Real-Time GPIO Performance of Modern Programming Languages for Embedded Linux Application Development

Using embedded Linux in real-time distributed systems is a common practice today as implementation of complex artificial intelligence (AI) and network functions is substantially simpler than on embedded systems built of Microcontrollers (MCU). However, access to peripherals, such as general purpose input/output pins (GPIO), analog to digital and digital to analog converters, etc. must be done on embedded Linux also in real-time. The paper investigates the performance of real-time peripheral access on Linux by implementing identical GPIO access semantics using relevant programming models, languages and GPIO access kernel support solutions. The investigation is done on the TI AM62x System on Chip (SoC) platform using Yocto generated Linux images. The investigated languages include BASH, Python, C, C# with various runtime environments, and RUST. The results show that modern managed languages with up-to-date run-time environments and peripheral support, such as C#, can readily compete with C, while they can keep the programming effort lower, significantly reducing human resource investment into software development.