Inter-Core Communication Performance Evaluation of a Multicore Microcontroller for Edge Computing Applications

Multicore microcontrollers are available on the market to provide the required real-time EDGE computing resources present in modern distributed embedded systems. These devices contain multiple processor cores interconnected with device-specific communication solutions to utilize the computing resources of the cores for applications. Ideally, the computing resources of the individual cores add together and can be fully utilized by the embedded software running on these devices. However, due to the inefficiencies of inter-core communication, practical applications can expect drastically lower available throughput, and the actual throughput depends on the selected inter-core communication software and its configuration substantially. It is paramount to have some information about this performance loss before application development. Unfortunately, there exists no common benchmark for the performance evaluation of these inter-core communication solutions nor quality performance data is available. In the paper, we introduce our initial performance evaluation of the STM32H745 two-core MCU using the ST-provided inter-core communication framework for the FreeRTOS operating system. As the memory architecture of the STM32H745 is quite complex, we investigate how memory allocation and caching influence the performance. The performance is evaluated as a function of individual data transfer sizes. We measure the delay and data rate of communication.