Applications › Education & research
A real-time kernel you can read in an afternoon, run on your PC and extend with your own scheduling algorithm: for courses, theses and experiments.
Large RTOSes bury scheduling logic under ports, drivers and configuration layers.
Students and labs need boards, debuggers and toolchains before the first task even runs.
Researchers need to implement and compare schedulers on the same kernel and the same task sets.
Clean C++ with no STL, exceptions or RTTI and minimal namespace use.
An x86 development mode lets you build and run examples without hardware.
Round-Robin, SWRR, fixed-priority with priority inheritance, preemption-threshold, RM, DM and EDF, with worst-case response time analysis for RM and DM.
Implement the C++ strategy interface to attach a new scheduling algorithm.
KERNEL_HRT mode with deadline monitoring and miss callbacks, plus SEGGER SystemView traces.
The same code targets Arm Cortex-M and RISC-V, with STM32 and Raspberry Pi Pico examples in the repository.
Clone the repository and run the x86 example to watch tasks being scheduled on a PC.
Create tasks with different periods and deadlines, switch between RM, DM and EDF, and compare the outcomes using deadline-miss callbacks.
Check schedulability with the worst-case response time analysis, then confirm it by measurement.
For research, implement a new strategy through the strategy interface and compare it with the built-in ones on identical task sets.
Move to a real board and use SystemView to compare theory with measurement.
Using STK in teaching or research? Let us know. The kernel is MIT-licensed, so no commercial license is needed for the open-source schedulers.
Questions about using STK in a course or research project? Get in touch.