Blog

Asymmetric Multiprocessing on Heterogeneous Multiprocessor Systems

Monday, 07 Mar 2022

Implementation of asymmetric multiprocessing using a practical example on the i.MX8X with OpenAMP

Author: David Kauschke, Mixed Mode GmbH
David Kauschke, Mixed Mode GmbH

Heterogenous multiprocessor systems on a chip (MPSoC) have become increasingly popular for industrial applications in recent years due to their high performance, lower cost and energy efficiency. In particular, the use of many different integrated processors running different operating systems poses multiple challenges. This architecture is also known as Asymmetric Multiprocessing (AMP). The two biggest challenges are lifecycle management (LCM) and inter-processor communication (IPC).

This article presents the design of heterogeneous MPSoCs and the use of different operating systems. A framework is selected as the proposed solution to the two challenges. This is followed by a presentation detailing the implementation of the developed AMP system with the selected OpenAMP framework on NXP i.MX8X MPSoC with embedded Linux on the ARM Cortex-A35 and FreeRTOS on the ARM Cortex-M4, using Variscite’s VAR-SOM-MX8X SoM. To evaluate the implemented system, the latency times on the iMX 8X are measured. The results show, among other things, that the maximum latency from Linux user space to FreeRTOS with the use of the RT patch is 628 µs. From the results, it can be concluded that the IPC between the processors is suitable for soft real-time.

Latest Posts
Software releases
June 2026 software updates
Stay current with software updates – Keep your Variscite SoMs performing at their best with our latest software releases.
 
16.06.2026
Newsroom
Variscite Keeps Embedded Products on Track During Global Component Shortages + SMARC
Global component shortages are putting embedded projects under pressure. As a leading global System on Modules (SoMs) developer and manufacturer, we keep our customers’ development timelines on track by applying a supply chain strategy built on buffered inventories, in-house production, established relationships with multiple suppliers, and more than two decades of operational experience.
As AI-driven demand accelerated through 2025, memory chip shortages developed as semiconductor manufacturers shifted capacity to large-volume buyers, leaving many sectors, including medical, industrial, edge/IoT, and robotics poorly served. Climbing prices and lengthening lead times, exacerbated by knock-on effects across other SoM components, have created two clear pressure points. SoM vendors relying on outsourced manufacturing have limited flexibility when availability drops, leaving customers facing the same delays. Separately, product developers who chose chip-down architectures over SoM-based designs must procure components on their own. Those buying in smaller quantities are routinely pushed down supplier priority lists, and in many cases cannot get supply at all.
21.06.2026