf3(Mbps)

38.6

39.6

2.60%

WiFi 5Ghz(ac) – iperf3(Mbps)

98.6

107

8.50%

External USB Drive Write – dd

35

155

342.90%

External USB Drive Read – dd

32

233

628.10%

Data Analysis

  • CPU performance improvement is mainly due to the use of A72 cores, with idle power consumption significantly increasing (35.4%) compared to using A53 cores, but peak power consumption (CPU full load) remains basically stable, aside from optimizations in chip process technology and internal structure.

  • DDR bandwidth has improved significantly (110%), mainly due to the use of LPDDR4.

  • The GPU has seen some optimization in actual OpenGL performance, partly because the GPU has 256M dedicated memory allocation.

  • Ethernet uses Gigabit PHY, resulting in a significant increase in network bandwidth.

  • WiFi throughput remains basically the same, continuing the Cypress solution.

  • USB read and write speeds have improved significantly due to the use of 3.0.

Summary: The hardware configuration and measured data are basically matched, indicating that version 4B has a significant overall performance improvement.

Part 5 Default Memory Allocation

Taking the 4G LPDDR4 version as an example, the GPU system automatically allocates 256M of dedicated memory, with the CPU system allocating the remaining 3840M.

Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

Part 6 Operating System

Official version: Raspbian, Raspbian is a Debian-based Linux distribution specifically for Raspberry Pi, the latest Raspbian is based on the upcoming Debian 10 Buster release, featuring the PIXEL graphical interface.

Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

Other supported operating systems include:

  • Raspbian Stretch Lite (official Raspberry Pi system without a graphical interface)

  • Ubuntu Core (Ubuntu version designed for IoT applications)

  • Ubuntu MATE (version for Raspberry Pi)

  • Android TV

  • Windows 10 ARM

  • Windows IoT (official Windows version for IoT by Microsoft)

  • RISC OS (unique open-source operating system designed specifically for ARM processors by the original ARM creators)

Part 7 Application Analysis

1. Overall Positioning: ARM Host

2. Maintaining the original educational positioning, and being more developer-friendly

  • Dual display output (for developers programming on dual screens)

  • The 4 GB LPDDR4 version can support most IDEs

  • The 4-core A72 combined with high-bandwidth LPDDR4 can serve directly as an edge intelligent platform for inference operations: object recognition/face recognition

  • 4K@60fps hardware decoding performance can be used for graphics and image development, with OpenGL ES 3.0 support enhancing rendering effects

3. Suitable products for Demo setup

  • Switch game console prototype, equipped with Lakka system

  • Commercial display devices, dual-screen/PIP

  • Robot prototypes (for products with lower real-time requirements)

  • Home network storage center/NAS (USB3.0 + Gigabit Ethernet)

  • Edge intelligent devices: face recognition/face detection

  • TV Box/set-top box (4K decoding), home multimedia center

  • Wireless soft router

  • Various small servers: BT download server, FTP file server, WEB server, RTMP live streaming server, etc.

4. Suitable for industrial class Demo setup

  • Industrial gateway (similar to E2D-Station), industrial collection devices (requiring expansion interfaces)

  • Industrial tablet/PC, ARM industrial control devices with screens (similar to E2D-HMI), small ARM edge servers

  • Remote intelligent data collection units (requiring expansion for 4G, Lora/NB-IoT and other transmission modules)

Part 8 Industrial IoT Application Analysis

1. Raspberry Pi 4 Advantages

  • High modular cost-effectiveness, lower cost than custom hardware when produced in small quantities.

  • Flexible development, fast prototyping speed.

  • ARM host positioning matches current ARM industrial control technology.

2. Raspberry Pi 4 Disadvantages

  • No onboard Flash/eMMC, requires SD card boot, with the image (bootloader, Kernel, file system) burned into the SD card, which is mechanically fixed and unreliable in industrial environments, easily causing contact failure and preventing normal device startup.

  • Power supply requires USB Type-C cable, module power supply needs a baseboard or other power supply connected via USB, which is unreliable in industrial scenarios, easily causing contact failure, and the power supply’s anti-surge/group pulse protection needs special treatment to meet industrial scene and related certification requirements.

  • The temperature range supported by the module is 0~50℃, while industrial application scenes recommend wide-temperature designs, so the overall product hardware heat dissipation design needs to be re-simulated and evaluated to ensure stability and reliability of the device under harsh temperature environments.

  • The ecosystem for supported industrial bus protocols has not yet been established, such as Modbus, EtherCAT, Profinet, etc., which currently lack case support for third-party software porting.

  • For industrial control scenarios requiring high real-time performance (such as servo control/motor control), the instruction cycle of ARM architecture is too long (SoC internal needs to go through multiple levels of cache to reach physical peripherals), the right diagram shows the PRU design architecture of Ti Sitara, where PRU has directly controllable peripheral units (MII/UART/GPIO), without CPU interference, achieving ns-level real-time data interaction.

Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

Image Source: Ti Official Website

Part 9 Others

Some core boards of Raspberry Pi, such as the Raspberry Pi Compute Module (BCM2837, 4-core A53), onboard SSD, can be used for industrial-grade product design solutions, but the interface board design and the entire productization process still need to meet the specific application scenario requirements.

Exclusive Release | Research Report on Raspberry Pi 4 Industrial IoT Applications

Part 10 Summary

The extreme open-source hardware price of Raspberry Pi has given countless technology enthusiasts around the world the opportunity to realize their ideas and creativity on this platform, creating a new technical social network in the fields of education and innovation, integrating the ideals and attitudes of makers, forming a very valuable technical sentiment. However, from the perspective of product design, Raspberry Pi still has a certain distance in understanding the specific scene applications of various vertical industries and related product certifications. If Raspberry Pi continues to deepen its exploration of various industry applications and productization R&D understanding, I believe Raspberry Pi can create greater universal value.

One More Thing

The Taobao Technical Department – IOT Team is currently focusing on promoting the technical application of ARM architecture in the industrial control field and the integrated development of hardware and software for industrial IoT devices. We welcome embedded partners to join us in advancing the development of industrial IoT through technological innovation. Please send your resume to:[email protected]

Leave a Comment

×