Computer-on-module standards: How COM-HPC and the like enable scalable embedded systems
Today, embedded systems need to combine high computing power, low latency, and the ability to process increasingly demanding AI workloads on one platform. At the same time, they are often in use over many years. That places high demands on the availability, maintainability, and long-term development of the hardware. Computer module standards such as COM-HPC, COM Express, and SMARC therefore rely on a strict separation of the computing unit from the application-specific carrier board. This makes it easier to reuse existing designs, simplifies the transition to new processor generations, and reduces the effort required for obsolescence and redesigns.
What computer module standards are there?
Computer modules, usually referred to as computer-on-modules (CoMs), are available on the market in both proprietary form factors and based on open standards. During the past few decades, various open computer module standards have become established. Among the leading standardization organizations are the PCI Computer Manufacturers Group (PICMG) and the Standardization Group for Embedded Technologies (SGET). The most important standards include:
- COM-HPC (Computer-on-Module for High-Performance Computing): The PICMG specification addresses modern high-performance embedded, client, and server systems with high data rates and high-speed interfaces. COM-HPC Mini adds a compact, credit card sized form factor for high-performance small-form-factor systems to the family.
- COM Express: This standard, also ratified by PICMG and well-established for many years, includes various module sizes and pinout types. COM Express Mini also offers a compact form factor that is suitable, for example, for x86-based embedded systems.
- SMARC (Smart Mobility ARChitecture): The SGET standard combines a compact form factor with interfaces that are particularly relevant for low-power ARM and energy-efficient applications. The current SMARC 2.2 specification is a logical evolution of the existing standard.
- OSM (Open Standard Module): With OSM, the SGET has created a standard for particularly small, solderable CoMs. OSM is designed for miniaturized embedded and IoT systems and is available in sizes 0, S, M, and L.
In addition, there are other standards such as CompactPCI Serial or MicroTCA for industrial backplane systems, as well as ModBlox7 for modular industrial PCs. Qseven was also successfully established on the market for many years. Which standard is suitable for a particular application depends on the respective system architecture. Developers must evaluate factors such as power requirements, available installation space, cooling design, required I/O interfaces, and desired interchangeability on a project-by-project basis.
Why do CoM standards simplify carrier board design?
The basic principle of a CoM is the functional separation of the computing platform from the application. The CPU, GPU, RAM, and a wide range of standard interfaces are located directly on the module. The carrier board, on the other hand, contains the application-specific functions and provides the required interfaces to the outside world. Within a CoM standard, the interfaces are assigned to fixed signal pins, allowing compatible modules from different manufacturers and processor generations to be used on the same standard-compliant carrier board.
That means manufacturers don’t have to develop a completely new CPU board with every new processor generation. Instead, the hardware design focuses on product-specific functions such as:
- Interfaces for Ethernet, fieldbuses, and communication
- Sensor and actuator connections
- Custom I/Os
- Application-specific extensions
Among other things, the CoM standards define the mechanical dimensions, connector types, signal assignment, and electrical requirements of the embedded module. For simple development of carrier boards, PICMG and SGET also provide their own carrier board design guides for each specification, as the current COM-HPC Carrier Design Guide shows.
How does modularity support design reuse?
COM Express is based, for example, on a modular architecture comprising a standardized compute module and an application-specific carrier board. That allows an existing carrier board to be combined with different COM-Express modules, provided that the respective interfaces and technical requirements are compatible.
Performance upgrades can be carried out by replacing the module without having to completely redesign the carrier board. PICMG outlines this architecture as the basis for a scalable upgrade path and for the reuse of existing technologies.
The modular approach can be used particularly in industrial automation, for scalable product families. Entire product families, for example, can be developed on the basis of the same carrier design by combining the latter with modules of varying performance levels. That allows computing power to be adapted to the respective requirements of an application. At the same time, developers can reuse existing designs for new applications, which significantly reduces the development effort and the associated one-time development and engineering costs.
Why life cycle support and interchangeability are crucial
Applications in industry, medical technology, or transport are often in use for ten years or longer. They include, for example:
- Machine controls in production facilities
- Control and visualization systems
- Medical analysis and imaging equipment
- Passenger information systems on railways
These systems need to be maintained and further developed throughout their entire life cycle, while semiconductor manufacturers replace the processor platforms in use with new generations in significantly shorter cycles.
This is exactly where modularization demonstrates one of its most important advantages. Ideally, if the processor technology changes or a manufacturer discontinues a module, then the entire system will not need to be redesigned or replaced. Instead, a compatible follow-up module can be installed on the existing carrier board. Open standards also reduce dependence on proprietary form factors and pinouts and enable different vendors to develop standard-compliant modules.
Congatec, for example, shows how this approach can be used in practice with a COM-HPC carrier board: The company specifies a long-term availability of at least seven years for embedded systems and, through the combination of carrier board and interchangeable modules, provides for future upgrade and update options. Advantech specifies product life cycles of seven to 15 years for its CoMs and additionally offers life cycle and migration services for embedded platforms. Modular architectures can therefore help ensure that existing system designs are used for longer and that the transition to new computing platforms is more predictable.
What does standardization mean for the embedded sector?
As demands on computing power increase, AI workloads become increasingly complex, and trends such as edge AI and modern high-speed interfaces emerge, the complexity of embedded systems continues to grow. At the same time, industrial applications demand long product life cycles and predictable migration paths. CoMs shift part of this complexity away from customer-specific hardware and toward standardized compute modules.
For developers and system architects, this turns the choice of the right standard into a fundamental architectural decision. Developers who consider not only the current CPU generation but also selection criteria such as
- product family
- carrier board
- upgrade paths
- expected service life
can use standard modules as the basis for embedded platforms that are scalable over the long term.
How can companies effectively combine edge, cloud, and OEE?
However, not every analysis should be carried out in the cloud. Safety-critical functions, such as monitoring threshold values, safe shutdown in critical states, or plausibility checks, must be carried out locally in the BMS or at the edge level. The cloud, on the other hand, is suitable for fleet comparisons, long-term trends, model training, and cross-location optimizations. A robust system architecture strategically combines both levels.
The benefit is apparent in the Overall Equipment Effectiveness (OEE). When applied to BESS, OEE comprises three dimensions: availability, performance, and quality. In other words, the storage system must be ready for use, deliver the expected performance, and provide energy in the required operating state. Predictive maintenance improves these dimensions by reducing unplanned outages, detecting degradation early on, and scheduling maintenance windows strategically.
The relevance of embedded computing for the electronica community
Embedded computing will play a key role at electronica 2026. In the edge lab LIVE! exhibition area, for example, developers can get first-hand information about the latest processor platforms, computer modules, and system solutions from embedded manufacturers.
In addition, forums such as the Embedded Systems Forum and the Embedded Developer Forum will address current issues related to embedded architectures, energy efficiency, and modern system design.