When it comes to computing hardware, the build-vs-buy decision has always been tricky. Custom hardware offers more opportunities for optimization, but off-the-shelf boards can provide a faster, easier design path. - What if there were a third option that provided the best of both worlds?
20 years ago, the industry’s leading hardware manufacturers joined forces to answer this question. The result was COM Express, which solves the customization dilemma by breaking single-board computers (SBCs) into two components: a standardized computer on module (CoM), also known as system on module (SoM), that holds the processor and other high-speed circuitry, and a customized carrier board that provides application-specific functionality like I/O.
This approach has been enormously successful, giving engineers the ability to optimize their hardware without a fully custom design. But COM Express has traditionally been limited to x86 architecture processors. While a logical choice 20 years ago, this restriction has become a serious design constraint. Today, that’s all changing with the introduction of COM Express modules that use the latest Arm processors to push new boundaries in integration, efficiency, and AI performance.
Advantages of bringing Arm to COM Express
The traditional focus on the x86 architecture was driven by pragmatic considerations. Features of x86 processors, such as their broad software support, inter-generational compatibility, and high-performance I/O, are critical for embedded systems. The x86 architecture I/O interfaces and power envelopes have aligned closely with COM Express.
However, ongoing advances in Arm processors have changed the balance of considerations. In response, manufacturers like SECO are pioneering the integration of Arm processors into COM Express modules. This move offers several benefits:
- Heterogeneous computing: The added circuit board area and input power capability of COM Express, as compared to more Arm-focused standards like SMARC, enables the combination of an Arm microprocessor with added accelerators such as artificial intelligence (AI) processors.
- Low energy consumption: Where Arm processors can satisfy the processing and peripherals needs of an application, they typically provide more energy-efficient operation.
- Integrated functions: Many Arm processors come with built-in peripherals and processing capabilities, such as neural processing units (NPUs), camera interfaces, and security engines, that may not be present in x86 processors.
- Microcontroller support: Many Arm SoCs include one or more microcontrollers, typically a variant of Cortex-M, that enable separate and simultaneous low-level control functions running real-time operating systems like FreeRTOS or Zephyr.
To leverage these benefits, with emphasis on heterogeneous computing, SECO has developed the COM Express module SOM-COMe-BT6-RK3588, which we will now examine in detail.
Bringing full Arm CPU Power to COM-Express Modules
With the SOM-COMe-BT6-RK3588, SECO has successfully integrated the Rockchip RK3588 Arm microprocessor and the Axelera AI Metis Artificial Intelligence Processing Unit (AIPU).
The RK3588 features four Cortex-A76, four Cortex-A55, and three Cortex-M0 processing cores, along with an integrated GPU (Arm Mali-G610 MC4), NPU, and embedded video encode and decode accelerators. The AIPU delivers a remarkable 214 TOPS of AI performance. In addition, the module supports up to 32 GB LPDDR5-3200 RAM for the microprocessor and 2 GB LPDDRx-2133 for the AIPU.
To ensure extensive connectivity, this COM Express Type 6 Basic module includes 1x Gigabit Ethernet, 2x USB Type-C with DP-Alt mode, 4x USB 5 Gbps ports, 1x USB 2.0 port, and up to four PCIe lanes, configured as 3 or 4 total ports. Additional interfaces include I2S, UART, SPI, and I2C. An onboard TPM 2.0 module facilitates security. The module adheres to the COM Express Type 6 Basic standard, measuring 125 mm x 95 mm.
By combining the RK3588 Arm processor with the Axelera Metis AIPU, the SOM-COMe-BT6-RK3588 facilitates high-performance multiple-sourced video and image detection and recognition operations at the edge, along with full application-level control.
Comprehensive Software Support
It’s not just about hardware – developers also need robust software support for CoM deployments. That’s where Clea, SECO’s edge-to-cloud software platform, comes in. Clea includes both cloud and edge software components to fully deploy and operate IoT infrastructure, orchestrating data analysis, communication, and fleet management across various levels of granularity. With Clea, organizations can operate, manage, and control a fleet of COM Express-based systems from a single IoT platform.
Clea OS is a Yocto-based embedded Linux distribution for edge devices that integrates IoT data management, communications, device management, real-time security, and other functions within a DevOps framework. Architected for portability across a variety of processor platforms with minimal recoding and porting, Clea OS additionally facilitates emerging cybersecurity regulation compliance, such as with Radio Equipment Directive (RED) Delegated Act (DA) on Cybersecurity (EN 18031).
Clea OS also integrates the Axelera AI Voyager SDK, which facilitates intelligent edge AI solutions. It offers tools for evaluating the performance, accuracy, and energy use of Metis and enables streamlined integration with optimized AI pipelines and direct hardware access via low-level APIs.
Driving Future Technology
The combination of the SOM-COMe-BT6-RK3588 and Clea is ideal for numerous industries. It excels in edge AI computing—especially computer vision—where real-time edge-based data processing minimizes reliance on cloud infrastructure. In healthcare, it supports AI-driven diagnostics and real-time patient monitoring. In automotive and transportation, it provides the power needed for real-time object detection, path planning, and navigation in autonomous vehicles.
Conclusion
Arm on COM Express is a novel yet promising approach that delivers benefits such as heterogeneous processing, enhanced power efficiency, added functionality, and higher scalability. It opens new opportunities for specialized use cases – especially AI edge computing and autonomous systems. In combination with Clea OS, the deployment of edge IoT systems, including enablement of edge AI applications, is streamlined. SECO is committed to making these advantages widely accessible by extending Arm-based COM Express support across multiple platforms.
Ready to start your COM Express design? Want to understand how to deploy edge AI or IoT-based systems? Connect with SECO experts today!