top of page
scheduling-code.webp
SCHEDULING DESIGN

Built around the next event.

QRT programs the hardware timer for the next scheduled event. Follow the path from timer expiry to application execution.

THE MECHANISM

The next event sets the timer.

Timed wakeups follow scheduled events. Application execution also depends on interrupt handling and the scheduler.

01

Choose the next timed event

Use the next scheduled event to determine when the hardware timer should expire.

02

Process the timer expiry

Handle the event and make any waiting work eligible to run.

03

Dispatch eligible work

The scheduler selects which ready thread runs. Other work and interrupt activity affect when execution begins.

pcb1.png

TIMING TERMS

Define the timing.
Then measure it.

Timer resolution

The granularity with which the configured timer can represent time.

Timer clock + configuration

Interrupt response

The interval between the timer event and the start of its interrupt handler.

Timer event → handler entry

Thread response

The interval between the timer event and the waiting thread resuming execution.

Timer event → thread resumes

EVALUATION

Put your workload on the timeline.

Evaluate the complete response path on your target, with the work your application runs.

State the target

Board, CPU clock, timer source, kernel version and build configuration.

01

Show the distribution

Defined endpoints, sample count, percentiles and maximum observed response.

03

Describe the load

Runnable threads, interrupt traffic, priorities and critical sections.

02

Make it reproducible

Test code, measurement method, raw traces and measurement overhead.

04

DESIGN QUESTIONS

The details engineers ask for.

Does timer resolution tell me when my thread will run?

Timer resolution describes the timer's granularity. Interrupt handling, scheduler decisions and competing work also

affect when the thread resumes. Measure those intervals separately.

Does a precise wakeup guarantee a deadline?

A completion deadline also depends on execution time, interference and the application's scheduling assumptions.

Evaluate the full workload against its deadlines.

Is the maximum observed response a worst-case bound?

A test reports the conditions and samples it covers. A worst-case bound needs analysis and explicit assumptions

beyond the maximum observed in a measurement run.

Get Started

Talk to the team

Reach out to the team for licensing, technical questions, or evaluation.

Thanks for submitting!

contact@preemptsystems.com
Salhojankatu 25d
33500 Tampere
FINLAND

bottom of page